xml_structure_count.py

import xml.etree.ElementTree as xml, sys

file = sys.argv[1]
tree = xml.parse(file)

#from lxml import etree
#import sys
#parser = etree.XMLParser(recover=True)
#tree = etree.parse(file,parser=parser)

root = tree.getroot()
max_level = -1
show_attributes = True

if len(sys.argv) > 2 and sys.argv[2]:
  max_level = int(sys.argv[2])

if len(sys.argv) > 3 and sys.argv[3]:
  show_attributes = bool(int(sys.argv[3]))

def count_tree(level, tag, tags):
  label = tag.tag
  #print 'start:'  + label
  if label not in tags:
    tags[label] = {}
    tags[label]['count'] = 0
    tags[label]['child'] = {}
    tags[label]['attributes'] = {}
  tags[label]['count'] += 1
  if show_attributes:
    for attribute in tag.attrib:
      if attribute not in tags[label]['attributes']:
        tags[label]['attributes'][attribute] = 0
      tags[label]['attributes'][attribute] += 1
  if max_level == -1 or level < max_level:
    for child in tag:
      tags[label]['child'] = count_tree(level+1, child, tags[label]['child'])
  return tags

def print_level(level,tags):
  for tag in tags:
    print str(level) + '-' * 4 * level + tag + ' : ' + str(tags[tag]['count'])
    for attribute in tags[tag]['attributes']:
      print str(level+1) + '-' * 4 * (level+1) + '@' + attribute + ' : ' + str(tags[tag]['attributes'][attribute])
    print_level(level+1,tags[tag]['child'])

def write_xml(file,level,tags):
  for tag in tags:
    file.write( '<' + tag + ' count="' + str(tags[tag]['count']) + '">')
    for attribute in tags[tag]['attributes']:
      file.write( '<_' + attribute + ' count="' + str(tags[tag]['attributes'][attribute]) + '"/>')
    write_xml(file,level+1,tags[tag]['child'])
    file.write( '</' + tag + '>')


all = {}
all = count_tree(0,root,all)
#print(all)
print_level(0, all)
outxml = open(file + '_xml_structure.xml', 'w')
print_level(0,all)
write_xml(outxml,0,all)
outxml.close()

#import json
#with open('result.json', 'w') as fp:
#    json.dump(all, fp)

#import dicttoxml
#xml = dicttoxml.dicttoxml(all)
#print(xml)