Other changes

• LP#1958539: The lxml.html.clean implementation suffered from several (only if used) security issues in the past and was now extracted into a separate library:

  https://github.com/fedora-python/lxml_html_clean

  Projects that use lxml without "lxml.html.clean" will not notice any difference, except that they won't have potentially vulnerable code installed. The module is available as an "extra" setuptools dependency "lxml[html_clean]", so that Projects that need "lxml.html.clean" will need to switch their requirements from "lxml" to "lxml[html_clean]", or install the new library themselves.

• The minimum CPU architecture for the Linux x86 binary wheels was upgraded to "sandybridge" (launched 2011), and glibc 2.28 / gcc 12 (manylinux_2_28) wheels were added.

• Built with Cython 3.0.10.

Bugs fixed

• LP#2048920: iterlinks() in lxml.html rejected bytes input in 5.1.0.
• High source line numbers from the parser are no longer truncated (up to a C long) when using libxml2 2.11 or later.

Other changes

• GH#407: A compatibility test was adapted to recent expat versions. Patch by Miro Hrončok.
• Binary wheels use the library versions libxml2 2.12.6 and libxslt 1.1.39.
• Windows binary wheels use the library versions libxml2 2.11.7 and libxslt 1.1.39.
• Built with Cython 3.0.9.

Features added

• Parsing ASCII strings is slightly faster.

Bugs fixed

• GH#349: The HTML Cleaner() interpreted an accidentally provided string parameter for the host_whitelist as list of characters and silently failed to reject any hosts. Passing a non-collection is now rejected.

Other changes

• Support for Python 2.7 and Python versions < 3.6 was removed.
• The wheel build was migrated to use cibuildwheel. Patch by Primož Godec.

Other changes

• GH#407: A compatibility test was adapted to recent expat versions. Patch by Miro Hrončok.
• Binary wheels use the library versions libxml2 2.12.6 and libxslt 1.1.39.
• Built with Cython 3.0.9.

Bugs fixed

• LP#2046208: Parsing non-BMP Python Unicode strings could fail on macOS.
• LP#2044225: When incrementally parsing broken HTML, reporting start events on missing structural tags failed and could lead to subsequent exceptions.
• LP#2045435: Some (not all) issues with stricter C compilers were resolved.
• The binary wheels in the 5.0.0 release did not validate cleanly (but installed ok).

Features added

• Character escaping in C14N2 serialisation now uses a single pass over the text instead of searching for each unescaped character separately.
• Early support for Python 3.13a2 was added.

Bugs fixed

• LP#1976304: The Element.addnext() method previously inserted the new element before existing tail text. The tail text of both sibling elements now stays on the respective elements.
• LP#1980767, GH#379: TreeBuilder.close() could fail with a TypeError after parsing incorrect input. Original patch by Enrico Minack.
• Element.itertext(with_tail=False) returned the tail text of comments and processing instructions, despite the explicit option.
• GH#370: A crash with recent libxml2 2.11.x versions was resolved. Patch by Michael Schlenker.
• A compile problem with recent libxml2 2.12.x versions was resolved.
• The internal exception handling in C callbacks was improved for Cython 3.0.
• The exception declarations of xmlInputReadCallback, xmlInputCloseCallback, xmlOutputWriteCallback and xmlOutputCloseCallback in tree.pxd were corrected to prevent running Python code or calling into the C-API with a live exception set.
• GH#385: The long deprecated unittest.m̀akeSuite() function is no longer used. Patch by Miro Hrončok.
• LP#1522052: A file-system specific test is now optional and should no longer fail on systems that don't support it.
• GH#392: Some tests were adapted for libxml2 2.13. Patch by Nick Wellnhofer.
• Contains all fixes from lxml 4.9.4.

Other changes

• LP#1742885: lxml no longer expands external entities (XXE) by default to prevent the security risk of loading arbitrary files and URLs. If this feature is needed, it can be enabled in a backwards compatible way by using a parser with the option resolve_entities=True. The new default is resolve_entities='internal'.

• With libxml2 2.10.4 and later (as provided by the lxml 5.0 binary wheels), parsing HTML tags with "prefixes" no longer builds a namespace dictionary in nsmap but considers the prefix:name string the actual tag name. With older libxml2 versions, since 2.9.11, the prefix was removed. Before that, the prefix was parsed as XML prefix.

  lxml 5.0 does not try to hide this difference but now changes the ElementPath implementation to let element.find("part1:part2") search for the tag part1:part2 in documents parsed as HTML, instead of looking only for part2.

• LP#2024343: The validation of the schema file itself is now optional in the ISO-Schematron implementation. This was done because some lxml distributions discard the RNG validation schema file due to licensing issues. The validation can now always be disabled with Schematron(..., validate_schema=False). It is enabled by default if available and disabled otherwise. The module constant lxml.isoschematron.schematron_schema_valid_supported can be used to detect whether schema file validation is available.

• Some redundant and long deprecated methods were removed: parser.setElementClassLookup(), xslt_transform.apply(), xpath.evaluate().

• Some incorrect declarations were removed from python.pxd. In general, this file should not be used by external Cython code. Use the C-API declarations provided by Cython itself instead.

• Binary wheels use the library versions libxml2 2.12.3 and libxslt 1.1.39.

• Built with Cython 3.0.7, updated to follow recent changes in Cython 3.1-dev.

Bugs fixed

• LP#2046398: Inserting/replacing an ancestor into a node's children could loop indefinitely.
• LP#1980767, GH#379: TreeBuilder.close() could fail with a TypeError after parsing incorrect input. Original patch by Enrico Minack.
• LP#1522052: A file-system specific test is now optional and should no longer fail on systems that don't support it.

Other changes

• Wheels include zlib 1.3, libxml2 2.10.3 and libxslt 1.1.39 (zlib 1.2.12, libxml2 2.10.3 and libxslt 1.1.37 on Windows).
• Built with Cython 0.29.37.

Bugs fixed

• LP#2008911: lxml.objectify accepted non-decimal numbers like ²²² as integers.
• A memory leak in lxml.html.clean was resolved by switching to Cython 0.29.34+.
• GH#348: URL checking in the HTML cleaner was improved. Patch by Tim McCormack.
• GH#371, GH#373: Some regex strings were changed to raw strings to fix Python warnings. Patches by Jakub Wilk and Anthony Sottile.

Other changes

• Wheels include zlib 1.2.13, libxml2 2.10.3 and libxslt 1.1.38 (zlib 1.2.12, libxml2 2.10.3 and libxslt 1.1.37 on Windows).
• Built with Cython 0.29.36 to adapt to changes in Python 3.12.

Bugs fixed

• CVE-2022-2309: A Bug in libxml2 2.9.1[0-4] could let namespace declarations from a failed parser run leak into later parser runs. This bug was worked around in lxml and resolved in libxml2 2.10.0. https://gitlab.gnome.org/GNOME/libxml2/-/issues/378

Other changes

• LP#1981760: Element.attrib now registers as collections.abc.MutableMapping.
• lxml now has a static build setup for macOS on ARM64 machines (not used for building wheels). Patch by Quentin Leffray.

Bugs fixed

• A crash was resolved when using iterwalk() (or canonicalize()) after parsing certain incorrect input. Note that iterwalk() can crash on valid input parsed with the same parser after failing to parse the incorrect input.

Bugs fixed

• GH#341: The mixin inheritance order in lxml.html was corrected. Patch by xmo-odoo.

Other changes

• Built with Cython 0.29.30 to adapt to changes in Python 3.11 and 3.12.
• Wheels include zlib 1.2.12, libxml2 2.9.14 and libxslt 1.1.35 (libxml2 2.9.12+ and libxslt 1.1.34 on Windows).
• GH#343: Windows-AArch64 build support in Visual Studio. Patch by Steve Dower.

Features added

• GH#337: Path-like objects are now supported throughout the API instead of just strings. Patch by Henning Janssen.
• The ElementMaker now supports QName values as tags, which always override the default namespace of the factory.

Bugs fixed

• GH#338: In lxml.objectify, the XSI float annotation "nan" and "inf" were spelled in lower case, whereas XML Schema datatypes define them as "NaN" and "INF" respectively. Patch by Tobias Deiminger.

Other changes

• Built with Cython 0.29.28.

Features added

• Chunked Unicode string parsing via parser.feed() now encodes the input data to the native UTF-8 encoding directly, instead of going through Py_UNICODE / wchar_t encoding first, which previously required duplicate recoding in most cases.

Bugs fixed

• The standard namespace prefixes were mishandled during "C14N2" serialisation on Python 3. See https://mail.python.org/archives/list/lxml@python.org/thread/6ZFBHFOVHOS5GFDOAMPCT6HM5HZPWQ4Q/
• lxml.objectify previously accepted non-XML numbers with underscores (like "1_000") as integers or float values in Python 3.6 and later. It now adheres to the number format of the XML spec again.
• LP#1939031: Static wheels of lxml now contain the header files of zlib and libiconv (in addition to the already provided headers of libxml2/libxslt/libexslt).

Other changes

• Wheels include libxml2 2.9.12+ and libxslt 1.1.34 (also on Windows).

Bugs fixed

• A vulnerability (GHSL-2021-1038) in the HTML cleaner allowed sneaking script content through SVG images (CVE-2021-43818).
• A vulnerability (GHSL-2021-1037) in the HTML cleaner allowed sneaking script content through CSS imports and other crafted constructs (CVE-2021-43818).

Features added

• GH#317: A new property system_url was added to DTD entities. Patch by Thirdegree.
• GH#314: The STATIC_* variables in setup.py can now be passed via env vars. Patch by Isaac Jurado.

Bugs fixed

• A vulnerability (CVE-2021-28957) was discovered in the HTML Cleaner by Kevin Chung, which allowed JavaScript to pass through. The cleaner now removes the HTML5 formaction attribute.

Bugs fixed

• A vulnerability (CVE-2020-27783) was discovered in the HTML Cleaner by Yaniv Nizry, which allowed JavaScript to pass through. The cleaner now removes more sneaky "style" content.

Bugs fixed

• A vulnerability was discovered in the HTML Cleaner by Yaniv Nizry, which allowed JavaScript to pass through. The cleaner now removes more sneaky "style" content.

Features added

• GH#310: lxml.html.InputGetter supports __len__() to count the number of input fields. Patch by Aidan Woolley.
• lxml.html.InputGetter has a new .items() method to ease processing all input fields.
• lxml.html.InputGetter.keys() now returns the field names in document order.
• GH-309: The API documentation is now generated using sphinx-apidoc. Patch by Chris Mayo.

Bugs fixed

• LP#1869455: C14N 2.0 serialisation failed for unprefixed attributes when a default namespace was defined.
• TreeBuilder.close() raised AssertionError in some error cases where it should have raised XMLSyntaxError. It now raises a combined exception to keep up backwards compatibility, while switching to XMLSyntaxError as an interface.

Bugs fixed

• Cleaner() now validates that only known configuration options can be set.
• LP#1882606: Cleaner.clean_html() discarded comments and PIs regardless of the corresponding configuration option, if remove_unknown_tags was set.
• LP#1880251: Instead of globally overwriting the document loader in libxml2, lxml now sets it per parser run, which improves the interoperability with other users of libxml2 such as libxmlsec.
• LP#1881960: Fix build in CPython 3.10 by using Cython 0.29.21.
• The setup options "--with-xml2-config" and "--with-xslt-config" were accidentally renamed to "--xml2-config" and "--xslt-config" in 4.5.1 and are now available again.

Bugs fixed

• LP#1570388: Fix failures when serialising documents larger than 2GB in some cases.
• LP#1865141, GH#298: QName values were not accepted by the el.iter() method. Patch by xmo-odoo.
• LP#1863413, GH#297: The build failed to detect libraries on Linux that are only configured via pkg-config. Patch by Hugh McMaster.

Features added

• A new function indent() was added to insert tail whitespace for pretty-printing an XML tree.

Bugs fixed

• LP#1857794: Tail text of nodes that get removed from a document using item deletion disappeared silently instead of sticking with the node that was removed.

Other changes

• MacOS builds are 64-bit-only by default. Set CFLAGS and LDFLAGS explicitly to override it.
• Linux/MacOS Binary wheels now use libxml2 2.9.10 and libxslt 1.1.34.
• LP#1840234: The package version number is now available as lxml.__version__.

Bugs fixed

• LP#1844674: itertext() was missing tail text of comments and PIs since 4.4.0.

Bugs fixed

• LP#1835708: ElementInclude incorrectly rejected repeated non-recursive includes as recursive. Patch by Rainer Hausdorf.

