Find and replace text

Your files and your privacy

The search, the replacement and the text they are applied to all stay in this browser tab. Nothing is uploaded and nothing is autosaved, so a configuration file or a block of source code pasted here leaves no copy once you reload the page or select Clear data.

What this tool does

Search a block of text and replace all matches at once, with optional regular expressions, case sensitivity and whole-word matching. An invalid pattern is reported as a message and your text is left untouched.

It finds every occurrence of a search term, replaces them all in a single pass, and reports how many replacements it made. Three switches change how the term is read: Regex treats it as a regular expression rather than literal characters, Case sensitive is on by default and can be turned off, and Whole word requires a word boundary at each end so a search for cat leaves concatenate alone. A pattern that will not compile produces an error message and no change at all, which matters because a half-typed pattern is a normal step towards a working one.

How to use it

1. Paste your text into the box, then type the search term and the replacement into the two fields.
2. Leave Regex off to match the term literally; every special character in it is escaped, so a full stop matches a full stop and nothing else.
3. Turn Regex on when you need a pattern, and use $1, $2 and so on in the replacement to reference its capture groups.
4. Select Replace all and read the count reported afterwards. If it is not what you expected, select Undo and refine the pattern.

Example use cases

• Swapping a renamed domain, product name or variable throughout a pasted file.
• Reformatting a column of dates with a capture-group pattern, turning 2024-01-02 into 02/01/2024.
• Replacing a word without touching the longer words that contain it, using whole-word matching.

Supported input and output

Accepts

• Plain text in the editor, plus a search term and a replacement string
• JavaScript regular expression syntax, as supported by your browser’s engine

Produces

• The rewritten text in the same box, with a count of replacements made
• A downloadable text.txt, or the text copied to your clipboard

What this tool will not do

• Replace all is the only action: no replace-one, no stepping through matches, no highlighting of what is about to change. Run it and read the count, or undo.
• Only two flags are used: global always, and case-insensitive when you clear the Case sensitive box. There is no multiline, dot-all, sticky or unicode flag, and JavaScript has no inline flag syntax, so a pattern beginning (?m) is rejected as an invalid group.
• Without the multiline flag, ^ and $ anchor to the start and end of the whole text rather than each line, and a full stop does not match a line break.
• The guard catches patterns that will not compile, not patterns that are slow. One that backtracks catastrophically still runs on the main thread and can lock the tab.
• Replacement covers the whole text; there is no way to limit it to a selection, a line range or the first few matches.

Common mistakes

• Putting a dollar sign in the replacement and expecting it to be literal. The replacement string is never escaped, even in literal mode, so $& inserts the matched text and $$ produces a single dollar sign — write $$ when you mean one.
• Turning on Whole word for a term containing an accent or punctuation. Word boundaries are ASCII-based, so searching for café or (cat) with Whole word on matches nothing and reports zero replacements rather than an error.
• Using Unicode property escapes in a pattern. Without the unicode flag \p{L} does not mean ‘any letter’ — it matches the literal characters p{L}, so the replacement lands somewhere you never intended.
• Anchoring with ^ to prefix every line. It matches only the very start of the text, so exactly one replacement is made and every other line is left alone.

Technical notes

In literal mode every regular-expression metacharacter in the search term is escaped before compilation, so it can only match itself; in regex mode the term is passed through as written. Whole word wraps whichever you get in \b(?:…)\b, as a non-capturing group, so an alternation such as cat|dog is bounded as a whole rather than binding the boundary to its first branch. Compilation happens inside a guard that returns an error value instead of throwing, and the engine’s own message is passed through without being double-prefixed. The count is taken by matching before replacing, then the standard string replacement runs — which is why the replacement honours $1 to $9, $& and $$ in both modes: only the search side is escaped.

Frequently asked questions

Which regular expression flavour does it use?

JavaScript’s own, compiled by the engine in the browser you are using. Patterns written for other flavours — PCRE, Python, or POSIX classes such as [:alpha:] — will either fail to compile or match something different.

Do capture groups work in the replacement?

Yes, in regex mode. Find (\d{4})-(\d{2})-(\d{2}) and replace with $3/$2/$1 to rewrite 2024-01-02 as 02/01/2024. A number beyond the groups in your pattern is left as literal text.

What happens if my regular expression is invalid?

You get the engine’s message explaining what was wrong, such as an unterminated group, and your text is left exactly as it was. Nothing is thrown, nothing is cleared, and you can keep editing the pattern.