# Manual Search Index

Use the generated page-level index before scanning a PDF. It covers every PDF currently bundled in `assets/manuals/` and returns exact physical PDF pages.

## Files

- `assets/indexes/xinje-manual-fulltext.sqlite`: SQLite FTS5 trigram page-text index.
- `assets/indexes/xinje-manual-toc.json`: machine-readable manual metadata and detected table-of-contents entries.
- `references/manual-chapter-index.md`: human-readable generated directory and printed-page map.

## Search

Run this before `search_pdf_keywords.py` for ordinary text questions:

```powershell
& '<python.exe>' -X utf8 scripts\search_xinje_manuals.py --index assets\indexes\xinje-manual-fulltext.sqlite '回原点'
```

Use `--manual DS5C` or another filename fragment to narrow results. Search results are candidates, not final authority for tables, wiring diagrams, images, or high-risk instructions; extract and visually verify the cited PDF page for those cases.

## Rebuild

Run after adding, replacing, or removing a PDF under `assets/manuals/`:

```powershell
& '<python.exe>' -X utf8 scripts\build_pdf_search_index.py --manual-dir assets\manuals --output-db assets\indexes\xinje-manual-fulltext.sqlite --output-toc-json assets\indexes\xinje-manual-toc.json --output-toc-md references\manual-chapter-index.md
```

The index is a search accelerator, not a Markdown rewrite of every PDF. Keep source PDFs for layout and visual validation.