✅ fresh

last synced 2026-06-17T22:39:35.514075+00:00 · coverage 100% (doc-indexer)

Validation by Beadloom doc_sync — same source as sync-check.

Doc Indexer (component)

Internal building block of the doc-sync domain.

Source: src/beadloom/doc_sync/doc_indexer.py

---

Overview

Scans the project's Markdown documentation, chunks it, and populates the SQLite docs / chunk tables (with content hashes). This index backs FTS5 search and is the doc half of every sync-check pair — the source-of-record for what documentation exists and whether it has changed.

Public surface

• index_docs(...) — scan the docs tree, chunk each file, and populate the docs / chunks tables; returns a DocIndexResult.
• chunk_markdown(text) — split Markdown into section chunks by H2 heading (capped at MAX_CHUNK_SIZE = 2000 chars per chunk).
• classify_section(heading) — map a heading to a section label (spec / invariants / constraints / api / tests / other) via _SECTION_RULES.
• DocIndexResult — the dataclass summarizing an index run.

Collaborators

Writes the docs / chunks tables consumed by the search (FTS5) feature and the context-builder (chunk collection). It is the doc half of every sync-check pair — the section labels it assigns drive chunk-priority ordering in the context bundle.

Component doc (BDL-051). Public surface verified against doc_indexer.py.