One of PHPs most glaring weaknesses is poor support for various character sets. As long as a string is ASCII text, no problems occur. However, there is no simple and transparent way, using the standard PHP string functions, to properly handle multi-byte encodings such as the BIG5 encoding used for the Chinese language.

This is a problem especially for XML parsers like DOMIT!, since according to the XML specification they are required to be unicode compliant.

One way of resolving this is using the mb_string extension. This will in the near future be integrated with DOMIT!, but the one disadvantage of such an approach is that mb_string must be installed separately from the base PHP distribution, which may not be possible with some web hosting providers.

The loadXML_utf8, parseXML_utf8, and saveXML_utf8 methods were originally intended to get around some of these problems, but they have not proved to be very effective and are currently not recommended for use.

A partial, though imperfect, solution to multibyte encodings has been introduced as of DOMIT! 0.95. This is the appendEntityTranslationTable method.

The appendEntityTranslationTable method allows you to encode non-ASCII characters as XML entities in your xml document, and provide a translation table at parse time to convert these entities into their equivalent characters. It assumes that the document being loaded has all non-ASCII characters converted into entities.

For example, the following code provides an (abridged) entity translation table for accented French characters:

$xmldoc =& new DOMIT_Document();
$xmldoc->appendEntityTranslationTable(array('é' => 'é'));
$xmldoc->loadXML('francais.xml', true);

When the document is saved using the saveXML method, all characters in the translation table are automatically converted back into entities.