def test_medium_minhashexcludes(self): #Minhash must exclude at least some misses!
testmgr = zTestDataManager.TestBookManager('samplebooks.zip','../testbooks/')
testmgr.unpack_archive(max=10)
testbook = testmgr.get_testbooks()[0]
e = zTestDataManager.ErrorMaker()
dupe = testmgr.make_error_dupes([testbook,], errormaker=e)[0]
books = testmgr.get_testbooks()
library = Library.Library()
for book in books:
mybook = Book(textfile=book)
mybook.initialize_text_data()
library.add_book(mybook)
dupe = library.get_book_textfile(dupe)
testablebook = library.get_book_textfile(testbook)
possiblematches = library.get_possible_matches(testablebook)
self.assertLess(len(possiblematches), 5, msg="minhash was not able to eliminate at least one possible book")
self.assertGreaterEqual(len(possiblematches), 1, msg="minhash didn't get at least one hit")
minverification = testmgr.verify_minhash_results(library)
testmgr.print_formatted_minhash_results(minverification)
def test_medium_minhashfindsmatches(self): #make certain minhash actually...works.
testmgr = zTestDataManager.TestBookManager('samplebooks.zip','../testbooks/')
testmgr.unpack_archive(max=2)
testbook = testmgr.get_testbooks()[0]
e = zTestDataManager.ErrorMaker()
dupe = testmgr.make_error_dupes([testbook,], errormaker=e)[0]
books = testmgr.get_testbooks()
library = Library.Library()
for book in books:
mybook = Book(textfile=book)
mybook.initialize_text_data()
library.add_book(mybook)
dupe = library.get_book_textfile(dupe)
testablebook = library.get_book_textfile(testbook)
possiblematches = library.get_possible_matches(testablebook)
found = False
for id, score in possiblematches:
if id == dupe.id:
found=True
self.assertTrue(found)