(???, from branches/stable/0.7.x)
(Jan 27 2013, from branches/stable/0.7.x)
* Add support for Python 3.1, 3.2 and 3.3 (via 2to3) and for PyPy. The
majority of the coding was done in a sprint run by the Cape Town Python
Users Group with financial assistance from the Python Software Foundation.
* Default input and output encodings changed from UTF-8 to None (i.e. unicode
* Skip Mako benchmarks if Mako isn't installed (rather than failing
(Jan 27 2012, from branches/stable/0.6.x)
* Security fix to enhance sanitizing of CSS in style attributes. Genshi's
`HTMLSanitizer` disallows style attributes by default (this remains
unchanged) and warns against such attacks in its documentation, but
the provided CSS santizing is now less lacking (see 455).
* Fix for error in how `HTMLFormFiller` would handle `textarea` elements if
no value was not supplied form them.
* The `HTMLFormFiller` now correctly handles check boxes and radio buttons
with an empty `value` attribute.
* Template `Context` objects now have a `.copy` method.
* Added a simple `tox.ini` file for using tox to test against multiple
verions of Python.
* Fix for bug in `QName` comparison (see 413).
* Fix for bug in handling of trailing events in match template matches
* Fix i18n namespace declaration in documentation (see 400).
* Fix for bug in caching of events in serializers by no longer caching
`(TEXT, Markup)` events (see 429).
* Fix handling of `None` by `Markup.escape` in `_speedups.c` (see 439).
* Fix handling of internal state by match templates (relevant when multiple
templates match the same part of the stream, see 370).
* Fix handling of multiple events between or on either side of start and end
tags in translated messages (see 404).
* Fix test failures caused by changes in HTMLParser in Python 2.7 (see 501).
* Fix infinite loop in interplotation lexing that was introduced by a change
in Python 2.7's tokenizer (see 540).
* Fix handling of processing instructions without data (see 368).
* Updated MANIFEST.in so as not to rely on build from Subersion 1.6.
(Apr 22 2010, from branches/stable/0.6.x)
* Support for Python 2.3 has been dropped.
* Rewrite of the XPath evaluation engine for better performance and improved
correctness. This is the result of integrating work done by Marcin Kurczych
during GSoC 2008.
* Updated the Python AST processing for template code evaluation to use the
`_ast` module instead of the deprecated `compiler` package, including an
adapter layer for Python 2.4. This, too, is the result of integrating work
done by Marcin Kurczych during GSoC 2008.
* Added caching in the serialization stage for improved performance in some
* Various improvements to the HTML sanitization filter.
* Fix problem with I18n filter that would get confused by expressions in
attribute values when inside an `i18n:msg` block (ticket 250).
* Fix problem with the transformation filter dropping events after the
selection (ticket 290).
* `for` loops in template code blocks no longer establish their own locals
scope, meaning you can now access variables assigned in the loop outside
of the loop, just as you can in regular Python code (ticket 259).
* Import statements inside function definitions in template code blocks no
longer result in an UndefinedError when the imported name is accessed
* Fixed handling of relative URLs with fragment identifiers containing colons
in the `HTMLSanitizer` (ticket 274).
* Added an option to the `HTMLFiller` to also populate password fields.
* Match template processing no longer produces unwanted duplicate output in
some cases (ticket 254).
* Templates instantiated without a loader now get an implicit loader based on
their file path, or the current directory as a fallback (ticket 320).
* Added documentation for the `TemplateLoader`.
* Enhanced documentation for internationalization.
(Jul 9 2008, from branches/stable/0.5.x)
* Fix problem with nested match templates not being applied when buffering
on the outer `py:match` is disabled. Thanks to Erik Bray for reporting the
problem and providing a test case!
* Fix problem in `Translator` filter that would cause the translation of
text nodes to fail if the translation function returned an object that was
not directly a string, but rather something like an instance of the
`LazyProxy` class in Babel (ticket 145).
* Fix problem with match templates incorrectly being applied multiple times.
