PrinceXML not working for unicode

Export an XHTML and open it in a web browser. Does it do the same thing?

Tools:

Open the generated PDF in one of those two tools and check if the fonts are embedded. The font-family property has several font names as a “fallback” system. If the first font is not found, or does not contain a character, then it tries the next font. It is possible the path to the tamil font is not found when exporting (open_basedir restriction in effect?)