Copying a paragraph out of Google Books should take two seconds. Often it doesn't, because the text on screen isn't text at all.
Sometimes a passage won't highlight — you hover, drag, and nothing happens. Other times what looks like text is a scanned page rendered as a flat image, invisible to your clipboard. So you retype it by hand, word by word, and for most people that isn't worth the effort.
I built a small browser script to close that gap. It runs Google Books OCR on the page you're already reading and hands you the text in a panel you can copy from with one click.
What Google Books OCR Does on the Page
OCR stands for optical character recognition. It's the same technology behind document scanners, the cheque-deposit feature in mobile banking apps, and the search function in a PDF reader — it looks at a picture of words and works out which characters those shapes represent. Once that's done, your browser holds actual text instead of a photograph of text.
My script does that conversion in the browser itself using the Tesseract.js OCR engine on GitHub, a JavaScript build of the long-running Tesseract project. Nothing gets uploaded to a server of mine, because there is no server of mine — the recognition happens on your machine.
The three-step flow
Click the floating button. Watch the text populate. Copy it.
That's the whole interaction. No file uploads, no accounts, no separate desktop application, and no waiting for a queue on someone else's hardware.
Installing the Tampermonkey Userscript
A userscript is a small piece of JavaScript that your browser runs automatically on particular websites. It needs a manager extension to hold it, and the usual choice is Tampermonkey.
Step one: add Tampermonkey to your browser
Install it from the official Tampermonkey site. It supports Chrome, Edge, Firefox, Safari and Opera. You'll get a small icon in your toolbar once it's active.
Step two: paste in the script
The script link opens as a plain text file in a new tab. Select everything on that tab and copy it. Then open the Tampermonkey icon, choose "Create a new script", delete the template that appears, paste in what you copied, and save with Ctrl+S (Cmd+S on a Mac).
You've now got the script installed and enabled. This is the step where non-technical readers usually stall, so it's worth saying plainly: you are not installing software, and you are not writing code. You are pasting text into a box and pressing save.
Step three: open any Google Books page
Reload a Google Books page. The floating button appears in the corner and stays there while you scroll.
What the Interface Gives You
Plenty of OCR tools already exist, and most of them feel like they were built in 2009 and never touched again. You upload a file, watch a spinner, and get back a wall of badly formatted text you then have to fix. I wanted the opposite of that experience.
A button that follows you
The button floats over the page rather than sitting at the top, so you never scroll away from it and never go hunting for it.
A sidebar that fills as it reads
Results open in a panel on the right side of the screen. Text appears in real time as recognition progresses, and a progress bar shows how far along it is. You aren't left staring at a blank box wondering whether something has broken.
One-click copy
When the pass finishes, one button puts everything on your clipboard. No dragging a selection across a panel, no right-clicking, no partial copies.
Where Tesseract Reads Cleanly and Where It Struggles
I'd rather set expectations properly than have you discover the limits mid-project.
Pages that come out well
Clean modern scans with a single column of reasonably sized type, printed in a standard serif or sans-serif face, generally come through in good shape. Twentieth-century print in decent condition is the comfortable case.
Pages that give it trouble
Faded or low-contrast scans, very small type, heavy show-through from the page behind, ornate or blackletter typefaces, and older printing with irregular inking all reduce accuracy. Tables and footnotes tend to lose their layout even when the individual words are recognised. Any OCR engine will produce a proportion of errors on difficult source material, so proofread anything you plan to quote.
Who Saves Time With This
Researchers and students
Hours go into pulling quotes, chasing citations and gathering source material inside Google Books. When the text won't select, that work stops dead. This removes that particular bottleneck.
Writers and journalists
Historical texts, public domain books and reference material live on Google Books in scanned form. Being able to lift a passage without retyping it keeps a drafting session moving.
Note-takers and knowledge workers
If you run Notion, Obsidian or Roam Research, you can move content from a scanned book into your own system directly, instead of breaking your concentration to transcribe a paragraph by hand.
Across a long project, that adds up to meaningful time saved — I'd rather not put a number on it, since it depends entirely on how many pages you're working through.
When the Button Doesn't Appear
No button on the page
Check that Tampermonkey is enabled and that the script shows as active in its dashboard. Then hard-refresh the page with Ctrl+Shift+R. If you installed the script while the tab was already open, it won't run until that tab reloads.
Empty or garbled output
An empty panel usually means the page hadn't finished rendering when the pass started. Let the page settle, then run it again. Garbled characters point at the source scan rather than the script — see the section above on difficult pages.
Multi-column pages
Where a page runs two or more columns, recognition often reads straight across the gutter, mixing lines from both columns into one run of text. The words are usually right; the order isn't. On those pages, extract, then reorder by hand.
Copyright, Terms of Service, and Your Call
This script is built for personal reading, learning and research. It makes your own note-taking faster. It is not a bulk downloader, it doesn't page through a book automatically, and it isn't a route to reproducing a whole title for redistribution.
Many books on Google Books remain under copyright, and copying and redistributing protected work without permission infringes it regardless of which tool did the copying. Whether any particular page is yours to extract from depends on that book's copyright status, on Google's terms of service, and on Google's own documentation on robots.txt rules for automated access. Those are your decisions to make, not mine.
Extract what you need for your own work. Cite your sources. Respect the authors whose words you're reading. It's a helper, not a loophole.
If You Need Something Like This for Your Own Workflow
This script solves one narrow problem on one site. Most of the work I do is the same idea pointed somewhere else: a page you need data out of, and no practical way to get it by hand.
Userscripts and Chrome extensions are one of six things I build — the others being Python and Playwright bots, no-code browser macros in UI.Vision RPA, AI-assisted workflows in Automa, cleaned output into CSV, SQLite, PostgreSQL, Google Sheets or Airtable, and scheduled API pipelines over REST, GraphQL and webhooks. You can read what each of those service lines covers if you're not sure which one fits your situation.
If you'd rather describe the problem and let me work out the approach, tell me what data you need and where it lives. I take every project through Fiverr. There's more on how I work and what I've built, and other write-ups on browser automation and data extraction if you want to see the range first.