* Includes from templates loaded via an absolute path now include the correct
file in nested directories as long if no search path has been configured
* Unbuffered match templates could result in parts of the matched content
being included in the output if the match template didn't actually consume
it via one or more calls to the `select()` function (ticket 243).
(Jun 9 2008, from branches/stable/0.5.x)
* Added include directive for text templates (ticket 115).
* Added new markup transformation filter contributed by Alec Thomas. This
provides gorgeous jQuery-inspired stream transformation capabilities based
on XPath expressions.
* When using HTML or XHTML serialization, the `xml:lang` attribute is
automatically translated to the `lang` attribute which HTML user agents
* Added support for the XPath 2 `matches()` function in XPath expressions,
which allow matching against regular expressions.
* Support for Python code blocks in templates can now be disabled
* Includes are now processed when the template is parsed if possible, but
only if the template loader is not set to do automatic reloading. Included
templates are basically inlined into the including template, which can
speed up rendering of that template a bit.
* Added new syntax for text templates, which is more powerful and flexible
with respect to white-space and line breaks. It also supports Python code
blocks. The old syntax is still available and the default for now, but in a
future release the new syntax will become the default, and some time after
that the old syntax will be removed.
* Added support for passing optimization hints to `<py:match>` directives,
which can speed up match templates in many cases, for example when a match
template should only be applied once to a stream, or when it should not be
* Text templates now default to rendering as plain text; it is no longer
necessary to explicitly specify the "text" method to the `render()` or
`serialize()` method of the generated markup stream.
* XInclude elements in markup templates now support the `parse` attribute;
when set to "xml" (the default), the include is processed as before, but
when set to "text", the included template is parsed as a text template using
the new syntax (ticket 101).
* Python code blocks inside match templates are now executed (ticket 155).
* The template engine plugin no longer adds the `default_doctype` when the
`fragment` parameter is `True`.
* The `striptags` function now also removes HTML/XML-style comments (ticket
* The `py:replace` directive can now also be used as an element, with an
attribute named `value` (ticket 144).
* The `TextSerializer` class no longer strips all markup in text by default,
so that it is still possible to use the Genshi `escape` function even with
text templates. The old behavior is available via the `strip_markup` option
of the serializer (ticket 146).
* Assigning to a variable named `data` in a Python code block no longer
breaks context lookup.
* The `Stream.render` now accepts an optional `out` parameter that can be
used to pass in a writable file-like object to use for assembling the
output, instead of building a big string and returning it.
* The XHTML serializer now strips `xml:space` attributes as they are only
allowed on very few tags.
* Match templates are now applied in a more controlled fashion: in the order
they are declared in the template source, all match templates up to (and
including) the matching template itself are applied to the matched content,
whereas the match templates declared after the matching template are only
applied to the generated content (ticket 186).
* The `TemplateLoader` class now provides an `_instantiate()` method that can
be overridden by subclasses to implement advanced template instantiation
logic (ticket 204).
* The search path of the `TemplateLoader` class can now contain ''load
functions'' in addition to path strings. A load function is passed the
name of the requested template file, and should return a file-like object
and some metadata. New load functions are supplied for loading from egg
package data, and loading from different loaders depending on the path
prefix of the requested filename (ticket 182).
* Match templates can now be processed without keeping the complete matched
content in memory, which could cause excessive memory use on long pages.
The buffering can be disabled using the new `buffer` optimization hint on
the `<py:match>` directive.
* Improve error reporting when accessing an attribute in a Python expression
raises an `AttributeError` (ticket 191).
* The `Markup` class now supports mappings for right hand of the `%` (modulo)
operator in the same way the Python string classes do, except that the
substituted values are escape. Also, the special constructor which took
positional arguments that would be substituted was removed. Thus the
`Markup` class now supports the same arguments as that of its `unicode`
base class (ticket 211).
* The `Template` class and its subclasses, as well as the interpolation API,
now take an `filepath` parameter instead of `basedir` (ticket 207).
* The `XHTMLSerializer` now has a `drop_xml_decl` option that defaults to
`True`. Setting it to `False` will cause any XML decl in the serialized
stream to be included in the output as it would for XML serialization.
