Text looks like the one thing computers cannot get wrong. It is just letters. Yet a character count comes out differently depending on who is counting, uppercasing a word can make it longer, and two lines can be identical on screen and refuse to match. These tools exist because of that gap — each one shows the rule behind the answer rather than handing you a number and hoping you do not ask.
Which one to use
| You want to… | Tool |
|---|---|
| Check a post fits X's 280 limit, with links and emoji counted properly | Word counter |
| Find out why a text message is being sent and billed as three | Word counter — SMS row |
| Know how much slack a word limit really leaves you | Word counter — both word rules |
| Uppercase Turkish text without turning i into the wrong letter | Case converter — Turkish |
| See whether uppercasing will change the length of a value | Case converter |
| Turn a heading into camelCase, snake_case or kebab-case | Case converter |
| Work out why two identical-looking values will not match | Text diff |
| Compare two files and see what really changed | Text diff |
The number depends on who is counting
Ask three programs how long a piece of text is and you can get three answers, all correct. X counts weight rather than characters, so an emoji costs two and a link costs twenty-three no matter how short it is. An SMS gateway counts encoding units, and a single character outside a small set drops the limit from 160 to 70 for the whole message. A word processor counts words, and disagrees with the next word processor about whether a hyphenated term is one word or two.
None of that is obscure, and almost no tool surfaces it. That is the thread running through this category: it is the same problem as an upload form rejecting a file that is exactly the size it allows — the number you were shown and the number being checked came from different rules.
Text carries things you cannot see
Copy a line out of a PDF, a spreadsheet or a web page and it can arrive with passengers: non-breaking spaces dressed as ordinary ones, zero-width characters with no width at all, a byte order mark at the front, Windows line endings that are two characters rather than one, and accented letters stored in either of two ways that render identically.
They break exact matching, they inflate character counts, and a single one of them can push a message into the expensive SMS encoding. Both the word counter and the text diff report them by name instead of cleaning them up quietly, because knowing what is there is the whole job.
Nothing is uploaded
Every tool here runs in your browser. Contracts, client copy, configuration files, names and addresses stay on your machine, which matters more for text than for most things — these are the files people are least comfortable pasting into a stranger's server. There is no size limit imposed by an upload, no account, and no queue.