Bugs fixed

• LP#1838252: The order of an OrderedDict was lost in 4.4.0 when passing it as attrib mapping during element creation.
• LP#1838521: The package metadata now lists the supported Python versions.

Features added

• Element.clear() accepts a new keyword argument keep_tail=True to clear everything but the tail text. This is helpful in some document-style use cases and for clearing the current element in iterparse() and pull parsing.
• When creating attributes or namespaces from a dict in Python 3.6+, lxml now preserves the original insertion order of that dict, instead of always sorting the items by name. A similar change was made for ElementTree in CPython 3.8. See https://bugs.python.org/issue34160
• Integer elements in lxml.objectify implement the __index__() special method.
• GH#269: Read-only elements in XSLT were missing the nsmap property. Original patch by Jan Pazdziora.
• ElementInclude can now restrict the maximum inclusion depth via a max_depth argument to prevent content explosion. It is limited to 6 by default.
• The target object of the XMLParser can have start_ns() and end_ns() callback methods to listen to namespace declarations.
• The TreeBuilder has new arguments comment_factory and pi_factory to pass factories for creating comments and processing instructions, as well as flag arguments insert_comments and insert_pis to discard them from the tree when set to false.
• A C14N 2.0 implementation was added as etree.canonicalize(), a corresponding C14NWriterTarget class, and a c14n2 serialisation method.

Bugs fixed

• 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 '../'.
• Assigning to Element child slices with negative step could insert the slice at the wrong position, starting too far on the left.
• Assigning to Element child slices with overly large step size could take very long, regardless of the length of the actual slice.
• Assigning to Element child slices of the wrong size could sometimes fail to raise a ValueError (like a list assignment would) and instead assign outside of the original slice bounds or leave parts of it unreplaced.
• The comment and pi events in iterwalk() were never triggered, and instead, comments and processing instructions in the tree were reported as start elements. Also, when walking an ElementTree (as opposed to its root element), comments and PIs outside of the root element are now reported.
• LP#1827833: The RelaxNG compact syntax support was broken with recent versions of rnc2rng.
• LP#1758553: The HTML elements source and track were added to the list of empty tags in lxml.html.defs.
• Registering a prefix other than "xml" for the XML namespace is now rejected.
• Failing to write XSLT output to a file could raise a misleading exception. It now raises IOError.

Other changes

• Support for Python 3.4 was removed.
• When using Element.find*() with prefix-namespace mappings, the empty string is now accepted to define a default namespace, in addition to the previously supported None prefix. Empty strings are more convenient since they keep all prefix keys in a namespace dict strings, which simplifies sorting etc.
• The ElementTree.write_c14n() method has been deprecated in favour of the long preferred ElementTree.write(f, method="c14n"). It will be removed in a future release.

Bugs fixed

• Fix leak of output buffer and unclosed files in _XSLTResultTree.write_output().

Bugs fixed

• Crash in 4.3.1 when appending a child subtree with certain text nodes.

Other changes

• Built with Cython 0.29.6.

Bugs fixed

• LP#1814522: Crash when appending a child subtree that contains unsubstituted entity references.

Other changes

• Built with Cython 0.29.5.

Features added

• The module lxml.sax is compiled using Cython in order to speed it up.
• GH#267: lxml.sax.ElementTreeProducer now preserves the namespace prefixes. If two prefixes point to the same URI, the first prefix in alphabetical order is used. Patch by Lennart Regebro.
• Updated ISO-Schematron implementation to 2013 version (now MIT licensed) and the corresponding schema to the 2016 version (with optional "properties").

Other changes

• GH#270, GH#271: Support for Python 2.6 and 3.3 was removed. Patch by hugovk.
• The minimum dependency versions were raised to libxml2 2.9.2 and libxslt 1.1.27, which were released in 2014 and 2012 respectively.
• Built with Cython 0.29.2.

Bugs fixed

• LP#1799755: Fix a DeprecationWarning in Py3.7+.
• Import warnings in Python 3.6+ were resolved.

Bugs fixed

• Javascript URLs that used URL escaping were not removed by the HTML cleaner. Security problem found by Omar Eissa. (CVE-2018-19787)

Features added

• GH#259: Allow using pkg-config for build configuration. Patch by Patrick Griffis.

Bugs fixed

• LP#1773749, GH#268: Crash when moving an element to another document with Element.insert(). Patch by Alexander Weggerle.

Bugs fixed

• Reverted GH#265: lxml links against zlib as a shared library again.

Bugs fixed

• GH#266: Fix sporadic crash during GC when parse-time schema validation is used and the parser participates in a reference cycle. Original patch by Julien Greard.
• GH#265: lxml no longer links against zlib as a shared library, only on static builds. Patch by Nehal J Wani.

Bugs fixed

• LP#1755825: iterwalk() failed to return the 'start' event for the initial element if a tag selector is used.
• LP#1756314: Failure to import 4.2.0 into PyPy due to a missing library symbol.
• LP#1727864, GH#258: Add "-isysroot" linker option on MacOS as needed by XCode 9.

Features added

• GH#255: SelectElement.value returns more standard-compliant and browser-like defaults for non-multi-selects. If no option is selected, the value of the first option is returned (instead of None). If multiple options are selected, the value of the last one is returned (instead of that of the first one). If no options are present (not standard-compliant) SelectElement.value still returns None.
• GH#261: The HTMLParser() now supports the huge_tree option. Patch by stranac.

Bugs fixed

• LP#1551797: Some XSLT messages were not captured by the transform error log.
• LP#1737825: Crash at shutdown after an interrupted iterparse run with XMLSchema validation.

Other changes

Features added

• ElementPath supports text predicates for current node, like "[.='text']".
• ElementPath allows spaces in predicates.
• Custom Element classes and XPath functions can now be registered with a decorator rather than explicit dict assignments.
• Static Linux wheels are now built with link time optimisation (LTO) enabled. This should have a beneficial impact on the overall performance by providing a tighter compiler integration between lxml and libxml2/libxslt.

Bugs fixed

• LP#1722776: Requesting non-Element objects like comments from a document with PythonElementClassLookup could fail with a TypeError.

Features added

• The ElementPath implementation is now compiled using Cython, which speeds up the .find*() methods quite significantly.
• The modules lxml.builder, lxml.html.diff and lxml.html.clean are also compiled using Cython in order to speed them up.
• xmlfile() supports async coroutines using async with and await.
• iterwalk() has a new method skip_subtree() that prevents walking into the descendants of the current element.
• RelaxNG.from_rnc_string() accepts a base_url argument to allow relative resource lookups.
• The XSLT result object has a new method .write_output(file) that serialises output data into a file according to the <xsl:output> configuration.

Bugs fixed

• GH#251: HTML comments were handled incorrectly by the soupparser. Patch by mozbugbox.

• LP#1654544: The html5parser no longer passes the useChardet option if the input is a Unicode string, unless explicitly requested. When parsing files, the default is to enable it when a URL or file path is passed (because the file is then opened in binary mode), and to disable it when reading from a file(-like) object.

  Note: This is a backwards incompatible change of the default configuration. If your code parses byte strings/streams and depends on character detection, please pass the option guess_charset=True explicitly, which already worked in older lxml versions.

• LP#1703810: etree.fromstring() failed to parse UTF-32 data with BOM.

• LP#1526522: Some RelaxNG errors were not reported in the error log.

• LP#1567526: Empty and plain text input raised a TypeError in soupparser.

• LP#1710429: Uninitialised variable usage in HTML diff.

• LP#1415643: The closing tags context manager in xmlfile() could continue to output end tags even after writing failed with an exception.

• LP#1465357: xmlfile.write() now accepts and ignores None as input argument.

• Compilation under Py3.7-pre failed due to a modified function signature.

Other changes

• The main module source files were renamed from lxml.*.pyx to plain *.pyx (e.g. etree.pyx) to simplify their handling in the build process. Care was taken to keep the old header files as fallbacks for code that compiles against the public C-API of lxml, but it might still be worth validating that third-party code does not notice this change.

Features added

• ElementTree.write() has a new option doctype that writes out a doctype string before the serialisation, in the same way as tostring().
• GH#220: xmlfile allows switching output methods at an element level. Patch by Burak Arslan.
• LP#1595781, GH#240: added a PyCapsule Python API and C-level API for passing externally generated libxml2 documents into lxml.
• GH#244: error log entries have a new property path with an XPath expression (if known, None otherwise) that points to the tree element responsible for the error. Patch by Bob Kline.
• The namespace prefix mapping that can be used in ElementPath now injects a default namespace when passing a None prefix.

Bugs fixed

• GH#238: Character escapes were not hex-encoded in the xmlfile serialiser. Patch by matejcik.
• GH#229: fix for externally created XML documents. Patch by Theodore Dubois.
• LP#1665241, GH#228: Form data handling in lxml.html no longer strips the option values specified in form attributes but only the text values. Patch by Ashish Kulkarni.
• LP#1551797: revert previous fix for XSLT error logging as it breaks multi-threaded XSLT processing.
• LP#1673355, GH#233: fromstring() html5parser failed to parse byte strings.

Other changes

• The previously undocumented docstring option in ElementTree.write() produces a deprecation warning and will eventually be removed.

Bugs fixed

• LP#1551797: revert previous fix for XSLT error logging as it breaks multi-threaded XSLT processing.
• LP#1673355, GH#233: fromstring() html5parser failed to parse byte strings.

Bugs fixed

• GH#218 was ineffective in Python 3.
• GH#222: lxml.html.submit_form() failed in Python 3. Patch by Jakub Wilk.

Bugs fixed

• Work around installation problems in recent Python 2.7 versions due to FTP download failures.
• GH#219: xmlfile.element() was not properly quoting attribute values. Patch by Burak Arslan.
• GH#218: xmlfile.element() was not properly escaping text content of script/style tags. Patch by Burak Arslan.

Features added

• GH#217: XMLSyntaxError now behaves more like its SyntaxError baseclass. Patch by Philipp A.
• GH#216: HTMLParser() now supports the same collect_ids parameter as XMLParser(). Patch by Burak Arslan.
• GH#210: Allow specifying a serialisation method in xmlfile.write(). Patch by Burak Arslan.
• GH#203: New option default_doctype in HTMLParser that allows disabling the automatic doctype creation. Patch by Shadab Zafar.
• GH#201: Calling the method .set('attrname') without value argument (or None) on HTML elements creates an attribute without value that serialises like <div attrname></div>. Patch by Daniel Holth.
• GH#197: Ignore form input fields in form_values() when they are marked as disabled in HTML. Patch by Kristian Klemon.

Bugs fixed

• GH#206: File name and line number were missing from XSLT error messages. Patch by Marcus Brinkmann.

Other changes

• Log entries no longer allow anything but plain string objects as message text and file name.
• zlib is included in the list of statically built libraries.

Features added

• GH#180: Separate option inline_style for Cleaner that only removes style attributes instead of all styles. Patch by Christian Pedersen.
• GH#196: Windows build support for Python 3.5. Contribution by Maximilian Hils.

Bugs fixed

• GH#199: Exclude file fields from FormElement.form_values (as browsers do). Patch by Tomas Divis.
• GH#198, LP#1568167: Try to provide base URL from Resolver.resolve_string(). Patch by Michael van Tellingen.
• GH#191: More accurate float serialisation in objectify.FloatElement. Patch by Holger Joukl.
• LP#1551797: Repair XSLT error logging. Patch by Marcus Brinkmann.

Features added

• GH#187: Now supports (only) version 5.x and later of PyPy. Patch by Armin Rigo.
• GH#181: Direct support for .rnc files in RelaxNG() if rnc2rng is installed. Patch by Dirkjan Ochtman.

Bugs fixed

• GH#189: Static builds honour FTP proxy configurations when downloading the external libs. Patch by Youhei Sakurai.
• GH#186: Soupparser failed to process entities in Python 3.x. Patch by Duncan Morris.
• GH#185: Rare encoding related TypeError on import was fixed. Patch by Petr Demin.

Bugs fixed

• Unicode string results failed XPath queries in PyPy.
• LP#1497051: HTML target parser failed to terminate on exceptions and continued parsing instead.
• Deprecated API usage in doctestcompare.

