Text Cleaner

One paste, seven passes

Text that has been somewhere else always arrives carrying the shape of where it came from. A paragraph copied out of a web page brings non-breaking spaces and the indentation of the HTML source. An export from a CMS brings the tags themselves. A chat log brings emoji, a PDF brings its line numbering, a European name list brings accents a legacy field cannot store. Cleaning any one of those is a two-second job. Cleaning all six at once, by hand, in the right order, is not — which is what this page is for.

Every switch starts on. Paste something and the default pass strips the markup, drops the emoji, folds the accents, removes the whole numbers, cuts the punctuation back to sentence endings and apostrophes, removes the symbols outside basic punctuation, and evens out the whitespace. That is a deliberately aggressive default, and it is meant to be edited: the switches are the first thing on the page precisely because the right answer depends on what you are cleaning. Untick Numbers if the dates matter. Untick Punctuation if a human has to read the result.

The order the passes run in is fixed and is not the order of the checkboxes: HTML tags, emoji, accents, numbers, punctuation, special characters, extra spaces. That sequence is doing real work. Markup has to go before punctuation, or <p> loses its angle brackets and leaves a stray p in the middle of the sentence. Accents have to be folded before any ASCII-only pass, or every accented letter is deleted rather than simplified. Whitespace has to be last, because every earlier step leaves gaps where the characters it removed used to be.

The counters underneath show the character count in, the character count out, how many characters went, and the word count either side. Everything runs in your browser: nothing you paste is uploaded, which is the point when the thing you are cleaning is a contract, a transcript, a customer email or an export you are not allowed to put through a web service.

Each pass, on its own page

Every switch above is the same code as the page below it, with more room for the settings that only matter for that one transform — and with the specifics of when it goes wrong.

• Remove Extra Spaces — collapse double spaces, trim line ends, and clear out the non-breaking and zero-width spaces that find-and-replace cannot see.
• Remove Punctuation — commas, quotes, brackets and dashes, with a choice about the apostrophes and hyphens inside words.
• Remove Special Characters — symbols, currency signs and control characters, cut back to letters and digits, to basic punctuation, or to plain ASCII.
• Remove Numbers — every digit, only whole numbers so mp3 and H2O survive, or only the “1.” at the start of each line.
• Remove Emojis — whole clusters, so flags, keycaps, skin tones and family sequences come out in one piece rather than in fragments.
• Remove HTML Tags — strip the markup, keep the line structure, decode the entities and drop the script bodies.
• Remove Accents — é ü ñ to plain letters, including ß, ø and ł, which normalisation alone silently destroys.

Text cleaner questions