How to generate java .properties files in python

[Arnaud Delobelle]

On 3 December 2011 23:51, Peter Otten <__peter__ at web.de> wrote:
> Arnaud Delobelle wrote:
>
>> I need to generate some java .properties files in Python (2.6 / 2.7).
>> It's a simple format to store key/value pairs e.g.
>>
>> blue=bleu
>> green=vert
>> red=rouge
>>
>> The key/value are unicode strings.  The annoying thing is that the
>> file is encoded in ISO 8859-1, with all non Latin1 characters escaped
>> in the form \uHHHH (same as how unicode characters are escaped in
>> Python).
>>
>> I thought I could use the "unicode_escape" codec.  But it doesn't work
>> because it escapes Latin1 characters with escape sequences of the form
>> \xHH, which is not valid in a java .properties file.
>>
>> Is there a simple way to achieve this? I could do something like this:
>>
>> def encode(u):
>>     """encode a unicode string in .properties format"""
>>     return u"".join(u"\\u%04x" % ord(c) if ord(c) > 0xFF else c for c
>> in u).encode("latin_1")
>>
>> but it would be quite inefficient as I have many to generate.
>
>>>> class D(dict):
> ...     def __missing__(self, key):
> ...             result = self[key] = u"\\u%04x" % key
> ...             return result
> ...
>>>> d = D(enumerate(map(unichr, range(256))))
>>>> u"ähnlich üblich nötig ΦΧΨ"
> u'\xe4hnlich \xfcblich n\xf6tig \u03a6\u03a7\u03a8'
>>>> u"ähnlich üblich nötig ΦΧΨ".translate(d)
> u'\xe4hnlich \xfcblich n\xf6tig \\u03a6\\u03a7\\u03a8'
>>>> u"ähnlich üblich nötig ΦΧΨ".translate(d).encode("latin1")
> '\xe4hnlich \xfcblich n\xf6tig \\u03a6\\u03a7\\u03a8'

A very nice solution - thanks, Peter.

-- 
Arnaud