Features added

• cleanup_namespaces() accepts a new argument keep_ns_prefixes that does not remove definitions of the provided prefix-namespace mapping from the tree.
• cleanup_namespaces() accepts a new argument top_nsmap that moves definitions of the provided prefix-namespace mapping to the top of the tree.
• LP#1490451: Element objects gained a cssselect() method as known from lxml.html. Patch by Simon Sapin.
• API functions and methods behave and look more like Python functions, which allows introspection on them etc. One side effect to be aware of is that the functions now bind as methods when assigned to a class variable. A quick fix is to wrap them in staticmethod() (as for normal Python functions).
• ISO-Schematron support gained an option error_finder that allows passing a filter function for picking validation errors from reports.
• LP#1243600: Elements in lxml.html gained a classes property that provides a set-like interface to the class attribute. Original patch by masklinn.
• LP#1341964: The soupparser now handles DOCTYPE declarations, comments and processing instructions outside of the root element. Patch by Olli Pottonen.
• LP#1421512: The docinfo of a tree was made editable to allow setting and removing the public ID and system ID of the DOCTYPE. Patch by Olli Pottonen.
• LP#1442427: More work-arounds for quirks and bugs in pypy and pypy3.
• lxml.html.soupparser now uses BeautifulSoup version 4 instead of version 3 if available.

Bugs fixed

• Memory errors that occur during tree adaptations (e.g. moving subtrees to foreign documents) could leave the tree in a crash prone state.
• Calling process_children() in an XSLT extension element without an output_parent argument failed with a TypeError. Fix by Jens Tröger.
• GH#162: Image data in HTML data URLs is considered safe and no longer removed by lxml.html.clean JavaScript cleaner.
• GH#166: Static build could link libraries in wrong order.
• GH#172: Rely a bit more on libxml2 for encoding detection rather than rolling our own in some cases. Patch by Olli Pottonen.
• GH#159: Validity checks for names and string content were tightened to detect the use of illegal characters early. Patch by Olli Pottonen.
• LP#1421921: Comments/PIs before the DOCTYPE declaration were not serialised. Patch by Olli Pottonen.
• LP#659367: Some HTML DOCTYPE declarations were not serialised. Patch by Olli Pottonen.
• LP#1238503: lxml.doctestcompare is now consistent with stdlib's doctest in how it uses + and - to refer to unexpected and missing output.
• Empty prefixes are explicitly rejected when a namespace mapping is used with ElementPath to avoid hiding bugs in user code.
• Several problems with PyPy were fixed by switching to Cython 0.23.

Bugs fixed

• An ElementTree compatibility test added in lxml 3.4.3 that failed in Python 3.4+ was removed again.

Bugs fixed

• Expression cache in ElementPath was ignored. Fix by Changaco.
• LP#1426868: Passing a default namespace and a prefixed namespace mapping as nsmap into xmlfile.element() raised a TypeError.
• LP#1421927: DOCTYPE system URLs were incorrectly quoted when containing double quotes. Patch by Olli Pottonen.
• LP#1419354: meta-redirect URLs were incorrectly processed by iterlinks() if preceded by whitespace.

Bugs fixed

• LP#1415907: Crash when creating an XMLSchema from a non-root element of an XML document.
• LP#1369362: HTML cleaning failed when hitting processing instructions with pseudo-attributes.
• CDATA() wrapped content was rejected for tail text.
• CDATA sections were not serialised as tail text of the top-level element.

Features added

• New htmlfile HTML generator to accompany the incremental xmlfile serialisation API. Patch by Burak Arslan.

Bugs fixed

• lxml.sax.ElementTreeContentHandler did not initialise its superclass.

Features added

• xmlfile(buffered=False) disables output buffering and flushes the content after each API operation (starting/ending element blocks or writes). A new method xf.flush() can alternatively be used to explicitly flush the output.
• lxml.html.document_fromstring has a new option ensure_head_body=True which will add an empty head and/or body element to the result document if missing.
• lxml.html.iterlinks now returns links inside meta refresh tags.
• New XMLParser option collect_ids=False to disable ID hash table creation. This can substantially speed up parsing of documents with many different IDs that are not used.
• The parser uses per-document hash tables for XML IDs. This reduces the load of the global parser dict and speeds up parsing for documents with many different IDs.
• ElementTree.getelementpath(element) returns a structural ElementPath expression for the given element, which can be used for lookups later.
• xmlfile() accepts a new argument close=True to close file(-like) objects after writing to them. Before, xmlfile() only closed the file if it had opened it internally.
• Allow "bytearray" type for ASCII text input.

Bugs fixed

Other changes

• LP#400588: decoding errors have become hard errors even in recovery mode. Previously, they could lead to an internal tree representation in a mixed encoding state, which lead to very late errors or even silently incorrect behaviour during tree traversal or serialisation.
• Requires Python 2.6, 2.7, 3.2 or later. No longer supports Python 2.4, 2.5 and 3.1, use lxml 3.3.x for those.
• Requires libxml2 2.7.0 or later and libxslt 1.1.23 or later, use lxml 3.3.x with older versions.

Bugs fixed

• Prevent tree cycle creation when adding Elements as siblings.
• LP#1361948: crash when deallocating Element siblings without parent.
• LP#1354652: crash when traversing internally loaded documents in XSLT extension functions.

Bugs fixed

• HTML cleaning could fail to strip javascript links that mix control characters into the link scheme.

Features added

• Source line numbers above 65535 are available on Elements when using libxml2 2.9 or later.

Bugs fixed

• lxml.html.fragment_fromstring() failed for bytes input in Py3.

Other changes

Bugs fixed

• LP#1287118: Crash when using Element subtypes with __slots__.

Other changes

• The internal classes _LogEntry and _Attrib can no longer be subclassed from Python code.

Bugs fixed

• The properties resolvers and version, as well as the methods set_element_class_lookup() and makeelement(), were lost from iterparse objects in 3.3.0.
• LP#1222132: instances of XMLSchema, Schematron and RelaxNG did not clear their local error_log before running a validation.
• LP#1238500: lxml.doctestcompare mixed up "expected" and "actual" in attribute values.
• Some file I/O tests were failing in MS-Windows due to non-portable temp file usage. Initial patch by Gabi Davar.
• LP#910014: duplicate IDs in a document were not reported by DTD validation.
• LP#1185332: tostring(method="html") did not use HTML serialisation semantics for trailing tail text. Initial patch by Sylvain Viollon.
• LP#1281139: .attrib value of Comments lost its mutation methods in 3.3.0. Even though it is empty and immutable, it should still provide the same interface as that returned for Elements.

Features added

Bugs fixed

• LP#1014290: HTML documents parsed with parser.feed() failed to find elements during tag iteration.
• LP#1273709: Building in PyPy failed due to missing support for PyUnicode_Compare() and PyByteArray_*() in PyPy's C-API.
• LP#1274413: Compilation in MSVC failed due to missing "stdint.h" standard header file.
• LP#1274118: iterparse() failed to parse BOM prefixed files.

Other changes

Features added

Bugs fixed

• The heuristic that distinguishes file paths from URLs was tightened to produce less false negatives.

Other changes

Features added

• The PEP 393 unicode parsing support gained a fallback for wchar strings which might still be somewhat common on Windows systems.

Bugs fixed

• Several error handling problems were fixed throughout the code base that could previously lead to exceptions being silently swallowed or not properly reported.
• The C-API function appendChild() is now deprecated as it does not propagate exceptions (its return type is void). The new function appendChildToElement() was added as a safe replacement.
• Passing a string into fromstringlist() raises an exception instead of parsing the string character by character.

Other changes

• Document cleanup code was simplified using the new GC features in Cython 0.20.

Features added

Bugs fixed

• The (empty) value returned by the attrib property of Entity and Comment objects was mutable.
• Element class lookup wasn't available for the new pull parsers or when using a custom parser target.
• Setting Element attributes on instantiation with both the attrib argument and keyword arguments could modify the mapping passed as attrib.
• LP#1266171: DTDs instantiated from internal/external subsets (i.e. through the docinfo property) lost their attribute declarations.

Other changes

• Built with Cython 0.20pre (gitrev 012ae82eb) to prepare support for Python 3.4.

Features added

• Unicode string parsing was optimised for Python 3.3 (PEP 393).

Bugs fixed

• HTML parsing of Unicode strings could misdecode the input on some platforms.
• Crash in xmlfile() when closing open elements out of order in an error case.

Other changes

Features added

• iterparse() supports the recover option.

Bugs fixed

• Crash in iterparse() for HTML parsing.
• Crash in target parsing with attributes.

Other changes

• The safety check in the read-only tree implementation (e.g. used by PythonElementClassLookup) raises a more appropriate ReferenceError for illegal access after tree disposal instead of an AssertionError. This should only impact test code that specifically checks the original behaviour.

Features added

• New option handle_failures in make_links_absolute() and resolve_base_href() (lxml.html) that enables ignoring or discarding links that fail to parse as URLs.
• New parser classes XMLPullParser and HTMLPullParser for incremental parsing, as implemented for ElementTree in Python 3.4.
• iterparse() enables recovery mode by default for HTML parsing (html=True).

Bugs fixed

• LP#1255132: crash when trying to run validation over non-Element (e.g. comment or PI).
• Error messages in the log and in exception messages that originated from libxml2 could accidentally be picked up from preceding warnings instead of the actual error.
• The ElementMaker in lxml.objectify did not accept a dict as argument for adding attributes to the element it's building. This works as in lxml.builder now.
• LP#1228881: repr(XSLTAccessControl) failed in Python 3.
• Raise ValueError when trying to append an Element to itself or to one of its own descendants, instead of running into an infinite loop.
• LP#1206077: htmldiff discarded whitespace from the output.
• Compressed plain-text serialisation to file-like objects was broken.
• lxml.html.formfill: Fix textarea form filling. The textarea used to be cleared before the new content was set, which removed the name attribute.

Other changes

• Some basic API classes use freelists internally for faster instantiation. This can speed up some iterparse() scenarios, for example.
• iterparse() was rewritten to use the new *PullParser classes internally instead of being a parser itself.

Features added

Bugs fixed

• Crash in xmlfile() when closing open elements out of order in an error case.
• Crash in target parsing with attributes.
• LP#1255132: crash when trying to run validation over non-Element (e.g. comment or PI).

Other changes

Features added

Bugs fixed

• Memory leak when creating an XPath evaluator in a thread.
• LP#1228881: repr(XSLTAccessControl) failed in Python 3.
• Raise ValueError when trying to append an Element to itself or to one of its own descendants.
• LP#1206077: htmldiff discarded whitespace from the output.
• Compressed plain-text serialisation to file-like objects was broken.

Other changes

Bugs fixed

• Fix support for Python 2.4 which was lost in 3.2.2.

Features added

Bugs fixed

• LP#1185701: spurious XMLSyntaxError after finishing iterparse().
• Crash in lxml.objectify during xsi annotation.

Other changes

• Return values of user provided element class lookup methods are now validated against the type of the XML node they represent to prevent API class mismatches.

Features added

• The methods apply_templates() and process_children() of XSLT extension elements have gained two new boolean options elements_only and remove_blank_text that discard either all strings or whitespace-only strings from the result list.

Bugs fixed

• When moving Elements to another tree, the namespace cleanup mechanism no longer drops namespace prefixes from attributes for which it finds a default namespace declaration, to prevent them from appearing as unnamespaced attributes after serialisation.
• Returning non-type objects from a custom class lookup method could lead to a crash.
• Instantiating and using subtypes of Comments and ProcessingInstructions crashed.

Other changes

Features added

Bugs fixed

• LP#690319: Leading whitespace could change the behaviour of the string parsing functions in lxml.html.
• LP#599318: The string parsing functions in lxml.html are more robust in the face of uncommon HTML content like framesets or missing body tags. Patch by Stefan Seelmann.
• LP#712941: I/O errors while trying to access files with paths that contain non-ASCII characters could raise UnicodeDecodeError instead of properly reporting the IOError.
• LP#673205: Parsing from in-memory strings disabled network access in the default parser and made subsequent attempts to parse from a URL fail.
• LP#971754: lxml.html.clean appends 'nofollow' to 'rel' attributes instead of overwriting the current value.
• LP#715687: lxml.html.clean no longer discards scripts that are explicitly allowed by the user provided whitelist. Patch by Christine Koppelt.

Other changes

Features added

Bugs fixed

• LP#1136509: Passing attributes through the namespace-unaware API of the sax bridge (i.e. the handler.startElement() method) failed with a TypeError. Patch by Mike Bayer.
• LP#1123074: Fix serialisation error in XSLT output when converting the result tree to a Unicode string.
• GH#105: Replace illegal usage of xmlBufLength() in libxml2 2.9.0 by properly exported API function xmlBufUse().

