…r this.

…d install bs3 under py3

* As STOP_WORDS are stored in unicode format we should keep our words candidates in unicode also to be able to compare candidates against dictionary correctly
* With some languages, short stopwords are linked to the next word in the sentance with no-breakable-space. To designate those stop words we should support nbsp when tokenizing.
 Russian is an example. So this fixes grangier#223

See

	https://github.com/vetal4444/python-goose/tree/python_3
	grangier#220

Conflicts:
	goose/text.py

Python 3 support

Html fetching is now done with requests

Using requests allows writing high-level code encapsulating network & html level (decoding gzip, etc)

1.0.28:

  * Move to requests as network library

Some special tags can be false positive, so we had to porcess them all to select best top node

Requests uses headers-preferred content encoding, but for HTML better choise is TAGS-preferred content encoding

Moving to requests as http library made test mocks, that used urllib mocking, incorrect
This commit fixes tests by using mock_requests library for mocking, instead of urllib one.

It is not clear why it was there in the first place, as valid html does not contain such header.

Again this is not connected to the test itself.

This benefits to automatic cookie handling, keep alive connection and may be some other features

After moving to requests http backend cookies are handled correctly.

Test url http://www.nytimes.com/2013/08/18/world/middleeast/pressure-by-us-failed-to-sway-egypts-leaders.html?hp checked working

Python 3.4, Python 3.5 added

* Requests used for images. Same http session is used for all requests.
* Analyze all possible text root nodes and select best one, do not stop on first text root node candidate
* Improve text selection filters

Config parameter is `known_context_patterns'
Default:

	{
		'known_context_patterns': [
		    {'attr': 'class', 'value': 'short-story'},
		    {'attr': 'itemprop', 'value': 'articleBody'},
		    {'attr': 'class', 'value': 'post-content'},
		    {'attr': 'class', 'value': 'g-content'},
		    {'tag': 'article'},
		]
	}

When performing network requests, use request timeout, provided by goose configuration

Swallowing errors makes it difficult to understand whether something went wrong with network, goose, or target resource.

So strict mode (now default) is introduced. With this mode goose will raise Exception instead of returning empty responses.