elementpath

• Fix external function registrations (issue #099)
• Add ExternalFunction and SchemaConstructor token classes

• Drop Python 3.9 compatibility and add Pyton 3.15 support
• Improve XPath sequence internal processing with a list derived type xlist
• Extensions and fixes for XSD datatypes
• Add XSequence datatype for external representation of XPath sequences

• Fix default/fixed values processing of schema annotated XML data (issue #94)

• Fix for XPath 1.0 processing of schema annotated XML data (issue #93)

• Fix for XPath 2.0 fn:node-name (issue #91)
• Workaround for processing arguments with multiple occurrences for external functions (issue #92)

• Fix XDM type labeling with element and xsi:type substitutions (issue #89)

• Replace SafeXMLParser with SafeExpatParser
• Drop compatibility with Python 3.8
• Restruct node trees with XPathNodeTree backbone
• Unicode categories fallback with a warning (issue #85)
• Rewriting decoder module (issue #87)
• Add extras subpackage for experiment XDM and XPath on Path objects

This is in fact a v5.0 pre-release, that includes changes in XDM structure, in order to split XML and XSD schema processing.

• Add full PSVI type labeling in XDM to solve type errors with XSD 1.1 assertions
• Add schema optional argument to dynamic context
• Add a RootToken as a proxy of the parsed token tree for compatibility with xmlschema<=3.4.3
• Extend XDM to split ElementTree/lxml processing from schema nodes and to allow future extensions

• Fix fragment argument usage (issue #81)
• Fix constructors nud() to skip argument check with XP31+ arrow operator (issue #83)

• Fix XsdAttributeGroupProtocol
• Improve Unicode support with installable UnicodeData.txt versions
• Extend names disambiguation with a fix for issue #78
• Refactor tree builders to fix document position of tails (issue #79)

• Fix and clean node trees iteration methods (issue #72)
• Fix missing raw string for '[^\r\n]' (pull request #76)
• Full and more specific type annotations

• Improve stand-alone XPath functions builder (issue #70)
• Update tokens and parsers __repr__
• Fix static typing protocols to work with etree and XSD elements

• Change the purpose of the evaluation with a dynamic schema context, that is now used only for labeling tokens with XSD types
• The static evaluation is now performed also when a schema is provided to the parser
• First tests with Python 3.13 pre-releases by adding a tox.ini testenv

• Fix dynamic context initialization with lxml a non-root element (issue #71)
• Fix XP30+ function fn:function-lookup
• Fix XP30+ fn:unparsed-text, fn:unparsed-text-lines and fn:unparsed-text-available

• Drop support for Python 3.7
• Add uri and fragment options to dynamic context
• Make context root node not mandatory (issue #63)
• Add function objects constructor (issue #70)

• Fix typed value of ElementNode instance if self.elem.text is None

• Fix select of prefixed names (issue #68)
• Fix zero length xs:base64Binary (pull request #69)

• Fix XP30+ fn:path (issue #67)
• Fix weak tests (issues #64 and #66)

• Add support for Python 3.12
• Fix self shortcut operator (adding is_schema_node() to node classes)

• Simplify type annotations for XSD datatypes
• Full test coverage of sequence type functions with bugfixes

• Refactor XPath function call (context=None only as keyword argument)
• Add external function support (issue #60)
• Some fixes to string representation and source property of tokens
• Extend documentation and tests
• Clean XSD datatypes hierarchy

• Fix packaging: include py.typed in package data
• Revert to comparison between xs:QName instances and strings
• Fix atomization with XSD typed data (list types have to be atomized)

• First XPath 3.1 implementation (without UCA collation support)

• Extend root concept to subtrees used as root (e.g. for XSD 1.1 assertions)
• Begin XPath 3.1 implementation adding XPathMap and XPathArray

• Fix of descendant path operator (issue #51)
• Add support for Python 3.11

• Transition to full XPath node implementation (more memory usage but better control and overall faster)
• Add etree.py module with a safe XML parser (ported from xmlschema)

• Fix unary path step operator (issue #46)
• Fix sphinx warnings 'reference target not found' (issue #45)

• Include PR #43 with fixes for XPathContext.iter_siblings() (issues #42 and #44)

• Fix for failed floats equality tests (issue #41)
• Static typing tested with mypy==0.950

• Add XPath 3.0 support
• Better use of lxml.etree features
• Full coverage of W3C tests
• Drop support for Python 3.6

• Fix type annotations and going strict on parsers and other public classes
• Add XPathConstructor token class (subclass of XPathFunction)
• Last release for Python 3.6

• Make ElementProtocol and LxmlElementProtocol runtime checkable (only for Python 3.8+)
• Type annotations for all package public APIs

• Add LxmlElementProtocol
• Add pytest env to tox.ini (testing issue #39)

• Add inline type annotations check support
• Add structural Protocol based type checks (effective for Python 3.8+)

• Add Python 3.10 in Tox and CI tests
• Apply slots to TDOP and regex classes

• Fix issue sissaschool/xmlschema#243 (assert with xsi:nil usage)
• First implementation of XPath 3.0 fn:format-integer

• Add function signatures at token registration
• Some fixes to XPath tokens and more XPath 3.0 implementations

• Optimize TDOP parser's tokenizer
• Resolve ambiguities with operators and statements that are also names
• Merge with XPath 3.0/3.1 develop (to be completed)

• Add tests and apply small fixes to TDOP parser
• Fix wildcard selection of attributes (issue #35)

• Extend tests for XPath 2.0 with minor fixes
  • Fix fn:round-half-to-even (issue #33)

• Extend tests for XPath 1.0/2.0 with minor fixes
• Fix for +/- prefix operators
• Fix for regex patterns anchors and binary datatypes

• Fix for issue #32 (test failure on missing locale setting)
• Extend tests for XPath 1.0 with minor fixes

New classes for XPath nodes

• Replaced named-tuple classes with a hierarchy of custom classes that bind instances (attribute nodes, text nodes, namespace nodes) with parent elements;
• Nodes are equal if they have also the same parent (same element instance).

• Increase the speed of path step selection on large trees

• More tests and small fixes to XSD builtin datatypes

Parser's tokenizer is now created at first usage of a parser. This reduce import time and avoid the build of tokenizers of unused parser version.

• Fix context handling in cycle statements

• Change constructor's label to 'constructor function'

• Add regex translator to package API

• More than 99% of W3C XPath 2.0 tests pass

Removed the change on Decimal default context (prec=30 instead of the default 28). Added regex transpiler for XPath/XQuery or XSD regular expressions.

About ~98% of W3C XPath 2.0 tests run successful.

• Fix XPathContext to let the subclasses replace the XPath nodes iterator function

• Fix tokenizer and parsers for ambiguities between symbols and names

• Improve XPath context and axes processing

• Integrate pull requests and fix bug on predicate selector

• Fix PyPy 3 tests on xs:base64Binary and xs:hexBinary

• Separated the tests of schema proxy API and other schemas based tests

• Multiple XSD type associations on a token

• Extend xs:untypedAtomic type usage

• Increase the tests coverage to 95%

• Fix for node kind tests

• Fix for issue #17

• Update test dependencies

• Add PyPy3 to tests

• Remove Python 2 support

• Add TextNode node type

• Fix for issue #15 and for errors related to PR #16

This is the last release compatible with Python 2 (2.7). The next release will be the v1.4.0, a "Python 3 only" release (3.5+).