html2pot and po2html

html2pot and po2html are command-line tools for translating web sites. They work in conjunction with all tools that can handle po files: GNU gettext, KDE's Lokalize, etc... The purpose is to separate the contents to be translated from the markup. These tools are published under GPL license version 2.

Advantages:

• avoid being annoyed by the HTML markup when translating web sites;
• be sure to have exactly the same structure of the web pages in the translation as in the original;
• translate only once a text that occurs at several places on several web pages;
• translate only the changed strings when the web site evolves;
• have the comfort of modern translation tools like Lokalize;
• have an exact estimation of the quantity of text to translate;
• translate at different rythms in many languages, do not stop updating the web site while translating;
• publish web sites that are only half-translated.

Requirements:

• To work, html2pot and po2html need a bash shell (as found on Linux, Mac OS X, etc), basic utilities like awk and sed, and a XSLT processor like libXSLT.
• html2pot and po2html have a major limitation / requirement. The web site must be "clean XHTML", i.e. validate against XHTML DTD. You can check that on W3C validator. Well, on one hand, that's not that much a requirement, but a guarantee of quality for your web site ;-).
• Another current limitation is that those scripts know little about PHP, javascript, and other stuff embedded in your web site. That could change in the future, however.

Work cycle:

1. Modify the web pages with your favourite HTML editing tools. Make sure they validate against XHTML.
2. Extract the po template: html2pot *.html > en.pot
3. Create first empty translation: cp en.pot de.po
   or merge with previous translations: msgmerge de.po en.pot > de.new; mv de.new de.po
4. Translate the new or fuzzy strings: kbabel de.po &
5. Publish the translation: for file in *.html; do po2html $file de.po > de/$file; done

Usage:

• html2pot file1.html file2.html ... > template.pot
  extract the strings to be translated
• po2html original.html translation.po > translation.html
  merge the translation with the HTML skeleton

Known bugs:

• XML and DTD declarations get lost during the transformation and you might want to re-add them manually.
• HTML comments like <-- --> get lost. It's a problem for embedded javascript code, for example.
• Whitespace at the beginning of preformatted lines in <pre> tags gets lost.
• Shell part is hard to maintain and slow. Needs to be rewritten from scratch (in C?).

Download:
Latest version is version number 0.3:

• Shell and XSLT sources (23 002 bytes)

Credits:
Many thanks to Stephan Kulow for the inspiration (xml2pot and po2xml scripts for DocBook)!

If you are illegally collecting email addresses, just follow this link (explanations here).

Last updated: 2011-01-11. Maintained by webmaster@bureau-cornavin.com.