Grapheme Cluster Viewer

Inspect browser-segmented grapheme clusters alongside their code points and UTF-16 offsets, including combining marks and emoji.

Text processing stays in this browser. Input is limited to 20,000 UTF-16 code units.

Uses Intl.Segmenter with grapheme granularity. Browser Unicode data can change boundaries. This is not a count of words, screen columns or glyphs.

Inspect groups that users often perceive as characters

A grapheme cluster can contain one code point or a sequence such as a letter with combining marks. Joined emoji can also form a cluster. This viewer uses the browser's Intl.Segmenter with grapheme granularity and shows the code points inside each returned segment. It is useful for investigating why a string's JavaScript length differs from a character count shown in an interface.

Try an accent and a joined emoji

Choose Load example (replaces input and settings) and then Process text. The example is A👨‍👩‍👧‍👦é, with a decomposed accent at the end. In a browser that recognizes the family sequence, the report contains three clusters:

Cluster UTF-16 offset UTF-16 units Contents
1 0 1 A
2 1 11 Four people joined by three U+200D characters
3 12 2 e followed by U+0301

The complete string contains ten code points and fourteen UTF-16 code units. Those counts answer different questions. The report's positions start at one, while offsets start at zero in the received string. The code-point column lists every value within the segment; it does not assign invented names to emoji sequences.

Understand what the boundaries mean

Grapheme segmentation is not a word count, a pixel-width calculation or a prediction of how many glyphs a font will draw. A browser may segment an emoji sequence even when the installed font cannot render it as one picture. This tool uses English as the segmenter's locale and does not provide language-specific editorial character-count rules.

The implementation relies on Intl.Segmenter; it shows a support error if that API is unavailable instead of silently switching to code-point counting. Unicode data and segmentation behavior can differ between browser versions, particularly for newer sequences. MDN describes Intl.Segmenter, while Unicode Standard Annex #29 defines text segmentation and grapheme boundaries. Record the browser version when reproducing a boundary difference.

Export a useful diagnostic report

The preview shows at most 200 rows; Copy full result and Download result include the full TSV report. JSON-quoted display cells escape tabs and newlines so those characters do not split report records. Input is limited to 20,000 UTF-16 code units. Unpaired surrogates are rejected, and pasted textarea line endings may be normalized to LF before inspection.

If a cluster looks unexpected, inspect its constituent values with the Unicode Code Point Inspector. Use the Unicode Normalization Tool only when you intend to change representation; this viewer itself leaves the input unchanged. A combining mark without a preceding base can appear in its own segment, which does not by itself prove that the source text is invalid.