* Add support for a protocol that would allow interoperability of different
Python packages that generate and/or consume markup, based on the special
`__html__()` method (ticket 202).
(Aug 14, 2007, from branches/stable/0.4.x)
* Fixed augmented assignment to local variables in Python code blocks.
* Fixed handling of nested function and class definitions in Python code
* Includes were not raising `TemplateNotFound` exceptions even when no
fallback has been specified. That has been corrected.
* The template loader now raises a `TemplateNotFound` error when a previously
cached template is removed or renamed, where it previously was passing up
* The Genshi I18n filter can be configured to only extract messages found in
`gettext` function calls, ignoring any text nodes and attribute values
(Jul 17 2007, from branches/stable/0.4.x)
* The I18n filter no longer extracts or translates literal strings in
attribute values that also contain expressions.
* Added `loader_callback` option to plugin interface, which allows specifying
a callback function that the template loader should invoke whenever a new
template is loaded (ticket 130). Note that the value for this option can
not be specified as a string, only as an actual function object, which means
it is not available for use through configuration files.
* The I18n filter now extracts messages from gettext functions even inside
ignored tags (ticket 132).
* The HTML sanitizer now strips any CSS comments in style attributes, which
could previously be used to hide malicious property values.
* The HTML sanitizer now also removes any HTML comments encountered, as those
may be used to hide malicious payloads targetting a certain "innovative"
browser that goes and interprets the content of specially prepared comments.
* Attribute access in template expressions no longer silently ignores
exceptions other than `AttributeError` raised in the attribute accessor.
(Jun 20 2007, from branches/stable/0.4.x)
* The `doctype` parameter of the markup serializers now also accepts the "name"
of the doctype as string, in addition to the `(name, pubid, sysid)` tuple.
* The I18n filter was not replacing the original attributes with the
translation, but instead adding a second attribute with the same name.
* `TextTemplate` can now handle unicode source (ticket 125).
* A `<?python ?>` processing instruction containing trailing whitespace no
longer causes a syntax error (ticket 127).
* The I18n filter now skips the content of elements that have an `xml:lang`
attribute with a fixed string value. Basically, `xml:lang` can now be used
as a flag to mark specific sections as not needing localization.
* Added plugin for message extraction via Babel (http://babel.edgewall.org/).
(May 21 2007, from branches/stable/0.4.x)
* Fix incorrect reference to translation function in the I18N filter.
* The `ET()` function now correctly handles attributes with a namespace.
* XML declarations are now processed internally, as well as written to the
output when XML serialization is used (ticket 111).
* Added the functions `encode()` and `get_serializer()` to the `genshi.output`
module, which provide a lower-level API to the functionality previously only
available through `Stream.render()` and `Stream.serialize()`.
* The `DocType` class now has a `get(name)` function that returns a `DOCTYPE`
tuple for a given string.
* Added frameset variants to the `DocType` constants for HTML 4.01 and XHTML
* Improved I18n extraction for pluralizable messages: for any translation
function with multiple string arguments (such as ``ngettext``), a single
item with a tuple of strings is yielded, instead an item for each string
* The `HTMLFormFiller` stream filter no longer alters form elements for which
the data element contains no corresponding item.
* Code in `<?python ?>` processing instructions no longer gets the special
treatment as Python code in template expressions, i.e. item and attribute
access are no longer interchangeable (which was broken in a number of ways
anyway, see ticket 113). This change does not affect expressions.
* Numerous fixes for the execution of Python code in `<?python ?>` processing
instructions (tickets 113 and 114).
* The `py:def` (and `def`) directive now supports "star args" (i.e. `*args`
and `**kwargs`) in the function declaration (ticket 116).
(Apr 16 2007, from branches/stable/0.4.x)
* New example applications for CherryPy and web.py.
* The template loader now uses a LRU cache to limit the number of cached
templates to a configurable maximum. Also, a race condition in the template
loader was fixed by adding locking.
* A new filter (genshi.filters.HTMLFormFiller) was added, which can populate
HTML forms based on a dictionary of values.
