Usage:
There are also some tools for experiments:
We are currently focusing on round-tripping of basic formatting like inline/bold, headings, lists, tables and links. Templates, citations and thumbnails are not expected to round-trip properly yet. Please report issues you see at :mw:Talk:Parsoid/Todo. Thanks!
'); }); var htmlSpecialChars = function ( s ) { return s.replace(/&/g,'&') .replace(/');
new WikitextSerializer({env: env}).serializeDOM(
p.tree.document.childNodes[0].childNodes[1],
function( c ) {
res.write( htmlSpecialChars( c ) );
});
res.write('');
res.write( "' + htmlSpecialChars( out ) + '
' + patch);
// Add a 'report issue' link
res.end('
'+
'' +
'Report a parser issue in this page at ' +
''+
'[[:mw:Talk:Parsoid/Todo]]
');
};
var parse = function ( req, res, cb, src ) {
var parser = parserPipelineFactory.makePipeline( 'text/x-mediawiki/full' );
parser.on('document', cb.bind( null, req, res, src ) );
try {
res.setHeader('Content-Type', 'text/html; charset=UTF-8');
parser.process( src );
} catch (e) {
console.log( e );
res.end( e );
}
};
/**
* Round-trip article testing
*/
app.get( new RegExp('/_rt/(?:(?:(?:' + env.interwikiRegexp + '):+)?(' + env.interwikiRegexp + '):)?(.*)') , function(req, res){
env.pageName = req.params[1];
if ( req.params[0] ) {
env.wgScriptPath = '/_rt/' + req.params[0] + ':';
env.wgScript = env.interwikiMap[req.params[0]];
} else {
env.wgScriptPath = '/_rt/';
env.wgScript = env.interwikiMap[config.defaultInterwiki];
}
if ( env.pageName === 'favicon.ico' ) {
res.end( 'no favicon yet..' );
return;
}
var target = env.resolveTitle( env.normalizeTitle( env.pageName ), '' );
console.log('starting parsing of ' + target);
var tpr = new TemplateRequest( env, target );
tpr.once('src', parse.bind( null, req, res, roundTripDiff ));
});
/**
* Round-trip article testing with newline stripping for editor-created HTML
* simulation
*/
app.get( new RegExp('/_rtve/(?:(?:(?:' + env.interwikiRegexp + '):+)?(' + env.interwikiRegexp + '):)?(.*)') , function(req, res){
env.pageName = req.params[1];
if ( req.params[0] ) {
env.wgScriptPath = '/_rtve/' + req.params[0] + ':';
env.wgScript = env.interwikiMap[req.params[0]];
} else {
env.wgScriptPath = '/_rtve/';
env.wgScript = env.interwikiMap[config.defaultInterwiki];
}
if ( env.pageName === 'favicon.ico' ) {
res.end( 'no favicon yet..' );
return;
}
var target = env.resolveTitle( env.normalizeTitle( env.pageName ), '' );
console.log('starting parsing of ' + target);
var tpr = new TemplateRequest( env, target ),
cb = function ( req, res, src, document ) {
// strip newlines from the html
var html = document.innerHTML.replace(/[\r\n]/g, ''),
p = new html5.Parser();
p.parse( html );
var newDocument = p.tree.document;
// monkey-patch document.body reference for now..
newDocument.body = newDocument.childNodes[0].childNodes[1];
roundTripDiff( req, res, src, newDocument );
};
tpr.once('src', parse.bind( null, req, res, cb ));
});
/**
* Form-based round-tripping for manual testing
*/
app.get(/\/_rtform\/(.*)/, function(req, res){
env.pageName = req.params[0];
res.setHeader('Content-Type', 'text/html; charset=UTF-8');
res.write( "Your wikitext:" );
textarea( res );
res.end('');
});
app.post(/\/_rtform\/(.*)/, function(req, res){
env.pageName = req.params[0];
res.setHeader('Content-Type', 'text/html; charset=UTF-8');
// we don't care about \r, and normalize everything to \n
parse( req, res, roundTripDiff, req.body.content.replace(/\r/g, ''));
});
/**
* Regular article parsing
*/
app.get(new RegExp( '/(?:(?:(?:' + env.interwikiRegexp + '):+)?(' + env.interwikiRegexp + '):)?(.*)' ), function(req, res){
env.pageName = req.params[1];
if ( req.params[0] ) {
env.wgScriptPath = '/' + req.params[0] + ':';
env.wgScript = env.interwikiMap[req.params[0]];
} else {
env.wgScriptPath = '/';
env.wgScript = env.interwikiMap[config.defaultInterwiki];
}
if ( env.pageName === 'favicon.ico' ) {
res.end( 'no favicon yet..');
return;
}
var target = env.resolveTitle( env.normalizeTitle( env.pageName ), '' );
console.log('starting parsing of ' + target);
var tpr = new TemplateRequest( env, target );
tpr.once('src', parse.bind( null, req, res, function ( req, res, src, document ) {
res.end(document.body.innerHTML);
}));
});
/**
* Regular article serialization using POST
*/
app.post(/\/(.*)/, function(req, res){
env.pageName = req.params[0];
env.wgScriptPath = '/';
res.setHeader('Content-Type', 'text/x-mediawiki; charset=UTF-8');
var p = new html5.Parser();
p.parse( req.body.content );
new WikitextSerializer({env: env}).serializeDOM(
p.tree.document.childNodes[0].childNodes[1],
res.write.bind( res ) );
res.end('');
});
console.log( ' - ' + instanceName + ' ready' );
module.exports = app;