def test_simple_links(self):
# make sure the shuffled sequence does not lose any elements
tp = TextParser(self.text)
tp.extract_urls()
self.assertEqual(
tp.urls,
[
"http://www.youtube.com/watch?v=zGXAVw3vF9A",
"http://www.slideshare.net/startuplessonslearned/2012-05-15-" "eric-ries-the-lean-startup-pwc-canada",
"http://www.nytimes.com/2010/04/25/business/25unboxed.html",
],
)
def test_complex_urls(self):
# make sure the shuffled sequence does not lose any elements
text = """
Check out this video of Eric Ries
http://bit.ly/abc from
Stanford University. It's a good start to understand Lean Startup.
Here is a presentation
you need to see as well
bit.ly/xyz
NYTIme recently covered him too
www.a.long.and.weird.url.com/yep
Steve
"""
tp = TextParser(text)
tp.extract_urls()
self.assertEqual(tp.urls, ["http://bit.ly/abc", "http://bit.ly/xyz", "http://www.a.long.and.weird.url.com/yep"])