Other changes

Features added

Bugs fixed

• LP#1160386: Write access to lxml.html.FormElement.fields raised an AttributeError in Py3.
• Illegal memory access during cleanup in incremental xmlfile writer.

Other changes

• The externally useless class lxml.etree._BaseParser was removed from the module dict.

Features added

• GH#89: lxml.html.clean allows overriding the set of attributes that it considers 'safe'. Patch by Francis Devereux.

Bugs fixed

• LP#1104370: copy.copy(el.attrib) raised an exception. It now returns a copy of the attributes as a plain Python dict.
• GH#95: When used with namespace prefixes, the el.find*() methods always used the first namespace mapping that was provided for each path expression instead of using the one that was actually passed in for the current run.
• LP#1092521, GH#91: Fix undefined C symbol in Python runtimes compiled without threading support. Patch by Ulrich Seidl.

Other changes

Features added

• New build-time option --with-unicode-strings for Python 2 that makes the API always return Unicode strings for names and text instead of byte strings for plain ASCII content.
• New incremental XML file writing API etree.xmlfile().
• E factory in lxml.objectify is callable to simplify the creation of tags with non-identifier names without having to resort to getattr().

Bugs fixed

• When starting from a non-namespaced element in lxml.objectify, searching for a child without explicitly specifying a namespace incorrectly found namespaced elements with the requested local name, instead of restricting the search to non-namespaced children.
• GH#85: Deprecation warnings were fixed for Python 3.x.
• GH#33: lxml.html.fromstring() failed to accept bytes input in Py3.
• LP#1080792: Static build of libxml2 2.9.0 failed due to missing file.

Other changes

• The externally useless class _ObjectifyElementMakerCaller was removed from the module API of lxml.objectify.
• LP#1075622: lxml.builder is faster for adding text to elements with many children. Patch by Anders Hammarquist.

Features added

Bugs fixed

• Fix crash during interpreter shutdown by switching to Cython 0.17.3 for building.

Other changes

Features added

Bugs fixed

• LP#1065924: Element proxies could disappear during garbage collection in PyPy without proper cleanup.
• GH#71: Failure to work with libxml2 2.6.x.
• LP#1065139: static MacOS-X build failed in Py3.

Other changes

Features added

Bugs fixed

• End-of-file handling was incorrect in iterparse() when reading from a low-level C file stream and failed in libxml2 2.9.0 due to its improved consistency checks.

Other changes

• The build no longer uses Cython by default unless the generated C files are missing. To use Cython, pass the option "--with-cython". To ignore the fatal build error when Cython is required but not available (e.g. to run special setup.py commands that do not actually run a build), pass "--without-cython".

Features added

• Python level access to (optional) libxml2 memory debugging features to simplify debugging of memory leaks etc.

Bugs fixed

• Fix a memory leak in XPath by switching to Cython 0.17.1.
• Some tests were adapted to work with PyPy.

Other changes

• The code was adapted to work with the upcoming libxml2 2.9.0 release.

Features added

• The .iter() method of elements now accepts tag arguments like "{*}name" to search for elements with a given local name in any namespace. With this addition, all combinations of wildcards now work as expected: "{ns}name", "{}name", "{*}name", "{ns}*", "{}*" and "{*}*". Note that "name" is equivalent to "{}name", but "*" is "{*}*". The same change applies to the .getiterator(), .itersiblings(), .iterancestors(), .iterdescendants(), .iterchildren() and .itertext() methods;the strip_attributes(), strip_elements() and strip_tags() functions as well as the iterparse() class. Patch by Simon Sapin.
• C14N allows specifying the inclusive prefixes to be promoted to top-level during exclusive serialisation.

Bugs fixed

• Passing long Unicode strings into the feed() parser interface failed to read the entire string.

Other changes

Features added

• Initial support for building in PyPy (through cpyext).
• DTD objects gained an API that allows read access to their declarations.
• xpathgrep.py gained support for parsing line-by-line (e.g. from grep output) and for surrounding the output with a new root tag.
• E-factory in lxml.builder accepts subtypes of known data types (such as string subtypes) when building elements around them.
• Tree iteration and iterparse() with a selective tag argument supports passing a set of tags. Tree nodes will be returned by the iterators if they match any of the tags.

Bugs fixed

• The .find*() methods in lxml.objectify no longer use XPath internally, which makes them faster in many cases (especially when short circuiting after a single or couple of elements) and fixes some behavioural differences compared to lxml.etree. Note that this means that they no longer support arbitrary XPath expressions but only the subset that the ElementPath language supports. The previous implementation was also redundant with the normal XPath support, which can be used as a replacement.
• el.find('*') could accidentally return a comment or processing instruction that happened to be in the wrong spot. (Same for the other .find*() methods.)
• The error logging is less intrusive and avoids a global setup where possible.
• Fixed undefined names in html5lib parser.
• xpathgrep.py did not work in Python 3.
• Element.attrib.update() did not accept an attrib of another Element as parameter.
• For subtypes of ElementBase that make the .text or .tail properties immutable (as in objectify, for example), inserting text when creating Elements through the E-Factory feature of the class constructor would fail with an exception, stating that the text cannot be modified.

Other changes

• The code base was overhauled to properly use 'const' where the API of libxml2 and libxslt requests it. This also has an impact on the public C-API of lxml itself, as defined in etreepublic.pxd, as well as the provided declarations in the lxml/includes/ directory. Code that uses these declarations may have to be adapted. On the plus side, this fixes several C compiler warnings, also for user code, thus making it easier to spot real problems again.
• The functionality of "lxml.cssselect" was moved into a separate PyPI package called "cssselect". To continue using it, you must install that package separately. The "lxml.cssselect" module is still available and provides the same interface, provided the "cssselect" package can be imported at runtime.
• Element attributes passed in as an attrib dict or as keyword arguments are now sorted by (namespaced) name before being created to make their order predictable for serialisation and iteration. Note that adding or deleting attributes afterwards does not take that order into account, i.e. setting a new attribute appends it after the existing ones.
• Several classes that are for internal use only were removed from the lxml.etree module dict: _InputDocument, _ResolverRegistry, _ResolverContext, _BaseContext, _ExsltRegExp, _IterparseContext, _TempStore, _ExceptionContext, __ContentOnlyElement, _AttribIterator, _NamespaceRegistry, _ClassNamespaceRegistry, _FunctionNamespaceRegistry, _XPathFunctionNamespaceRegistry, _ParserDictionaryContext, _FileReaderContext, _ParserContext, _PythonSaxParserTarget, _TargetParserContext, _ReadOnlyProxy, _ReadOnlyPIProxy, _ReadOnlyEntityProxy, _ReadOnlyElementProxy, _OpaqueNodeWrapper, _OpaqueDocumentWrapper, _ModifyContentOnlyProxy, _ModifyContentOnlyPIProxy, _ModifyContentOnlyEntityProxy, _AppendOnlyElementProxy, _SaxParserContext, _FilelikeWriter, _ParserSchemaValidationContext, _XPathContext, _XSLTResolverContext, _XSLTContext, _XSLTQuotedStringParam
• Several internal classes can no longer be inherited from: _InputDocument, _ResolverRegistry, _ExsltRegExp, _ElementUnicodeResult, _IterparseContext, _TempStore, _AttribIterator, _ClassNamespaceRegistry, _XPathFunctionNamespaceRegistry, _ParserDictionaryContext, _FileReaderContext, _PythonSaxParserTarget, _TargetParserContext, _ReadOnlyPIProxy, _ReadOnlyEntityProxy, _OpaqueDocumentWrapper, _ModifyContentOnlyPIProxy, _ModifyContentOnlyEntityProxy, _AppendOnlyElementProxy, _FilelikeWriter, _ParserSchemaValidationContext, _XPathContext, _XSLTResolverContext, _XSLTContext, _XSLTQuotedStringParam, _XSLTResultTree, _XSLTProcessingInstruction

Features added

Bugs fixed

• Passing long Unicode strings into the feed() parser interface failed to read the entire string.

Other changes

Features added

Bugs fixed

• Crash when merging text nodes in element.remove().
• Crash in sax/target parser when reporting empty doctype.

Other changes

Features added

Bugs fixed

• Crash when building an nsmap (Element property) with empty namespace URIs.
• Crash due to race condition when errors (or user messages) occur during threaded XSLT processing.
• XSLT stylesheet compilation could ignore compilation errors.

Other changes

Features added

• lxml.html.tostring() gained new serialisation options with_tail and doctype.

Bugs fixed

• Fixed a crash when using iterparse() for HTML parsing and requesting start events.
• Fixed parsing of more selectors in cssselect. Whitespace before pseudo-elements and pseudo-classes is significant as it is a descendant combinator. "E :pseudo" should parse the same as "E *:pseudo", not "E:pseudo". Patch by Simon Sapin.
• lxml.html.diff no longer raises an exception when hitting 'img' tags without 'src' attribute.

Other changes

Features added

• lxml.objectify.deannotate() has a new boolean option cleanup_namespaces to remove the objectify namespace declarations (and generally clean up the namespace declarations) after removing the type annotations.
• lxml.objectify gained its own SubElement() function as a copy of etree.SubElement to avoid an otherwise redundant import of lxml.etree on the user side.

Bugs fixed

• Fixed the "descendant" bug in cssselect a second time (after a first fix in lxml 2.3.1). The previous change resulted in a serious performance regression for the XPath based evaluation of the translated expression. Note that this breaks the usage of some of the generated XPath expressions as XSLT location paths that previously worked in 2.3.1.
• Fixed parsing of some selectors in cssselect. Whitespace after combinators ">", "+" and "~" is now correctly ignored. Previously it was parsed as a descendant combinator. For example, "div> .foo" was parsed the same as "div>* .foo" instead of "div>.foo". Patch by Simon Sapin.

Other changes

Features added

• New option kill_tags in lxml.html.clean to remove specific tags and their content (i.e. their whole subtree).
• pi.get() and pi.attrib on processing instructions to parse pseudo-attributes from the text content of processing instructions.
• lxml.get_include() returns a list of include paths that can be used to compile external C code against lxml.etree. This is specifically required for statically linked lxml builds when code needs to compile against the exact same header file versions as lxml itself.
• Resolver.resolve_file() takes an additional option close_file that configures if the file(-like) object will be closed after reading or not. By default, the file will be closed, as the user is not expected to keep a reference to it.

Bugs fixed

• HTML cleaning didn't remove 'data:' links.
• The html5lib parser integration now uses the 'official' implementation in html5lib itself, which makes it work with newer releases of the library.
• In lxml.sax, endElementNS() could incorrectly reject a plain tag name when the corresponding start event inferred the same plain tag name to be in the default namespace.
• When an open file-like object is passed into parse() or iterparse(), the parser will no longer close it after use. This reverts a change in lxml 2.3 where all files would be closed. It is the users responsibility to properly close the file(-like) object, also in error cases.
• Assertion error in lxml.html.cleaner when discarding top-level elements.
• In lxml.cssselect, use the xpath 'A//B' (short for 'A/descendant-or-self::node()/B') instead of 'A/descendant::B' for the css descendant selector ('A B'). This makes a few edge cases like "div *:last-child" consistent with the selector behavior in WebKit and Firefox, and makes more css expressions valid location paths (for use in xsl:template match).
• In lxml.html, non-selected <option> tags no longer show up in the collected form values.
• Adding/removing <option> values to/from a multiple select form field properly selects them and unselects them.

Other changes

• Static builds can specify the download directory with the --download-dir option.

Features added

• When looking for children, lxml.objectify takes '{}tag' as meaning an empty namespace, as opposed to the parent namespace.

Bugs fixed

• When finished reading from a file-like object, the parser immediately calls its .close() method.
• When finished parsing, iterparse() immediately closes the input file.
• Work-around for libxml2 bug that can leave the HTML parser in a non-functional state after parsing a severely broken document (fixed in libxml2 2.7.8).
• marque tag in HTML cleanup code is correctly named marquee.

Other changes

• Some public functions in the Cython-level C-API have more explicit return types.

Features added

Bugs fixed

• Crash in newer libxml2 versions when moving elements between documents that had attributes on replaced XInclude nodes.
• XMLID() function was missing the optional parser and base_url parameters.
• Searching for wildcard tags in iterparse() was broken in Py3.
• lxml.html.open_in_browser() didn't work in Python 3 due to the use of os.tempnam. It now takes an optional 'encoding' parameter.

Other changes

Features added

Bugs fixed

• Crash in XSLT when generating text-only result documents with a stylesheet created in a different thread.

