Unicode Character Inspector

Every character in this tool gets a row showing: the character itself, the codepoint in U+HHHH hex, the UTF-8 byte sequence, an HTML decimal entity (&#NNNN;), a CSS escape (\HHHH), and the Unicode block it belongs to. Useful for debugging mojibake, finding the exact codepoint of a confusing character (is that a hyphen-minus or an em dash?), or seeing how many bytes your string takes in UTF-8 storage. Handles surrogate pairs correctly using Array.from for proper codepoint iteration.

Surrogate pairs are the reason the iteration method matters. JavaScript strings are sequences of 16-bit units, so anything above U+FFFF, which includes every emoji and a great deal of CJK, occupies two of them. Indexing such a string with [0] or measuring it with .length counts halves rather than characters, and slicing between the halves produces a lone surrogate that renders as a replacement box. Array.from walks codepoints instead, which is why a four-byte character shows here as one row rather than two broken ones.

A codepoint is still not a character in the sense a reader means. An emoji with a skin tone is a base plus a modifier, a family emoji is several people joined by zero-width joiners, and an accented letter may be one precomposed codepoint or a letter plus a combining mark. All of those are one thing on screen and several rows here, which is exactly the information you want when a string compares unequal to one that looks identical.

The UTF-8 byte count is where storage limits bite. ASCII takes one byte, Latin letters with accents and Greek and Cyrillic take two, most CJK and the rest of the BMP take three, and emoji take four. A database column declared as 255 characters in UTF-8 may reserve 1020 bytes, and a field limited to 255 bytes holds far fewer than 255 characters of anything but English.

Mojibake has a readable signature once the bytes are visible. Text that shows ’ where an apostrophe belongs is UTF-8 read as Windows-1252, and the row for those characters makes the original bytes plain.