Changelog

All notable changes to pandera-ui are documented here.

[0.3.0] — 2026-04-26

--export markdown / --export html — generate schema docs without starting a server; embeds into README, Sphinx, or MkDocs
--coverage flag — print documentation coverage (% of schemas and columns with title/description); also available as GET /api/coverage for CI quality gates
--watch / -w flag — auto-reload schemas on .py file changes using watchdog (install via pandera-ui[watch])
CoverageReport Pydantic model in pandera_ui.models
pandera_ui._export module with to_markdown() and to_html() functions
pandera_ui._coverage module with compute_coverage() and format_coverage() functions
watch optional extra: pip install pandera-ui[watch]
MkDocs documentation site

rich optional extra: pip install pandera-ui[rich] enables spinner and summary table in CLI
examples/quickstart.py — standalone Python API demo

scanner.py split into _extract_runtime.py and _extract_ast.py for clarity
_FIELD_CHECK_MAP and _CHECK_STAT_MAP unified — stat map is now derived, removing duplication
Renamed internal _cli → cli_app in cli.py

CI: id-token: write permission moved to dedicated upload-coverage job (was granted to all matrix runners)
CI: setup-uv updated from v4 → v5

Two-pass schema extraction: runtime import first, static AST fallback when imports fail
FastAPI server serving a single-page vanilla JS UI
Supports pa.DataFrameSchema and pa.DataFrameModel subclasses
Per-column metadata: name, dtype, nullable, required, checks with parameters, title, description
Per-schema metadata: name, coerce, title, description, index, source file, variable/class name
CLI (pandera-ui) with --json mode for CI/tooling integration
Python API: scan_project() returns a list of Pydantic SchemaMetadata models
4 UI languages: EN, RU, FR, DE
Dark / light theme with localStorage persistence
Filter by schema type, sort by name / file / columns, full-text search
Docker image and docker-compose support