Mailman 3 October 2013 - lxml - The Python XML Toolkit

Attribute access inconsistency when using namespace
by Wichert Akkerman 14 Oct '13

14 Oct '13

I am receiving XML document using namespace, but I’m having problems trying to access attributes. All our input matches an XSD, but not everyone uses namespace the same way in documents delivered to us: some documents use a default namespace, which results in the attribute being present in the attrib dictionary without any namespace declarations. Other documents use explicit namespace everywhere, which results in attributes always using the namespace in the attrib dictionary, even if the namespace is the exact same as the namespace of the element itself. Here is a small test case demonstrating the problem: from lxml import etree NS = 'http://www.example.com/ns1' def find_id(root): elem = root.find('{%s}elem' % NS) import pdb ; pdb.set_trace() return elem.attrib.get('id') input1 = '''<root xmlns="http://www.example.com/ns1"> <elem id="1"/> </root>''' print 'Result with default namespace: %s' % find_id(etree.fromstring(input1)) input2 = '''<e:root xmlns:e="http://www.example.com/ns1"> <e:elem e:id="1"/> </e:root>''' print 'Result with explicit namespace: %s' % find_id(etree.fromstring(input2)) Which results in this output: Result with default namespace: 1 Result with explicit namespace: None While I would like to have “1” as result for both inputs. Is there a simple/official way to get the right attribute? The only option I can see is to always look at the element’s nsmap to find possible matches: def get_attribute(elem, namespace, attr): for (prefix, ns) in elem.nsmap.items(): if ns == namespace: key = '%s%s' % ('' if prefix is None else '{%s}' % ns, attr) value = elem.attrib.get(key) if value is not None: return value return None This works, but is not very efficient or pretty. Wichert.

2 1

Porting back to previous version of python-lxml
by gaurav parida 03 Oct '13

03 Oct '13

Hello, I installed python-lxml in my Ubuntu 13.04 64-bit. How can I roll back to a previous version of python-lxml? Thanks, gaurav.p

1 0

Schema validation error since Oct 1st
by Christian Heimes 02 Oct '13

02 Oct '13

Hi, just a friendly reminder. If you see a new error message like XMLSchemaParseError: Opening and ending tag mismatch: br line 30 and h1, line 31, column 128 then it's not your fault. You can blame it on the US Congress. The Library Of Congress has shut down so XSD schema links like http://www.loc.gov/standards/xlink/xlink.xsd redirect to a HTTP page. In order to fix the issue in your code you have to implement a resolver. Example: from lxml import etree class XSDResolver(etree.Resolver): """See http://lxml.de/resolvers.html#uri-resolvers """ xsd_map = { 'http://www.loc.gov/standards/xlink/xlink.xsd': '/path/to/xlink.xsd', } def resolve(self, url, id, context): xsdfile = self.xsd_map.get(url) if xsdfile: return self.resolve_filename(xsdfile, context) # use next resolver return None parser = etree.XMLParser() parser.resolvers.add(XSDResolver()) tree = etree.parse(xmlfile, parser) HTH Christian

1 0

lxml.objectify drops leading whitespace
by Wolfgang Schnerring 02 Oct '13

02 Oct '13

Hello, lxml.objectify drops leading whitespace before a tag: >>> lxml.etree.tostring(lxml.objectify.fromstring(' <a>foo</a>')) '<a>foo</a>' while lxml.etree handles the same situation just fine: >>> lxml.etree.tostring(lxml.etree.fromstring(' <a>foo</a>')) ' <a>foo</a>' What's the reason for the different behaviour? Is this intentional? Most importantly, how can I make lxml.objectify preserve whitespace correctly? Thanks for your help, Wolfgang

2 2