I'm interested in selectively parsing Mediawiki XML markup to generate a customized HTML page that's some subset of the HTML produced by the actual PHP Mediawiki render engine.
I want it for BzReader, an offline Mediawiki compressed dump reader written in C#. So a C# parser would be ideal, but any good code would help.
Of course, if no one has done it before, I guess it's time to start a project maintaining a free and separate Mediawiki parser, based on Mediawiki's own parser, but less tightly integrated with Mediawiki itself.
So, does anyone know of any base I could begin with, that would be better than hacking from the Mediawiki PHP code?
There is a list of parsers on http://www.mediawiki.org/wiki/Alternative_parsers, but a c# parser is not included there...
Update
Bare in mind Screwturn doesn't stick to the Mediawiki syntax but uses its own variation which does vary a bit.
The Mediawiki syntax doesn't lend itself to LALR parser (or even LL*) as it has a lot of ambiguities in its definition, and also allows HTML. There's a discussion on that in this question, you're essentially stuck with writing your own parser and tokenizer rather than simply writing a BNF file for it and then using ANTLR/Gold/Irony.
Roadkill Wiki uses a Creole parser for its Mediawiki parsing, but with limited support.
Screwturn is released under the GPL license, and has a C# parser:
- Screwturn license
- Screwturn source download (unfortunately there's no web svn)
The class you are after is Core.Formatter which has lots of regexs to do its work:
public static class Formatter {
}
It's not the nicest looking code "but it works".
I had some words to say about Mediawiki templates here. Interesting that there's a list of alternative parsers now, I'll have to investigate that.
来源：https://stackoverflow.com/questions/324758/open-source-parser-code-for-mediawiki-markup