I have many rows in a database that contains XML and I’m trying to write a Python script to count instances of a particular node attribute.

My tree looks like:

<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>

How can I access the attributes "1" and "2" in the XML using Python?

I suggest ElementTree. There are other compatible implementations of the same API, such as lxml, and cElementTree in the Python standard library itself; but, in this context, what they chiefly add is even more speed — the ease of programming part depends on the API, which ElementTree defines.

First build an Element instance root from the XML, e.g. with the XML function, or by parsing a file with something like:

import xml.etree.ElementTree as ET
root = ET.parse('thefile.xml').getroot()

Or any of the many other ways shown at ElementTree. Then do something like:

for type_tag in root.findall('bar/type'):
    value = type_tag.get('foobar')
    print(value)

And similar, usually pretty simple, code patterns.

minidom is the quickest and pretty straight forward.

XML:

<data>
    <items>
        <item name="item1"></item>
        <item name="item2"></item>
        <item name="item3"></item>
        <item name="item4"></item>
    </items>
</data>

Python:

from xml.dom import minidom
xmldoc = minidom.parse('items.xml')
itemlist = xmldoc.getElementsByTagName('item')
print(len(itemlist))
print(itemlist[0].attributes['name'].value)
for s in itemlist:
    print(s.attributes['name'].value)

Output:

4
item1
item1
item2
item3
item4

You can use BeautifulSoup:

from bs4 import BeautifulSoup

x="""<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>"""

y=BeautifulSoup(x)
>>> y.foo.bar.type["foobar"]
u'1'

>>> y.foo.bar.findAll("type")
[<type foobar="1"></type>, <type foobar="2"></type>]

>>> y.foo.bar.findAll("type")[0]["foobar"]
u'1'
>>> y.foo.bar.findAll("type")[1]["foobar"]
u'2'

There are many options out there. cElementTree looks excellent if speed and memory usage are an issue. It has very little overhead compared to simply reading in the file using readlines.

The relevant metrics can be found in the table below, copied from the cElementTree website:

library                         time    space
xml.dom.minidom (Python 2.1)    6.3 s   80000K
gnosis.objectify                2.0 s   22000k
xml.dom.minidom (Python 2.4)    1.4 s   53000k
ElementTree 1.2                 1.6 s   14500k  
ElementTree 1.2.4/1.3           1.1 s   14500k  
cDomlette (C extension)         0.540 s 20500k
PyRXPU (C extension)            0.175 s 10850k
libxml2 (C extension)           0.098 s 16000k
readlines (read as utf-8)       0.093 s 8850k
cElementTree (C extension)  --> 0.047 s 4900K <--
readlines (read as ascii)       0.032 s 5050k   

As pointed out by @jfs, cElementTree comes bundled with Python:

• Python 2: from xml.etree import cElementTree as ElementTree.
• Python 3: from xml.etree import ElementTree (the accelerated C version is used automatically).

I suggest xmltodict for simplicity.

It parses your XML to an OrderedDict;

>>> e = '<foo>
             <bar>
                 <type foobar="1"/>
                 <type foobar="2"/>
             </bar>
        </foo> '

>>> import xmltodict
>>> result = xmltodict.parse(e)
>>> result

OrderedDict([(u'foo', OrderedDict([(u'bar', OrderedDict([(u'type', [OrderedDict([(u'@foobar', u'1')]), OrderedDict([(u'@foobar', u'2')])])]))]))])

>>> result['foo']

OrderedDict([(u'bar', OrderedDict([(u'type', [OrderedDict([(u'@foobar', u'1')]), OrderedDict([(u'@foobar', u'2')])])]))])

>>> result['foo']['bar']

OrderedDict([(u'type', [OrderedDict([(u'@foobar', u'1')]), OrderedDict([(u'@foobar', u'2')])])])

lxml.objectify is really simple.

