Some while ago a (now-ex) colleague introduced me to a Python class he called DictObject. Data stored in this object can be accessed via attributes or array suffixes, thus:

foo.bar is the same as foo["bar"].

I've taken this further, building in some support for nesting, and for parsing either JSON style objects or element tree XML that was itself parsed by LXML. Now, I can use LXML to parse this XML:

<foo>
  <bar a="123" b="223">
    <baz>the end</baz>
  </bar>
</foo>

and it will return an object such that:

foo.bar.a = "123"
foo.bar.baz = "the end"

Likewise, this code:

a = {"a":{"b": 3, "c":4, "x":["aa", "bb", "cc"]}, "d":123}
b = DictObject(json = a)
print b.a.x[2]

produces "cc".

This makes working with restful web services much faster to code with, whether they return JSON or XML.

Here's the class itself.

from lxml import etree

class DictObject(dict):
  def __init__(self, xml = None, json = None):
    dict.__init__(self)
    self.__dict__ = self
    
    if json is not None:
      for key, val in json.items():
        if isinstance(val, dict):
          val = DictObject(json=val)
        self[key]=val
    if xml is not None:
      for attrib in xml.attrib:
        self[attrib] = xml.attrib[attrib]
      for child in xml.getchildren():
        if len(child.attrib)==0 and len(child.getchildren())==0:
          self[child.tag]=child.text
        else:
          self[child.tag] = DictObject(xml = child)