Read all the contents in ini file into dictionary with Python

I managed to get an answer, but I expect there should be a better one.

dictionary = {}
for section in config.sections():
    dictionary[section] = {}
    for option in config.options(section):
        dictionary[section][option] = config.get(section, option)

I suggest subclassing ConfigParser.ConfigParser (or SafeConfigParser, &c) to safely access the "protected" attributes (names starting with single underscore -- "private" would be names starting with two underscores, not to be accessed even in subclasses...):

import ConfigParser

class MyParser(ConfigParser.ConfigParser):

    def as_dict(self):
        d = dict(self._sections)
        for k in d:
            d[k] = dict(self._defaults, **d[k])
            d[k].pop('__name__', None)
        return d

This emulates the usual logic of config parsers, and is guaranteed to work in all versions of Python where there's a ConfigParser.py module (up to 2.7, which is the last of the 2.* series -- knowing that there will be no future Python 2.any versions is how compatibility can be guaranteed;-).

If you need to support future Python 3.* versions (up to 3.1 and probably the soon forthcoming 3.2 it should be fine, just renaming the module to all-lowercase configparser instead of course) it may need some attention/tweaks a few years down the road, but I wouldn't expect anything major.

I know that this question was asked 5 years ago, but today I've made this dict comprehension thingy:

parser = ConfigParser()
parser.read(filename)
confdict = {section: dict(parser.items(section)) for section in parser.sections()}

The instance data for ConfigParser is stored internally as a nested dict. Instead of recreating it, you could just copy it.

>>> import ConfigParser
>>> p = ConfigParser.ConfigParser()
>>> p.read("sample_config.ini")
['sample_config.ini']
>>> p.__dict__
{'_defaults': {}, '_sections': {'A': {'y': '2', '__name__': 'A', 'z': '3', 'x': '1'}, 'B':         {'y': '2', '__name__': 'B', 'z': '3', 'x': '1'}}, '_dict': <type 'dict'>}
>>> d = p.__dict__['_sections'].copy()
>>> d
{'A': {'y': '2', '__name__': 'A', 'z': '3', 'x': '1'}, 'B': {'y': '2', '__name__': 'B', 'z': '3', 'x': '1'}}

Edit:

Alex Martelli's solution is cleaner, more robust, and prettier. While this was the accepted answer, I'd suggest using his approach instead. See his comment to this solution for more info.