Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.

Werkzeug 3.0.1 and 2.3.8 include a security fix: Slow multipart parsing for large parts potentially enabling DoS attacks.
https://github.com/pallets/werkzeug/commit/b1916c0c083e0be1c9d887ee2f3d696922bfc5c1

Affected versions of Werkzeug are potentially vulnerable to resource exhaustion when parsing file data in forms. Applications using 'werkzeug.formparser.MultiPartParser' to parse 'multipart/form-data' requests (e.g. all flask applications) are vulnerable to a relatively simple but effective resource exhaustion (denial of service) attack. A specifically crafted form submission request can cause the parser to allocate and block 3 to 8 times the upload size in main memory. There is no upper limit; a single upload at 1 Gbit/s can exhaust 32 GB of RAM in less than 60 seconds.

Werkzeug 2.2.3 includes a fix for CVE-2023-25577: Prior to version 2.2.3, Werkzeug's multipart form data parser will parse an unlimited number of parts, including file parts. Parts can be a small amount of bytes, but each requires CPU time to parse and may use more memory as Python data. If a request can be made to an endpoint that accesses 'request.data', 'request.form', 'request.files', or 'request.get_data(parse_form_data=False)', it can cause unexpectedly high resource usage. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests. The amount of RAM required can trigger an out of memory kill of the process. Unlimited file parts can use up memory and file handles. If many concurrent requests are sent continuously, this can exhaust or kill all available workers.
https://github.com/pallets/werkzeug/security/advisories/GHSA-xg9f-g7g7-2323

Flask-Admin 1.5.3 includes a fix for CVE-2018-16516: Helpers.py in Flask-Admin 1.5.2 has Reflected XSS via a crafted URL.
https://github.com/flask-admin/flask-admin/pull/1699

Lxml 4.2.5 includes a fix for CVE-2018-19787: lxml/html/clean.py in the lxml.html.clean module does not remove javascript: URLs that use escaping, allowing a remote attacker to conduct XSS attacks, as demonstrated by "j a v a s c r i p t:" in Internet Explorer. This is a similar issue to CVE-2014-3146.

Lxml version 4.6.3 includes a fix for CVE-2021-28957: An XSS vulnerability was discovered in python-lxml's clean module versions before 4.6.3. When disabling the safe_attrs_only and forms arguments, the Cleaner class does not remove the formation attribute allowing for JS to bypass the sanitizer. A remote attacker could exploit this flaw to run arbitrary JS code on users who interact with incorrectly sanitized HTML.
https://bugs.launchpad.net/lxml/+bug/1888153

Lxml 4.6.2 includes a fix for CVE-2020-27783: A XSS vulnerability was discovered in python-lxml's clean module. The module's parser didn't properly imitate browsers, which caused different behaviors between the sanitizer and the user's page. A remote attacker could exploit this flaw to run arbitrary HTML/JS code.

In lxml before 4.4.0, when writing to file paths that contain the URL escape character '%', the file path could wrongly be mangled by URL unescaping and thus write to a different file or directory.  Code that writes to file paths that are provided by untrusted sources, but that must work with previous versions of lxml, should best either reject paths that contain '%' characters, or otherwise make sure that the path does not contain maliciously injected '%XX' URL hex escapes for paths like '../'.
https://github.com/lxml/lxml/commit/0245aba002f069a0b157282707bdf77418d1b5be

Lxml 4.6.5 includes a fix for CVE-2021-43818: Prior to version 4.6.5, the HTML Cleaner in lxml.html lets certain crafted script content pass through, as well as script content in SVG files embedded using data URIs. Users that employ the HTML cleaner in a security relevant context should upgrade to lxml 4.6.5 to receive a patch.

Lxml 4.9.1 includes a fix for CVE-2022-2309: NULL Pointer Dereference allows attackers to cause a denial of service (or application crash). This only applies when lxml is used together with libxml2 2.9.10 through 2.9.14. libxml2 2.9.9 and earlier are not affected. It allows triggering crashes through forged input data, given a vulnerable code sequence in the application. The vulnerability is caused by the iterwalk function (also used by the canonicalize function). Such code shouldn't be in wide-spread use, given that parsing + iterwalk would usually be replaced with the more efficient iterparse function. However, an XML converter that serialises to C14N would also be vulnerable, for example, and there are legitimate use cases for this code sequence. If untrusted input is received (also remotely) and processed via iterwalk function, a crash can be triggered.

