# Parsing

See which parser KuraDB selects for each changed file and which files never enter the pipeline.

## Parser dispatch

A changed non-directory entry is matched by lowercase extension first, then by text sniffing:

| Input | Parser path | Chunking behavior |
|---|---|---|
| `.pdf` | `go_pkg_parser.PDF` | Document parser chunks |
| `.docx` | `go_pkg_parser.Docx` | Document parser chunks |
| `.pptx` | `go_pkg_parser.PPTX` | Presentation parser chunks |
| `.csv`, `.tsv` | `go_pkg_parser.CSV` through `parseTabular` | Header-aware groups of five data rows |
| `.xlsx` | `go_pkg_parser.XLSX` through `parseTabular` | Header-aware groups of five data rows |
| Other valid UTF-8 text | `go_pkg_parser.Markdown` | Markdown/text chunks |

Tabular chunks render each row as `[n] header=value, ...`, using `colN` when a header cell is empty and dropping cells beyond the header width. A sheet with fewer than two rows (header plus data) produces no chunks.

If a parser returns an error, the failure is logged and nothing is written, so rows from the previous successful parse stay active.

## Skipped inputs

| Rule | Examples |
|---|---|
| Excluded by name | `.DS_Store` |
| Image extensions | `.jpg`, `.png`, `.gif`, `.webp`, `.heic`, `.svg`, `.tiff`, `.avif` |
| Video and audio extensions | `.mp4`, `.mov`, `.mkv`, `.mp3`, `.wav`, `.flac`, `.m4a`, `.opus` |
| Archive and package extensions | `.zip`, `.tar`, `.gz`, `.7z`, `.rar`, `.iso`, `.dmg`, `.deb`, `.apk` |
| Binary, bytecode, and library extensions | `.exe`, `.dll`, `.so`, `.dylib`, `.class`, `.jar`, `.pyc`, `.wasm` |
| Database, design, and font extensions | `.db`, `.sqlite`, `.psd`, `.sketch`, `.fig`, `.ttf`, `.woff2` |
| Fails text sniffing | Empty files, files with a NUL byte or invalid UTF-8 in the first 8,192 bytes |

Skipping keeps image base64 and unsupported binary data out of the text embedding pipeline. A skipped file is not dismissed: if a previously indexed text file is emptied, it fails sniffing and its earlier chunks remain searchable until the file is deleted or receives new text.