Man page for cvt-utf8


cvt-utf8 – convert between UTF-8 and Unicode, and analyse Unicode


cvt-utf8 [flags] [hex UTF-8 bytes, U+codepoints, SGML entities]


cvt-utf8 is a tool for manipulating and analysing UTF-8 and Unicode data. Its functions include:

By default, cvt-utf8 expects to receive character data on the command line (as a mixture of UTF-8 bytes, Unicode code points and SGML numeric character entities), and it will print out a verbose analysis of the input data. If you need it to read UTF-8 from standard input or to write pure UTF-8 to standard output, you can do so using command-line options.


Read UTF-8 data from standard input and analyse that, instead of expecting hex numbers on the command line.
Write well-formed UTF-8 to standard output, instead of writing a long analysis of the input data.
Look up each code point in the Unihan database as well as the main Unicode character database.


In cvt-utf8's native mode, it simply analyses input Unicode or UTF-8 data. For example, you can give a list of Unicode code points...

$ cvt-utf8 U+20ac U+31 U+30
U-000020AC  E2 82 AC          EURO SIGN
U-00000031  31                DIGIT ONE
U-00000030  30                DIGIT ZERO

... and cvt-utf8 gives you the UTF-8 encodings plus the character definitions.

If it's more convenient, you can specify those characters as SGML numeric entity references (for example if you're cutting and pasting out of a web page):

$ cvt-utf8 '€' '–'
U-000020AC  E2 82 AC          EURO SIGN
U-00002013  E2 80 93          EN DASH

Alternatively, you can supply a list of UTF-8 bytes...

$ cvt-utf8 D0 A0 D1 83 D1 81 D1 81 D0 BA D0 B8 D0 B9
U-00000420  D0 A0             CYRILLIC CAPITAL LETTER ER
U-00000443  D1 83             CYRILLIC SMALL LETTER U
U-00000441  D1 81             CYRILLIC SMALL LETTER ES
U-00000441  D1 81             CYRILLIC SMALL LETTER ES
U-0000043A  D0 BA             CYRILLIC SMALL LETTER KA
U-00000438  D0 B8             CYRILLIC SMALL LETTER I
U-00000439  D0 B9             CYRILLIC SMALL LETTER SHORT I

... and you get back the same output format, including the UTF-8 code points.

If you supply malformed data, cvt-utf8 will break it down for you and identify the malformed pieces and any correctly formed characters:

$ cvt-utf8 A9 FE 45 C2 80 90 0A
            A9                (unexpected continuation byte)
            FE                (invalid UTF-8 byte)
U-00000045  45                LATIN CAPITAL LETTER E
U-00000080  C2 80             <control>
            90                (unexpected continuation byte)
U-0000000A  0A                <control>

If you need the UTF-8 encoding of a particular character, you can use the -o option to cause the UTF-8 to be written to standard output:

$ cvt-utf8 -o U+20AC >> my-utf8-file.txt

If you have UTF-8 data in a file or output from another program, you can use the -i option to have cvt-utf8 analyse it. This works particularly well if you also have my xcopy program, which can be told to extract UTF-8 data from the X selection and write it to its standard output. With these two programs working together, if you ever have trouble identifying some text in a UTF-8-supporting web browser such as Mozilla, you can simply select the text in question, switch to a terminal window, and type

$ xcopy -u -r | cvt-utf8 -i

If the text is in Chinese, you can get at least a general idea of its meaning by using the -h option to print the meaning of each ideograph from the Unihan database. For example, if you pass in the Chinese text meaning ‘Traditional Chinese’:

$ cvt-utf8 -h U+7E41 U+9AD4 U+4E2D U+6587
U-00007E41  E7 B9 81          <han> complicated, complex, difficult
U-00009AD4  E9 AB 94          <han> body; group, class, body, unit
U-00004E2D  E4 B8 AD          <han> central; center, middle; in the
                              midst of; hit (target); attain
U-00006587  E6 96 87          <han> literature, culture, writing


In order to print the official name of each character, cvt-utf8 requires a file mapping code points to names. This file is in DBM database format, for rapid lookup.

This database file is accessed using the Python anydbm module, so its precise file name will vary depending on what flavours of DBM you have installed. The name Python knows it by is ‘unicode’; it may actually be called ‘unicode.db’ or something similar.

cvt-utf8 generates this DBM file itself starting from the Unicode Character Database, in the form of the file UnicodeData.txt supplied by It supports two administrative options for this purpose:

cvt-utf8 --build /path/to/UnicodeData.txt /path/to/unicode

Given a copy of UnicodeData.txt on disk, this mode will create the DBM file and store it in a place of your choice.

cvt-utf8 --fetch-build /path/to/unicode

If you have a direct Internet connection, this will automatically download the text file from and process it straight into the DBM file.

There is a second DBM file, known to Python as unihan, which is required to support the -h option. This one is built from the Unihan Database, distributed by as a zip file containing a text file Unihan.txt.

If you already have Unihan.txt on your system, you can build cvt-utf8's unihan DBM file like this:

cvt-utf8 --build-unihan /path/to/Unihan.txt /path/to/unihan

Or, again, cvt-utf8 can automatically download it from, unpack the zip file on the fly, and write the DBM straight out:

cvt-utf8 --fetch-build-unihan /path/to/unihan

cvt-utf8 expects to find these database files in one of the following locations:


If either of these files is not found, cvt-utf8 will still perform the rest of its functions.


cvt-utf8 is free software, distributed under the MIT licence. Type cvt-utf8 --licence to see the full licence text.

[cvt-utf8 version 20230903.c678881]