Apache Module mod_charset_lite
Summary
mod_charset_lite
allows the server to change the character set of responses before sending them to the client. In an EBCDIC environment, Apache always translates HTTP protocol content (e.g. response headers) from the code page of the Apache process locale to ISO-8859-1, but not the body of responses. In any environment, mod_charset_lite
can be used to specify that response bodies should be translated. For example, if files are stored in EBCDIC, then mod_charset_lite
can translate them to ISO-8859-1 before sending them to the client.
This module provides a small subset of configuration mechanisms implemented by Russian Apache and its associated mod_charset
.
Common Problems
Invalid character set names
The character set name parameters of CharsetSourceEnc
and CharsetDefault
must be acceptable to the translation mechanism used by APR on the system where mod_charset_lite
is deployed. These character set names are not standardized and are usually not the same as the corresponding values used in http headers. Currently, APR can only use iconv(3), so you can easily test your character set names using the iconv(1) program, as follows:
iconv -f charsetsourceenc-value -t charsetdefault-value
Mismatch between character set of content and translation rules
If the translation rules don't make sense for the content, translation can fail in various ways, including:
- The translation mechanism may return a bad return code, and the connection will be aborted.
- The translation mechanism may silently place special characters (e.g., question marks) in the output buffer when it cannot translate the input buffer.
CharsetDefault Directive
The CharsetDefault
directive specifies the charset that content in the associated container should be translated to.
The value of the charset argument must be accepted as a valid character set name by the character set support in APR. Generally, this means that it must be supported by iconv.
Example
<Directory /export/home/trawick/apacheinst/htdocs/convert>
CharsetSourceEnc UTF-16BE
CharsetDefault ISO-8859-1
</Directory>
CharsetOptions Directive
The CharsetOptions
directive configures certain behaviors of mod_charset_lite
. Option can be one of
DebugLevel=n
- The
DebugLevel
keyword allows you to specify the level of debug messages generated by mod_charset_lite
. By default, no messages are generated. This is equivalent to DebugLevel=0
. With higher numbers, more debug messages are generated, and server performance will be degraded. The actual meanings of the numeric values are described with the definitions of the DBGLVL_ constants near the beginning of mod_charset_lite.c
. ImplicitAdd | NoImplicitAdd
- The
ImplicitAdd
keyword specifies that mod_charset_lite
should implicitly insert its filter when the configuration specifies that the character set of content should be translated. If the filter chain is explicitly configured using the AddOutputFilter
directive, NoImplicitAdd
should be specified so that mod_charset_lite
doesn't add its filter. TranslateAllMimeTypes | NoTranslateAllMimeTypes
- Normally,
mod_charset_lite
will only perform translation on a small subset of possible mimetypes. When the TranslateAllMimeTypes
keyword is specified for a given configuration section, translation is performed without regard for mimetype.
CharsetSourceEnc Directive
The CharsetSourceEnc
directive specifies the source charset of files in the associated container.
The value of the charset argument must be accepted as a valid character set name by the character set support in APR. Generally, this means that it must be supported by iconv.
Example
<Directory /export/home/trawick/apacheinst/htdocs/convert>
CharsetSourceEnc UTF-16BE
CharsetDefault ISO-8859-1
</Directory>
The character set names in this example work with the iconv translation support in Solaris 8.