Andrej Andrej - 12 days ago 4
Python Question

Parse xml with lxml - extract element value

Let's suppose we have the XML file with the structure as follows.

<?xml version="1.0" ?>
<searchRetrieveResponse xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.loc.gov/zing/srw/ http://www.loc.gov/standards/sru/sru1-1archive/xml-files/srw-types.xsd" xmlns="http://www.loc.gov/zing/srw/">
<records xmlns:ns1="http://www.loc.gov/zing/srw/">
<record>
<recordData>
<record xmlns="">
<datafield tag="000">
<subfield code="a">123</subfield>
<subfield code="b">456</subfield>
</datafield>
<datafield tag="001">
<subfield code="a">789</subfield>
<subfield code="b">987</subfield>
</datafield>
</record>
</recordData>
</record>
<record>
<recordData>
<record xmlns="">
<datafield tag="000">
<subfield code="a">123</subfield>
<subfield code="b">456</subfield>
</datafield>
<datafield tag="001">
<subfield code="a">789</subfield>
<subfield code="b">987</subfield>
</datafield>
</record>
</recordData>
</record>
</records>
</searchRetrieveResponse>


I need to parse out:


  • The content of the "subfield" (e.g. 123 in the example above) and

  • Attribute values (e.g. 000 or 001)



I wonder how to do that using lxml and XPath. Pasted below is my initial code and I kindly ask someone to explain me, how to parse out values.

import urllib, urllib2
from lxml import etree

url = "https://dl.dropbox.com/u/540963/short_test.xml"
fp = urllib2.urlopen(url)
doc = etree.parse(fp)
fp.close()

ns = {'xsi':'http://www.loc.gov/zing/srw/'}

for record in doc.xpath('//xsi:record', namespaces=ns):
print record.xpath("xsi:recordData/record/datafield[@tag='000']", namespaces=ns)

Answer

I would be more direct in your XPath: go straight for the elements you want, in this case datafield.

>>> for df in doc.xpath('//datafield'):
        # Iterate over attributes of datafield
        for attrib_name in df.attrib:
                print '@' + attrib_name + '=' + df.attrib[attrib_name]

        # subfield is a child of datafield, and iterate
        subfields = df.getchildren()
        for subfield in subfields:
                print 'subfield=' + subfield.text

Also, lxml appears to let you ignore the namespace, maybe because your example only uses one namespace?

Comments