Flask 2.2.5 and 2.3.2 include a fix for CVE-2023-30861: When all of the following conditions are met, a response containing data intended for one client may be cached and subsequently sent by the proxy to other clients. If the proxy also caches 'Set-Cookie' headers, it may send one client's 'session' cookie to other clients. The severity depends on the application's use of the session and the proxy's behavior regarding cookies. The risk depends on all these conditions being met:
1. The application must be hosted behind a caching proxy that does not strip cookies or ignore responses with cookies.
2. The application sets 'session.permanent = True'
3. The application does not access or modify the session at any point during a request.
4. 'SESSION_REFRESH_EACH_REQUEST' enabled (the default).
5. The application does not set a 'Cache-Control' header to indicate that a page is private or should not be cached.
This happens because vulnerable versions of Flask only set the 'Vary: Cookie' header when the session is accessed or modified, not when it is refreshed (re-sent to update the expiration) without being accessed or modified.
https://github.com/pallets/flask/security/advisories/GHSA-m2qf-hxjv-5gpq

Pillow 8.1.1 includes a fix for CVE-2021-25293: There is an out-of-bounds read in SGIRleDecode.c.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In libImaging/Jpeg2KDecode.c in Pillow before 7.1.0, there are multiple out-of-bounds reads via a crafted JP2 file.

Pillow 10.0.1 updates its C dependency 'libwebp' to 1.3.2 to include a fix for a high-risk vulnerability.
https://pillow.readthedocs.io/en/stable/releasenotes/10.0.1.html

Pillow 9.0.0 ensures JpegImagePlugin stops at the end of a truncated file to avoid Denial of Service attacks.
https://github.com/python-pillow/Pillow/pull/5921
https://github.com/advisories/GHSA-4fx9-vc88-q2xc

Pillow 9.0.0 excludes carriage return in PDF regex to help prevent ReDoS.
https://github.com/python-pillow/Pillow/pull/5912
https://github.com/python-pillow/Pillow/commit/43b800d933c996226e4d7df00c33fcbe46d97363

Pillow before 9.0.1 allows attackers to delete files because spaces in temporary pathnames are mishandled.

Pillow 8.1.1 includes a fix for CVE-2021-25292: The PDF parser allows a regular expression DoS (ReDoS) attack via a crafted PDF file because of a catastrophic backtracking regex.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow 8.1.1 includes a fix for CVE-2021-25290: In TiffDecode.c, there is a negative-offset memcpy with an invalid size.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

An issue was discovered in Pillow before 8.2.0. PSDImagePlugin.PsdImageFile lacked a sanity check on the number of input layers relative to the size of the data block. This could lead to a DoS on Image.open prior to Image.load.

Pillow 8.2.0 includes a fix for CVE-2021-25288: There is an out-of-bounds read in J2kDecode, in j2ku_gray_i.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow before 9.2.0 performs Improper Handling of Highly Compressed GIF Data (Data Amplification).

Pillow 8.1.0 fixes TIFF OOB Write error. CVE-2020-35654 #5175.

Pillow 8.1.0 includes a fix for SGI Decode buffer overrun. CVE-2020-35655 #5173.

In Pillow before 7.1.0, there are two Buffer Overflows in libImaging/TiffDecode.c.

Pillow 8.1.1 includes a fix for CVE-2021-25291: In TiffDecode.c, there is an out-of-bounds read in TiffreadRGBATile via invalid tile boundaries.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICO container, and thus an attempted memory allocation can be very large.

Pillow is potentially vulnerable to DoS attacks through PIL.ImageFont.ImageFont.getmask(). A decompression bomb check has also been added to the affected function.

Pillow 8.1.1 includes a fix for CVE-2021-27922: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for an ICNS container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow version 8.2.0 includes a fix for CVE-2021-28678: For BLP data, BlpImagePlugin did not properly check that reads (after jumping to file offsets) returned data. This could lead to a DoS where the decoder could be run a large number of times on empty data.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28678-fix-blp-dos

Pillow 8.0.1 updates 'FreeType' used in binary wheels to v2.10.4 to include a security fix.

In libImaging/SgiRleDecode.c in Pillow through 7.0.0, a number of out-of-bounds reads exist in the parsing of SGI image files, a different issue than CVE-2020-5311.

In libImaging/PcxDecode.c in Pillow before 7.1.0, an out-of-bounds read can occur when reading PCX files where state->shuffle is instructed to read beyond state->buffer.

Pillow before 7.1.0 has multiple out-of-bounds reads in libImaging/FliDecode.c.

