Unicode Code Point Inspector

What this tool does

This Unicode code point inspector walks a pasted string one code point at a time. Each row shows an index, a glyph or the words not visible, the code point as U+ hex, the UTF-8 bytes of that scalar, and a name only when the character is on the fixed list shipped with the page. Every other row says No name in this tool. The summary counts code points, UTF-16 code units, and UTF-8 bytes. The UTF-8 total uses TextEncoder on the whole string.

A surrogate pair, such as a single emoji above the basic multilingual plane, is one row. A lone surrogate, a half of a pair with no partner, is an error row. That row has no UTF-8 bytes. The name cell says it is not a Unicode scalar. TextEncoder replaces a lone surrogate with U+FFFD when it counts bytes for the whole string, and the summary says so when that happens. The page does not pretend the lone surrogate has a UTF-8 encoding of its own.

Names exist for tab, line feed, carriage return, ordinary space, no-break space, soft hyphen, the Ogham space mark, the Mongolian vowel separator, the en and em quads and spaces, the thin and hair spaces, the zero-width space, joiner, and non-joiner, the directional marks and embeddings, line and paragraph separators, the narrow no-break space, the word joiner, the directional isolates, the ideographic space, and U+FEFF. Those rows say not visible, because a blank glyph would hide the character you are trying to find. Characters outside that list keep their glyph and the sentence No name in this tool.

The split is by code point, not by grapheme cluster. An emoji sequence that uses a zero-width joiner stays several rows, and the joiner row is named. The page is not a spoofing detector. It does not tell you that two letters look alike, and it does not fetch a name from a service. Markup in the text is text. A tag is a series of code points, not a node the page inserts.

How to use

1. Paste or type the text. The field is a textarea.
2. Choose Inspect. Enter inside the textarea does not run Inspect, so you can type a line break.
3. Copy writes the summary and the rows. Clear empties the field and the table.

Empty text is an error and clears the table. More than 20,000 UTF-16 code units, or more than 2,000 code points, is an error and clears the table. The page does not show a partial list. Nothing is stored.

Example

The letter A is one code point, U+0041, UTF-8 byte 41, one UTF-16 unit, and one UTF-8 byte. The name cell says No name in this tool. A no-break space is not visible, U+00A0, UTF-8 bytes c2 a0, named NO-BREAK SPACE. A grinning face emoji is one code point, U+1F600, two UTF-16 units, and UTF-8 bytes f0 9f 98 80, with no name in this tool.

A family emoji built with zero-width joiners is more than one row. One of those rows is the joiner, named ZERO WIDTH JOINER. A string of 2,000 code points is accepted. A string of 2,001 is rejected and the table is cleared. A string that is only a lone surrogate shows one error row with an empty UTF-8 cell, and the UTF-8 total is 3 because TextEncoder writes the replacement character.

Limits

• Code points, not grapheme clusters. Emoji sequences stay split.
• Names only for the fixed whitespace and invisible list. Not a complete name database and not a spoofing check.
• Caps are 20,000 UTF-16 code units and 2,000 code points. A textarea turns CR into LF before Inspect.

The text stays in the browser. There is no network lookup and no storage. If you need a carriage return to survive as U+000D, this textarea cannot show it, because the browser normalizes that character to a line feed before the script reads the value.