Turn noisy PDFs into clean text and structured JSON — entirely in your browser.
Visit live site ↗
What it is
PDF2Data is a free, browser-only tool that extracts PDFs into two clean formats: de-noised plain text, and structured JSON with tables, fields, and roles. Everything runs locally — nothing is uploaded — which makes it both private and instant. It's tuned for app-summary documents but works as a general PDF-to-text converter too.
Why it exists
Dense PDFs are painful to feed into AI assistants — they bloat the context window and confuse field names. PDF2Data turns that wall of text into clean, queryable structure you can hand to an AI, and lets you keep only the tables you actually need, which slashes token count and improves accuracy.
Who it's for
Developers who paste documents into AI assistants — and anyone who needs fast, private PDF-to-text extraction.
What it does
- Clean text extraction that strips repeated headers and footers
- Automatic structured-JSON output for recognized documents
- An interactive JSON editor with tree, table, and text views
- A table picker to export only what you need
- A token estimator to preview output size before you copy
- Runs 100% in the browser — files never leave your device
How it works
What made it interesting to build
The hardest part was rebuilding table columns from a PDF without any server-side extractor. PDFs only give you the position of each text fragment, so the parser infers column boundaries from the header row and snaps each value into the right column — handling wrapped names and multi-line cells. It was validated against a reference parser on a real, large document to be sure the structure came out right. A small, separate worker handles notifications so the main tool stays a drag-and-drop static site.