uconv - convert data from one encoding to another

uconv [ -h, -?, --help ] [ -V, --version ] [ -s, --silent ] [ -v, --verbose ] [ -l, --list | -l, --list-code code | --default-code | -L, --list-transliterators ] [ --canon ] [ -x"transliteration ] [ --to-callback callback | -c ] [ --from-callback callback | -i ] [ --callback callback ] [ --fallback | --no-fallback ] [ -b, --block-size size ] [ -f, --from-code encoding ] [ -t, --to-code encoding ] [ --add-signature ] [ --remove-signature ] [ -o, --output file ] [ file... ]

uconv converts, or transcodes, each given file (or its standard input if no file is specified) from one encoding to another. The transcoding is done using Unicode as a pivot encoding (i.e. the data are first transcoded from their original encoding to Unicode, and then from Unicode to the destination encoding).

If an encoding is not specified or is -, the default encoding is used. Thus, calling uconv with no encoding provides an easy way to validate and sanitize data files for further consumption by tools requiring data in the default encoding.

When calling uconv, it is possible to specify callbacks that are used to handle invalid characters in the input, or characters that cannot be transcoded to the destination encoding. Some encodings, for example, offer a default substitution character that can be used to represent the occurence of such characters in the input. Other callbacks offer a useful visual representation of the invalid data.

uconv can also run the specified transliteration on the transcoded data, in which case transliteration will happen as an intermediate step, after the data have been transcoded to Unicode. The transliteration can be either a list of semicolon-separated transliterator names, or an arbitrarily complex set of rules in the ICU transliteration rules format.

For transcoding purposes, uconv options are compatible with those of iconv(1), making it easy to replace it in scripts. It is not necessarily the case, however, that the encoding names used by uconv and ICU are the same as the ones used by iconv(1). Also, options that provide informational data, such as the -l, --list one offered by some iconv(1) variants such as GNU's, produce data in a slightly different and easier to parse format.

-h, -?, --help Print help about usage and exit.

-V, --version Print the version of uconv and exit.

-s, --silent Suppress messages during execution.

-v, --verbose Display extra informative messages during execution.

-l, --list List all the available encodings and exit.

-l, --list-code code List only the code encoding and exit. If code is not a proper encoding, exit with an error.

--default-code

List only the name of the default encoding and exit.

-L, --list-transliterators List all the available transliterators and exit.

--canon

If used with -l, --list or --default-code, the list of encodings is produced in a format compatible with convrtrs.txt(5). If used with -L, --list-transliterators, print only one transliterator name per line.

-x transliteration

Run the given transliteration on the transcoded Unicode data, and use the transliterated data as input for the transcoding to the the destination encoding.

--to-callback callback

Use callback to handle characters that cannot be transcoded to the destination encoding. See section CALLBACKS for details on valid callbacks.

-c

Omit invalid characters from the output. Same as --to-callback skip.

--from-callback callback

Use callback to handle characters that cannot be transcoded from the original encoding. See section CALLBACKS for details on valid callbacks.

-i

Ignore invalid sequences in the input. Same as --from-callback skip.

--callback callback

Use callback to handle both characters that cannot be transcoded from the original encoding and characters that cannot be transcoded to the destination encoding. See section CALLBACKS for details on valid callbacks.

--fallback

Use the fallback mapping when transcoding from Unicode to the destination encoding.

--no-fallback

Do not use the fallback mapping when transcoding from Unicode to the destination encoding. This is the default.

-b, --block-size size Read input in blocks of size bytes at a time. The default block size is 4096.

-f, --from-code encoding Set the original encoding of the data to encoding.

-t, --to-code encoding Transcode the data to encoding.

--add-signature

Add a U+FEFF Unicode signature character (BOM) if the output charset supports it and does not add one anyway.

--remove-signature

Remove a U+FEFF Unicode signature character (BOM).

-o, --output file Write the transcoded data to file.

uconv supports specifying callbacks to handle invalid data. Callbacks can be set for both directions of transcoding: from the original encoding to Unicode, with the --from-callback option, and from Unicode to the destination encoding, with the --to-callback option.

The following is a list of valid callback names, alonmg with a description of their behavior. The list of callbacks actually supported by uconv is displayed when it is called with -h, --help.

substitute

Write the the encoding's substitute sequence, or the Unicode replacement character U+FFFD when transcoding to Unicode.

skip

Ignore the invalid data.

stop

Stop with an error when encountering invalid data. This is the default callback.

escape

Same as escape-icu.

escape-icu

Replace the missing characters with a string of the format %Uhhhh for plane 0 characters, and %Uhhhh%Uhhhh for planes 1 and above characters, where hhhh is the hexadecimal value of one of the UTF-16 code units representing the character. Characters from planes 1 and above are written as a pair of UTF-16 surrogate code units.

escape-java

Replace the missing characters with a string of the format \uhhhh for plane 0 characters, and \uhhhh\uhhhh for planes 1 and above characters, where hhhh is the hexadecimal value of one of the UTF-16 code units representing the character. Characters from planes 1 and above are written as a pair of UTF-16 surrogate code units.

escape-c

Replace the missing characters with a string of the format \uhhhh for plane 0 characters, and \Uhhhhhhhh for planes 1 and above characters, where hhhh and hhhhhhhh are the hexadecimal values of the Unicode codepoint.

escape-xml

Same as escape-xml-hex.

escape-xml-hex

Replace the missing characters with a string of the format &#xhhhh;, where hhhh is the hexadecimal value of the Unicode codepoint.

escape-xml-dec

Replace the missing characters with a string of the format &#xnnnn;, where nnnn is the decimal value of the Unicode codepoint.

escape-unicode

Replace the missing characters with a string of the format {U+hhhh}, where hhhh is the hexadecimal value of the Unicode codepoint. That hexadecimal string is of variable length and can use from 4 to 6 digits. This is the format universally used to denote a Unicode codepoint in the litterature, delimited by curly braces for easy recognition of those substitutions in the output.

Convert data from a given encoding to the platform encoding:

$ uconv -f encoding

Check if a file contains valid data for a given encoding:

$ uconv -f encoding -c file >/dev/null

Convert a UTF-8 file to a given encoding and ensure that the resulting text is good for any version of HTML:

$ uconv -f utf-8 -t encoding \

--callback escape-xml-dec file

Display the names of the Unicode code points in a UTF-file:

$ uconv -f utf-8 -x any-name file

Print the name of a Unicode code point whose value is known (U+30AB in this example):

$ echo '\u30ab' | uconv -x 'hex-any; any-name'; echo

{KATAKANA LETTER KA}{LINE FEED}

$

(The names are delimited by curly braces. Also, the name of the line terminator is also displayed.)

Normalize UTF-8 data using Unicode NFKC, remove all control characters, and map Katakana to Hiragana:

$ uconv -f utf-8 -t utf-8 \

-x '::nfkc; [:Cc:] >; ::katakana-hiragana;'

uconv does report errors as occuring at the first invalid byte encountered. This may be confusing to users of GNU iconv(1), which reports errors as occuring at the first byte of an invalid sequence. For multi-byte character sets or encodings, this means that uconv error positions may be at a later offset in the input stream than would be the case with GNU iconv(1).

The reporting of error positions when a transliterator is used may be inaccurate or unavailable, in which case uconv will report the offset in the output stream at which the error occured.

Jonas Utterstroem

Yves Arrouye

3.4.1

Copyright (C) 2000-2005 IBM, Inc. and others.

iconv(1)