Pillow version 8.2.0 includes a fix for CVE-2021-28677: For EPS data, the readline implementation used in EPSImageFile has to deal with any combination of \r and \n as line endings. It used an accidentally quadratic method of accumulating lines while looking for a line ending. A malicious EPS file could use this to perform a DoS of Pillow in the open phase, before an image was accepted for opening.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28677-fix-eps-dos-on-open

Pillow 9.0.0 includes a fix for CVE-2022-22815: path_getbbox in path.c in Pillow before 9.0.0 improperly initializes ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 9.0.1 includes a fix for CVE-2022-22817: PIL.ImageMath.eval in Pillow before 9.0.0 allows evaluation of arbitrary expressions, such as ones that use the Python exec method. A first patch was issued for version 9.0.0 but it did not prevent builtins available to lambda expressions.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.1.html#security

Pillow 8.1.1 includes a fix for CVE-2021-27921: Pillow before 8.1.1 allows attackers to cause a denial of service (memory consumption) because the reported size of a contained image is not properly checked for a BLP container, and thus an attempted memory allocation can be very large.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

In Pillow before 8.1.0, PcxDecode has a buffer over-read when decoding a crafted PCX file because the user-supplied stride value is trusted for buffer calculations.

Pillow version 8.2.0 includes a fix for CVE-2021-28676: For FLI data, FliDecode did not properly check that the block advance was non-zero, potentially leading to an infinite loop on load.
https://lists.fedoraproject.org/archives/list/package-announce@lists.fedoraproject.org/message/MQHA5HAIBOYI3R6HDWCLAGFTIQP767FL/
https://github.com/python-pillow/Pillow/pull/5377
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-28676-fix-fli-dos

Pillow 8.2.0 includes a fix for CVE-2021-25287: There is an out-of-bounds read in J2kDecode, in j2ku_graya_la.
https://pillow.readthedocs.io/en/stable/releasenotes/8.2.0.html#cve-2021-25287-cve-2021-25288-fix-oob-read-in-jpeg2kdecode

Pillow 9.0.0 includes a fix for CVE-2022-22816: path_getbbox in path.c in Pillow before 9.0.0 has a buffer over-read during initialization of ImagePath.Path.
https://pillow.readthedocs.io/en/stable/releasenotes/9.0.0.html#fixed-imagepath-path-array-handling

Pillow 8.3.0 includes a fix for CVE-2021-34552: Pillow through 8.2.0 and PIL (also known as Python Imaging Library) through 1.1.7 allow an attacker to pass controlled parameters directly into a convert function to trigger a buffer overflow in Convert.c
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.0.html#buffer-overflow
https://pillow.readthedocs.io/en/stable/releasenotes/index.html

Pillow 10.3.0 introduces a security update addressing CVE-2024-28219 by replacing certain functions with strncpy to prevent buffer overflow issues.

Pillow is affected by an arbitrary code execution vulnerability. If an attacker has control over the keys passed to the environment argument of PIL.ImageMath.eval(), they may be able to execute arbitrary code.
https://pillow.readthedocs.io/en/stable/releasenotes/10.2.0.html

Pillow 10.0.0 includes a fix for CVE-2023-44271: Denial of Service that uncontrollably allocates memory to process a given task, potentially causing a service to crash by having it run out of memory. This occurs for truetype in ImageFont when textlength in an ImageDraw instance operates on a long text argument.
https://github.com/python-pillow/Pillow/pull/7244

Pillow 8.1.1 includes a fix for CVE-2021-25289: TiffDecode has a heap-based buffer overflow when decoding crafted YCbCr files because of certain interpretation conflicts with LibTIFF in RGBA mode. NOTE: this issue exists because of an incomplete fix for CVE-2020-35654.
https://pillow.readthedocs.io/en/stable/releasenotes/8.1.1.html

Pillow from 5.2.0 and before 8.3.2 is vulnerable to Regular Expression Denial of Service (ReDoS) via the getrgb function.
https://github.com/python-pillow/Pillow/commit/9e08eb8f78fdfd2f476e1b20b7cf38683754866b
https://pillow.readthedocs.io/en/stable/releasenotes/8.3.2.html

An oversight in how the Jinja sandboxed environment detects calls to str.format allows an attacker who controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to store a reference to a malicious string's format method, then pass that to a filter that calls it. No such filters are built-in to Jinja, but could be present through custom filters in an application. After the fix, such indirect calls are also handled by the sandbox.