Other changes

• repr() of Element objects shows the hex ID with leading 0x (following ElementTree 1.3).

Features added

• Keyword argument namespaces in lxml.cssselect.CSSSelector() to pass a prefix-to-namespace mapping for the selector.
• New function lxml.etree.register_namespace(prefix, uri) that globally registers a namespace prefix for a namespace that newly created Elements in that namespace will use automatically. Follows ElementTree 1.3.
• Support 'unicode' string name as encoding parameter in tostring(), following ElementTree 1.3.
• Support 'c14n' serialisation method in ElementTree.write() and tostring(), following ElementTree 1.3.
• The ElementPath expression syntax (el.find*()) was extended to match the upcoming ElementTree 1.3 that will ship in the standard library of Python 3.2/2.7. This includes extended support for predicates as well as namespace prefixes (as known from XPath).
• During regular XPath evaluation, various ESXLT functions are available within their namespace when using libxslt 1.1.26 or later.
• Support passing a readily configured logger instance into PyErrorLog, instead of a logger name.
• On serialisation, the new doctype parameter can be used to override the DOCTYPE (internal subset) of the document.
• New parameter output_parent to XSLTExtension.apply_templates() to append the resulting content directly to an output element.
• XSLTExtension.process_children() to process the content of the XSLT extension element itself.
• ISO-Schematron support based on the de-facto Schematron reference 'skeleton implementation'.
• XSLT objects now take XPath object as __call__ stylesheet parameters.
• Enable path caching in ElementPath (el.find*()) to avoid parsing overhead.
• Setting the value of a namespaced attribute always uses a prefixed namespace instead of the default namespace even if both declare the same namespace URI. This avoids serialisation problems when an attribute from a default namespace is set on an element from a different namespace.
• XSLT extension elements: support for XSLT context nodes other than elements: document root, comments, processing instructions.
• Support for strings (in addition to Elements) in node-sets returned by extension functions.
• Forms that lack an action attribute default to the base URL of the document on submit.
• XPath attribute result strings have an attrname property.
• Namespace URIs get validated against RFC 3986 at the API level (required by the XML namespace specification).
• Target parsers show their target object in the .target property (compatible with ElementTree).

Bugs fixed

• API is hardened against invalid proxy instances to prevent crashes due to incorrectly instantiated Element instances.
• Prevent crash when instantiating CommentBase and friends.
• Export ElementTree compatible XML parser class as XMLTreeBuilder, as it is called in ET 1.2.
• ObjectifiedDataElements in lxml.objectify were not hashable. They now use the hash value of the underlying Python value (string, number, etc.) to which they compare equal.
• Parsing broken fragments in lxml.html could fail if the fragment contained an orphaned closing '</div>' tag.
• Using XSLT extension elements around the root of the output document crashed.
• lxml.cssselect did not distinguish between x[attr="val"] and x [attr="val"] (with a space). The latter now matches the attribute independent of the element.
• Rewriting multiple links inside of HTML text content could end up replacing unrelated content as replacements could impact the reported position of subsequent matches. Modifications are now simplified by letting the iterlinks() generator in lxml.html return links in reversed order if they appear inside the same text node. Thus, replacements and link-internal modifications no longer change the position of links reported afterwards.
• The .value attribute of textarea elements in lxml.html did not represent the complete raw value (including child tags etc.). It now serialises the complete content on read and replaces the complete content by a string on write.
• Target parser didn't call .close() on the target object if parsing failed. Now it is guaranteed that .close() will be called after parsing, regardless of the outcome.

Other changes

• Official support for Python 3.1.2 and later.
• Static MS Windows builds can now download their dependencies themselves.
• Element.attrib no longer uses a cyclic reference back to its Element object. It therefore no longer requires the garbage collector to clean up.
• Static builds include libiconv, in addition to libxml2 and libxslt.

Bugs fixed

• Crash in newer libxml2 versions when moving elements between documents that had attributes on replaced XInclude nodes.
• Import fix for urljoin in Python 3.1+.

Bugs fixed

• Crash in XSLT when generating text-only result documents with a stylesheet created in a different thread.

Bugs fixed

• Fixed several Python 3 regressions by building with Cython 0.11.3.

Features added

• Support for running XSLT extension elements on the input root node (e.g. in a template matching on "/").

Bugs fixed

• Crash in XPath evaluation when reading smart strings from a document other than the original context document.
• Support recent versions of html5lib by not requiring its XHTMLParser in htmlparser.py anymore.
• Manually instantiating the custom element classes in lxml.objectify could crash.
• Invalid XML text characters were not rejected by the API when they appeared in unicode strings directly after non-ASCII characters.
• lxml.html.open_http_urllib() did not work in Python 3.
• The functions strip_tags() and strip_elements() in lxml.etree did not remove all occurrences of a tag in all cases.
• Crash in XSLT extension elements when the XSLT context node is not an element.

Bugs fixed

• Static build of libxml2/libxslt was broken.

Features added

Bugs fixed

• The resolve_entities option did not work in the incremental feed parser.
• Looking up and deleting attributes without a namespace could hit a namespaced attribute of the same name instead.
• Late errors during calls to SubElement() (e.g. attribute related ones) could leave a partially initialised element in the tree.
• Modifying trees that contain parsed entity references could result in an infinite loop.
• ObjectifiedElement.__setattr__ created an empty-string child element when the attribute value was rejected as a non-unicode/non-ascii string
• Syntax errors in lxml.cssselect could result in misleading error messages.
• Invalid syntax in CSS expressions could lead to an infinite loop in the parser of lxml.cssselect.
• CSS special character escapes were not properly handled in lxml.cssselect.
• CSS Unicode escapes were not properly decoded in lxml.cssselect.
• Select options in HTML forms that had no explicit value attribute were not handled correctly. The HTML standard dictates that their value is defined by their text content. This is now supported by lxml.html.
• XPath raised a TypeError when finding CDATA sections. This is now fully supported.
• Calling help(lxml.objectify) didn't work at the prompt.
• The ElementMaker in lxml.objectify no longer defines the default namespaces when annotation is disabled.
• Feed parser failed to honour the 'recover' option on parse errors.
• Diverting the error logging to Python's logging system was broken.

Other changes

Features added

• New helper functions strip_attributes(), strip_elements(), strip_tags() in lxml.etree to remove attributes/subtrees/tags from a subtree.

Bugs fixed

• Namespace cleanup on subtree insertions could result in missing namespace declarations (and potentially crashes) if the element defining a namespace was deleted and the namespace was not used by the top element of the inserted subtree but only in deeper subtrees.
• Raising an exception from a parser target callback didn't always terminate the parser.
• Only {true, false, 1, 0} are accepted as the lexical representation for BoolElement ({True, False, T, F, t, f} not any more), restoring lxml <= 2.0 behaviour.

Other changes

Features added

• Injecting default attributes into a document during XML Schema validation (also at parse time).
• Pass huge_tree parser option to disable parser security restrictions imposed by libxml2 2.7.

Bugs fixed

• The script for statically building libxml2 and libxslt didn't work in Py3.
• XMLSchema() also passes invalid schema documents on to libxml2 for parsing (which could lead to a crash before release 2.6.24).

Other changes

Features added

• Support for standalone flag in XML declaration through tree.docinfo.standalone and by passing standalone=True/False on serialisation.

Bugs fixed

• Crash when parsing an XML Schema with external imports from a filename.

Features added

• Support strings and instantiable Element classes as child arguments to the constructor of custom Element classes.
• GZip compression support for serialisation to files and file-like objects.

Bugs fixed

• Deep-copying an ElementTree copied neither its sibling PIs and comments nor its internal/external DTD subsets.
• Soupparser failed on broken attributes without values.
• Crash in XSLT when overwriting an already defined attribute using xsl:attribute.
• Crash bug in exception handling code under Python 3. This was due to a problem in Cython, not lxml itself.
• lxml.html.FormElement._name() failed for non top-level forms.
• TAG special attribute in constructor of custom Element classes was evaluated incorrectly.

Other changes

• Official support for Python 3.0.1.
• Element.findtext() now returns an empty string instead of None for Elements without text content.

Features added

• XSLT.strparam() class method to wrap quoted string parameters that require escaping.

Bugs fixed

• Memory leak in XPath evaluators.
• Crash when parsing indented XML in one thread and merging it with other documents parsed in another thread.
• Setting the base attribute in lxml.objectify from a unicode string failed.
• Fixes following changes in Python 3.0.1.
• Minor fixes for Python 3.

Other changes

• The global error log (which is copied into the exception log) is now local to a thread, which fixes some race conditions.
• More robust error handling on serialisation.

Bugs fixed

• Potential memory leak on exception handling. This was due to a problem in Cython, not lxml itself.
• iter_links (and related link-rewriting functions) in lxml.html would interpret CSS like url("link") incorrectly (treating the quotation marks as part of the link).
• Failing import on systems that have an io module.

Bugs fixed

• Potential memory leak on exception handling. This was due to a problem in Cython, not lxml itself.
• Failing import on systems that have an io module.

Features added

• Allow lxml.html.diff.htmldiff to accept Element objects, not just HTML strings.

Bugs fixed

• Crash when using an XPath evaluator in multiple threads.
• Fixed missing whitespace before Link:... in lxml.html.diff.

Other changes

• Export lxml.html.parse.

Bugs fixed

• Crash when using an XPath evaluator in multiple threads.

Bugs fixed

• Crash when using an XPath evaluator in multiple threads.

Features added

• Support for XSLT result tree fragments in XPath/XSLT extension functions.
• QName objects have new properties namespace and localname.
• New options for exclusive C14N and C14N without comments.
• Instantiating a custom Element classes creates a new Element.

Bugs fixed

• XSLT didn't inherit the parse options of the input document.
• 0-bytes could slip through the API when used inside of Unicode strings.
• With lxml.html.clean.autolink, links with balanced parenthesis, that end in a parenthesis, will be linked in their entirety (typical with Wikipedia links).

Other changes

Features added

Bugs fixed

• Ref-count leaks when lxml enters a try-except statement while an outside exception lives in sys.exc_*(). This was due to a problem in Cython, not lxml itself.
• Parser Unicode decoding errors could get swallowed by other exceptions.
• Name/import errors in some Python modules.
• Internal DTD subsets that did not specify a system or public ID were not serialised and did not appear in the docinfo property of ElementTrees.
• Fix a pre-Py3k warning when parsing from a gzip file in Py2.6.
• Test suite fixes for libxml2 2.7.
• Resolver.resolve_string() did not work for non-ASCII byte strings.
• Resolver.resolve_file() was broken.
• Overriding the parser encoding didn't work for many encodings.

Other changes

Bugs fixed

• Ref-count leaks when lxml enters a try-except statement while an outside exception lives in sys.exc_*(). This was due to a problem in Cython, not lxml itself.

Features added

• lxml.etree now tries to find the absolute path name of files when parsing from a file-like object. This helps custom resolvers when resolving relative URLs, as lixbml2 can prepend them with the path of the source document.

Bugs fixed

• Memory problem when passing documents between threads.
• Target parser did not honour the recover option and raised an exception instead of calling .close() on the target.

Other changes

Bugs fixed

• Memory problem when passing documents between threads.
• Target parser did not honour the recover option and raised an exception instead of calling .close() on the target.

Features added

Bugs fixed

• Crash when parsing XSLT stylesheets in a thread and using them in another.
• Encoding problem when including text with ElementInclude under Python 3.

Other changes

Features added

• lxml.html.rewrite_links() strips links to work around documents with whitespace in URL attributes.

Bugs fixed

• Crash when parsing XSLT stylesheets in a thread and using them in another.
• CSS selector parser dropped remaining expression after a function with parameters.

Other changes

Features added

• Smart strings can be switched off in XPath (smart_strings keyword option).
• lxml.html.rewrite_links() strips links to work around documents with whitespace in URL attributes.

Bugs fixed

• Custom resolvers were not used for XMLSchema includes/imports and XInclude processing.
• CSS selector parser dropped remaining expression after a function with parameters.

Other changes

• objectify.enableRecursiveStr() was removed, use objectify.enable_recursive_str() instead
• Speed-up when running XSLTs on documents from other threads

Features added

• Pickling ElementTree objects in lxml.objectify.

Bugs fixed

• Descending dot-separated classes in CSS selectors were not resolved correctly.
• ElementTree.parse() didn't handle target parser result.
• Potential threading problem in XInclude.
• Crash in Element class lookup classes when the __init__() method of the super class is not called from Python subclasses.

Other changes

• Non-ASCII characters in attribute values are no longer escaped on serialisation.

