7.4.3 The filter

The filter can be any program that reads a translation from standard input and writes a modified translation to standard output. A frequently used filter is `sed'.

Note: It is your responsibility to ensure that the filter can cope with input encoded in the translation catalog's encoding. If the filter wants input in a particular encoding, you can in a first step convert the translation catalog to that encoding using the `msgconv' program, before invoking `msgfilter'. If the filter wants input in the locale's encoding, but you want to avoid the locale's encoding, then you can first convert the translation catalog to UTF-8 using the `msgconv' program and then make `msgfilter' work in an UTF-8 locale, by using the LC_ALL environment variable.

Note: Most translations in a translation catalog don't end with a newline character. For this reason, it is important that the filter recognizes its last input line even if it ends without a newline, and that it doesn't add an undesired trailing newline at the end. The `sed' program on some platforms is known to ignore the last line of input if it is not terminated with a newline. You can use GNU sed instead; it does not have this limitation.