Added published date to the extractor+article

[parhammmm]

Nothing fancy just checks the meta tags and then if nothing is there checks the URL for the published date.

[codelucas]

Thanks for this PR, will read & test it tomorrow.

[adamlwgriffiths]

You could also parse the article HTML too using dateutil if no date is in the meta tags.
http://stackoverflow.com/a/3276459/1591957

Not sure if those tags would be trimmed by newspaper's extractor though.

[parhammmm]

It would be hard to identify the correct date, there's usually more than one date on a page. After digging around I found only in a small number of cases the date was neither in the meta or the URL.

[codelucas]

@SkinnyP Thanks again for this PR. After reading it over more and testing it locally my thoughts are:

1.) Reference: https://github.com/skinnyp/newspaper/blob/master/newspaper/article.py#L466
L466-L467 are not needed because the set_published_date method will only be called internally and the returned value from extractors.py is always of datetime.datetime. Also, remove the import datetime from articles.py.

2.) Reference: https://github.com/skinnyp/newspaper/blob/master/newspaper/extractors.py#L226
I see your strategy for trying to guess out the date from the article's URL. Smart stuff, this exact same technique is used by newspaper to guess if a URL is a valid news URL or not.

However, for something as important as publishing date, we can't afford to give any incorrect value. Having a URL guesser like this can result in bogus values which are hard to debug. Please remove lines 226-232 of your extractors.py.

Otherwise nice work! After fixing the two nits above, add 1-2 unit tests of this working and i'll be ready to accept this pull request. Thanks again!

[codelucas]

^ Oops! I just accidentally merged this commit. I then reverted that commit in a seperate commit and now this entire issue can't be re-opened.

Oh well, @SkinnyP, note my above comment and just submit a new pull request with the new changes!
Thanks!