* The set of permitted tag and attribute names for the HTMLSanitizer can now
be configured per instance.
* The template engine plugin now supports a range of options for
configuration, for example to set the default serialization method, the
default output encoding, or the default DOCTYPE.
* The ElementTree adaptation function `ET()` has moved into the `genshi.input`
* Allow `when` directives to omit the test expression as long as the
associated choose directive does have one. In that case, the when branch is
followed if the expression of the choose directive evaluates to a truth
* Unsuccessful attribute or item lookups now return `Undefined` objects for
nicer error messages.
* Split up the `genshi.template` module into multiple modules inside the new
* Results of expression evaluation are no longer implicitly called if they
* Instances of the `genshi.core.Attrs` class are now immutable (they are
subclasses of `tuple` instead of `list`).
* `MarkupTemplate`s can now be instantiated from markup streams, in addition
to strings and file-like objects (ticket 69).
* Improve handling of incorrectly nested tags in the HTML parser.
* Template includes can now be nested inside fallback content.
* Expressions can now contain dict literals (ticket 37).
* It is now possible to have one or more escaped dollar signs in front of a
full expression (ticket 92).
* The `Markup` class is now available by default in template expressions
* The handling of namespace declarations in XML/XHTML output has been improved.
* The `Attrs` class no longer automatically wraps all attribute names in
`QName` objects. This is now the responsibility of whoever is instantiating
`Attrs` objects (for example, stream filters and generators).
* Python code blocks are now supported using the `<?python ?>` processing
instruction (ticket 84).
* The way errors in template expressions are handled can now be configured. The
option `LenientLookup` provides the same forgiving mode used in previous
Genshi versions, while `StrictLookup` raises exceptions when undefined
variables or members are accessed. The lenient mode is still the default in
this version, but that may change in the future. (ticket 88)
* If a variable is not necessarily defined at the top level of the template
data, the new built-in functions `defined(key)` and `value_of(key, default)`
can be used so that the template also works in strict lookup mode. These
functions were previously only available when using Genshi via the template
engine plugin (for compatibility with Kid).
* `style` attributes are no longer allowed by the `HTMLSanitizer` by default.
If they are explicitly added to the set of safe attributes, any unicode
escapes in the attribute value are now handled properly.
* Namespace declarations on conditional elements (for example using a `py:if`
directive`) are no longer moved to the following element when the element
originally carrying the declaration is removed from the stream (ticket 107).
* Added basic built-in support for internationalizing templates by providing
a new `Translator` class that can both extract localizable strings from a
stream, and replace those strings with their localizations at render time.
The code for this was largely taken from previous work done by Matt Good
and David Fraser.
(Dec 11 2006, from branches/stable/0.3.x)
* The builder API now accepts streams as children of elements and fragments.
(Nov 22 2006, from branches/stable/0.3.x)
* Fix XPath traversal in match templates. Previously, `div/p` would be treated
the same as `div//p`, i.e. it would match all descendants and not just the
* Preserve whitespace in HTML `<pre>` elements also when they contain child
* Match templates no longer match their own output (ticket 77).
* Blank lines before directives in text templates are now preserved as
expected (ticket 62).
(Nov 2 2006, from branches/stable/0.3.x)
* The encoding of HTML and XML files, as well as markup and text templates,
can now be specified. Also, the encoding specified in XML declarations is
now respected unless an expiclit encoding is requested.
* Expressions used as arguments for `py:with`, `py:def`, and `py:for`
directives can now contain non-ASCII strings.
(Oct 16 2006, from branches/stable/0.3.x)
* Fixed bug introduced in 0.3.2 that broke the parsing of templates which
declare the same namespace more than once in a nested fashion.
* Fixed the parsing of HTML entity references inside attribute values, both
in the `XMLParser` and the `HTMLParser` classes.
* Some changes to usage of absolute vs. relative template paths to ensure that
the filenamed-keyed cache employed by the TemplateLoader doesn't mix up
templates with the same name, but from different subdirectories.
