parse.js

var XmlStream = require('xml-stream'),
    stream    = new XmlStream(process.stdin);

var currentPageTitle = null,
    currentPageId = null,
    rx = {
        storting: /^85\.88\.(6[4-9]|[7-8]\d|9[0-5])\./,
        regjering: /^132\.150\./
    },
    count = 0;

stream.on('endElement: page > id', function (obj) {
    count++;
    currentPageId = obj.$text;

    if (count % 5 == 0) {
        process.stderr.write(count + '\n');
    }
});

stream.on('endElement: page > title', function (obj) {
    currentPageTitle = obj.$text;
});

stream.on('endElement: page > revision', function (revision) {
    if (!revision.contributor.ip) {
        return;
    }

    var source;

    if (revision.contributor.ip.match(rx.storting)) {
        source = 'Stortinget';
    } else if (revision.contributor.ip.match(rx.regjering)) {
        source = 'Regjeringskontorene';
    } else {
        return;
    }

    var rev = {source: source};

    rev.title = currentPageTitle;
    rev.page_id = +currentPageId;
    rev.revision_id = +revision.id;
    rev.timestamp = new Date(revision.timestamp).getTime();
    rev.contributor_ip = revision.contributor.ip;
    rev.comment = revision.comment;

    console.log(JSON.stringify(rev));
});