xmlschema

• Add arguments validation for schemas and validation methods (by validation contexts)
• Add custom XPath parser for find/findall/iterfind APIs on schemas for match
  singleton sequence also if position is a number greater than 1 in predicate
  expression (issue #468)
• Improve build of XSD elements and groups, using a three-state built flag for
  components
• Extend and fix memory tests (Python 3.14+ seems to consume more memory)
• Drop support for Python 3.9 and add development support for Python 3.15

• Add arguments validation for schemas and validation methods (by validation contexts)
• Add SchemaSettings dataclass for storing read-only settings for schema instances and for managing package default settings
• Add block argument to XMLResource class (issue #464)
• Add MAX_SCHEMA_SOURCES package limit (1000, applied to XsdGlobals global maps instances)
• Add MAX_XML_ELEMENTS package limit (1,000,000, applied to non-lazy XMLResource instances)
• Reduce MAX_XML_DEPTH limit to 1000 and apply it to all XMLResource instances
• Fix for substitute match in case of unexpected child (issue #461)

• Add XPath selectors: ElementSelector and ElementPathSelectors
• Add selector argument to XMLResource and iterparse filters
• Fix element substitution (issue #452)
• Restore v3.4.5 builder methods as helpers (issue #451)
• Add overall_min_occurs and overall_max_occurs helpers to XsdElement (issue #449)
• Extend error message for XML syntax errors (issue #448)
• Fix query/fragment like for filenames (issue #447)

• Fix XSD identity update with xsi:type substitution (issue #446)

• Extend support of lxml.etree through iterparse optional argument
• Rework internal decoding/encoding methods, replacing generators with normal functions
• Replace kwargs with a validation context (DecodeContext and EncodeContext)
• Add SchemaLoader class and variants (related to #419)
• Slicing of XsdGlobals with parent maps
• Fix XsdUnion validation performance issue (related to #442)
• Drop support for Python 3.8

• Fix xs:all groups occurs check (issue #437)

• Migrate project metadata to pyproject.toml by @afuetterer (PR #436)
• Fix static typing errors with mypy==1.15.0 and elementpath==4.8.0

• Fix incorrect validation error for substitution group with abstract head (issue #417)
• Refactor XSD identities processing using also typed XPath 2.0+ for getting values (issue #418)
• Clean tag retrieval during encode for some converter types (Abdera, BadgerFish and GData)

• Fix other failing URL normalization tests
• Avoid the use of sys.version_info for checking results, better to extend the check to more values.

• Fix failing URL normalization tests (issue #416)
• Disable protocols checking with elementpath v4.5.0

• Extended ModelVisitor to make it usable as an helper class for generating content (issues #405 and #408)

• Fix UNC path tests (issues #405 and #408)

• Update validation errors with logging stack trace in debug mode
• Improve locations parsing and URL encoding

• Rewrite the validation of openContent using InterleavedModelVisitor and SuffixedModelVisitor
• Fix validation of XSD 1.1 'all' nested models

• Improve ModelVisitor and particle occurs checking
• Fix interleave mode with XSD 1.1 open content (issue #397)
• Fix for export/download of XSD sources with commented-out imports/include (issue #387)

• Add download_schemas() to package API (#387)
• Fix issue with facets on list types (#396)

• Add GData converter (issue #388/PR #391)
• Fix static typing protocols usage
• Extend XSD annotations parsing (issue #366)

• Use XPath subtree as fragment for xs:assert (issue #386)
• Fix in XMLSchemaProxy definition and usage for providing a base-uri to schema nodes (issue #379)
• Module xpath.py splitted to a subpackage for including all the custom XPath 1.0/2.0 parsers and related classes to XPath
• Add support for Python 3.13 (pre-releases)

• Hotfix release for broken requirement
• Set python-requires metadata to >=3.8 (pull request #382)
• Upgrade GitHub Actions (pull request #381)

WARNING: this release has been yanked in PyPI due to a broken requirement, use v3.0.1.

• Namespace declarations processing modes through the option xmlns_processing for converters
• Decode/validate from XML document with dynamic schema load
• XMLResource class enhancement for a better XML resources processing
• Improve lazy resources iteration removing preceding elements (thin_mode option)
• Add uri_mapper option for using relocated or URN-addressed schemas
• Drop support for Python 3.7

• Fix slowness of key selectors introduced by v2.5.0 (issue #378)
• Remove redundant wheel dep from pyproject.toml and unnecessary build deps from tox.ini (PR #368)

• Fix identity keys tracking with additional full XPath checks on XML data
• Rewrite schema exports using relative paths

• Improve schema export using XSD source encoding
• Add XML signature and encryption to local fallback schemas (issue #357)

• Meta-schema elements and groups ignore xsi:type attributes (issue #350)
• Use the meta-schemas only for validating XSD sources otherwise create dummy schemas

