Changelog

0.1.0

First release. Flask works end to end; FastAPI and Django/DRF are supported for the
parts that make sense on each.

Schema builder (s) mirroring the Node helper method-for-method, producing JSON
that is byte-identical to Node's — same values, same key order.
Route registry, config normalisation, and an OpenAPI 3.1 exporter with $ref
dedup, anonymous-schema promotion, security schemes, tags, callbacks, webhooks and
status-keyed responses.
Verified against the Node reference by the shared
conformance suite, which
runs the same fixtures through both implementations on every push.

Flask — reads app.url_map, so there is no router to replace. Path parameters are
typed from Flask's own converters. @define_route attaches metadata and returns your
function unchanged. Introspection is lazy, so mount order does not matter.
FastAPI / Starlette — consumes app.openapi() rather than deriving a second spec,
and adds drift. FastAPI's Swagger UI is left alone; DocTreen defaults to /doctreen.
Django / DRF — walks the URLconf and reads DRF serializers. A ViewSet expands into
one documented route per action, each carrying its own docstring.
Pydantic v2 and DRF serializers as schema sources.

Request validation with the standard 422 envelope, checking body, query and path in
one pass. Optional write-back pushes coerced values at the handler.
Response assertion — a development-time check that never touches the body, and is
status-aware: a 409 is checked against the schema declared for 409.
Schema drift detection with sampling, rolling buckets, a pluggable store, webhook
and callback hooks, announce_routes, and release attribution.

Docs UI sharing its entire visual surface with the Node package; the stylesheet and
browser script are extracted verbatim by scripts/sync_ui_assets.py.
Integration flows as a shared document format, plus a pytest plugin that runs
*.flow.json files as tests, and a doctreen-flow CLI.

Requires Python 3.9+. Zero runtime dependencies; everything else is an optional extra.
On Python 3.9, write Pydantic fields as Optional[str] rather than str | None —
Pydantic evaluates annotations at runtime.
The in-memory drift store is per-process. Under gunicorn or uWSGI each worker keeps its
own; use a shared drift.store in those deployments.