feat!: re-home source record URLs onto the upstream entry and name inlined Study after config section

[SkyeAv]

Derived builds now place a section's source record URLs on the upstream repository's supporting sources entry (infores:pubmed-central for PMC sections) instead of the primary graph-infores entry, and the inlined supporting Study is named after the table config file plus its 1-based section index, so edge provenance names the exact repository record and configuration it came from.

Source record URL placement

• Re-homed placement: sections without a provenance.override emit their source.url values as source_record_urls on the supporting entry matching upstream_resource_ids(provenance.repo) (Tcode._provenance_ops in src/tablassert/lib.py): infores:pubmed-central for PMC sections, infores:pubmed for PMID sections. The primary entry (the graph-level infores, e.g. infores:multiomics-kp) keeps upstream_resource_ids and gains no section record URLs: the primary is the transforming resource, not the downloadable record, the same rationale the upstream_source_record_urls override already carried.
• Linkouts contract preserved: build-kg --linkouts keeps the per-edge https://linkouts.skyelanegoetz.com/edges/{edge_id} template on the primary entry; retrieval_sources now passes primary urls through in the upstream_urls branch, and the post-dedup sweep still resolves {edge_id} in every sources[].source_record_urls string.
• Override sections byte-identical: override.upstream_source_record_urls re-homes exactly as specified (the primary emits nothing), override.sources emits its template verbatim, any other override keeps the legacy primary-entry emission, and --linkouts still never injects into override sections.

Study naming

• Name: Study.name in has_supporting_studies is now <table-config.yaml> SECTION <n> (1-based position in document order) instead of the worksheet name or a source data filename like data.parquet. ingests.to_sections stamps section_index into each expanded section and Tcode carries it as a field; Study.id is unchanged (publication CURIE, or the config stem when unidentified).
• Store keys: the stamp rides the content-addressed section hash, so duplicate identical sections in one config now get distinct caches, and pre-upgrade parquets naturally miss instead of serving stale provenance.

Docs

• docs/configuration/table.md, docs/configuration/advanced-example.md, src/tablassert/models.py: the BaseSource.url description, the three url field rows, the sources derivation paragraph, and the Study naming prose are restated for the new behavior; CHANGELOG.md gains Unreleased Breaking Changes entries.

Testing

• make check -> ruff clean, uv run pyright 0 errors, uv run pytest 1217 passed + 1 skipped, cargo test ok, cargo clippy --all-targets -- -D warnings clean.
• uv run mkdocs build --strict -> pass.
• New pins: test_tcode_collect_rehomes_pmc_record_urls_onto_pubmed_central (the reported PMC10010437 S3 xlsx case), test_tcode_collect_rehomes_pmid_record_urls_onto_pubmed, test_tcode_collect_names_study_after_config_file_and_section_index, test_tcode_collect_names_override_study_after_config_section; override byte-identity stays pinned by the existing test_production_patterns.py and test_models.py corpus.

Caveats

• Breaking: consumers reading sources[primary].source_record_urls from derived builds must read the upstream supporting entry instead; override-authored output is untouched.
• Section cache parquets from before this change are content-addressed and will not be reused (one-time rebuild per section).

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Configuration used: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: 0aa02946-54e8-4b13-b3a3-6abf421d8184

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}