def possible_disambiguation_pages_simple(self, title):
"""
get the possible disambiguation page just by using MySql Full Text indexing
:param title: string
:return: [(title, match_score)]
"""
spaced_title = normalizeTitle(title)
spaced_title_boolean = " ".join(
['+{}'.format(w.strip()) for w in spaced_title.split()])
query = '''
select name, match(spaced_title) against ("%s") as score from Page
where match(spaced_title) against ("%s" in boolean mode)
order by score desc limit 15;
''' % (spaced_title, spaced_title_boolean)
return [q[0].lower() for q in self.fetch_all(query)]
def test_no_namespace(self):
self.assertEqual(normalizeTitle("python"), "Python")
self.assertEqual(normalizeTitle("python 3"), "Python 3")
self.assertEqual(normalizeTitle("python__3"), "Python 3")
def test_not_known_namespace(self):
self.assertEqual(normalizeTitle("Category: Births"),
"Category: Births")
self.assertEqual(normalizeTitle("_category: births___"),
"Category: Births")
def test_known_namespace(self):
self.assertEqual(normalizeTitle("Template: Births"),
"Template:Births")
self.assertEqual(normalizeTitle(" template: births_"),
"Template:Births")
def test_no_namespace(self):
self.assertEqual(normalizeTitle("python"), "Python")
self.assertEqual(normalizeTitle("python 3"), "Python 3")
self.assertEqual(normalizeTitle("python__3"), "Python 3")
def test_not_known_namespace(self):
self.assertEqual(normalizeTitle("Category: Births"), "Category: Births")
self.assertEqual(normalizeTitle("_category: births___"), "Category: Births")
def test_known_namespace(self):
self.assertEqual(normalizeTitle("Template: Births"), "Template:Births")
self.assertEqual(normalizeTitle(" template: births_"), "Template:Births")