Find and Replace Text Online

How it works

Both modes compile down to the same thing: a regular expression executed by the browser's own engine. In plain-text mode, your search string is first escaped so that characters like ., $, and ( lose their special meanings — the pattern that runs matches exactly the characters you typed. In regex mode your pattern is passed through untouched, and the replacement string keeps its full substitution grammar: $1 for groups, $& for the whole match, $<name> for named groups.

The whole-word option does not use the textbook \b boundary. \b defines a word as ASCII letters, digits, and underscore, which makes it wrong for most of the world's text — it would happily match "cafe" inside "café" because é looks like a boundary to it. Instead the tool wraps the search in Unicode-aware lookarounds, so a word is bounded by anything that is not a letter, digit, or underscore in any script.

Counting is a separate pass from replacing. The engine reports every match first, then performs the substitution natively — so the count you see and the substitution semantics you get are both exactly the platform's, with no approximation layered in between.

Choosing between the two modes

Plain text is the right default. It cannot surprise you: what you type is what changes, and nothing else. Reach for regular expressions when the thing you want to change is a shape rather than a string — every date, every price, every trailing "(draft)" whatever precedes it. The moment you find yourself running plain replacements repeatedly with tiny variations, that repetition is usually one regex.

When you'd use this

Fixing a name misspelled forty times in a document that only exists as an email. Reordering date formats before a spreadsheet import, as in the first example. Stripping a prefix from every line of a log excerpt. Renaming a variable in a code snippet where "whole words only" keeps count from becoming counter. Doing any of these on a machine where you cannot install anything.

Read the count before you take the output

The match count is the check that makes a bulk replacement safe, and it costs nothing to look at.

A replacement is one action affecting an unknown number of places, and the failure is silent in both directions. Too many matches and you have changed something you did not mean to; too few and the job is half done and looks finished. Neither announces itself, because you are reading the output for the places you expected to change.

So form an expectation first. If you believe a name appears about forty times and the count says 41, that is worth ten seconds — perhaps it appears in a URL or an email address you did not think of. If it says 200, stop: the search is matching inside longer words, and whole-word mode is probably what you wanted. If it says 3, the text may use a different spelling or a non-breaking space where you typed a plain one.

The count is also how you test a regex safely. Write the pattern, look at the number, and only then run it. A pattern that matches everything usually reports a count that is obviously wrong before it has done any damage.

Order matters when you run several replacements

Chained replacements are not independent, and the second one sees the output of the first.

Swap A for B and then B for C, and everything that was originally A is now C too — the second pass cannot tell your new Bs from the ones that were already there. This is the standard way a two-step rename produces a result nobody wanted, and it is easy to miss because each step is individually correct.

The usual fix is to make the intermediate value one that cannot occur in the text. Replace A with a marker no document would contain, do the other substitutions, then replace the marker last. Ugly, and reliable.

Where possible, do it in one pass instead. A regex alternation with a single replacement handles cases the chain gets wrong, because each position in the text is considered once and only once.

And with plain-text replacements, the more specific one goes first. Replacing a short string before a longer one that contains it means the longer one no longer exists by the time its turn comes.