pyxsd documentation

Documentation for pyxsd, a Python library and CLI that maps XML documents into Python object trees according to an XML Schema (XSD). It reports validation issues, queries and exports the bound tree (xpath / find / to_dict / to_json), runs user-defined transforms, and writes XML. The base install has two small, pure-Python dependencies, elementpath and defusedxml; defusedxml is used only when the opt-in parser profiles are selected. pyxsd is an XSD 1.1 processor with an optional XSD 1.0 mode.

For explicit local input sequences, see Sequential multi-document parsing: reuse one compiled schema for parsing or export a new manifested Parquet dataset with the optional Arrow adapter. To attach methods and validation checks to generated classes, see Mixin extensions. The API reference covers schema introspection and filtered stubs; Validation covers resource policies and diagnostic locations; XML security and untrusted input covers parser profiles, limits, remote loading, and untrusted-input boundaries.

Contents

• Quickstart
  • Requirements
  • Installation
  • Command line
  • Library use
  • What pyxsd validates
  • Next steps
• Command-line reference
  • Flags
  • Exit codes
  • Schema location
  • Type stubs
  • Examples
• End-to-end demo walkthrough
  • The pieces
  • Command line
  • Library
  • The transform
  • Crystallography examples
• Architecture
  • Pipeline
  • Resource resolution and the parser boundary
  • The ElementRepresentative system
  • Validation reporting
  • Overlay classes (experimental)
• MCP integration
  • Install and launch
  • Persistent store and session inputs
  • Basic workflow
  • Tool catalog
  • Artifacts
  • Schema exploration and diagnostics
  • Projected records (optional Arrow)
  • Schema comparison and migration evidence
  • Packaged skill discovery
  • Local transformation previews
  • Worker and resource model
  • Audit records and restart behavior
• MCP worker deployment
  • Runtime choices
  • Store and transport
  • Local boundaries
• Data model
  • Node structure
  • Writers
  • Dict and JSON export
  • XPath projection
  • Generated classes
  • Type stubs
  • Value conversions
• Validation
  • Resource and parser policies
  • Report API
  • Issue codes
  • Strict mode
  • Reports and transformed documents
  • Structured and bounded reports
  • Known limitations (schema legality)
• XML security and untrusted input
  • Parser profiles
  • DTDs, entities, and Expat
  • Resource policies
  • Remote XSD loading
  • XPath and denial-of-service limits
  • MCP servers
  • What these controls do not cover
• Sequential multi-document parsing
  • Sources and outcomes
  • Error timing and policies
  • Resources and limits
  • Runnable validation example
  • Optional Parquet dataset export
• Parse modes and binding
  • Using a mode
  • Presets
  • Policy fields
  • Namespaces
  • Why not just loosen the report?
  • Extending the policy
• Mixin extensions
  • Passing a registry to compile
  • Registering a mixin
  • The five selectors
  • Path selectors
  • Ordering and dedup
  • Where mixins sit in the MRO
  • The __pyxsd_validate__ hook
  • Example: a URI type with a method and a hook
  • What extensions are not
• Transforms
  • Using transforms
  • Writing transforms
  • The Transform class
• Supported features
  • Built-in type lattice
  • Structural schema elements
  • Instance semantics
  • Composition and identity
  • Validation reporting
  • Extending and typing generated classes
  • XSD versions (1.0 and 1.1)
  • Known gaps
• XML and XSD validation
  • Well-formed XML is the starting point
  • DTDs can affect the information XSD sees
  • Schema compilation and instance assessment
  • XML versions and XSD versions are independent
  • Reports, PSVI, and a Boolean verdict
  • What XSTS establishes
  • Primary specifications
• XSTS methodology and results
  • Corpus and case selection
  • What the adapters execute
  • Outcome definitions
  • Interpreting residual failures
  • Why extra-suite.xml is separate
  • Two distinct xmlschema roles
  • Reproducing reports and baseline gates
  • Measurement limits
  • Source data
  • Outcomes and checked agreement
  • Schema and instance splits
  • Failure directions
  • Combined-profile applicability
  • Cross-profile failure relationships
  • Residual dispositions
• pyxsd compared with xmlschema
  • At a glance
  • Where pyxsd is a good fit
  • Where xmlschema is a better fit
  • Honest limitations of pyxsd
  • Using them together
  • See also
• pyxsd compared with libxml2
  • Scope, APIs, and resources
  • Preliminary macOS libxml2 2.9.13 probe
  • What was not measured
  • See also
• API reference
  • Schema and documents
  • Parser and resource policies
  • Source spans
  • Mixin extensions
  • Type stubs for generated classes
  • Batch parsing
  • Querying a document
  • Errors
  • Modes and the node protocol
  • Transforms
• Optional Pydantic and Arrow integrations
  • Generate models without instance XML
  • Arrow tables and Parquet
  • Validation, supported subset, and errors
  • Runnable examples
• Data workflow examples
  • Install
  • Shared preparation
  • DuckDB workflow
  • Polars workflow
  • Order assessment
  • Caveats and failure modes
  • Why analysis stays outside the library
• Skills
  • Packaged skills
  • Using skills
  • Install with npx skills
  • Evaluation status
• Migrating from pyxsd 0.1 to 1.0
  • Why the architecture changed
  • Breaking changes
  • Upgrade guide
  • History
• Project history
  • Program origins
• Contributing
  • Reporting issues
  • Development setup
  • Pull requests
  • Quality bar
  • Releases
  • Contribution terms

Project

• License
• Source

Indices and tables

• Index

• Module Index

• Search Page