Features added

• Major overhaul of tools/xpathgrep.py script.
• Pickling ElementTree objects in lxml.objectify.
• Support for parsing from file-like objects that return unicode strings.
• New function etree.cleanup_namespaces(el) that removes unused namespace declarations from a (sub)tree (experimental).
• XSLT results support the buffer protocol in Python 3.
• Polymorphic functions in lxml.html that accept either a tree or a parsable string will return either a UTF-8 encoded byte string, a unicode string or a tree, based on the type of the input. Previously, the result was always a byte string or a tree.
• Support for Python 2.6 and 3.0 beta.
• File name handling now uses a heuristic to convert between byte strings (usually filenames) and unicode strings (usually URLs).
• Parsing from a plain file object frees the GIL under Python 2.x.
• Running iterparse() on a plain file (or filename) frees the GIL on reading under Python 2.x.
• Conversion functions html_to_xhtml() and xhtml_to_html() in lxml.html (experimental).
• Most features in lxml.html work for XHTML namespaced tag names (experimental).

Bugs fixed

• ElementTree.parse() didn't handle target parser result.
• Crash in Element class lookup classes when the __init__() method of the super class is not called from Python subclasses.
• A number of problems related to unicode/byte string conversion of filenames and error messages were fixed.
• Building on MacOS-X now passes the "flat_namespace" option to the C compiler, which reportedly prevents build quirks and crashes on this platform.
• Windows build was broken.
• Rare crash when serialising to a file object with certain encodings.

Other changes

• Non-ASCII characters in attribute values are no longer escaped on serialisation.
• Passing non-ASCII byte strings or invalid unicode strings as .tag, namespaces, etc. will result in a ValueError instead of an AssertionError (just like the tag well-formedness check).
• Up to several times faster attribute access (i.e. tree traversal) in lxml.objectify.

Features added

Bugs fixed

• Incorrect evaluation of el.find("tag[child]").
• Windows build was broken.
• Moving a subtree from a document created in one thread into a document of another thread could crash when the rest of the source document is deleted while the subtree is still in use.
• Rare crash when serialising to a file object with certain encodings.

Other changes

• lxml should now build without problems on MacOS-X.

Features added

• All parse functions in lxml.html take a parser keyword argument.
• lxml.html has a new parser class XHTMLParser and a module attribute xhtml_parser that provide XML parsers that are pre-configured for the lxml.html package.

Bugs fixed

• Moving a subtree from a document created in one thread into a document of another thread could crash when the rest of the source document is deleted while the subtree is still in use.
• Passing an nsmap when creating an Element will no longer strip redundantly defined namespace URIs. This prevented the definition of more than one prefix for a namespace on the same Element.

Other changes

• If the default namespace is redundantly defined with a prefix on the same Element, the prefix will now be preferred for subelements and attributes. This allows users to work around a problem in libxml2 where attributes from the default namespace could serialise without a prefix even when they appear on an Element with a different namespace (i.e. they would end up in the wrong namespace).

Features added

Bugs fixed

• Resolving to a filename in custom resolvers didn't work.
• lxml did not honour libxslt's second error state "STOPPED", which let some XSLT errors pass silently.
• Memory leak in Schematron with libxml2 >= 2.6.31.

Other changes

Features added

• Error logging in Schematron (requires libxml2 2.6.32 or later).
• Parser option strip_cdata for normalising or keeping CDATA sections. Defaults to True as before, thus replacing CDATA sections by their text content.
• CDATA() factory to wrap string content as CDATA section.

Bugs fixed

• Resolving to a filename in custom resolvers didn't work.
• lxml did not honour libxslt's second error state "STOPPED", which let some XSLT errors pass silently.
• Memory leak in Schematron with libxml2 >= 2.6.31.
• lxml.etree accepted non well-formed namespace prefix names.

Other changes

• Major cleanup in internal moveNodeToDocument() function, which takes care of namespace cleanup when moving elements between different namespace contexts.
• New Elements created through the makeelement() method of an HTML parser or through lxml.html now end up in a new HTML document (doctype HTML 4.01 Transitional) instead of a generic XML document. This mostly impacts the serialisation and the availability of a DTD context.

Features added

Bugs fixed

• Hanging thread in conjunction with GTK threading.
• Crash bug in iterparse when moving elements into other documents.
• HTML elements' .cssselect() method was broken.
• ElementTree.find*() didn't accept QName objects.

Other changes

Features added

• New event types 'comment' and 'pi' in iterparse().
• XSLTAccessControl instances have a property options that returns a dict of access configuration options.
• Constant instances DENY_ALL and DENY_WRITE on XSLTAccessControl class.
• Extension elements for XSLT (experimental!)
• Element.base property returns the xml:base or HTML base URL of an Element.
• docinfo.URL property is writable.

Bugs fixed

• Default encoding for plain text serialisation was different from that of XML serialisation (UTF-8 instead of ASCII).

Other changes

• Minor API speed-ups.
• The benchmark suite now uses tail text in the trees, which makes the absolute numbers incomparable to previous results.
• Generating the HTML documentation now requires Pygments, which is used to enable syntax highlighting for the doctest examples.

Most long-time deprecated functions and methods were removed:

• etree.clearErrorLog(), use etree.clear_error_log()

• etree.useGlobalPythonLog(), use etree.use_global_python_log()

• etree.ElementClassLookup.setFallback(), use etree.ElementClassLookup.set_fallback()

• etree.getDefaultParser(), use etree.get_default_parser()

• etree.setDefaultParser(), use etree.set_default_parser()

• etree.setElementClassLookup(), use etree.set_element_class_lookup()

  Note that parser.setElementClassLookup() has not been removed yet, although parser.set_element_class_lookup() should be used instead.

• xpath_evaluator.registerNamespace(), use xpath_evaluator.register_namespace()

• xpath_evaluator.registerNamespaces(), use xpath_evaluator.register_namespaces()

• objectify.setPytypeAttributeTag, use objectify.set_pytype_attribute_tag

• objectify.setDefaultParser(), use objectify.set_default_parser()

Features added

• soupparser.parse() allows passing keyword arguments on to BeautifulSoup.
• fromstring() method in lxml.html.soupparser.

Bugs fixed

• lxml.html.diff didn't treat empty tags properly (e.g., <br>).
• Handle entity replacements correctly in target parser.
• Crash when using iterparse() with XML Schema validation.
• The BeautifulSoup parser (soupparser.py) did not replace entities, which made them turn up in text content.
• Attribute assignment of custom PyTypes in objectify could fail to correctly serialise the value to a string.

Other changes

• lxml.html.ElementSoup was replaced by a new module lxml.html.soupparser with a more consistent API. The old module remains for compatibility with ElementTree's own ElementSoup module.
• Setting the XSLT_CONFIG and XML2_CONFIG environment variables at build time will let setup.py pick up the xml2-config and xslt-config scripts from the supplied path name.
• Passing --with-xml2-config=/path/to/xml2-config to setup.py will override the xml2-config script that is used to determine the C compiler options. The same applies for the --with-xslt-config option.

Features added

• Support passing base_url to file parser functions to override the filename of the file(-like) object.

Bugs fixed

• The prefix for objectify's pytype namespace was missing from the set of default prefixes.
• Memory leak in Schematron (fixed only for libxml2 2.6.31+).
• Error type names in RelaxNG were reported incorrectly.
• Slice deletion bug fixed in objectify.

Other changes

• Enabled doctests for some Python modules (especially lxml.html).
• Add a method argument to lxml.html.tostring() (method="xml" for XHTML output).
• Make it clearer that methods like lxml.html.fromstring() take a base_url argument.

Features added

• Child iteration in lxml.pyclasslookup.
• Loads of new docstrings reflect the signature of functions and methods to make them visible in API docs and help()

Bugs fixed

• The module lxml.html.builder was duplicated as lxml.htmlbuilder
• Form elements would return None for form.fields.keys() if there was an unnamed input field. Now unnamed input fields are completely ignored.
• Setting an element slice in objectify could insert slice-overlapping elements at the wrong position.

Other changes

• The generated API documentation was cleaned up and disburdened from non-public classes etc.
• The previously public module lxml.html.setmixin was renamed to lxml.html._setmixin as it is not an official part of lxml. If you want to use it, feel free to copy it over to your own source base.
• Passing --with-xslt-config=/path/to/xslt-config to setup.py will override the xslt-config script that is used to determine the C compiler options.

Features added

• Passing the unicode type as encoding to tostring() will serialise to unicode. The tounicode() function is now deprecated.
• XMLSchema() and RelaxNG() can parse from StringIO.
• makeparser() function in lxml.objectify to create a new parser with the usual objectify setup.
• Plain ASCII XPath string results are no longer forced into unicode objects as in 2.0beta1, but are returned as plain strings as before.
• All XPath string results are 'smart' objects that have a getparent() method to retrieve their parent Element.
• with_tail option in serialiser functions.
• More accurate exception messages in validator creation.
• Parse-time XML schema validation (schema parser keyword).
• XPath string results of the text() function and attribute selection make their Element container accessible through a getparent() method. As a side-effect, they are now always unicode objects (even ASCII strings).
• XSLT objects are usable in any thread - at the cost of a deep copy if they were not created in that thread.
• Invalid entity names and character references will be rejected by the Entity() factory.
• entity.text returns the textual representation of the entity, e.g. &amp;.
• New properties position and code on ParseError exception (as in ET 1.3)
• Rich comparison of element.attrib proxies.
• ElementTree compatible TreeBuilder class.
• Use default prefixes for some common XML namespaces.
• lxml.html.clean.Cleaner now allows for a host_whitelist, and two overridable methods: allow_embedded_url(el, url) and the more general allow_element(el).
• Extended slicing of Elements as in element[1:-1:2], both in etree and in objectify
• Resolvers can now provide a base_url keyword argument when resolving a document as string data.
• When using lxml.doctestcompare you can give the doctest option NOPARSE_MARKUP (like # doctest: +NOPARSE_MARKUP) to suppress the special checking for one test.
• Separate feed_error_log property for the feed parser interface. The normal parser interface and iterparse continue to use error_log.
• The normal parsers and the feed parser interface are now separated and can be used concurrently on the same parser instance.
• fromstringlist() and tostringlist() functions as in ElementTree 1.3
• iterparse() accepts an html boolean keyword argument for parsing with the HTML parser (note that this interface may be subject to change)
• Parsers accept an encoding keyword argument that overrides the encoding of the parsed documents.
• New C-API function hasChild() to test for children
• annotate() function in objectify can annotate with Python types and XSI types in one step. Accompanied by xsiannotate() and pyannotate().
• ET.write(), tostring() and tounicode() now accept a keyword argument method that can be one of 'xml' (or None), 'html' or 'text' to serialise as XML, HTML or plain text content.
• iterfind() method on Elements returns an iterator equivalent to findall()
• itertext() method on Elements
• Setting a QName object as value of the .text property or as an attribute will resolve its prefix in the respective context
• ElementTree-like parser target interface as described in http://effbot.org/elementtree/elementtree-xmlparser.htm
• ElementTree-like feed parser interface on XMLParser and HTMLParser (feed() and close() methods)
• Reimplemented objectify.E for better performance and improved integration with objectify. Provides extended type support based on registered PyTypes.
• XSLT objects now support deep copying
• New makeSubElement() C-API function that allows creating a new subelement straight with text, tail and attributes.
• XPath extension functions can now access the current context node (context.context_node) and use a context dictionary (context.eval_context) from the context provided in their first parameter
• HTML tag soup parser based on BeautifulSoup in lxml.html.ElementSoup
• New module lxml.doctestcompare by Ian Bicking for writing simplified doctests based on XML/HTML output. Use by importing lxml.usedoctest or lxml.html.usedoctest from within a doctest.
• New module lxml.cssselect by Ian Bicking for selecting Elements with CSS selectors.
• New package lxml.html written by Ian Bicking for advanced HTML treatment.
• Namespace class setup is now local to the ElementNamespaceClassLookup instance and no longer global.
• Schematron validation (incomplete in libxml2)
• Additional stringify argument to objectify.PyType() takes a conversion function to strings to support setting text values from arbitrary types.
• Entity support through an Entity factory and element classes. XML parsers now have a resolve_entities keyword argument that can be set to False to keep entities in the document.
• column field on error log entries to accompany the line field
• Error specific messages in XPath parsing and evaluation NOTE: for evaluation errors, you will now get an XPathEvalError instead of an XPathSyntaxError. To catch both, you can except on XPathError
• The regular expression functions in XPath now support passing a node-set instead of a string
• Extended type annotation in objectify: new xsiannotate() function
• EXSLT RegExp support in standard XPath (not only XSLT)

