Stop Using Slow PDF Parsers | LiteParse v2.0 (Rust) Tutorial

LiteParse v2.0 just dropped — rewritten in Rust, runs everywhere (Python, Node, CLI, even WASM), and parses PDFs and images locally with no LLM and no API key. In this video I walk through it end-to-end: what changed in v2, how it compares to other PDF parsing options, how to install it with uv, how to fix the most common Tesseract OCR error, how to parse PDFs and handwritten images, how to use the CLI, and how to wire it into Claude Code as a skill. Everything is hands-on — code, terminal, real errors and how I fixed them. 🔗 Repo with all code from this video: https://github.com/sudarshan-koirala/liteparse 📌 TIMESTAMPS 00:00 LiteParse Overview 01:53 Why LiteParse v2 03:08 PDF Parsing Landscape 05:24 Docs OCR Options 06:48 Clone Repo Setup 08:16 Install Dependencies UV 10:11 Notebook Environment Tips 12:50 Parse First PDF Fast 13:45 Fix Tesseract OCR 16:48 Save Markdown Output 18:24 Handwritten Image Parsing 20:15 OCR Quality Tweaks 21:40 Tuning OCR Settings 22:01 Extracting PDF Screenshots 23:39 Claude Code Skill Setup 24:53 Fixing Missing Skills 29:06 Parsing PDFs via Agent 32:39 CLI Parsing with uv 34:39 Saving Output and OCR Errors 36:04 Parsing PDFs from URLs 39:28 Wrap Up and Recommendations 📚 RESOURCES - LiteParse v2.0 launch blog (LlamaIndex): https://www.llamaindex.ai/blog/liteparse-v2-0-runs-everywhere - LiteParse official docs: https://developers.llamaindex.ai/liteparse/ - Getting Started: https://developers.llamaindex.ai/liteparse/getting_started/ - Python library usage: https://developers.llamaindex.ai/liteparse/guides/library-usage/ - CLI reference: https://developers.llamaindex.ai/liteparse/cli-reference/ - LiteParse on PyPI: https://pypi.org/project/liteparse/ - LlamaParse agent skills (Claude Code): https://github.com/run-llama/llamaparse-agent-skills - uv (Python package manager): https://docs.astral.sh/uv/ - Tesseract OCR: https://github.com/tesseract-ocr/tesseract - Claude Code skills docs: https://code.claude.com/docs/en/skills --- Github link: https://github.com/sudarshan-koirala/liteparse 🧰 WHAT YOU'LL LEARN - What changed in LiteParse v2 (Rust core, runs anywhere, ~100x faster on small docs) - Where LiteParse fits vs PyMuPDF, Docling, and LLM parsers like LlamaParse - Setting up the project with uv (uv sync, uv add liteparse) - Fixing the dreaded "TESSDATA_PREFIX" Tesseract error on macOS - Parsing PDFs in Python and saving to .txt / .md - OCR for scanned PDFs and handwritten images - Tuning OCR (DPI, language packs, preprocessing) - Extracting page screenshots for vision pipelines - Using the lit CLI with uv (parse, batch-parse, screenshot) - Why curl | lit parse - doesn't work today, and what to do instead - Installing LiteParse as a Claude Code skill (.claude/skills/) - Common gotchas (skill folder location, restart-on-first-add, project vs personal scope) 🤝 SOCIALS - YouTube: https://www.youtube.com/@datasciencebasics - Subscribe for more AI engineering, RAG, Databricks, and developer-tooling tutorials If this helped you, a 👍 and subscribe means a lot — and drop a comment with what you'd like me to build next (a full RAG pipeline using LiteParse is on the shortlist). ----------------------------------------------------------------------------------------- 🤝 Connect with me: 📺 Youtube: https://www.youtube.com/@datasciencebasics?sub_confirmation=1 👔 LinkedIn: https://www.linkedin.com/in/sudarshan-koirala/ 🐦 Twitter: https://twitter.com/mesudarshan 🔉Medium: https://medium.com/@sudarshan-koirala 💼 Consulting: https://topmate.io/sudarshan_koirala #LiteParse #LlamaIndex #PDFParsing #RAG #AIEngineering