How do I run XPath queries in QT? - c++

Question

How do I run an XPath query in QT?
I need to sort out certain tags with specific values in a certain attribute. The QXmlQuery documentation is anything but legible.
The schema I'm parsing is the Rhythmbox DB format:
<rhythmdb version="1.6">
<entry type="ignore">
<title></title>
<genre></genre>
<artist></artist>
<album></album>
<location>file:///mnt/disk/music/Cover.jpg</location>
<mountpoint>file:///mnt/disk</mountpoint>
<mtime>1222396828</mtime>
<date>0</date>
<mimetype>application/octet-stream</mimetype>
<mb-trackid></mb-trackid>
<mb-artistid></mb-artistid>
<mb-albumid></mb-albumid>
<mb-albumartistid></mb-albumartistid>
<mb-artistsortname></mb-artistsortname>
</entry>
<entry type="song">
<title>Bar</title>
<genre>Foobared Music</genre>
<artist>Foo</artist>
<album>The Great big Bar</album>
<track-number>1</track-number>
<disc-number>1</disc-number>
<duration>208</duration>
<file-size>8694159</file-size>
<location>file:///media/disk/music/01-Foo_-_Bar.ogg
<mountpoint>file:///media/disk
<mtime>1216995840</mtime>
<first-seen>1250478814</first-seen>
<last-seen>1250478814</last-seen>
<bitrate>301</bitrate>
<date>732677</date>
<mimetype>application/x-id3</mimetype>
<mb-trackid></mb-trackid>
<mb-artistid></mb-artistid>
<mb-albumid></mb-albumid>
<mb-albumartistid></mb-albumartistid>
<mb-artistsortname></mb-artistsortname>
</entry>
</rhythmdb>
This is your basic XML Schema which has a collection of structured entries. My intention was to filter out the entries with the type 'ignore'.

Answer 1

The relevant documentation is at: http://qt-project.org/doc/qt-4.8/qxmlquery.html#running-xpath-expressions.
The solution I came to was to use QXmlQuery to generate an XML file then parse it again using QDomDocument.
RhythmboxTrackModel::RhythmboxTrackModel()
{
QXmlQuery query;
QXmlQuery entries;
QString res;
QDomDocument rhythmdb;
/*
* Try and open the Rhythmbox DB. An API call which tells us where
* the file is would be nice.
*/
QFile db(QDir::homePath() + "/.gnome2/rhythmbox/rhythmdb.xml");
if ( ! db.exists()) {
db.setFileName(QDir::homePath() + "/.local/share/rhythmbox/rhythmdb.xml");
if ( ! db.exists())
return;
}
if (!db.open(QIODevice::ReadOnly | QIODevice::Text))
return;
/*
* Use QXmlQuery to execute and XPath query. Check the version to
* make sure.
*/
query.setFocus(&db);
query.setQuery("rhythmdb[#version='1.6']/entry[#type='song']");
if ( ! query.isValid())
return;
query.evaluateTo(&res);
db.close();
/*
* Parse the result as an XML file. These shennanigans actually
* reduce the load time from a minute to a matter of seconds.
*/
rhythmdb.setContent("" + res + "");
m_entryNodes = rhythmdb.elementsByTagName("entry");
for (int i = 0; i < m_entryNodes.count(); i++) {
QDomNode n = m_entryNodes.at(i);
QString location = n.firstChildElement("location").text();
m_mTracksByLocation[location] = n;
}
qDebug() << rhythmdb.doctype().name();
qDebug() << "RhythmboxTrackModel: m_entryNodes size is" << m_entryNodes.size();
}
In case anyone is wondering, this is my code taken from a recent branch of the Mixxx project, specifically the features_looping branch.
The things I dislike about this solution are:
Parsing the XML twice
Concatenating the result with a starting and ending tag.

Answer 2

If it fits your parsing requirements, you can use the SAX-based reader instead of a DOM-based one. Using QXmlSimpleReader with a sub-classed QXmlDefaultHandler, you can get access to each element of your XPath query as well as its attributes as the document is scanned. I think this approach would be faster than a DOM-based one; you don't have to read anything twice and it's already built into Qt. There is an example here: http://www.digitalfanatics.org/projects/qt_tutorial/chapter09.html under "Reading Using SAX."