PDF to Text — OCR, Copy & Export to Markdown

Can't copy text from a PDF? Getting garbled or gibberish characters when you paste? PDF to Text fixes it — right in Chrome, with nothing uploaded. It extracts clean, structured text from any PDF, including the hard ones: scanned documents (with OCR), and PDFs with broken Unicode or scrambled fonts where Chrome's own copy-paste produces junk. No Acrobat, no account, no server. And because it runs entirely in your browser, it works great on Chromebook, where desktop PDF tools don't exist. How it works: 1. Open any .pdf in Chrome 2. Click the PDF to Text toolbar icon, then "Open in PDF to Text" 3. A WebAssembly engine extracts the text locally in your browser 4. Copy the clean text, press "Copy for AI" to paste into ChatGPT or Claude, or download as Markdown What it handles that others don't: • Broken Unicode and scrambled-font PDFs — retail receipts, web-captured pages, and exports where normal copy-paste gives gibberish • Scanned documents — built-in OCR for image-only pages • Web-to-PDF exports (Google Docs, Chrome "Save as PDF") that break naive extractors • Large, modern PDFs (cross-reference streams) that many tools fail to open • Government forms and publications — IRS instructions, agency PDFs, and other documents with unusual structure Features: • "Copy for AI": one click copies structured Markdown (headings, frontmatter, table of contents) ready to paste into any AI assistant • Three output formats: plain text, basic Markdown, and structured Markdown with YAML frontmatter and token counts • One-click copy, copy as Markdown, or download as a .md file • Readable text view: proportional type, comfortable line length, no monospace wall • Recent extraction history in the popup • Agentic API: automation tools like Playwright, Puppeteer, and Claude in Chrome can read window.__glyph.markdown.structured directly • Use the same engine from the terminal or with AI coding agents: https://jordancoin.github.io/pdf-to-text/install.html Honest about the hard cases: When a character genuinely can't be decoded, PDF to Text shows a visible replacement mark (�) instead of silently inventing a wrong letter. You always know exactly which characters were recovered and which weren't — no fake confidence, no hidden garbage. Privacy: Your PDFs never leave your device. Extraction runs locally in WebAssembly; there is no server that sees your documents. The extension sends a handful of anonymous counters after an extraction (character count, page count, timing, version) and when you close the viewer (whether you copied or downloaded, how long it was open, OS). No PDF content, no URLs, no filenames, no identifiers, no cookies, no third-party analytics. You can switch these counters off with one toggle in the popup. Full policy: https://jordancoin.github.io/pdf-to-text/privacy.html