How to use the unicode normalization comparer
- Type or paste your text.
- Optionally add a second text to compare with the first.
- Read the four normal forms of the first text with their code points, and which ones the text is already in.
- If you added a second text, read whether the two are identical, canonically equivalent or compatibility equivalent.
Formula
NFD: decompose characters into a base letter and combining marks
NFC: decompose, then compose again where a single character exists
NFKD and NFKC: also replace compatibility characters, then decompose or compose
canonically equivalent: NFC forms are equal; compatibility equivalent: NFKC forms are equal
The page uses the Unicode data built into your browser. Other software may use a different Unicode version, which only matters for very new characters.
One letter written two ways
The word café can be typed as four code points, with the single character U+00E9, or as five, with the letter e (U+0065) and a combining acute accent (U+0301). They look the same but are not identical. NFC turns the five code point form into the four code point form, so the two are canonically equal.
The text final ＡＢＣ holds a ligature (U+FB01) and full width capitals. It is already in NFC and NFD, but NFKC turns it into final ABC, which has 9 code points instead of 8. The Korean syllable 한 is one code point (U+D55C) in NFC and three (U+1112 U+1161 U+11AB) in NFD.
Why strings that look the same differ
Unicode lets many accented letters be written either as one precomposed character or as a letter followed by a combining mark. Text typed on different systems, copied from a PDF or stored in a file name can use either one. A program that compares strings code point by code point sees them as different, which can break searches, sorting and logins.
Normalizing both sides to the same form, usually NFC, before comparing makes the result match what people see. The compatibility forms go further and fold characters such as ligatures, circled digits and full width letters into plain ones. That is useful for searching but loses information, so do not use it to store text.
Related tools
To see how many bytes each form takes, use the Text Byte Layout Inspector. To repair text that shows odd symbols instead of accents, use the Text Encoding Fixer. To compare sizes against a limit, use the Text Byte Size Calculator.
Assumptions and limits
- Text is limited to 5,000 characters for each box.
- Code points are shown in the U+XXXX form.
- The result depends on the Unicode version of your browser.
- Combining marks may look misplaced in the results table. This is a display matter and the code points are exact.
- The texts shown when the page opens are examples.
Frequently asked questions
Which form should I store?
NFC is the most common choice for storing and exchanging text, because it keeps text short and matches what most keyboards produce.
What is the difference between canonical and compatibility equivalence?
Canonical equivalence means the same character written two ways, such as an accented letter. Compatibility equivalence also covers characters that differ in form, such as a ligature and its two letters.
Why does the count of code points change?
Decomposing splits a precomposed character into a letter and a mark, so the count goes up. Composing joins them and the count goes down.
Does the page change the text I typed?
No. It shows the normal forms next to your original and marks whether each one changed. Your text is never sent anywhere.