<p>此代码使用<a href="http://www.crummy.com/software/BeautifulSoup/" rel="nofollow">BeautifulSoup</a>。JSON结构需要一些额外的信息才能成为一个像样的XML,所以我添加了一些“product”标记来封装每个产品。</p>
<pre><code>from copy import deepcopy
import json
import re
import urllib
from BeautifulSoup import BeautifulStoneSoup, Tag, NavigableString
results = json.load(urllib.urlopen("http://www.kimonolabs.com/api/4v1t8lgk?apikey=880a9bdbfb8444e00e2505e1533970a7"))
class named_list(list):
def __init__(self, name, *args, **kwargs):
self.name = name
super(named_list, self).__init__(*args, **kwargs)
def __repr__(self):
return "<named_list %s: %s>" % (self.name, super(named_list, self).__repr__())
def dict_recurse(obj, obj_parent, nameless_default = "product"):
olist = []
if isinstance(obj, list):
for x in obj:
olist.extend(dict_recurse(x, obj))
elif isinstance(obj, dict):
for x in sorted(obj.keys()):
olist.append(named_list(x, dict_recurse(obj[x], obj)))
if isinstance(obj_parent, list):
olist = [named_list(nameless_default, olist)]
else:
olist.append(obj)
return olist
def tag_tree_build(ilist, top_tag, soup):
if isinstance(ilist, named_list):
cur_tag = Tag(soup, ilist.name)
for q in ilist:
tag_tree_build(q, cur_tag, soup)
top_tag.insert(len(top_tag), cur_tag)
elif isinstance(ilist, list):
for q in ilist:
tag_tree_build(q, top_tag, soup)
else:
cur_str = NavigableString(re.sub("\s+", " ", unicode(ilist)))
top_tag.insert(len(top_tag), cur_str)
nested = dict_recurse(results, results)
soup = BeautifulStoneSoup()
tag_tree_build(nested, soup, soup)
file("finnish.xml", "w").write(soup.prettify())
file("finnish.json", "w").write(json.dumps(results, sort_keys = True, indent = 4))
</code></pre>