python 一个XML解析
'''Version:0.5
author;wujianqiang
date:2016/11/13
'''
import xml.etree.ElementTree as ET
from collections import Counter
from pprint import pprint
import json
from fnmatch import fnmatch
import os
class XMLParse(object):
def __init__(self):
self.data = {}
def _open_file(self,filename):
return ET.parse(filename).getroot()
def _type_select(self,root, sybol='-'):
my_type = {}
flag =False
attrib_data = root.attrib
if len(attrib_data) > 0:
flag = True
data = {sybol + key: value for key, value in attrib_data.items()}
my_type.update(data)
if len(root):
myset =
mydata = Counter(myset)
for key, value in mydata.items():
if value > 1:
my_type = []
elif value == 1:
my_type = {}
return my_type
else:
if flag:
if root.text:
my_type['#text'] = root.text
return my_type
else:
return root.text
def _par(self,root, data, sybol='-'):
utdata = self._type_select(root, sybol)
if isinstance(data, dict):
data = utdata
elif isinstance(data, list):
data.append(utdata)
if len(root) > 0:
for item in root:
self._par(item, utdata, sybol)
def getdata_fromFile(self,filename, outfile=None,sybol='-'):
root = self._open_file(filename)
self.data = {}
if root.attrib:
data = {sybol + key: value for key, value in root.attrib.items()}
self.data.update(data)
self._par(root, self.data,sybol)
if outfile:
file = open(outfile,'w')
json.dump(self.data,file)
file.close()
return True
else:
return self.data
def getdata_fromString(self, stringdata, outfile=None, sybol='-'):
root = ET.fromstring(stringdata)
self.data = {}
if root.attrib:
data = {sybol + key: value for key, value in root.attrib.items()}
self.data.update(data)
self._par(root, self.data, sybol)
if outfile:
file = open(outfile, 'w')
json.dump(self.data, file)
file.close()
return True
else:
return self.data
def _create_sub_Element(self,parnent, data, tag=None, attrib='-', text='#text'):
if isinstance(data, dict):
for key, value in data.items():
if fnmatch(key, '%s*' % (attrib,)):
parnent.attrib.update({key: value})
elif fnmatch(key, text):
parnent.text = value
else:
if type(value) == str:
a = ET.SubElement(parnent, key)
a.text = value
elif isinstance(value, dict):
a = ET.SubElement(parnent, key)
self._create_sub_Element(a, value)
else:
self._create_sub_Element(parnent, value, tag=key)
elif isinstance(data, list):
for item in data:
l = ET.SubElement(parnent, tag)
self._create_sub_Element(l, item, tag)
else:
pass
def _create_root_Element(self,data):
root_key = data.keys()
root = ET.Element(root_key)
etre = ET.ElementTree(root)
return , etre]
def _create_Element(self, data, attr='-', text='#text'):
root, mydata, tree = self._create_root_Element(data)
self._create_sub_Element(root, mydata, attrib=attr, text=text)
return tree
def create_xml(self,data,outfile=None):
if data:
root = self._create_Element(data)
filename = outfile if outfile else '.temp'
root.write(filename)
if not outfile:
f = open(filename)
data = f.readlines()
f.close()
os.remove(filename)
return data
else:
return None
if __name__ == '__main__':
xdata = XMLParse()
dat = {'data': {'-fictional': 'http://characters.example.com',
'country': [{'-name': 'Liechtenstein',
'gdppc': '141100',
'neighbor': [{'-direction': 'E',
'-name': 'Austria'},
{'-direction': 'W',
'-name': 'Switzerland'}],
'rank': '1',
'year': '2008'},
{'-name': 'dfd',
'gdppc': '14100',
'neighbor': {'-direction': 'E', '-name': 'wuul'},
'rank': '3',
'year': {'#text': '208', '-d': 'sd'}}]}}
# pprint(xdata.getdata_fromFile('myxml2.xml',sybol='-'))
pprint(xdata.create_xml(dat))
页:
[1]