• Improve sequence/all restriction checks for XSD 1.1
• Add schema argument to Wsdl11Document

• Add support for Python 3.12
• Detach content iteration methods from ModelVisitor

• Fix mixed content extension with empty content (issue #337)
• Fix lru_cache() usage on global maps caching

• Fix mixed content extension without explicit mixed attribute (issue #334)

• Refine string serialization of XML resources and data elements
• Switch to use elementpath v4
• Fix sequence_type property for XSD types
• Remove XsdElement.get_attribute(): unused and doesn't work as expected

• Fix schema_path usage in XMLSchemaBase.iter_errors()
• Add allow_empty option to XMLSchemaBase validation API

• Add to_etree() to document API
• Improve generic encoding with wildcards
• Clean document API and schema decoding

• Add use_location_hints argument to document API for giving the option of ignoring XSI schema locations hints
• Fix import from locations hints with namespace mismatch (issue #324)

• Add keep_empty and element_hook options to main iter_decode() method
• Fix default namespace mapping in BadgerFishConverter
• Fix type restriction check if restricted particle has maxOccurs==0 (issue #323)

• Fix XSD 1.1 assertions effective scope
• Add support for Python 3.11

• Remove warnings during the build of the package using package_data specs in setup.py
• Fix decoding with process_namespaces=False and xsi:type in XML instance
• Refactor DataElement.get(), restore DataElement.set() (issue #314)
• Add map_attribute_names argument to DataElementConverter

• Refactor XPath interface for the full XPath node implementation of elementpath v3.0
• Fix BadgerFishConverter with mixed content (issue #315)
• Improve get() and set() of DataElement (issue #314)

• Fix invalid element not detected with empty particle (issue #306)
• Fix Sphinx warnings (issue #305)

• Fix 'replace_existing' argument usage in XsdElement.get_binding method (issue #300)
• Add Russian full translation (from PR #303 and #304)

• Protect converter calls in iter_decode()/iter_encode()
• Extend XSD type matching for code generators

• Add localization for XSD parsing and validation related error messages
• Add full Italian translation
• Add Russian partial translation (from PR #293)

• Add 'nonlocal' option to defuse argument of XMLResource (also for schema classes)
• Add 'none' option to allow argument of XMLResource
• Fix too strict parsing on XSD annotations (issue 287)
• Drop support for Python 3.6

• Fix for global simple type naming (issue #278)
• Use mypy==0.930 for static typing checks

• Improve error reporting for encoded data (issue #275)
• Fix attribute duplicates in attribute group (issue #276)
• Add process_skipped optional argument to decoding/encoding

• Add iter_decode() to document level API
• Enhance XMLResource class adding usage of pathlib.Path objects for source and base_url arguments
• Fix for issue #273

• Fix for issues #266 and #268
• Fix type annotation of XMLSchema source argument (issue #230)

• Near compliance with strict type annotations
• Removed ModelGroup class, merged with XsdGroup
• Some optimizations and fixes from static analysis

• Refactor XMLSchemaMeta deprecating BUILDERS attribute
• Extend type annotations to package API
• Add static typing tests with checked mypy runs

• Activate mypy checks for package
• Fix for issues #257 and #259

• Make XSD annotations parsing lazy
• Add lazy annotations to schema instances
• Add get_annotation() method to multiple-facets classes (issue #255)

• Add testing config for Python 3.10 (Tox and CI)
• Fix internal _PurePath class with Python 3.10 (issue #251)
• Remove redundant xmlns="" declaration when encoding with lxml (issue #252)

• Refactor normalize_url() using pathlib.PurePath
• Support UNC paths (issue #246)
• Fix API docs (issue #248)
• Fix encoding of non-root elements (PR #250)

• Fix for issue #245 (key/keyref with dynamic types)
• Change default decoding of mixed content with only text to a string instead of a dictionary (issue #242)

• Add multi-source initialization and add_schema() to schema class
• Add bytes strings to accepted XML sources

XML data bindings and code generators

• XML data bindings and code generators are now considered stable
• Add arguments 'max_depth' and 'extra_validator' to validation methods
• Enhance decoding with 'value_hook' argument

• Fix schema logger (issue #228)
• Fix ElementPathMixin to avoid conflicts when schema is used by an XPath 3 parser

• Improve empty content checking
• Fix simple content restriction of xs:complexType
• Fix facets retrieving for xs:complexType with simple content

Dynamic XML data binding and code generators (experimental)

• Add experimental XML data bindings with a DataBindingConverter
• Add experimental PythonGenerator for static codegen with Jinja2
• Optimize NamespaceView read-only mapping

Decoding XML to data elements with schema bindings

• Add DataElement class for creating objects with schema bindings
• Add DataElementConverter for decode to structured objects
• Add an experimental abstract base class for building jinja2 based code generators (jinja2 as an optional dependency)