Taking your sample text:

from lxml import objectify
from collections import defaultdict

count = defaultdict(int)

root = objectify.fromstring(text)

for item in root.bar.type:
    count[item.attrib.get("foobar")] += 1

print dict(count)

Output:

{'1': 1, '2': 1}

Python has an interface to the expat XML parser.

xml.parsers.expat

It’s a non-validating parser, so bad XML will not be caught. But if you know your file is correct, then this is pretty good, and you’ll probably get the exact info you want and you can discard the rest on the fly.

stringofxml = """<foo>
    <bar>
        <type arg="value" />
        <type arg="value" />
        <type arg="value" />
    </bar>
    <bar>
        <type arg="value" />
    </bar>
</foo>"""
count = 0
def start(name, attr):
    global count
    if name == 'type':
        count += 1

p = expat.ParserCreate()
p.StartElementHandler = start
p.Parse(stringofxml)

print count # prints 4

I might suggest declxml.

Full disclosure: I wrote this library because I was looking for a way to convert between XML and Python data structures without needing to write dozens of lines of imperative parsing/serialization code with ElementTree.

With declxml, you use processors to declaratively define the structure of your XML document and how to map between XML and Python data structures. Processors are used to for both serialization and parsing as well as for a basic level of validation.

Parsing into Python data structures is straightforward:

import declxml as xml

xml_string = """
<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>
"""

processor = xml.dictionary('foo', [
    xml.dictionary('bar', [
        xml.array(xml.integer('type', attribute='foobar'))
    ])
])

xml.parse_from_string(processor, xml_string)

Which produces the output:

{'bar': {'foobar': [1, 2]}}

You can also use the same processor to serialize data to XML

data = {'bar': {
    'foobar': [7, 3, 21, 16, 11]
}}

xml.serialize_to_string(processor, data, indent='    ')

Which produces the following output

<?xml version="1.0" ?>
<foo>
    <bar>
        <type foobar="7"/>
        <type foobar="3"/>
        <type foobar="21"/>
        <type foobar="16"/>
        <type foobar="11"/>
    </bar>
</foo>

If you want to work with objects instead of dictionaries, you can define processors to transform data to and from objects as well.

import declxml as xml

class Bar:

    def __init__(self):
        self.foobars = []

    def __repr__(self):
        return 'Bar(foobars={})'.format(self.foobars)


xml_string = """
<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>
"""

processor = xml.dictionary('foo', [
    xml.user_object('bar', Bar, [
        xml.array(xml.integer('type', attribute='foobar'), alias='foobars')
    ])
])

xml.parse_from_string(processor, xml_string)

Which produces the following output

{'bar': Bar(foobars=[1, 2])}

Just to add another possibility, you can use untangle, as it is a simple xml-to-python-object library. Here you have an example:

Installation:

pip install untangle

Usage:

Your XML file (a little bit changed):

<foo>
   <bar name="bar_name">
      <type foobar="1"/>
   </bar>
</foo>

Accessing the attributes with untangle:

import untangle

obj = untangle.parse('/path_to_xml_file/file.xml')

print obj.foo.bar['name']
print obj.foo.bar.type['foobar']

The output will be:

bar_name
1

More information about untangle can be found in “untangle“.

Also, if you are curious, you can find a list of tools for working with XML and Python in “Python and XML“. You will also see that the most common ones were mentioned by previous answers.

Here a very simple but effective code using cElementTree.

try:
    import cElementTree as ET
except ImportError:
  try:
    # Python 2.5 need to import a different module
    import xml.etree.cElementTree as ET
  except ImportError:
    exit_err("Failed to import cElementTree from any known place")      

def find_in_tree(tree, node):
    found = tree.find(node)
    if found == None:
        print "No %s in file" % node
        found = []
    return found  

# Parse a xml file (specify the path)
def_file = "xml_file_name.xml"
try:
    dom = ET.parse(open(def_file, "r"))
    root = dom.getroot()
except:
    exit_err("Unable to open and parse input definition file: " + def_file)

# Parse to find the child nodes list of node 'myNode'
fwdefs = find_in_tree(root,"myNode")

This is from “python xml parse“.