Bugs fixed

• Missing import in lxml.html.clean.
• Some Python 2.4-isms prevented lxml from building/running under Python 2.3.
• XPath on ElementTrees could crash when selecting the virtual root node of the ElementTree.
• Compilation --without-threading was buggy in alpha5/6.
• Memory leak in the parse() function.
• Minor bugs in XSLT error message formatting.
• Result document memory leak in target parser.
• Target parser failed to report comments.
• In the lxml.html iter_links method, links in <object> tags weren't recognized. (Note: plugin-specific link parameters still aren't recognized.) Also, the <embed> tag, though not standard, is now included in lxml.html.defs.special_inline_tags.
• Using custom resolvers on XSLT stylesheets parsed from a string could request ill-formed URLs.
• With lxml.doctestcompare if you do <tag xmlns="..."> in your output, it will then be namespace-neutral (before the ellipsis was treated as a real namespace).
• AttributeError in feed parser on parse errors
• XML feed parser setup problem
• Type annotation for unicode strings in DataElement()
• lxml failed to serialise namespace declarations of elements other than the root node of a tree
• Race condition in XSLT where the resolver context leaked between concurrent XSLT calls
• lxml.etree did not check tag/attribute names
• The XML parser did not report undefined entities as error
• The text in exceptions raised by XML parsers, validators and XPath evaluators now reports the first error that occurred instead of the last
• Passing '' as XPath namespace prefix did not raise an error
• Thread safety in XPath evaluators

Other changes

• Exceptions carry only the part of the error log that is related to the operation that caused the error.
• XMLSchema() and RelaxNG() now enforce passing the source file/filename through the file keyword argument.
• The test suite now skips most doctests under Python 2.3.
• make clean no longer removes the .c files (use make realclean instead)
• Minor performance tweaks for Element instantiation and subelement creation
• Various places in the XPath, XSLT and iteration APIs now require keyword-only arguments.
• The argument order in element.itersiblings() was changed to match the order used in all other iteration methods. The second argument ('preceding') is now a keyword-only argument.
• The getiterator() method on Elements and ElementTrees was reverted to return an iterator as it did in lxml 1.x. The ET API specification allows it to return either a sequence or an iterator, and it traditionally returned a sequence in ET and an iterator in lxml. However, it is now deprecated in favour of the iter() method, which should be used in new code wherever possible.
• The 'pretty printed' serialisation of ElementTree objects now inserts newlines at the root level between processing instructions, comments and the root tag.
• A 'pretty printed' serialisation is now terminated with a newline.
• Second argument to lxml.etree.Extension() helper is no longer required, third argument is now a keyword-only argument ns.
• lxml.html.tostring takes an encoding argument.
• The module source files were renamed to "lxml.*.pyx", such as "lxml.etree.pyx". This was changed for consistency with the way Pyrex commonly handles package imports. The main effect is that classes now know about their fully qualified class name, including the package name of their module.
• Keyword-only arguments in some API functions, especially in the parsers and serialisers.
• Tag name validation in lxml.etree (and lxml.html) now distinguishes between HTML tags and XML tags based on the parser that was used to parse or create them. HTML tags no longer reject any non-ASCII characters in tag names but only spaces and the special characters <>&/"'.
• lxml.etree now emits a warning if you use XPath with libxml2 2.6.27 (which can crash on certain XPath errors)
• Type annotation in objectify now preserves the already annotated type by default to prevent losing type information that is already there.
• element.getiterator() returns a list, use element.iter() to retrieve an iterator (ElementTree 1.3 compatible behaviour)
• objectify.PyType for None is now called "NoneType"
• el.getiterator() renamed to el.iter(), following ElementTree 1.3 - original name is still available as alias
• In the public C-API, findOrBuildNodeNs() was replaced by the more generic findOrBuildNodeNsPrefix
• Major refactoring in XPath/XSLT extension function code
• Network access in parsers disabled by default

Bugs fixed

• Backported decref crash fix from 2.0
• Well hidden free-while-in-use crash bug in ObjectPath

Other changes

• The test suites now run gc.collect() in the tearDown() methods. While this makes them take a lot longer to run, it also makes it easier to link a specific test to garbage collection problems that would otherwise appear in later tests.

Features added

Bugs fixed

• lxml.etree could crash when adding more than 10000 namespaces to a document
• lxml failed to serialise namespace declarations of elements other than the root node of a tree

Features added

• The ElementMaker in lxml.builder now accepts the keyword arguments namespace and nsmap to set a namespace and nsmap for the Elements it creates.
• The docinfo on ElementTree objects has new properties internalDTD and externalDTD that return a DTD object for the internal or external subset of the document respectively.
• Serialising an ElementTree now includes any internal DTD subsets that are part of the document, as well as comments and PIs that are siblings of the root node.

Bugs fixed

• Parsing with the no_network option could fail

Other changes

• lxml now raises a TagNameWarning about tag names containing ':' instead of an Error as 1.3.3 did. The reason is that a number of projects currently misuse the previous lack of tag name validation to generate namespace prefixes without declaring namespaces. Apart from the danger of generating broken XML this way, it also breaks most of the namespace-aware tools in XML, including XPath, XSLT and validation. lxml 1.3.x will continue to support this bug with a Warning, while lxml 2.0 will be strict about well-formed tag names (not only regarding ':').
• Serialising an Element no longer includes its comment and PI siblings (only ElementTree serialisation includes them).

Features added

• ElementTree compatible parser ETCompatXMLParser strips processing instructions and comments while parsing XML
• Parsers now support stripping PIs (keyword argument 'remove_pis')
• etree.fromstring() now supports parsing both HTML and XML, depending on the parser you pass.
• Support base_url keyword argument in HTML() and XML()

Bugs fixed

• Parsing from Python Unicode strings failed on some platforms
• Element() did not raise an exception on tag names containing ':'
• Element.getiterator(tag) did not accept Comment and ProcessingInstruction as tags. It also accepts Element now.

Features added

Bugs fixed

• "deallocating None" crash bug

Features added

• objectify.DataElement now supports setting values from existing data elements (not just plain Python types) and reuses defined namespaces etc.
• E-factory support for lxml.objectify (objectify.E)

Bugs fixed

• Better way to prevent crashes in Element proxy cleanup code
• objectify.DataElement didn't set up None value correctly
• objectify.DataElement didn't check the value against the provided type hints
• Reference-counting bug in Element.attrib.pop()

Features added

• Module lxml.pyclasslookup module implements an Element class lookup scheme that can access the entire tree in read-only mode to help determining a suitable Element class
• Parsers take a remove_comments keyword argument that skips over comments
• parse() function in objectify, corresponding to XML() etc.
• Element.addnext(el) and Element.addprevious(el) methods to support adding processing instructions and comments around the root node
• Element.attrib was missing clear() and pop() methods
• Extended type annotation in objectify: cleaner annotation namespace setup plus new deannotate() function
• Support for custom Element class instantiation in lxml.sax: passing a makeelement function to the ElementTreeContentHandler will reuse the lookup context of that function
• '.' represents empty ObjectPath (identity)
• Element.values() to accompany the existing .keys() and .items()
• collectAttributes() C-function to build a list of attribute keys/values/items for a libxml2 node
• DTD validator class (like RelaxNG and XMLSchema)
• HTML generator helpers by Fredrik Lundh in lxml.htmlbuilder
• ElementMaker XML generator by Fredrik Lundh in lxml.builder.E
• Support for pickling objectify.ObjectifiedElement objects to XML
• update() method on Element.attrib
• Optimised replacement for libxml2's _xmlReconsiliateNs(). This allows lxml a better handling of namespaces when moving elements between documents.

Bugs fixed

• Removing Elements from a tree could make them lose their namespace declarations
• ElementInclude didn't honour base URL of original document
• Replacing the children slice of an Element would cut off the tails of the original children
• Element.getiterator(tag) did not accept Comment and ProcessingInstruction as tags
• API functions now check incoming strings for XML conformity. Zero bytes or low ASCII characters are no longer accepted (AssertionError).
• XSLT parsing failed to pass resolver context on to imported documents
• passing '' as namespace prefix in nsmap could be passed through to libxml2
• Objectify couldn't handle prefixed XSD type names in xsi:type
• More ET compatible behaviour when writing out XML declarations or not
• More robust error handling in iterparse()
• Documents lost their top-level PIs and comments on serialisation
• lxml.sax failed on comments and PIs. Comments are now properly ignored and PIs are copied.
• Possible memory leaks in namespace handling when moving elements between documents

Other changes

• major restructuring in the documentation

Bugs fixed

• Build fixes for MS compiler
• Item assignments to special names like element["text"] failed
• Renamed ObjectifiedDataElement.__setText() to _setText() to make it easier to access
• The pattern for attribute names in ObjectPath was too restrictive

Features added

• Rich comparison of QName objects
• Support for regular expressions in benchmark selection
• get/set emulation (not .attrib!) for attributes on processing instructions
• ElementInclude Python module for ElementTree compatible XInclude processing that honours custom resolvers registered with the source document
• ElementTree.parser property holds the parser used to parse the document
• setup.py has been refactored for greater readability and flexibility
• --rpath flag to setup.py to induce automatic linking-in of dynamic library runtime search paths has been renamed to --auto-rpath. This makes it possible to pass an --rpath directly to distutils; previously this was being shadowed.

Bugs fixed

• Element instantiation now uses locks to prevent race conditions with threads
• ElementTree.write() did not raise an exception when the file was not writable
• Error handling could crash under Python <= 2.4.1 - fixed by disabling thread support in these environments
• Element.find*() did not accept QName objects as path

Other changes

• code cleanup: redundant _NodeBase super class merged into _Element class Note: although the impact should be zero in most cases, this change breaks the compatibility of the public C-API

Features added

• Data elements in objectify support repr(), which is now used by dump()
• Source distribution now ships with a patched Pyrex
• New C-API function makeElement() to create new elements with text, tail, attributes and namespaces
• Reuse original parser flags for XInclude
• Simplified support for handling XSLT processing instructions

Bugs fixed

• Parser resources were not freed before the next parser run
• Open files and XML strings returned by Python resolvers were not closed/freed
• Crash in the IDDict returned by XMLDTDID
• Copying Comments and ProcessingInstructions failed
• Memory leak for external URLs in _XSLTProcessingInstruction.parseXSL()
• Memory leak when garbage collecting tailed root elements
• HTML script/style content was not propagated to .text
• Show text xincluded between text nodes correctly in .text and .tail
• 'integer * objectify.StringElement' operation was not supported

Features added

• XSLT profiling support (profile_run keyword)
• countchildren() method on objectify.ObjectifiedElement
• Support custom elements for tree nodes in lxml.objectify

Bugs fixed

• lxml.objectify failed to support long data values (e.g., "123L")
• Error messages from XSLT did not reach XSLT.error_log
• Factories objectify.Element() and objectify.DataElement() were missing attrib and nsmap keyword arguments
• Changing the default parser in lxml.objectify did not update the factories Element() and DataElement()
• Let lxml.objectify.Element() always generate tree elements (not data elements)
• Build under Windows failed ('0' bug in patched Pyrex version)

Features added

• Comments and processing instructions return '<!-- comment -->' and '<?pi-target content?>' for repr()
• Parsers are now the preferred (and default) place where element class lookup schemes should be registered. Namespace lookup is no longer supported by default.
• Support for Python 2.5 beta
• Unlock the GIL for deep copying documents and for XPath()
• New compact keyword argument for parsing read-only documents
• Support for parser options in iterparse()
• The namespace axis is supported in XPath and returns (prefix, URI) tuples
• The XPath expression "/" now returns an empty list instead of raising an exception
• XML-Object API on top of lxml (lxml.objectify)
• Customizable Element class lookup:
  • different pre-implemented lookup mechanisms
  • support for externally provided lookup functions
• Support for processing instructions (ET-like, not compatible)
• Public C-level API for independent extension modules
• Module level iterwalk() function as 'iterparse' for trees
• Module level iterparse() function similar to ElementTree (see documentation for differences)
• Element.nsmap property returns a mapping of all namespace prefixes known at the Element to their namespace URI
• Reentrant threading support in RelaxNG, XMLSchema and XSLT
• Threading support in parsers and serializers:
  • All in-memory operations (tostring, parse(StringIO), etc.) free the GIL
  • File operations (on file names) free the GIL
  • Reading from file-like objects frees the GIL and reacquires it for reading
  • Serialisation to file-like objects is single-threaded (high lock overhead)
• Element iteration over XPath axes:
  • Element.iterdescendants() iterates over the descendants of an element
  • Element.iterancestors() iterates over the ancestors of an element (from parent to parent)
  • Element.itersiblings() iterates over either the following or preceding siblings of an element
  • Element.iterchildren() iterates over the children of an element in either direction
  • All iterators support the tag keyword argument to restrict the generated elements
• Element.getnext() and Element.getprevious() return the direct siblings of an element

Bugs fixed

• filenames with local 8-bit encoding were not supported
• 1.1beta did not compile under Python 2.3
• ignore unknown 'pyval' attribute values in objectify
• objectify.ObjectifiedElement.addattr() failed to accept Elements and Lists
• objectify.ObjectPath.setattr() failed to accept Elements and Lists
• XPathSyntaxError now inherits from XPathError
• Threading race conditions in RelaxNG and XMLSchema
• Crash when mixing elements from XSLT results into other trees, concurrent XSLT is only allowed when the stylesheet was parsed in the main thread
• The EXSLT regexp:match function now works as defined (except for some differences in the regular expression syntax)
• Setting element.text to '' returned None on request, not the empty string
• iterparse() could crash on long XML files
• Creating documents no longer copies the parser for later URL resolving. For performance reasons, only a reference is kept. Resolver updates on the parser will now be reflected by documents that were parsed before the change. Although this should rarely become visible, it is a behavioral change from 1.0.

