Python: Decent config file format

Question

I'd like to use a configuration file format which supports key value pairs and nestable, repeatable structures, and which is as light on syntax as possible. I'm imagining something along the lines of:

cachedir = /var/cache
mail_to = [email protected]

job {
   name = my-media
   frequency = 1 day
   source {
      from = /home/michael/Images

   source { }
   source { }       
}

job { }

I'd be happy with something using significant-whitespace as well.

JSON requires too many explicit syntax rules (quoting, commas, etc.). YAML is actually pretty good, but would require the jobs to be defined as a YAML list, which I find slightly awkward to use.

Answer 1

I think YAML is great for this purpose, actually:

jobs:
 - name: my-media
   ...

 - name: something else
   ...

Or, as a dict instead of list:

jobs:
  my-media:
    frequency: 1 day
    ...
  something-else:
    frequency: 2 day
    ...

Another thing to consider, which you might not have, is using Python source for the configuration. You can nest Python dicts and lists in a very readable manner and it provides multiple unexpected benefits. Django uses Python source for its settings files, for example.

Answer 2

Why re-invent the wheel? You can make use of:

http://docs.python.org/library/configparser.html

Answer 3

You can use the config system of red-dove.

http://www.red-dove.com/config-doc/

Answer 4

I think you should check libconfig library http://www.hyperrealm.com/libconfig/. There should be somewhere python bindings for it.

Another solution is to use json format which is already provided by python itself. Search documentation for JSON module.

Answer 5

As Python's built-in configparser module does not seem to support nested sections, I'd first try ConfigObj. (See an introductory tutorial here). According to its homepage, this is the set of features worth mentioning:

• Nested sections (subsections), to any level
• List values
• Multiple line values
• String interpolation (substitution)
• Integrated with a powerful validation system
  • including automatic type checking/conversion
  • repeated sections
  • and allowing default values
• When writing out config files, ConfigObj preserves all comments and the order of members and sections
• Many useful methods and options for working with configuration files (like the 'reload' method)
• Full Unicode support

ConfigObj is used by Bazaar, Trac, IPython, matplotlib and many other large Python projects, so it seems pretty mature and stable to me (although I never used it myself).