document-parse

name

document-parse

description

Use this skill to READ / EXTRACT / PARSE / OCR existing documents — pull text and structure out of PDF, DOCX, PPTX, XLSX, EPUB, HTML, images (JPG/PNG/TIFF/WebP), scanned-document PDFs, or any unstructured file. Triggers: 'parse this PDF', 'extract text', 'OCR this image', 'text aus scan', 'lies das Dokument', 'convert PDF to markdown', 'erkenne Text', 'transcribe document', 'spatial layout extraction', plus any time the user provides a file and asks what's inside. Runs LiteParse locally — Tesseract.js OCR built-in, optional external OCR server (EasyOCR / PaddleOCR) for higher accuracy. Output is clean markdown ready for downstream LLM processing. Do NOT use for: (1) CREATING a new PDF from scratch — use the `pdf` skill. (2) Filling forms in an existing PDF — see `pdf` skill `forms.md`. (3) Pure speech-to-text from audio/video — use the `stt` skill.

Document to Text

Parse unstructured documents (PDF, DOCX, PPTX, XLSX, images, and more) locally with LiteParse (fast local OCR).

Parse a Single File

# Basic text extraction
lit parse document.pdf

# JSON output saved to a file
lit parse document.pdf --format json -o output.json

# Specific page range
lit parse document.pdf --target-pages "1-5,10,15-20"

# Disable OCR (faster, text-only PDFs)
lit parse document.pdf --no-ocr

# Use an external HTTP OCR server for higher accuracy
lit parse document.pdf --ocr-server-url http://localhost:8828/ocr

# Higher DPI for better quality
lit parse document.pdf --dpi 300

Batch Parse a Directory

lit batch-parse ./input-directory ./output-directory

# Only process PDFs, recursively
lit batch-parse ./input ./output --extension .pdf --recursive

Generate Page Screenshots

Screenshots are useful for LLM agents that need to see visual layout or to extract pages as screenshots.

# All pages
lit screenshot document.pdf -o ./screenshots

# Specific pages
lit screenshot document.pdf --pages "1,3,5" -o ./screenshots

# High-DPI PNG
lit screenshot document.pdf --dpi 300 --format png -o ./screenshots

# Page range
lit screenshot document.pdf --pages "1-10" -o ./screenshots

---

Key Options Reference

OCR Options

Option	Description
(default)	Tesseract.js — zero setup, built-in
--ocr-language fra	Set OCR language (ISO code)
--ocr-server-url <url>	Use external HTTP OCR server (EasyOCR, PaddleOCR, custom)
--no-ocr	Disable OCR entirely

Output Options

Option	Description
--format json	Structured JSON with bounding boxes
--format text	Plain text (default)
-o <file>	Save output to file

Performance / Quality Options

Option	Description
--dpi <n>	Rendering DPI (default: 150; use 300 for high quality)
--max-pages <n>	Limit pages parsed
--target-pages <pages>	Parse specific pages (e.g. "1-5,10")
--no-precise-bbox	Disable precise bounding boxes (faster)
--skip-diagonal-text	Ignore rotated/diagonal text
--preserve-small-text	Keep very small text that would otherwise be dropped

Repeated Use: Using a Config File

For repeated use with consistent options, generate a liteparse.config.json:

{
  "ocrLanguage": "en",
  "ocrEnabled": true,
  "maxPages": 1000,
  "dpi": 150,
  "outputFormat": "json",
  "preciseBoundingBox": true,
  "skipDiagonalText": false,
  "preserveVerySmallText": false
}

Use with:

lit parse document.pdf --config liteparse.config.json

---

Non-ASCII / Internationalization

OCR is the main source of Umlaut/diacritic corruption — Tesseract defaults to English and will misread ä as a or ö as 6. Always set the language explicitly for non-English documents:

# Single language
lit parse vertrag.pdf --ocr-language deu

# Multiple languages (e.g. German + English in one document)
lit parse mixed.pdf --ocr-language "deu+eng"

Common ISO codes: deu (German), fra (French), spa (Spanish), ita (Italian), por (Portuguese), nld (Dutch), pol (Polish), rus (Russian), jpn (Japanese), chi_sim/chi_tra (Chinese). For higher accuracy on Umlaut-heavy material, route to a stronger backend via --ocr-server-url (PaddleOCR, EasyOCR).

Reading the output: lit writes UTF-8. When loading the result in Python use open(path, encoding="utf-8") — never rely on the platform default. If you see ä / ö / ü in output, the file was decoded as Latin-1; re-read with explicit UTF-8.

Embedded text PDFs: --no-ocr skips OCR entirely and pulls the embedded text layer, which is already correctly encoded — preferred when the PDF is digital-native.

Supported Input Formats

Category	Formats
PDF	.pdf
Word	.doc, .docx, .docm, .odt, .rtf
PowerPoint	.ppt, .pptx, .pptm, .odp
Spreadsheets	.xls, .xlsx, .xlsm, .ods, .csv, .tsv
Images	.jpg, .jpeg, .png, .gif, .bmp, .tiff, .webp, .svg

LiteParse auto-converts these formats to PDF before parsing.