Features added

• List-like Element.extend() method

Bugs fixed

• Crash in tail handling in Element.replace()

Features added

• Element.replace(old, new) method to replace a subelement by another one

Bugs fixed

• Crash when mixing elements from XSLT results into other trees
• Copying/deepcopying did not work for ElementTree objects
• Setting an attribute to a non-string value did not raise an exception
• Element.remove() deleted the tail text from the removed Element

Features added

• Support for setting a custom default Element class as opposed to namespace specific classes (which still override the default class)

Bugs fixed

• Rare exceptions in Python list functions were not handled
• Parsing accepted unicode strings with XML encoding declaration in certain cases
• Parsing 8-bit encoded strings from StringIO objects raised an exception
• Module function initThread() was removed - useless (and never worked)
• XSLT and parser exception messages include the error line number

Features added

• Repeated calls to Element.attrib now efficiently return the same instance

Bugs fixed

• Document deallocation could crash in certain garbage collection scenarios
• Extension function calls in XSLT variable declarations could break the stylesheet and crash on repeated calls
• Deep copying Elements could lose namespaces declared in parents
• Deep copying Elements did not copy tail
• Parsing file(-like) objects failed to load external entities
• Parsing 8-bit strings from file(-like) objects raised an exception
• xsl:include failed when the stylesheet was parsed from a file-like object
• lxml.sax.ElementTreeProducer did not call startDocument() / endDocument()
• MSVC compiler complained about long strings (supports only 2048 bytes)

Features added

• Element.getiterator() and the findall() methods support finding arbitrary elements from a namespace (pattern {namespace}*)
• Another speedup in tree iteration code
• General speedup of Python Element object creation and deallocation
• Writing C14N no longer serializes in memory (reduced memory footprint)
• PyErrorLog for error logging through the Python logging module
• Element.getroottree() returns an ElementTree for the root node of the document that contains the element.
• ElementTree.getpath(element) returns a simple, absolute XPath expression to find the element in the tree structure
• Error logs have a last_error attribute for convenience
• Comment texts can be changed through the API
• Formatted output via pretty_print keyword in serialization functions
• XSLT can block access to file system and network via XSLTAccessControl
• ElementTree.write() no longer serializes in memory (reduced memory footprint)
• Speedup of Element.findall(tag) and Element.getiterator(tag)
• Support for writing the XML representation of Elements and ElementTrees to Python unicode strings via etree.tounicode()
• Support for writing XSLT results to Python unicode strings via unicode()
• Parsing a unicode string no longer copies the string (reduced memory footprint)
• Parsing file-like objects reads chunks rather than the whole file (reduced memory footprint)
• Parsing StringIO objects from the start avoids copying the string (reduced memory footprint)
• Read-only 'docinfo' attribute in ElementTree class holds DOCTYPE information, original encoding and XML version as seen by the parser
• etree module can be compiled without libxslt by commenting out the line include "xslt.pxi" near the end of the etree.pyx source file
• Better error messages in parser exceptions
• Error reporting also works in XSLT
• Support for custom document loaders (URI resolvers) in parsers and XSLT, resolvers are registered at parser level
• Implementation of exslt:regexp for XSLT based on the Python 're' module, enabled by default, can be switched off with 'regexp=False' keyword argument
• Support for exslt extensions (libexslt) and libxslt extra functions (node-set, document, write, output)
• Substantial speedup in XPath.evaluate()
• HTMLParser for parsing (broken) HTML
• XMLDTDID function parses XML into tuple (root node, ID dict) based on xml:id implementation of libxml2 (as opposed to ET compatible XMLID)

Bugs fixed

• Memory leak in Element.__setitem__
• Memory leak in Element.attrib.items() and Element.attrib.values()
• Memory leak in XPath extension functions
• Memory leak in unicode related setup code
• Element now raises ValueError on empty tag names
• Namespace fixing after moving elements between documents could fail if the source document was freed too early
• Setting namespace-less tag names on namespaced elements ('{ns}t' -> 't') didn't reset the namespace
• Unknown constants from newer libxml2 versions could raise exceptions in the error handlers
• lxml.etree compiles much faster
• On libxml2 <= 2.6.22, parsing strings with encoding declaration could fail in certain cases
• Document reference in ElementTree objects was not updated when the root element was moved to a different document
• Running absolute XPath expressions on an Element now evaluates against the root tree
• Evaluating absolute XPath expressions (/*) on an ElementTree could fail
• Crashes when calling XSLT, RelaxNG, etc. with uninitialized ElementTree objects
• Removed public function initThreadLogging(), replaced by more general initThread() which fixes a number of setup problems in threads
• Memory leak when using iconv encoders in tostring/write
• Deep copying Elements and ElementTrees maintains the document information
• Serialization functions raise LookupError for unknown encodings
• Memory deallocation crash resulting from deep copying elements
• Some ElementTree methods could crash if the root node was not initialized (neither file nor element passed to the constructor)
• Element/SubElement failed to set attribute namespaces from passed attrib dictionary
• tostring() adds an XML declaration for non-ASCII encodings
• tostring() failed to serialize encodings that contain 0-bytes
• ElementTree.xpath() and XPathDocumentEvaluator were not using the ElementTree root node as reference point
• Calling document('') in XSLT failed to return the stylesheet

Features added

• Speedup for Element.makeelement(): the new element reuses the original libxml2 document instead of creating a new empty one
• Speedup for reversed() iteration over element children (Py2.4+ only)
• ElementTree compatible QName class
• RelaxNG and XMLSchema accept any Element, not only ElementTrees

Bugs fixed

• str(xslt_result) was broken for XSLT output other than UTF-8
• Memory leak if write_c14n fails to write the file after conversion
• Crash in XMLSchema and RelaxNG when passing non-schema documents
• Memory leak in RelaxNG() when RelaxNGParseError is raised

Features added

• lxml.sax.ElementTreeContentHandler checks closing elements and raises SaxError on mismatch
• lxml.sax.ElementTreeContentHandler supports namespace-less SAX events (startElement, endElement) and defaults to empty attributes (keyword argument)
• Speedup for repeatedly accessing element tag names
• Minor API performance improvements

Bugs fixed

• Memory deallocation bug when using XSLT output method "html"
• sax.py was handling UTF-8 encoded tag names where it shouldn't
• lxml.tests package will no longer be installed (is still in source tar)

Features added

• Error logging API for libxml2 error messages
• Various performance improvements
• Benchmark script for lxml, ElementTree and cElementTree
• Support for registering extension functions through new FunctionNamespace class (see doc/extensions.txt)
• ETXPath class for XPath expressions in ElementTree notation ('//{ns}tag')
• Support for variables in XPath expressions (also in XPath class)
• XPath class for compiled XPath expressions
• XMLID module level function (ElementTree compatible)
• XMLParser API for customized libxml2 parser configuration
• Support for custom Element classes through new Namespace API (see doc/namespace_extensions.txt)
• Common exception base class LxmlError for module exceptions
• real iterator support in iter(Element), Element.getiterator()
• XSLT objects are callable, result trees support str()
• Added MANIFEST.in for easier creation of RPM files.
• 'getparent' method on elements allows navigation to an element's parent element.
• Python core compatible SAX tree builder and SAX event generator. See doc/sax.txt for more information.

Bugs fixed

• Segfaults and memory leaks in various API functions of Element
• Segfault in XSLT.tostring()
• ElementTree objects no longer interfere, Elements can be root of different ElementTrees at the same time
• document('') works in XSLT documents read from files (in-memory documents cannot support this due to libxslt deficiencies)

Features added

• Support for copy.deepcopy() on elements. copy.copy() works also, but does the same thing, and does not create a shallow copy, as that makes no sense in the context of libxml2 trees. This means a potential incompatibility with ElementTree, but there's more chance that it works than if copy.copy() isn't supported at all.
• Increased compatibility with (c)ElementTree; .parse() on ElementTree is supported and parsing of gzipped XML files works.
• implemented index() on elements, allowing one to find the index of a SubElement.

Bugs fixed

• Use xslt-config instead of xml2-config to find out libxml2 directories to take into account a case where libxslt is installed in a different directory than libxslt.
• Eliminate crash condition in iteration when text nodes are changed.
• Passing 'None' to tostring() does not result in a segfault anymore, but an AssertionError.
• Some test fixes for Windows.
• Raise XMLSyntaxError and XPathSyntaxError instead of plain python syntax errors. This should be less confusing.
• Fixed error with uncaught exception in Pyrex code.
• Calling lxml.etree.fromstring('') throws XMLSyntaxError instead of a segfault.
• has_key() works on attrib. 'in' tests also work correctly on attrib.
• INSTALL.txt was saying 2.2.16 instead of 2.6.16 as a supported libxml2 version, as it should.
• Passing a UTF-8 encoded string to the XML() function would fail; fixed.

Features added

• parameters (XPath expressions) can be passed to XSLT using keyword parameters.
• Simple XInclude support. Calling the xinclude() method on a tree will process any XInclude statements in the document.
• XMLSchema support. Use the XMLSchema class or the convenience xmlschema() method on a tree to do XML Schema (XSD) validation.
• Added convenience xslt() method on tree. This is less efficient than the XSLT object, but makes it easier to write quick code.
• Added convenience relaxng() method on tree. This is less efficient than the RelaxNG object, but makes it easier to write quick code.
• Make it possible to use XPathEvaluator with elements as well. The XPathEvaluator in this case will retain the element so multiple XPath queries can be made against one element efficiently. This replaces the second argument to the .evaluate() method that existed previously.
• Allow registerNamespace() to be called on an XPathEvaluator, after creation, to add additional namespaces. Also allow registerNamespaces(), which does the same for a namespace dictionary.
• Add 'prefix' attribute to element to be able to read prefix information. This is entirely read-only.
• It is possible to supply an extra nsmap keyword parameter to the Element() and SubElement() constructors, which supplies a prefix to namespace URI mapping. This will create namespace prefix declarations on these elements and these prefixes will show up in XML serialization.

Bugs fixed

• Killed yet another memory management related bug: trees created using newDoc would not get a libxml2-level dictionary, which caused problems when deallocating these documents later if they contained a node that came from a document with a dictionary.
• Moving namespaced elements between documents was problematic as references to the original document would remain. This has been fixed by applying xmlReconciliateNs() after each move operation.
• Can pass None to 'dump()' without segfaults.
• tostring() works properly for non-root elements as well.
• Cleaned out the tostring() method so it should handle encoding correctly.
• Cleaned out the ElementTree.write() method so it should handle encoding correctly. Writing directly to a file should also be faster, as there is no need to go through a Python string in that case. Made sure the test cases test both serializing to StringIO as well as serializing to a real file.

Features added

• Changed setup.py so that library_dirs is also guessed. This should help with compilation on the Mac OS X platform, where otherwise the wrong library (shipping with the OS) could be picked up.
• Tweaked setup.py so that it picks up the version from version.txt.

Bugs fixed

• Do the right thing when handling namespaced attributes.
• fix bug where tostring() moved nodes into new documents. tostring() had very nasty side-effects before this fix, sorry!