libxmljs_sax_parser_example.js

/** Testing sax push parsing
 * 
 * The output file should have almost identical
 * to the input file
 *
 * Known issues with this example output
 *
 * - Doesn't escape entities. You can do it
 *   manually on the character content.
 * - Doesn't replicate self-closing tags.
 *   They get an end element instead
 **/

var sys = require('sys')
        , fs = require('fs')
        , libxmljs = require('libxmljs');

var d = new Date();

// input xml file
var infile = fs.createReadStream('input.xml', {encoding:'utf8'});

// output xml file
var outfile = fs.createWriteStream('output.xml', {encoding:'utf8'});

var funcs = {
    onStartDocument: function() {
        console.log('starting...');
        outfile.write('<?xml version="1.0" encoding="utf-8"?>\n');
    }
    , onEndDocument: function() {
        outfile.end();
        console.log('done.');
    }
    , onStartElementNS: function(elem, attrs, prefix, uri, namespaces) {
        var sys = require('sys')
            , out = []
            , aname, aprefix, aurl, aval;

        out.push('<' + (prefix ? prefix + ':' : '') + elem);

        if(attrs.length) {
            for(var i=0, len=attrs.length; i<len; i++) {
                out.push(' ');
                aname = attrs[i][0]
                , aprefix = attrs[i][1]
                , aurl = attrs[i][2]
                , aval = attrs[i][3];
                out.push((aprefix ? aprefix + ':' : '')
                         + aname + '=' + '"' + (aval || '') + '"');
            }
        }

        if(namespaces.length) {
            for(i=0, len=namespaces.length; i<len; i++) {
                out.push(' ');
                aname = namespaces[i][0];
                aval = namespaces[i][1];
                out.push('xmlns:'
                         + aname + '=' + '"' + (aval || '') + '"');
            }
        }

        out.push('>');
        outfile.write(out.join(''));
    }
    , onEndElementNS: function(elem, prefix, uri) {
        outfile.write('</' + (prefix ? prefix + ':' : '') + elem + '>');
    }
    , onCharacters: function(chars) {
        outfile.write(chars || '');
    }
    , onCdata: function(chars) {
        outfile.write('<![CDATA[');
        outfile.write(chars || '');
        outfile.write(']]>');
    }
    , onComment: function(chars) {
        outfile.write('<!--');
        outfile.write(chars || '');
        outfile.write('-->');
    }
    , onWarning: function(warning) {
        console.warn(warning);
    }
    , onError: function(error) {
        console.error('ERROR: ' + error);
    }
}
var p = new libxmljs.SaxPushParser(function (cb) {
    var keys = Object.keys(funcs)
        , key, val;
    for(var i=0;i<keys.length; i++) {
        key = keys[i]
        , val = funcs[key];
        cb[key](val);
    }
});

infile.on('data', function(chunk) {
    if(chunk) p.push(chunk);
});
infile.on('end', function() {
    console.log('TIME: ' + (new Date() - d) + ' ms');
});