def setUp(self):
self.__class__.index1 = Index(10010,
"Index 1",
"www.index1.com",
OccurrenceList(["www.index2.com",
"www.index3.com"]))
self.__class__.index2 = Index(10011,
"Index 2",
"www.index2.com",
OccurrenceList(["www.index1.com",
"www.index3.com"]))
self.__class__.index3 = Index(10010,
"Index 3",
"www.index3.com",
OccurrenceList(["www.index1.com",
"www.index2.com"]))
def test_6_add_and_search_multiple_words(self):
# Arrange
trie = self.__class__.trie
words = [
"bear", "bell", "bid", "bull", "buy", "sell", "stock", "stop",
"beat", "belly", "bully", "sea", "stoop", "gamma", "game"
]
for i in range(len(words)):
value = OccurrenceList()
value.append(
Index("1000{0}".format(i), "Test {0}".format(i),
"www.test{0}.com".format(i), []))
result = trie.add(words[i], value)
self.assertTrue(result, "Word {0} is not added".format(words[i]))
print(trie)
# Act and Assert
for i in range(len(words)):
result = trie.search(words[i])
self.assertIsNotNone(result,
"Word {0} is not found".format(words[i]))
self.assertTrue(words[i].endswith(result.key),
"Suffix not matching with {0}".format(words[i]))
self.assertTrue(result.value[0].index.endswith(str(i)),
"Value suffix not matching with {0}".format(i))
def test_4_add_with_existing_prefix(self):
# Arrange
trie = self.__class__.trie
# Act
value = OccurrenceList()
value.append(Index("10041", "Test 3", "www.test3.com", []))
result = trie.add("bell", value)
# Assert
self.assertTrue(result, "Word is not added")
print(trie)
def test_3_add_again(self):
# Arrange
trie = self.__class__.trie
# Act
value = OccurrenceList()
value.append(Index("10023", "Test 2", "www.test2.com", []))
result = trie.add("bear", value)
# Assert
self.assertTrue(result, "Word is not added")
print(trie)
def populate_web(self):
"""
Populates the web and the inverted index keyword
dictionary with the urls provided
"""
for url in self.urls:
page = scrape(url)
keywords = get_keywords(page.text)
index = len(self.web)
self.web.append(Index(index, page.title, page.url, page.links_to))
for word in keywords:
value = OccurrenceList()
value.append(index)
self.keywords.add(word.lower(), value)
def test_5_add_multiple_words(self):
# Arrange
trie = self.__class__.trie
words = [
"bear", "bell", "bid", "bull", "buy", "sell", "stock", "stop",
"beat", "sea", "stoop"
]
# Act and Assert
for i in range(len(words)):
value = OccurrenceList()
value.append(
Index("1000{0}".format(i), "Test {0}".format(i),
"www.test{0}.com".format(i), []))
result = trie.add(words[i], value)
self.assertTrue(result, "Word {0} is not added".format(words[i]))
print(trie)
def test_9_validate_to_json(self):
# Arrange
trie = self.__class__.trie
words = [
"bear", "bell", "bid", "bull", "buy", "sell", "stock", "stop",
"beat", "belly", "bully", "sea", "stoop", "gamma", "game"
]
for i in range(len(words)):
value = OccurrenceList()
value.append(
Index("1000{0}".format(i), "Test {0}".format(i),
"www.test{0}.com".format(i), []))
result = trie.add(words[i], value)
self.assertTrue(result, "Word {0} is not added".format(words[i]))
# Act
json_string = trie.to_json()
# Assert
self.assertIsNotNone(json_string, "to_json returns None")
self.assertIs(type(json_string), type(str()),
"to_json doesn't return string")