def test_get_synonyms(self):
original_google_url = "http://www.google.com"
original_google_page = IndexedPage(url=original_google_url)
original_google_page.save()
other_google_urls = [
"www.google.com", "https://www.google.com", "google.com",
"http://google.com"
]
noice_urls = ["facebook.com", "math.org"]
for url in other_google_urls:
IndexedPage(url=url, original_page=original_google_page).save()
for url in noice_urls:
IndexedPage(url=url).save()
# test function from original
expected = set(other_google_urls)
actual = set([e.url for e in original_google_page.get_synonyms()])
self.assertSetEqual(expected, actual)
# test function from all secondary pages.
for url in other_google_urls:
secondary_page = IndexedPage.objects.get(pk=url)
expected = set(other_google_urls) ^ set([url]) | set(
[original_google_url])
actual = set([e.url for e in secondary_page.get_synonyms()])
self.assertSetEqual(expected, actual)
def test_get_words_with_repetition(self):
url = "http://www.google.com"
google_page = IndexedPage(url=url)
google_page.save()
word_list = ["google", "feeling", "lucky", "search"] * 3
for word in word_list:
WordFromIndexedPage(word=word,
indexed_page=google_page,
offsets_in_indexedPage=str([1])).save()
self.assertEqual(len(google_page.words.all()), len(set(word_list)))
all_words = map(lambda element: element.word, google_page.get_words())
for word in word_list:
self.assertIn(word, all_words)
def test_get_words_no_words(self):
url = "http://www.google.com"
google_page = IndexedPage(url=url)
google_page.save()
word_list = []
for word in word_list:
WordFromIndexedPage(word=word,
indexed_page=google_page,
offsets_in_indexedPage=str([1])).save()
self.assertEqual(len(google_page.words.all()), len(word_list))
all_words = map(lambda element: element.word, google_page.get_words())
for word in word_list:
self.assertIn(word, all_words)
def test_set_and_check_attributes_duplicated_locations(self):
url = "http://www.google.com"
google_page = IndexedPage(url=url)
google_page.save()
our_word = WordFromIndexedPage(word="google", indexed_page=google_page)
locations_list = [2, 5, 7, 7, 2, 2, 11]
our_word.set_offsets(locations_list)
our_word.save()
# checking if the locations are retrievable
retrieved_offsets = our_word.get_offsets()
for location in locations_list:
self.assertIn(location, retrieved_offsets)
self.assertEqual(len(locations_list), len(retrieved_offsets))
def test_set_and_check_attributes_duplicated_locations(self):
url = "http://www.google.com"
google_page = IndexedPage(url = url)
google_page.save()
our_word = WordFromIndexedPage(word = "google", indexed_page = google_page)
locations_list = [2, 5, 7, 7, 2, 2, 11]
our_word.set_offsets(locations_list)
our_word.save()
#checking if the locations are retrievable
retrieved_offsets = our_word.get_offsets()
for location in locations_list:
self.assertIn(location, retrieved_offsets)
self.assertEqual(len(locations_list), len(retrieved_offsets))
def test_get_words_with_repetition(self):
url = "http://www.google.com"
google_page = IndexedPage(url=url)
google_page.save()
word_list = ["google", "feeling", "lucky", "search"] * 3
for word in word_list:
WordFromIndexedPage(
word=word,
indexed_page=google_page,
offsets_in_indexedPage=str([1])).save()
self.assertEqual(len(google_page.words.all()), len(set(word_list)))
all_words = map(lambda element: element.word, google_page.get_words())
for word in word_list:
self.assertIn(word, all_words)
def test_get_words_no_words(self):
url = "http://www.google.com"
google_page = IndexedPage(url=url)
google_page.save()
word_list = []
for word in word_list:
WordFromIndexedPage(
word=word,
indexed_page=google_page,
offsets_in_indexedPage=str([1])).save()
self.assertEqual(len(google_page.words.all()), len(word_list))
all_words = map(lambda element: element.word, google_page.get_words())
for word in word_list:
self.assertIn(word, all_words)
def test_get_synonyms(self):
original_google_url = "http://www.google.com"
original_google_page = IndexedPage(url=original_google_url)
original_google_page.save()
other_google_urls = [
"www.google.com", "https://www.google.com", "google.com",
"http://google.com"
]
noice_urls = ["facebook.com", "math.org"]
for url in other_google_urls:
IndexedPage(url=url, original_page=original_google_page).save()
for url in noice_urls:
IndexedPage(url=url).save()
# test function from original
expected = set(other_google_urls)
actual = set([e.url for e in original_google_page.get_synonyms()])
self.assertSetEqual(expected, actual)
# test function from all secondary pages.
for url in other_google_urls:
secondary_page = IndexedPage.objects.get(pk=url)
expected = set(other_google_urls) ^ set([url]) | set(
[original_google_url])
actual = set([e.url for e in secondary_page.get_synonyms()])
self.assertSetEqual(expected, actual)
def test_synonyms_logic_invariant(self):
original_google_url = "http://www.google.com"
original_google_page = IndexedPage(url=original_google_url)
original_google_page.save()
other_google_urls = [
"www.google.com",
"https://www.google.com",
"google.com",
"http://google.com"]
for url in other_google_urls:
IndexedPage(url=url, original_page=original_google_page).save()
# Make sure they all know their original
for page in IndexedPage.objects.all():
if not page == original_google_page:
self.assertEqual(original_google_page, page.original_page, "original not set!")
else:
self.assertIsNone(page.original_page, "original_page's orginal_page should be none")
def test_synonyms_logic_invariant(self):
original_google_url = "http://www.google.com"
original_google_page = IndexedPage(url=original_google_url)
original_google_page.save()
other_google_urls = [
"www.google.com", "https://www.google.com", "google.com",
"http://google.com"
]
for url in other_google_urls:
IndexedPage(url=url, original_page=original_google_page).save()
# Make sure they all know their original
for page in IndexedPage.objects.all():
if not page == original_google_page:
self.assertEqual(original_google_page, page.original_page,
"original not set!")
else:
self.assertIsNone(
page.original_page,
"original_page's orginal_page should be none")
from searchEngine.models import WordFromIndexedPage, IndexedPage
googlePage = IndexedPage(url="http://www.google.com")
googlePage.save()
print googlePage
googleWord = "google"
googleWord2 = "google2"
googleWordLocation = WordFromIndexedPage(indexedPage=googlePage,
word=googleWord)
googleWordLocation.set_offsets([1])
googleWordLocation.save()
googleWord2Location = WordFromIndexedPage(indexedPage=googlePage,
word=googleWord2)
googleWord2Location.set_offsets([1])
googleWord2Location.save()
print "--" * 100
print "googlePage.words:", googlePage.get_words()
print len(googlePage.words.all())
from searchEngine.models import WordFromIndexedPage, IndexedPage googlePage = IndexedPage(url="http://www.google.com") googlePage.save() print googlePage googleWord = "google" googleWord2 = "google2" googleWordLocation = WordFromIndexedPage(indexedPage=googlePage, word=googleWord) googleWordLocation.set_offsets([1]) googleWordLocation.save() googleWord2Location = WordFromIndexedPage(indexedPage=googlePage, word=googleWord2) googleWord2Location.set_offsets([1]) googleWord2Location.save() print "--"* 100 print "googlePage.words:", googlePage.get_words() print len(googlePage.words.all())
