Text, data and web tools that show their work

Start with the counter

Paste anything and watch it measure characters, words, sentences and UTF-8 bytes at once. It is the same engine the full character and word counter uses.

The byte figure is the one that surprises people: Vietnamese, Thai and emoji characters take two to four bytes each, so a field that looks well under its limit can still be over it.

English Tiếng Việt ไทย Emoji

Text to measure Hello world. This sentence counts too.

Characters0

Words0

Sentences0

UTF-8 bytes0

🔒 Everything runs on this device — nothing is sent anywhere. Open the full tool →

Code playgrounds

Run HTML, JavaScript, TypeScript, Python, Lua and SQL in the browser. Each page states what runs locally, what is downloaded, and where the sandbox ends.

</> HTML, CSS & JavaScript Playground Write HTML, CSS and JavaScript and see the rendered result instantly in a sandboxed preview, with console output and uncaught errors shown live. JS JavaScript Playground Run JavaScript in a sandboxed browser tab with console output, uncaught errors and top-level await — no account, upload or server. TS TypeScript Playground Write TypeScript, strip its types with Sucrase and run the JavaScript in a sandboxed tab — execution without a server, with type-checking limits explained. Py Python Playground Run Python in your browser with Pyodide — standard library, stdin and tracebacks, with no account and no code upload. Lua Lua Playground Run Lua 5.3 in a sandboxed browser tab through Fengari, with print output and errors shown live — no install and no upload. SQL SQL Playground (SQLite) Run SQL against an in-memory SQLite database — create tables, insert rows and see result sets, with no server and no upload.

Browse by family

Each family is a hub page with the tools it contains and the rule behind each one.

See all 108 tools →

Read before you convert

The rules behind the tools, and the record of what broke while building them.

Guide · Text Invisible characters that break lookups Zero-width spaces, BOMs and soft hyphens: why identical strings compare unequal. Read the guide → Guide · Encoding Base64 is not encryption — and not URL encoding What each encoding is for, what it does not protect, and when the wrong one corrupts data. Read the guide → Guide · Data CSV quoting, the rule that corrupts quietly Commas, quotes and newlines inside fields: what RFC 4180 actually requires. Read the guide → Build log What the tests caught Six failures that reached the released build, and how each was fixed and verified. Open the log →

What these tools are for

Values that look correct on screen are frequently wrong as data. A Thai ID number written with Thai digits will not fill a column that expects 0–9. A value copied from a web page carries a non-breaking space, so a lookup silently fails. Two strings that render identically differ because one uses a precomposed character and the other a base letter plus a combining mark. A JSON payload arrives as one unreadable line, and a Unix timestamp means a different instant depending on the zone it is read in. These are not exotic problems; they are the ordinary friction of moving text and values between systems built by different people.

Each tool does one of those jobs, states the rule it applies and shows a worked example. That last part matters: a converter you cannot check is a converter you have to trust, and you should not have to.

How we verify accuracy

Every tool ships with an automated test set that includes the examples shown on its page. Where a rule has an authoritative definition, the test asserts that definition rather than our own expectation — the Unicode digit blocks, the Unicode normalization forms, the canonical forms of Roman numerals. The methodology page lists the standard behind each tool, and the editorial policy describes how content is written, reviewed and corrected, including how we use automation.

The longer explanations live in the guides: invisible characters that break lookups, Base64 versus URL encoding, CSV quoting rules, why decoding a JWT is not verifying it, CIDR subnets, and UTF-8 bytes versus characters.

The decisions and bugs behind the tools are recorded in the build log: what broke, what changed, and how the fix was checked before it shipped.

Frequently asked questions