Support schema v2.0.0 (citations, provenance, hatch_schema_version, breaking field changes)

[LittleCoinCoin]

Context

Hatch-Schemas PR #29 introduced package/v2.0.0/hatch_pkg_metadata_schema.json. This issue tracks the corresponding validator work needed to support packages using hatch_schema_version: "2.0.0".

The implementation follows the established pattern: add a hatch_validator/package/v2_0_0/ module, wire it into the validator chain, and register it in the factory.

---

Breaking changes from v1.2.2 → v2.0.0

These are changes that affect existing validation logic and require deliberate handling, not just schema passthrough.

1. Version field rename: package_schema_version → hatch_schema_version

The field used to route packages to the correct validator has been renamed. The current chain-of-responsibility dispatcher reads metadata.get("package_schema_version", "") (see v1_2_2/validator.py:67). The v2.0.0 validator must read hatch_schema_version instead.

Note: Packages with the old field name should still be routed to the appropriate ≤1.2.2 validator. The dispatch key used for routing must not break backward compatibility.

2. author → authors (renamed + stricter)

The singular author field is now authors, an array with minItems: 1. Each item requires name and allows an optional email. Any accessor or dependency logic reading the author field must be updated.

3. tools[].description → tools[].desc

The description field inside each tool entry was renamed from description to desc. The tools_validation module must be updated accordingly.

4. Docker dependency: version_constraint → tag, digest now required

• digest (sha256:<64 hex chars>) is now a required field on docker dependencies.
• version_constraint is replaced by the optional tag field.
• The docker dependency validator must enforce digest presence and stop looking for version_constraint.

5. version_constraint is optional for all dependency types

Previously version_constraint was required on some dependency objects. In v2.0.0 only name is required for hatch, python, and system dependencies. The validator must not error when version_constraint is absent.

---

New fields requiring validation

6. provenance (optional object)

{
  "git_sha": "a1b2c3d",      // ^[0-9a-f]{7,40}$
  "build_env": "conda-lock"  // enum: conda-lock | pip-compile | manual
}

The object is optional, but if present it must satisfy anyOf: [required: git_sha, required: build_env] — at least one of the two fields must be provided. additionalProperties: false.

7. citations (optional array)

Each citation item requires format (enum) and value, with an optional note. The value field has format-specific patterns enforced via allOf/if-then:

format	value pattern / constraint
doi	^10\.\d{4,}/\S+$
arxiv	^\d{4}\.\d{4,5}(v\d+)? or https://arxiv.org/abs/...
pmid	^\d+$
isbn	^(97[89])?\d{9}[\dX]$
url	URI format
bibtex, ris, csl-json, formatted	free-form string

These conditional validations are handled by JSON Schema itself, but any custom citation-level checks belong in a new citations_validation.py strategy.

---

Non-breaking schema changes (no custom logic needed)

Enforced by the fetched JSON schema, no custom Python validation required:

• version now validates full SemVer.
• name follows reverse-DNS format with a single / separator.
• description.maxLength increased to 200 (was 100).
• license and tags are no longer required fields.
• additionalProperties: false added to nested objects — stricter schema surfaces unexpected fields automatically.

---

Implementation checklist

• Create hatch_validator/package/v2_0_0/ with:
  • __init__.py
  • accessor.py — reads hatch_schema_version; handles authors (array); handles tools[].desc
  • schema_validation.py — fetches package schema at tag schemas-package-v2.0.0
  • dependency_validation.py — enforces digest required on docker; tag replaces version_constraint for docker; version_constraint optional everywhere
  • tools_validation.py — reads desc instead of description per tool
  • entry_point_validation.py — unchanged from v1.2.2, delegate or copy
  • citations_validation.py (if custom logic beyond JSON Schema is desired)
  • provenance_validation.py (if custom logic beyond JSON Schema is desired)
  • validator.py — dispatches on hatch_schema_version == "2.0.0"
• Register v2.0.0 in hatch_validator/core/validator_factory.py
• Update hatch_validator/schemas/schema_fetcher.py if the release tag prefix changes for v2.0.0
• Add test fixtures using examples/v2.0.0/server_v2.0.0_example.json (valid) and server_v2.0.0_invalid_example.json (invalid) from Hatch-Schemas
• Add tests/test_package_validator_for_v2_0_0.py covering: schema routing, citations, provenance, docker digest, tools desc, authors array

---

Reference files

• Schema: package/v2.0.0/hatch_pkg_metadata_schema.json
• Valid example: examples/v2.0.0/server_v2.0.0_example.json
• Invalid example: examples/v2.0.0/server_v2.0.0_invalid_example.json
• Prior art: hatch_validator/package/v1_2_2/ (most recent validator module to use as template)