Free text cleaner, seventeen operations
Text cleaner with a receipt for every change
This is a checklist rather than a single button: seventeen clean-up operations, each with its own tick box, and a badge on each one counting the places it found in your text before you commit to it. It is free, wants no signup and runs on documents of any size, with four presets named after where the mess came from — Word, a PDF, a list bound for a spreadsheet, or the strictest plain-letters output. The table underneath records what every operation changed and what it cost, so you can hand the result on knowing exactly what left.
- 100% free
- No signup
- 17 operations
- 4 one-click presets
- Counts every change
Start from where the mess came from
Undoes what autocorrect added: curly quotes, long dashes, non-breaking spaces, soft hyphens.
What each operation caught
Paste something above and this list fills in with the operations that actually found work to do.
How to clean up messy pasted text
Three steps, and the badges tell you what was wrong before you change anything.
Paste the text and say where it came from
Drop the passage into the top box. The keyboard shortcut reaches it from anywhere here — ⌘V, or Ctrl+V on Windows — so you never have to click in first. Then hit the preset that matches its origin: Word and Google Docs for typography autocorrect changed, a PDF for lines broken at the margin, a spreadsheet list for bullets and repeats, or plain letters for the strictest possible output.
Adjust the tick boxes and watch the badges
Every operation the preset switched on shows a badge with the number of places it found in your text, so you can see at a glance that there really were 14 non-breaking spaces and 3 curly apostrophes hiding in there. Tick and untick freely — the clean version and all the badges recompute on the same keystroke, and switching a box off never damages what an earlier one did.
Audit the summary before you paste over the original
The list at the foot of the tool names each operation that actually did something, how many places it touched and how many characters that cost. Copy the clean text or save it as cleaned-text.txt; the four tiles above the table give characters in, characters out, characters gone and lines gone.
Technical specifications
| Operations available | 17, grouped into invisible characters, spacing, typography, line layout and characters to drop |
|---|---|
| Presets | Four: pasted from Word or Google Docs (8 operations), pasted from a PDF (7), a list for a spreadsheet (9), and down to plain letters (13) |
| Invisible characters recognised | 24 by name — 14 lookalike spaces from the non-breaking one to the ideographic, and 10 widthless marks including the byte order mark, the soft hyphen and the direction marks |
| Typography reversed | 12 quotation marks folded to " and ', 7 dash characters folded to -, and the single-character ellipsis expanded to three periods |
| Order of execution | Fixed pipeline, not tick order: invisible characters, then typography, then spacing, then line layout, then character removal — so each stage sees the output of the last |
| What the preview reports | A badge per operation with the number of places it touched, plus a table of characters lost per operation and four before-and-after totals |
| Speed | A 100,000-word manuscript runs all 13 operations of the strictest preset in under 500 ms; typing is never blocked because the pass runs off the keystroke |
| Price | Free, no signup, no cap on document size or on how many you clean |
Frequently asked questions
What is actually wrong with text I copy from Word?
Autocorrect has silently rewritten it while you typed. Word converts straight quotes into curly ones, a double hyphen into an em dash, three periods into a single ellipsis character, and it inserts non-breaking spaces to hold pairs of words together. None of that is visible in Word, and all of it causes trouble the moment the text leaves: curly quotes break CSV files and code, the ellipsis character is missing from some fonts, and the non-breaking space defeats a search for the ordinary kind. The first preset undoes exactly those substitutions and nothing else.
Does the order the operations run in matter?
Very much, and it is fixed rather than left to the order you ticked the boxes. Unusual spaces are converted before anything measures a line, because a line consisting of one non-breaking space is not blank until it has been normalised. Wrapped lines are rejoined before blank lines are counted, since rejoining changes how many lines there are. The character-level operations run last, so the gentler ones have already done their work when the destructive ones start. This is why one pass here is not the same as running seven tools in whatever order you happened to open them.
Why did my em dashes turn into hyphens?
Because the dash operation is switched on, and it is deliberately blunt. Em and en dashes are correct in prose and wrong nearly everywhere else — filenames, form fields, command lines and older systems either reject them or render them as a question mark in a box. If you are cleaning text that is going back into a document rather than into a machine, untick that one box and the dashes survive while everything else is still swept.
Will this change any of my actual words?
Only if you ask it to, and the two operations that can are both off in three of the four presets. Flattening accents rewrites café as cafe, and removing punctuation takes the marks out from between your words. Everything else moves or deletes characters that were never part of the writing: spacing, invisible marks, line structure and typographic substitutes. The characters-gone tile tells you the size of the change before you commit to it.
Can it keep my bold and italics?
No, and nothing that works on plain text can. What you paste here is the plain-text version of your selection; the formatting was already discarded by the clipboard before this page saw anything. That is often the point — stripping formatting is why people paste into a plain box in the first place — but if you need the styling preserved, clean the text here and reapply the emphasis in your editor afterwards.
I only want one of these things done. Is that worse than doing all of them?
Not at all, and there are dedicated pages for the ones people ask for on their own. Switch everything off with the last button and tick the single operation you want, or use the page built around it: rejoining broken lines gets its own controls for the hyphen rule and the column width, and removing repeated entries gets a keep-first-or-last choice plus a list of which lines went. The results are identical either way, because both run the same code.
How big a document can it take?
There is no limit, and a book-length manuscript is not a problem. A 100,000-word file runs the full seventeen-operation preset in under half a second, and typing stays responsive throughout because the recomputation happens off the keystroke. Nothing is uploaded, so the size of your text has no effect on anyone but you.
What a paste drags in behind it
Text arriving from somewhere else is never only text. A paragraph out of Word carries curly quotes and em dashes that autocorrect installed while you were looking elsewhere, plus non-breaking spaces holding a figure to its unit. A paragraph out of a web page carries zero-width spaces marking where a long address may wrap, and often a byte order mark at the very front. A paragraph out of a PDF carries breaks at every margin and columns of ordinary spaces standing in for a table. A list out of an email carries bullet characters, trailing spaces on every line and the same entry twice because two people replied. None of it is visible, all of it survives the copy, and each source leaves a different signature — which is why the presets here are named after origins rather than after operations.
Doing these fixes one at a time in different tools produces different results depending on the order, and the differences are not subtle. Delete blank lines before converting unusual spaces and the lines holding a single non-breaking space survive, because they were not blank yet. Collapse runs of spaces before expanding tabs and a tab that should have become four spaces becomes one. Strip punctuation before repairing hyphenated line breaks and the hyphens are gone, so the words can never be rejoined. The pipeline here is fixed for exactly this reason: invisible characters are normalised first, typography second, spacing third, line structure fourth, and the destructive character operations last, when everything else has already settled.
The badges exist because cleaning text is an act of trust, and a tool that silently rewrites a document does not deserve any. Seeing that the passage held 14 non-breaking spaces and 6 curly apostrophes is often more useful than the cleaned output itself: it tells you the file came out of Word rather than a plain editor, and it explains why a search for a phrase kept failing. When one operation is all you need, the section it belongs to goes deeper than a tick box can — the line break remover lets you set the column width and the hyphen rule yourself, and removing duplicate lines there gives you a choice of which copy survives and a list of every entry that repeated.
Nothing you paste here leaves the browser
All seventeen operations are plain JavaScript functions running inside this tab, so a contract, a manuscript or a client list can be cleaned without any of it crossing the network. No text is uploaded, no copy is kept and the tick boxes are not remembered between visits. Once the page has loaded it needs no connection at all, which is worth knowing if the document is the sort you would rather not put through a website.