(Oct 12 2006, from branches/stable/0.3.x)
* Exceptions from templates now contain the absolute path to the template file
when a search path is used. This enables tracebacks to display the line in
* The template engine plugin now provides three different types: "genshi" and
"genshi-markup" map to markup templates, while "genshi-text" maps to text
* Fixed the namespace context used by XPath patterns in py:match templates.
The were erroneously using the namespace context of the elements being
matched, where they should rather use the context in which they were
* The contents of `<script>` and `<style>` elements are no longer escaped when
serializing to HTML but declaring the XHTML namespace in the template.
* Improved error message raised when using the `py:for` directive on an object
that is not iterable (ticket 60).
* Fixed the XPath function `starts-with()` which was always returning true
(Sep 22 2006, from branches/stable/0.3.x)
* Includes and user-defined filters were not getting the correct context data
when used inside a match template (ticket 56).
* XPath patterns using the union operator (`|`) were returning only partial
results in some cases.
(Sep 17 2006, from branches/stable/0.3.x)
* The project name was changed from "Markup" to "Genshi". See UPGRADE.txt
for upgrade instructions.
* Expression evaluation now differentiates between undefined variables and
variables that are defined but set to `None`. This also means that local
variables can override built-ins even if the local variable are set to
`None` (ticket 36).
* The plugin implementation now makes more functions available for use inside
templates: XML(), HTML(), defined(), and value-of(). These are mostly for
migration from Kid.
* The parsing of `py:with` directives has been improved: you can now assign
to multiple names, and semicolons inside string literals are treated as
* Generator expressions can now be used in template expressions (ticket 16).
* Added serialization to plain text (ticket 41).
* Calling functions from template expressions with *args and/or **kwargs
now works correctly (ticket 42).
* The `TemplateLoader` class and the `Template` instances that it manages are
now thread-safe, as they no longer keep any state related to the current
* Stream filters and serializers can now be applied using the "pipe" (bitwise
or) operator "|".
* The execution of named template functions (`py:def`) no longer silently
swallows TypeError exceptions (ticket 44).
* The XML Parser now correctly handles unicode input (ticket 43).
* HTML entities can now be used in templates without having to declare an HTML
* Error reporting on undefined entities has been fixed.
* Fix traversal of substreams by XPath expressions. For example, the path
'*/*' no longer matches non-immediate child elements, and '*' no longer
matches the attributes of descendants.
* Fixes for `py:match` which would get confused when it should be applied
to multiple elements (ticket 49).
* Using namespace prefixes in XPath expressions is now supported.
* Expressions now correctly handle slices (ticket 51).
* A simple text-based template language is now included for generating
plain text output (ticket 47).
(Aug 22 2006, from branches/stable/0.2.x)
* XPath syntax errors now contain position info (ticket 20).
* Errors in expression evaluation now contain the correct line number in the
template (ticket 22).
* <script> and <style> element contents are no longer escaped when using HTML
serialization (ticket 24).
* In some cases expressions in templates did not get interpolated (ticket
* CDATA sections are now passed through the pipeline and serialized correctly.
That allows using <script> or <style> elements in XHTML output that is still
compatible with HTML user agents.
* The XHTML serializer now correctly handles elements in foreign namespaces
(such as SVG or MathML).
* Fixed relative includes in templates on Windows (ticket 27).
* Output can be encoded using legacy codecs such as ISO-8859-1. Any character
not representable in the chosen encoding gets replaced by the corresponding
XML character reference.
* String literals in XPath expressions that contain spaces are now parsed
* Added support for the XPath functions boolean(), ceiling(), concat(),
contains(), false(), floor(), normalize-space(), number(), round(),
starts-with(), string-length(), substring(), substring-after(),
substring-before(), translate(), and true().
* Non-ASCII characters in expressions should now be handled correctly (ticket
* Default values for arguments of template functions now also work with
constants and complex expressions (they only worked for string or number
* XPath expressions in now support XPath variables ($var) in predicates
* Expressions in templates can now span multiple lines if they are enclosed
in curly braces.
* py:def macros can now be invoked from within expressions inside attribute
values (ticket 34).
(Aug 3 2006, from branches/stable/0.1.x)
* First public release