xml2dict.py

#!/usr/bin/env python
#coding: utf8
#
#
#   xml2dict
#   ============
#
#   Convert XML to dict.
#

from xml.etree import cElementTree as ElementTree
from collections import OrderedDict

def xml2dict(xmlstring, strip=True, dict_constructor=OrderedDict):
    '''Convert XML to JSON

    <foo />                => {'foo': None}
    <foo>bar</foo>         => {'foo': 'bar'}
     <foo id="1">bar</foo> => {'foo': {'#text': 'bar', '@id': '1'}}
     <foo id="1">             {'foo': {'a': '1',     
        foo                            '#text': 'foo', 
        <a>1</a>           =>          'b': {'#tail': 'bar',  
        <b>2</b>                             '#text': '2'}, 
        bar                            '@id': '1'}}         
    </foo>                    
                                  
                               
    '''
    
    def parse_item(item, strip, dict_constructor):
        '''Parse every single XML item'''

        d = dict_constructor()
        for k, v in item.attrib.items():
            d['@'+k] = v

        for child in item:
            cd = parse_item(child, strip=strip, dict_constructor=dict_constructor)
            tag = child.tag
            v = cd[tag]
            
            if tag in d:
                try: 
                    d[tag].append(v)
                except AttributeError:
                    d[tag] = [d[tag], v]
            else:
                d[tag] = v

        if strip:
            if item.tail:
                item.tail = item.tail.strip()
            if item.text:
                item.text = item.text.strip()

        if item.tail:
            d['#tail'] = item.tail 

        if d and item.text:
            d['#text'] = item.text
        if not d:
            d = item.text

        return {item.tag: d} 

    try:
        xml = ElementTree.fromstring(xmlstring)
        return parse_item(xml, strip=strip, dict_constructor=dict_constructor)
    except ElementTree.ParseError:
        return None
    


if __name__ == '__main__':
    import sys, json
    xmlstring = sys.stdin.read()
    d = xml2dict(xmlstring)
    json.dump(d, sys.stdout, indent=2)