def test_remove_links():
text = """
Wikisyntax is the [[syntax (to be parsed)|syntax]] used on
[[Wikipedia]].{{citation needed|date=October 2014}}
We have to parse it, and we use every [[hack]] in the
[[text]][[book]] that we can find.
"""
# Note space in "text book", inserted to match extract_links output (see
# test above). If we don't do this, probabilities won't add up properly.
expected = """
Wikisyntax is the syntax used on Wikipedia.
We have to parse it, and we use every hack in the
text book that we can find.
"""
assert_equal(remove_links(clean_text(text)).split(), expected.split())
def test_remove_links():
text = """
Wikisyntax is the [[syntax (to be parsed)|syntax]] used on
[[Wikipedia]].{{citation needed|date=October 2014}}
We have to parse it, and we use every [[hack]] in the
[[text]][[book]] that we can find.
"""
# Note space in "text book", inserted to match extract_links output (see
# test above). If we don't do this, probabilities won't add up properly.
expected = """
Wikisyntax is the syntax used on Wikipedia.
We have to parse it, and we use every hack in the
text book that we can find.
"""
assert_equal(remove_links(clean_text(text)).split(), expected.split())
def test_clean_text():
out = clean_text(unclosed_table).splitlines()
expected = (unclosed_table.splitlines()[:4] + [''] +
unclosed_table.splitlines()[-1:])
assert_equal(out, expected)
def test_clean_text():
out = clean_text(unclosed_table).splitlines()
expected = unclosed_table.splitlines()[:4] + [""] + unclosed_table.splitlines()[-1:]
assert_equal(out, expected)
