如何使用命名前缀更改XML元素的属性

Question

I'm having real trouble trying to edit an XML attribute within a string that contains a single XML element with a name prefix.

I'm trying to use code as per below:

import xml.etree.ElementTree as ET

def replace_xml_label(xml):
    element = ET.fromstring(xml)
    element.set('label', 'new_test_label')
    return ET.tostring(element).decode('ascii')

xml_1 = '<abc label="test label">test_value</abc>'
xml_2 = '<abc:option label="test label">test_value</abc:option>'

For xml_1 , I get output as expected:

print(replace_xml_label(xml_1))
<abc label="new_test_label">test_value</abc>

However, the style of XML element I need to work with has a name prefix similar to xml_2 , which raises a ParseError :

print(replace_xml_label(xml_2))
Traceback (most recent call last):
 ... in XML parser.feed(text)
xml.etree.ElementTree.ParseError: unbound prefix: line 1, column 0

My expected output would be:

<abc:option label="new_test_label">test_value</abc:option>

I suspect the error is related to the lack of a defined namespace, but have been unable to successfully define one (eg using ET.register_namespace('abc', 'my-ns') .

Trying to modify the string in-place to define a namespace:

# ...doesn't raise an exception, but the output isn't in the format I need
xml_3 = xml_2.replace('<abc:option', '<abc:option xmlns:abc="my-ns"')
print(replace_xml_label(xml_3))  
<ns0:option xmlns:ns0="myns" label="new_test_label">test_value</ns0:option>

# replacing the output afterwards works, but by this point I may as well have used a regular expression!
print(replace_xml_label(xml_3).replace('ns0', 'abc').replace(' xmlns:abc="my-ns"',''))
<abc:option label="new_test_label">test_value</abc:option>

Am I doing something wrong, missing something obvious, or simply using the wrong tool?

I'd prefer to use what's available in the standard Python 3.4+ library.

Answer 1

Sure, the problem was due to undeclared prefix. XML requires all namespace prefixes in use to be properly declared, otherwise the document doesn't qualify to be XML, hence can't normally be parsed using XML parser library. So the ultimate solution is to fix at the side that currently produce XML-like document to be producing well-formed XML.

One possible workaround to fix this at the parsing side is, by wrapping the string with a parent element that contains declaration of the undeclared prefix, for example :

xml_2 = '<abc:option label="test label">test_value</abc:option>'

parent = '<foo xmlns:abc="bar">{}</foo>'
wellformed_xml = parent.format(xml_2)

result = replace_xml_label(wellformed_xml)
print(result)

Answer 2

It seems my main problem as suspected, (and confirmed by har07 ) was the undeclared namespace prefix.

Since it's not possible to fix at source, and the output needs to be in the format I specified, it seems the best workaround is to temporarily convert to simple tags before processing, and convert back to the original tags afterwards.

xml_2 = '<abc:option label="test label">test_value</abc:option>'

# Convert first and last occurrences of tag to be without namespace.
original_tag = 'abc:option'
temp_tag = 'abc'
valid_xml = temp_tag.join(
    xml_2.replace(original_tag, temp_tag, 1).rsplit(original_tag, 1)
)

# Replace label
modified_xml = replace_xml_label(valid_xml)

# Convert first and last occurrences of tag to re-add namespace.
output_xml = original_tag.join(
    modified_xml.replace(temp_tag, original_tag, 1).rsplit(temp_tag, 1)
)
print(output_xml)