Prior to 3.1.6, an oversight in how the Jinja sandboxed environment interacts with the |attr filter allows an attacker that controls the content of a template to execute arbitrary Python code. To exploit the vulnerability, an attacker needs to control the content of a template. Whether that is the case depends on the type of application using Jinja. This vulnerability impacts users of applications which execute untrusted templates. Jinja's sandbox does catch calls to str.format and ensures they don't escape the sandbox. However, it's possible to use the |attr filter to get a reference to a string's plain format method, bypassing the sandbox. After the fix, the |attr filter no longer bypasses the environment's attribute lookup. This vulnerability is fixed in 3.1.6.

This affects the package jinja2 from 0.0.0 and before 2.11.3. The ReDoS vulnerability is mainly due to the '_punctuation_re regex' operator and its use of multiple wildcards. The last wildcard is the most exploitable as it searches for trailing punctuation. This issue can be mitigated by Markdown to format user content instead of the urlize filter, or by implementing request timeouts and limiting process memory.

A vulnerability in the Jinja compiler allows an attacker who can control both the content and filename of a template to execute arbitrary Python code, bypassing Jinja's sandbox protections. To exploit this vulnerability, an attacker must have the ability to manipulate both the template's filename and its contents. The risk depends on the application's specific use case. This issue affects applications that render untrusted templates where the attacker can determine the template filename, potentially leading to severe security breaches.

Jinja is an extensible templating engine. The `xmlattr` filter in affected versions of Jinja accepts keys containing non-attribute characters. XML/HTML attributes cannot contain spaces, `/`, `>`, or `=`, as each would then be interpreted as starting a separate attribute. If an application accepts keys (as opposed to only values) as user input, and renders these in pages that other users see as well, an attacker could use this to inject other attributes and perform XSS. The fix for CVE-2024-22195 only addressed spaces but not other characters. Accepting keys as user input is now explicitly considered an unintended use case of the `xmlattr` filter, and code that does so without otherwise validating the input should be flagged as insecure, regardless of Jinja version. Accepting _values_ as user input continues to be safe.

Jinja is an extensible templating engine. Special placeholders in the template allow writing code similar to Python syntax. It is possible to inject arbitrary HTML attributes into the rendered HTML template, potentially leading to Cross-Site Scripting (XSS). The Jinja `xmlattr` filter can be abused to inject arbitrary HTML attribute keys and values, bypassing the auto escaping mechanism and potentially leading to XSS. It may also be possible to bypass attribute validation checks if they are blacklist-based.

Urllib3's ProxyManager ensures that the Proxy-Authorization header is correctly directed only to configured proxies. However, when HTTP requests bypass urllib3's proxy support, there's a risk of inadvertently setting the Proxy-Authorization header, which remains ineffective without a forwarding or tunneling proxy. Urllib3 does not recognize this header as carrying authentication data, failing to remove it during cross-origin redirects. While this scenario is uncommon and poses low risk to most users, urllib3 now proactively removes the Proxy-Authorization header during cross-origin redirects as a precautionary measure. Users are advised to utilize urllib3's proxy support or disable automatic redirects to handle the Proxy-Authorization header securely. Despite these precautions, urllib3 defaults to stripping the header to safeguard users who may inadvertently misconfigure requests.

Urllib3 1.25.9 includes a fix for CVE-2020-26137: Urllib3 before 1.25.9 allows CRLF injection if the attacker controls the HTTP request method, as demonstrated by inserting CR and LF control characters in the first argument of putrequest(). NOTE: this is similar to CVE-2020-26116.
https://github.com/python/cpython/issues/83784
https://github.com/urllib3/urllib3/pull/1800

Urllib3 1.26.5 includes a fix for CVE-2021-33503: When provided with a URL containing many @ characters in the authority component, the authority regular expression exhibits catastrophic backtracking, causing a denial of service if a URL were passed as a parameter or redirected to via an HTTP redirect.
https://github.com/advisories/GHSA-q2q7-5pp4-w6pg

Affected versions of urllib3 are vulnerable to an HTTP redirect handling vulnerability that fails to remove the HTTP request body when a POST changes to a GET via 301, 302, or 303 responses. This flaw can expose sensitive request data if the origin service is compromised and redirects to a malicious endpoint, though exploitability is low when no sensitive data is used. The vulnerability affects automatic redirect behavior. It is fixed in versions 1.26.18 and 2.0.7; update or disable redirects using redirects=False.
This vulnerability is specific to Python's urllib3 library.

