LiteParse standalone OSS PDF parsing tool

LiteParse is a standalone OSS PDF parsing tool.

Verified

Core facts

Read by a script from the official API, not rewritten by a model.

License
Apache-2.0
Languages
Rust
Maintainer
run-llama
Verified

What it is

LiteParse is a standalone OSS PDF parsing tool. It provides spatial text parsing with bounding boxes, without proprietary LLM features or cloud dependencies. The project says everything runs locally on your machine, which matches the default built-in Tesseract OCR; if you plug in an HTTP OCR server, that server's API receives the file and language parameters, so whether the file stays on your machine depends on where the server runs. The project is a Rust workspace with the core library and language-specific binding crates.

Main features

  1. Output formats are Markdown, JSON, and Text.

  2. It can be used from Rust, Node.js/TypeScript, Python, or the browser (WASM).

  3. Supported platforms are Linux, macOS (Intel/ARM), and Windows.

  4. It can generate page screenshots for LLM agents.

  5. Markdown output is structured Markdown with headings, tables, lists, images, and links.

  6. In Python and Node.js, Worker Pool Mode parses in persistent worker processes for true parallelism (PDFium otherwise serializes concurrent parses).

Getting started

  1. All versions except WASM ship with the same lit CLI.

  2. liteparse can be used as an agent skill, downloaded with the skills CLI tool.

  3. Batch parsing parses an entire directory of documents.

  4. The complexity check always prints per-page JSON to stdout and a human-readable verdict to stderr.

From the README · bash
npx skills add run-llama/llamaparse-agent-skills --skill liteparse
From the README · bash
# Basic parsing
lit parse document.pdf

# Parse to Markdown — headings, tables, lists, images, links
lit parse document.pdf --format markdown -o output.md

# Parse with specific format
lit parse document.pdf --format json -o output.json

# Parse specific pages
lit parse document.pdf --target-pages "1-5,10,15-20"

# Parse without OCR
lit parse document.pdf --no-ocr

# Include page-scoped vector path data in JSON
lit parse document.pdf --format json --extract-vector-graphics

# Include rich per-item PDF text metadata
lit parse document.pdf --format json --extract-text-metadata

# Include page annotations in structured JSON
lit parse document.pdf --format json --extract-annotations

# Include AcroForm widget fields and values (repairs orphaned widgets in memory)
lit parse document.pdf --format json --extract-form-fields

# Parse a remote PDF
curl -sL https://example.com/report.pdf | lit parse -
From the README · bash
lit batch-parse ./input-directory ./output-directory
From the README · bash
export TESSDATA_PREFIX=/path/to/tessdata
lit parse document.pdf --ocr-language eng

Who it is for

The built-in OCR is Tesseract, bundled with the library. HTTP OCR servers can also be plugged in, such as EasyOCR, PaddleOCR, or a custom one. Complexity detection checks whether a document needs OCR or heavier parsing.

The Markdown output mode is purely heuristics and rule-based. In Markdown output, complex documents may not render perfectly. Vector path output is opt-in because path-heavy PDFs can produce large payloads.

In JavaScript/WASM, the structure tree extraction option adds a page-scoped structure tree. LiteParse supports automatic conversion of various document formats to PDF before parsing. Office documents are handled via LibreOffice.

Supported image formats are jpg, jpeg, png, gif, bmp, tiff, webp, and svg. The license is Apache 2.0. The credits list PDFium for PDF rendering and text extraction.

The credits list Tesseract as the OCR engine (via tesseract-rs). The credits list napi-rs for Node.js native bindings. The credits list PyO3 for Python native bindings.

The credits list wasm-bindgen for WebAssembly bindings.

Sources

This page is written from the official sources below. Open them to check the original text.

Verified ·

This page is written by AI from the official sources above and checked automatically. This site has not tested it. Found a mistake? Tell us.