Saving XML files using ElementTree

Question

I'm trying to develop simple Python (3.2) code to read XML files, do some corrections and store them back. However, during the storage step ElementTree adds this namespace nomenclature. For example:

<ns0:trk>
  <ns0:name>ACTIVE LOG</ns0:name>
<ns0:trkseg>
<ns0:trkpt lat="38.5" lon="-120.2">
  <ns0:ele>6.385864</ns0:ele>
  <ns0:time>2011-12-10T17:46:30Z</ns0:time>
</ns0:trkpt>
<ns0:trkpt lat="40.7" lon="-120.95">
  <ns0:ele>5.905273</ns0:ele>
  <ns0:time>2011-12-10T17:46:51Z</ns0:time>
</ns0:trkpt>
<ns0:trkpt lat="43.252" lon="-126.453">
  <ns0:ele>7.347168</ns0:ele>
  <ns0:time>2011-12-10T17:52:28Z</ns0:time>
</ns0:trkpt>
</ns0:trkseg>
</ns0:trk>

The code snippet is below:

def parse_gpx_data(gpxdata, tzname=None, npoints=None, filter_window=None,
                   output_file_name=None):
        ET = load_xml_library();

    def find_trksegs_or_route(etree, ns):
        trksegs=etree.findall('.//'+ns+'trkseg')
        if trksegs:
            return trksegs, "trkpt"
        else: # try to display route if track is missing
            rte=etree.findall('.//'+ns+'rte')
            return rte, "rtept"

    # try GPX10 namespace first
    try:
        element = ET.XML(gpxdata)
    except ET.ParseError as v:
        row, column = v.position
        print ("error on row %d, column %d:%d" % row, column, v)

    print ("%s" % ET.tostring(element))
    trksegs,pttag=find_trksegs_or_route(element, GPX10)
    NS=GPX10
    if not trksegs: # try GPX11 namespace otherwise
        trksegs,pttag=find_trksegs_or_route(element, GPX11)
        NS=GPX11
    if not trksegs: # try without any namespace
        trksegs,pttag=find_trksegs_or_route(element, "")
        NS=""

    # Store the results if requested
    if output_file_name:
        ET.register_namespace('', GPX11)
        ET.register_namespace('', GPX10)
        ET.ElementTree(element).write(output_file_name, xml_declaration=True)

    return;

I have tried using the register_namespace, but with no positive result. Are there any specific changes for this version of ElementTree 1.3?

Johnsten

Tell me if I understood your question, you'd like to have <trk> instead of <ns0:trk> and so on? – Visibility

Correct. I'd like to have <trk> instead of <ns0:trk> and so on. – Johnsten

This is not a real solution but since it seems that you load a string, have you tried to remove the namespace with some regexp? After that if you load and save without everything should be ok. – Visibility

Hi Rik. I'll do it everything else fails. I'd like to configure ElementTree not to print it in the first place. – Johnsten

Answer 1

In order to avoid the ns0 prefix the default namespace should be set before reading the XML data.

ET.register_namespace('', "http://www.topografix.com/GPX/1/1")
ET.register_namespace('', "http://www.topografix.com/GPX/1/0")

Johnsten

Looks like not before. I'm able to read XML file and get namespace and only after that set register_namespace. tree = ET.parse(str(udx_path)) root = tree.getroot() ns = { # extract namespace of root element 'udx': root.tag[1:root.tag.index('}')] } ET.register_namespace('', root.tag[1:root.tag.index('}')]) – Tobi

This is not complete way to preserve difference in parsed and output ElementTree string (if using ElementTree.tostring(root)). singingsingh is complete. – Quatrain

Just register before printing can be good enough. – Dilorenzo

Singingsingh's explanation is more appropriate to the question. – Siana

Answer 2

You need to register all your namespaces before you parse xml file.

For example: If you have your input xml like this and Capabilities is the root of your Element tree.

<Capabilities xmlns="http://www.opengis.net/wmts/1.0"
    xmlns:ows="http://www.opengis.net/ows/1.1"
    xmlns:xlink="http://www.w3.org/1999/xlink"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:gml="http://www.opengis.net/gml"
    xsi:schemaLocation="http://www.opengis.net/wmts/1.0 http://schemas.opengis.net/wmts/1.0/wmtsGetCapabilities_response.xsd"
    version="1.0.0">

Then you have to register all the namespaces i.e attributes present with xmlns like this:

ET.register_namespace('', "http://www.opengis.net/wmts/1.0")
ET.register_namespace('ows', "http://www.opengis.net/ows/1.1")
ET.register_namespace('xlink', "http://www.w3.org/1999/xlink")
ET.register_namespace('xsi', "http://www.w3.org/2001/XMLSchema-instance")
ET.register_namespace('gml', "http://www.opengis.net/gml")

Coleslaw

This answer is the complete one. – Uproot

Answer 3

If you try to print the root, you will see something like this: http://www.host.domain/path/to/your/xml/namespace}RootTag' at 0x0000000000558DB8>

So, to avoid the ns0 prefix, you have to change the default namespace before parsing the XML data as below:

ET.register_namespace('', "http://www.host.domain/path/to/your/xml/namespace")

Mohenjodaro

Answer 4

It seems that you have to declare your namespace, meaning that you need to change the first line of your xml from:

<ns0:trk>

to something like:

<ns0:trk xmlns:ns0="uri:">

Once did that you will no longer get ParseError: for unbound prefix: ..., and:

elem.tag = elem.tag[(len('{uri:}'):]

will remove the namespace.

Visibility

Hi Rik. The example XML I showed is the output. The input XML, which parses fine, doesn't have the 'ns0:' prefix. It is just standard GPX code. – Johnsten

If the line element = ET.XML(gpxdata) gives you an element with ns0 then the "problem" is in gpxdata, in which case you have to options: "fix" the gpxdata or find out why the standard parser does that and build a new one for ET.XML. – Visibility

The original gpxdata doesn't have any ns0 entries. However, your hint, Rik, kind of lead me to the solution. Basically, the ET.register_namespace('', GPX11) ET.register_namespace('', GPX10) should be done before reading, i.e. ET.XML. – Johnsten

Answer 5

Or you could regex it away:

def remove_xml_namespace(xml_str: str) -> str:
    xml_str = re.sub(r"<([^:]+):(\w+).+(?=xmlns)[^>]+>([\s\S]*)</(\1):(\2)>", r"\3", xml_str)
    # replace namespace elements from end tag
    xml_str = re.sub(r"</[^:]*:", r"</", xml_str)
    # replace namespace from start tags
    xml_str = re.sub(r"<[^/][^:]*:([^/>]*)(/?)>", r"<\1\2>", xml_str)
    return xml_str

Redcoat