Urllib3 1.26.17 and 2.0.5 include a fix for CVE-2023-43804: Urllib3 doesn't treat the 'Cookie' HTTP header special or provide any helpers for managing cookies over HTTP, that is the responsibility of the user. However, it is possible for a user to specify a 'Cookie' header and unknowingly leak information via HTTP redirects to a different origin if that user doesn't disable redirects explicitly.
https://github.com/urllib3/urllib3/security/advisories/GHSA-v845-jxx5-vc9f

Gunicorn version 21.2.0 does not properly validate the value of the 'Transfer-Encoding' header as specified in the RFC standards, which leads to the default fallback method of 'Content-Length,' making it vulnerable to TE.CL request smuggling. This vulnerability can lead to cache poisoning, data exposure, session manipulation, SSRF, XSS, DoS, data integrity compromise, security bypass, information leakage, and business logic abuse.

Gunicorn fails to properly validate Transfer-Encoding headers, leading to HTTP Request Smuggling (HRS) vulnerabilities. By crafting requests with conflicting Transfer-Encoding headers, attackers can bypass security restrictions and access restricted endpoints. This issue is due to Gunicorn's handling of Transfer-Encoding headers, where it incorrectly processes requests with multiple, conflicting Transfer-Encoding headers, treating them as chunked regardless of the final encoding specified. This vulnerability allows for a range of attacks including cache poisoning, session manipulation, and data exposure.

A time-based vulnerability in Gunicorn affected versions allows an attacker to disrupt service by manipulating the system clock, causing premature worker timeouts and potential denial-of-service (DoS) conditions. The issue stems from the use of time.time() in the worker timeout logic, which can be exploited if an attacker gains the ability to change the system time.

Gunicorn 20.0.1 fixes chunked encoding support to prevent http request smuggling attacks.
https://github.com/benoitc/gunicorn/issues/2176
https://github.com/p4k03n4t0r/http-request-smuggling#request-smuggling-using-mitmproxy-and-gunicorn

Werkzeug is a comprehensive WSGI web application library. The debugger in affected versions of Werkzeug can allow an attacker to execute code on a developer's machine under some circumstances. This requires the attacker to get the developer to interact with a domain and subdomain they control, and enter the debugger PIN, but if they are successful it allows access to the debugger even if it is only running on localhost. This also requires the attacker to guess a URL in the developer's application that will trigger the debugger.

Affected versions of Werkzeug are vulnerable to Path Traversal (CWE-22) on Windows systems running Python versions below 3.11. The safe_join() function failed to properly detect certain absolute paths on Windows, allowing attackers to potentially access files outside the intended directory. An attacker could craft special paths starting with "/" that bypass the directory restrictions on Windows systems. The vulnerability exists in the safe_join() function which relied solely on os.path.isabs() for path validation. This is exploitable on Windows systems by passing paths starting with "/" to safe_join(). To remediate, upgrade to the latest version which includes additional path validation checks. 
NOTE: This vulnerability specifically affects Windows systems running Python versions below 3.11 where ntpath.isabs() behavior differs.

Pallets Werkzeug before 0.15.3, when used with Docker, has insufficient debugger PIN randomness because Docker containers share the same machine id.

In Pallets Werkzeug before 0.15.5, SharedDataMiddleware mishandles drive names (such as C:) in Windows pathnames.

Werkzeug 2.2.3 includes a fix for CVE-2023-23934: Browsers may allow "nameless" cookies that look like '=value' instead of 'key=value'. A vulnerable browser may allow a compromised application on an adjacent subdomain to exploit this to set a cookie like '=__Host-test=bad' for another subdomain. Werkzeug prior to 2.2.3 will parse the cookie '=__Host-test=bad' as __Host-test=bad'. If a Werkzeug application is running next to a vulnerable or malicious subdomain which sets such a cookie using a vulnerable browser, the Werkzeug application will see the bad cookie value but the valid cookie key.
https://github.com/pallets/werkzeug/security/advisories/GHSA-px8h-6qxv-m22q

Werkzeug is a comprehensive WSGI web application library. If an upload of a file that starts with CR or LF and then is followed by megabytes of data without these characters: all of these bytes are appended chunk by chunk into internal bytearray and lookup for boundary is performed on growing buffer. This allows an attacker to cause a denial of service by sending crafted multipart data to an endpoint that will parse it. The amount of CPU time required can block worker processes from handling legitimate requests.