#83 NLTK_DATA environment variable parsed incorrectly on Windows

[Chris Gahan]

On Windows, there's a problem handling the NLTK_DATA and NLTK_CORPORA environment variables, since it's treating them as colon-separated search paths and splitting them on colons (which doesn't work when your path is "C:\stuff").

I fixed it by splitting on os.pathsep instead of ':'.

Here's the lines I changed in nltk\data.py:

# User-specified locations:
path += [d for d in os.environ.get('NLTK_CORPORA', '').split(':') if d]
path += [d for d in os.environ.get('NLTK_DATA', '').split(':') if d]

I just changed them to:

# User-specified locations:
path += [d for d in os.environ.get('NLTK_CORPORA', '').split(os.pathsep) if d]
path += [d for d in os.environ.get('NLTK_DATA', '').split(os.pathsep) if d]

[Steven Bird]

Logged In: YES
user_id=195736
Originator: NO

Assigned to cumber