Content translation/cxserver

cxserver is a Nodejs based stateless server powering the Content Translation. It does the following:

1. Fetch the page html for given title and language from a configured parsoid instance, segment the html content to sections and sentences.
2. Provide the MT backend. Expose a generic MT API, the backend of that act as bridge to multiple Machine Translation services hosted outside cxserver such as Apertium. It also transfers the mark from source HTML to translated HTML in case MT services are capable of only plain text translation.
3. Provide Dictionary API for dictionary look up - either through the dictd, json dictionary clients bundled in cxserver or by contacting third party services.
4. Tools listing api for a language pair - listing what are the translation tools available for the language pair.

CXServer does not handle any user information, does not require any authentication. Does not contact any MediaWiki database directly.

API Documentation

• https://cxserver.wikimedia.org/?doc

See also

• mediawiki/services/cxserver (browse)
• Content translation/cxserver/Logging
• Content translation/cxserver/Setup
• Deployment how-to