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_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)
Esempio n. 5
0
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())
Esempio n. 6
0
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())