XML:

<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>

Python code:

import xml.etree.cElementTree as ET

tree = ET.parse("foo.xml")
root = tree.getroot() 
root_tag = root.tag
print(root_tag) 

for form in root.findall("./bar/type"):
    x=(form.attrib)
    z=list(x)
    for i in z:
        print(x[i])

Output:

foo
1
2

xml.etree.ElementTree vs. lxml

These are some pros of the two most used libraries I would have benefit to know before choosing between them.

xml.etree.ElementTree:

1. From the standard library: no needs of installing any module

lxml

1. Easily write XML declaration: for instance do you need to add standalone="no"?
2. Pretty printing: you can have a nice indented XML without extra code.
3. Objectify functionality: It allows you to use XML as if you were dealing with a normal Python object hierarchy.node.

import xml.etree.ElementTree as ET
data = '''<foo>
           <bar>
               <type foobar="1"/>
               <type foobar="2"/>
          </bar>
       </foo>'''
tree = ET.fromstring(data)
lst = tree.findall('bar/type')
for item in lst:
    print item.get('foobar')

This will print the value of the foobar attribute.

I find the Python xml.dom and xml.dom.minidom quite easy. Keep in mind that DOM isn’t good for large amounts of XML, but if your input is fairly small then this will work fine.

There’s no need to use a lib specific API if you use python-benedict. Just initialize a new instance from your XML and manage it easily since it is a dict subclass.

Installation is easy: pip install python-benedict

from benedict import benedict as bdict

# data-source can be an url, a filepath or data-string (as in this example)
data_source = """
<foo>
   <bar>
      <type foobar="1"/>
      <type foobar="2"/>
   </bar>
</foo>"""

data = bdict.from_xml(data_source)
t_list = data['foo.bar'] # yes, keypath supported
for t in t_list:
   print(t['@foobar'])

It supports and normalizes I/O operations with many formats: Base64, CSV, JSON, TOML, XML, YAML and query-string.

It is well tested and open-source on GitHub.

#If the xml is in the form of a string as shown below then
from lxml  import etree, objectify
'''sample xml as a string with a name space {http://xmlns.abc.com}'''
message =b'<?xml version="1.0" encoding="UTF-8"?>\r\n<pa:Process xmlns:pa="http://xmlns.abc.com">\r\n\t<pa:firsttag>SAMPLE</pa:firsttag></pa:Process>\r\n'  # this is a sample xml which is a string


print('************message coversion and parsing starts*************')

message=message.decode('utf-8') 
message=message.replace('<?xml version="1.0" encoding="UTF-8"?>\r\n','') #replace is used to remove unwanted strings from the 'message'
message=message.replace('pa:Process>\r\n','pa:Process>')
print (message)

print ('******Parsing starts*************')
parser = etree.XMLParser(remove_blank_text=True) #the name space is removed here
root = etree.fromstring(message, parser) #parsing of xml happens here
print ('******Parsing completed************')


dict={}
for child in root: # parsed xml is iterated using a for loop and values are stored in a dictionary
    print(child.tag,child.text)
    print('****Derving from xml tree*****')
    if child.tag =="{http://xmlns.abc.com}firsttag":
        dict["FIRST_TAG"]=child.text
        print(dict)


### output
'''************message coversion and parsing starts*************
<pa:Process xmlns:pa="http://xmlns.abc.com">

    <pa:firsttag>SAMPLE</pa:firsttag></pa:Process>
******Parsing starts*************
******Parsing completed************
{http://xmlns.abc.com}firsttag SAMPLE
****Derving from xml tree*****
{'FIRST_TAG': 'SAMPLE'}'''

If the source is an xml file, say like this sample

<pa:Process xmlns:pa="http://sssss">
        <pa:firsttag>SAMPLE</pa:firsttag>
    </pa:Process>

you may try the following code

from lxml import etree, objectify
metadata = 'C:\\Users\\PROCS.xml' # this is sample xml file the contents are shown above
parser = etree.XMLParser(remove_blank_text=True) # this line removes the  name space from the xml in this sample the name space is --> http://sssss
tree = etree.parse(metadata, parser) # this line parses the xml file which is PROCS.xml
root = tree.getroot() # we get the root of xml which is process and iterate using a for loop
for elem in root.getiterator():
    if not hasattr(elem.tag, 'find'): continue  # (1)
    i = elem.tag.find('}')
    if i >= 0:
        elem.tag = elem.tag[i+1:]

dict={}  # a python dictionary is declared
for elem in tree.iter(): #iterating through the xml tree using a for loop
    if elem.tag =="firsttag": # if the tag name matches the name that is equated then the text in the tag is stored into the dictionary
        dict["FIRST_TAG"]=str(elem.text)
        print(dict)

Output would be

{'FIRST_TAG': 'SAMPLE'}