def print_examples(out=sys.stdout):
"""Print examples of the snippet creation to `out`."""
for (description, doc, query, chars, sents) in DATA:
snippet = highlight_doc(doc, query, chars, sents)
print >> out, '#### %s ####' % description
print >> out, 'DOCUMENT: """%s"""' % doc
print >> out, 'QUERY: """%s"""' % query
print >> out, 'SNIPPET: """%s"""' % snippet
print >> out, '\n'
def test(self):
doc = 'Dog! Cat! The sentence with rat is too long.'
query = 'rat'
snippet = snippets.highlight_doc(doc, query, max_chars=10, max_sents=1)
assert snippet in 'Dog!', 'Cat!'
def test_two(self):
doc = 'I love pepperoni. Pepperoni. Pepperoni.'
query = 'pepperoni'
snippet = snippets.highlight_doc(doc, query, max_sents=2)
assert snippet.count('.') == 2
def test_with_tags(self):
doc = 'I love pepperoni pizza. Pizza!'
query = 'pizza'
snippet = snippets.highlight_doc(doc, query, max_chars=35)
assert snippet == '[[HIGHLIGHT]]Pizza[[ENDHIGHLIGHT]]!'
def test_no_tags(self):
doc = 'I love pepperoni pizza. LOL!'
query = 'pizza'
snippet = snippets.highlight_doc(doc, query, max_chars=10)
assert snippet == 'LOL!'
def test_mixed_both(self):
doc = 'pEpPeRoNi PiZzA'
query = 'PePpErOnI pIzZa'
snippet = snippets.highlight_doc(doc, query)
assert snippet == '[[HIGHLIGHT]]pEpPeRoNi PiZzA[[ENDHIGHLIGHT]]'
def test_mixed_query(self):
doc = 'pepperoni pizza is good pizza.'
query = 'pEppeRonI pIzZa'
snippet = snippets.highlight_doc(doc, query)
assert snippet == '[[HIGHLIGHT]]pepperoni pizza[[ENDHIGHLIGHT]] is good [[HIGHLIGHT]]pizza[[ENDHIGHLIGHT]].'
def test_multiple_query_matches_2(self):
doc = 'pizza pepperoni olive pizza olive pizza'
query = 'pepperoni olive pizza'
snippet = snippets.highlight_doc(doc, query)
assert snippet == '[[HIGHLIGHT]]pizza[[ENDHIGHLIGHT]] [[HIGHLIGHT]]pepperoni olive pizza[[ENDHIGHLIGHT]] [[HIGHLIGHT]]olive pizza[[ENDHIGHLIGHT]]'
def test_multiple_query_matches_1(self):
doc = 'Their specialty pizza is deep dish pizza.'
query = 'deep dish pizza'
snippet = snippets.highlight_doc(doc, query)
assert snippet == 'Their specialty [[HIGHLIGHT]]pizza[[ENDHIGHLIGHT]] is [[HIGHLIGHT]]deep dish pizza[[ENDHIGHLIGHT]].'
def test_multi_word_query(self):
doc = 'I really love deep dish pizza.'
query = 'deep dish pizza'
snippet = snippets.highlight_doc(doc, query)
assert snippet == 'I really love [[HIGHLIGHT]]deep dish pizza[[ENDHIGHLIGHT]].'
def test_(self):
doc = query = 'pepperoni pizza'
snippet = snippets.highlight_doc(doc, query)
assert snippet == '[[HIGHLIGHT]]pepperoni pizza[[ENDHIGHLIGHT]]'
