feat(doc-symbols): Python objects.inv provider + seed_doc_symbols CLI #17

Merged

raisondetredev merged 1 commit from refs/pull/17/head into main

2026-06-25 20:40:50 +02:00

raisondetredev commented

2026-06-25 20:37:38 +02:00

(Migrated from codeberg.org)

Part of the multi-repo doc-symbol feature — app project-halkyon/halkyon-learn#685, backend project-halkyon/halkyon-infra#65, curator project-halkyon/halkyon-curator#17.

The curator pipeline for "Link to docs": per-ecosystem providers that build (ecosystem, fqn) -> official_url rows and upsert them into the doc_symbols catalog.

DocSymbolProvider interface + registry, DocSymbolRow (snake_case toJson + trigram search_blob with dotted-suffix aliases), upsertDocSymbols (idempotent (ecosystem, fqn) upsert), and bin/seed_doc_symbols.dart (clone of seed_dictionary: service-role key from env, --dry-run, dedup).
Phase 1 = Python: parsePythonObjectsInv parses Sphinx v2 objects.inv (pandas, numpy, scipy, scikit-learn, seaborn, matplotlib, statsmodels, stdlib) — the authoritative name→URL map. Tests: fixture parse + search_blob/toJson.

Dart/Flutter, Go, Rust, R providers slot into the registry next (re-seed; no app/infra change). Run after the doc_symbols table exists.

> Part of the multi-repo **doc-symbol** feature — app project-halkyon/halkyon-learn#685, backend project-halkyon/halkyon-infra#65, curator project-halkyon/halkyon-curator#17. The curator pipeline for **"Link to docs"**: per-ecosystem providers that build `(ecosystem, fqn) -> official_url` rows and upsert them into the `doc_symbols` catalog. - `DocSymbolProvider` interface + registry, `DocSymbolRow` (snake_case toJson + trigram `search_blob` with dotted-suffix aliases), `upsertDocSymbols` (idempotent `(ecosystem, fqn)` upsert), and `bin/seed_doc_symbols.dart` (clone of `seed_dictionary`: service-role key from env, `--dry-run`, dedup). - **Phase 1 = Python**: `parsePythonObjectsInv` parses Sphinx v2 `objects.inv` (pandas, numpy, scipy, scikit-learn, seaborn, matplotlib, statsmodels, stdlib) — the authoritative name→URL map. Tests: fixture parse + search_blob/toJson. Dart/Flutter, Go, Rust, R providers slot into the registry next (re-seed; no app/infra change). Run after the `doc_symbols` table exists.