jprops

Parser for Java .properties files.

Supports Python 2.6, 2.7, and 3.3+

Installation

Use pip to install jprops from PyPI:

pip install jprops

Usage

Reading properties

Use jprops.load_properties to read a properties file and return a normal Python dict:

import jprops
with open('mine.properties') as fp:
  properties = jprops.load_properties(fp)

You can provide a custom "mapping" to load the properties into a different data strucuture. For example, if you would like to keep the properties in the same order they originally appeared, you can use an "ordered dict", such as collections.OrderedDict. The "mapping" can be any type or function that accepts an iterable of (key, value) pairs:

import collections
with open('mine.properties') as fp:
  properties = jprops.load_properties(fp, collections.OrderedDict)

load_properties is just a wrapper to iter_properties, which you can use directly if you want to process the properties lazily without loading them all into a data structure:

with open('mine.properties') as fp:
  for key, value in jprops.iter_properties(fp):
    if key.startswith('foo'):
      print key, value

Writing properties

Use jprops.store_properties to write a dict, dict-like object, or any iterable of (key, value) pairs to a file:

x = {'y': '1', 'z': '2'}
with open('out.properties', 'w') as fp:
  jprops.store_properties(fp, x)

By default jprops follows the Java convention of writing a timestamp comment to the beginning of the file, like:

#Thu Oct 06 19:08:50 EDT 2011
y=1
z=2

You can suppress writing the timestamp comment by passing timestamp=False.

You can provide a custom header comment that appears before the timestamp. Multi-line comments are handled appropriately, continuing the comment across lines:

jprops.store_properties(fp, {'x': '1'}, comment='Hello\nworld!')

#Hello
#world!
#Thu Oct 06 19:17:21 EDT 2011
x=1

You can also use write_comment and write_property for finer-grained control over writing a properties file:

with open('out.properties', 'w') as fp:
  jprops.write_comment(fp, 'the hostname:')
  jprops.write_property(fp, 'host', 'localhost')
  jprops.write_comment(fp, 'the port number:')
  jprops.write_property(fp, 'port', '443')

#the hostname:
host=localhost
#the port number:
port=443

Comments

By default, comments in the input will be ignored, but they can be included by iter_properties by passing comments=True. The comments will be included with jprops.COMMENT as a sentinal value in place of the key:

with open('in.properties') as fp:
  props = list(jprops.iter_properties(fp, comments=True))
for k, v in props:
  if k is jprops.COMMENT:
    print 'comment:', v

jprops doesn't include any special data structures for preserving comments, but you can manipulate the properties before writing them back out. For example this is one simple pattern for altering properties while writing the ouput:

updates = {'one': '1', 'two': '2', 'to_remove': None}

with open('out.properties', 'w') as fp:
  for key, value in props:
    # updates.pop will return and remove the value for the key, or return
    # the original `value` if it doesn't exist
    value = updates.pop(key, value)
    # skip keys set to `None` in `updates`
    if value is not None:
      # write_property handles jprops.COMMENT as the key so you don't have to
      # check whether to use write_comment
      jprops.write_property(fp, key, value)
  # since the existing keys have already been popped, use store_properties
  # to write the remaining updates
  jprops.store_properties(fp, updates, timestamp=False)

File encodings and Unicode

Files opened in binary mode such as open(filename, 'rb') or open(filename, 'wb') will use the latin-1 encoding and escape unicode characters in the format \uffff for compatibility with the Java Properties byte stream encoding.

Starting with version 2.0, files opened with other text encodings are also supported:

with io.open('sample.properties', encoding='utf-8') as fp:
  props = jprops.load_properties(fp)

This works with the built-in open function, codecs.open, or io.open. Other file-like objects that extend io.TextIOBase or have a non-empty encoding property will be read or written as unicode text values, otherwise they will be considered binary and read or written as latin-1 encoded bytes.

Authors

Matt Good (matt@matt-good.net)