Update nltk version

[ssaamm]

nltk 2.0.5 doesn't seem to install. How about we update to a version that works?

Tests pass with these changes:

$ python tests/unit_tests.py
        testing function 'test_config_build'
        [OK] in 'test_config_build' 0.01 sec
.       testing function 'test_chinese_fulltext_extract'
Building prefix dict from <OBSCURED>/lib/python2.7/site-packages/jieba/dict.txt ...
Loading model from cache /var/folders/mt/4fq_3q8s6hzg69gy03lq40zm0000gn/T/jieba.cache
Loading model cost 0.455691099167 seconds.
Prefix dict has been built succesfully.
        [OK] in 'test_chinese_fulltext_extract' 0.93 sec
        testing function 'test_arabic_fulltext_extract'
        [OK] in 'test_arabic_fulltext_extract' 0.32 sec
        testing function 'test_spanish_fulltext_extract'
        [OK] in 'test_spanish_fulltext_extract' 41.24 sec
.       testing function 'test_valid_urls'
        [OK] in 'test_valid_urls' 0.00 sec
.       testing function 'test_url'
        [OK] in 'test_url' 0.00 sec
        testing function 'test_download_html'
        [OK] in 'test_download_html' 0.00 sec
        testing function 'test_pre_download_parse'
You must download() an article before parsing it!
        [OK] in 'test_pre_download_parse' 0.00 sec
        testing function 'test_parse_html'
        [OK] in 'test_parse_html' 0.31 sec
        testing function 'test_meta_type_extraction'
        [OK] in 'test_meta_type_extraction' 0.00 sec
        testing function 'test_meta_extraction'
        [OK] in 'test_meta_extraction' 0.00 sec
        testing function 'test_pre_download_nlp'
You must download and parse an article before parsing it!
        [OK] in 'test_pre_download_nlp' 0.00 sec
        testing function 'test_pre_parse_nlp'
You must download and parse an article before parsing it!
        [OK] in 'test_pre_parse_nlp' 0.00 sec
        testing function 'test_nlp_body'
        [OK] in 'test_nlp_body' 0.00 sec
.       testing function 'test_hot_trending'
        [OK] in 'test_hot_trending' 0.31 sec
        testing function 'test_popular_urls'
        [OK] in 'test_popular_urls' 0.00 sec
.
----------------------------------------------------------------------
Ran 5 tests in 43.129s

OK

[ssaamm]

@codelucas, the Travis failure looks unrelated to these changes. Am I misunderstanding?

[ssaamm]

@codelucas or @yprez , have either of you gotten the chance to look this over?

[yprez]

@ssaamm I think that travis is not configured on the py2 branch, that's why it's failing.
Is this version fully compatible with Python2.7? If so, I see no reason not to merge this in.
Thanks.

[ssaamm]

@yprez ah, that makes sense. Thank you for clarifying.

Yes, (a) NLTK says they're 2.7 compatible and (b) the tests seem to work when I run them with a 2.7 interpreter.

[yprez]

Sounds good enough to me. Thanks!

[MichaelBaumgarn]

so newspaper will never be compatible with python 3?

[codelucas]

@MichaelBaumgarn Newspaper already is compatible with python 3 and has been for some time.

We maintain two separate branches in this repo, 'master' is purely python 3 while 'python-2-head' (this branch) is purely python 2.

These separate branches are separate projects in pip, download the python 3 version with pip3 install newspaper3k or python 2 version via pip install newspaper

[MichaelBaumgarn]

yes my bad newspaper3k did the trick, I found this via google (nltk==2.0.5 bug)