doc_converter.js

process.on('uncaughtException', function uncaughtException(e) {
    if(outfile && outfile.writable) {
        outfile.end();
        outfile.destroy();
    }
    console.error(e.message);
    console.error(e.stack);
})

var outfile;

var fs = require('fs')
    , _ = require('underscore')._
    , xml = require('libxmljs')
    , dominiq  = require('node-sizzle/dominiq')
    , sizzFact = require('node-sizzle');

_ = require('underscorex').extend(_);
xml = require('xmlx').extend(xml);

Object.getAllPropertyNames = function getAllPropertyNames(obj) {
    var keys = [];
    while(obj && obj !== Object) {
        keys = keys.concat(Object.getOwnPropertyNames(obj));
        obj = Object.getPrototypeOf(obj);
    }
    return keys;
}

var pages = ['element'
             , 'htmldocument'
             , 'saxparser'
             , 'syntaxerror'
             , 'attribute'
             , 'saxpushparser'
             , 'xmldocument'
             , 'namespace'
             , 'parser'];


_.each(pages, function pageLoop(page, idx, pages) {
    var page = page;
    fs.readFile(page, 'utf8', function fileProcessor(err, htmlstr) {
        if(err) throw err;

    console.warn(page);

        var doc = xml.parseHtmlString(htmlstr);

        if(doc) {
            outfile = fs.createWriteStream(page + '.md', {encoding:'utf8'
                                                              , mode: 0775});
            outfile.writeln = function writeln(msg) { this.write(msg + '\n'); };

            var content = doc.get('//body//*[@id="content"]');

            var title = content.get('h1');
            outfile.writeln('# ' + title.text() + '\n');
            var desc = title.nextElement();
            outfile.writeln(desc.text() + '\n');

            var children = content.childElements();
            _.each(children, function childLoop(child, idx) {
                console.warn('child: ' + child.name());
                var cur, cur2;
                switch(child.name()) {
                case 'h2':
                    if(child.text().indexOf('Instantiation') !== -1) {
                        outfile.writeln('### Constructor\n');
                    }
                    if(child.text().indexOf('Methods') !== -1) {
                        outfile.writeln('### Methods\n');
                    }
                case 'p':
                    if(child.get('*[@class="CodeRay"]')) {
                        cur = '';
                        if(child.text().indexOf('new') === -1) cur = page + '.';
                        outfile.writeln('#### ' + cur + child.text() + '\n');
                    }
                    break;
                case 'blockquote':
                    cur = child.get('p');
                    if(cur) {
                        cur = cur.text();
                        cur2 = cur.indexOf('returns');
                        if(cur2 !== -1) cur = cur.slice(0, cur2);
                        outfile.writeln('>' + cur + '\n');
                    }

                    cur = child.get('h3');
                    if(cur && (cur.text().indexOf('Arguments') !== -1)) {
                        outfile.writeln('>**args**  ');
                        cur = child.find('ul/li');
                        if(cur) {
                            for(var i=0, len=cur.length; i<len; i++) {
                                cur2 = cur[i].get('strong');
                                outfile.write('*' + cur2.text() + '* ');
                                cur2 = cur[i].text();
                                cur2 = cur2.slice(cur2.indexOf('-'));
                                outfile.writeln(cur2 + '  ');
                            }
                            outfile.writeln('\n');
                        }
                    }

                    cur = child.get('p/strong');
//                    console.log(child.toString(), String(cur));
                    if(cur && (cur.text().indexOf('returns') !== -1)) {
                        outfile.write('>**returns** ');
                        while(cur = cur.nextSibling()) {
                            outfile.write(cur.text());
                        }
                        outfile.writeln('\n');
                    }
                    break;
                }
            });

            outfile.end();
            console.warn(page, ' done');
        } else { console.log('errors in %s:\n%j', page, doc) }
    });
});