class TxtDictionaryTest(unittest.TestCase):
dictionary: TxtDictionary
def setUp(self) -> None:
self.dictionary = TxtDictionary()
def test_GetCorrectForm(self):
for i in range(self.dictionary.size()):
self.assertTrue(
len(
self.dictionary.getCorrectForm(
self.dictionary.getWordWithIndex(i).getName())) == 0)
def test_Morphology(self):
self.assertEqual("ab", self.dictionary.getWord("ab").getMorphology())
self.assertEqual("çarp+HcH+lHk",
self.dictionary.getWord("çarpıcılık").getMorphology())
self.assertEqual(
"aciz+lAş+yAbil+mA",
self.dictionary.getWord("âcizleşebilme").getMorphology())
self.assertEqual(
"ak+Hş+GAn+lAş+DHr+HCH+lHk",
self.dictionary.getWord("akışkanlaştırıcılık").getMorphology())
def test_PrepareTrie(self):
trie = self.dictionary.prepareTrie()
self.assertTrue(Word("ben") in trie.getWordsWithPrefix("bana"))
self.assertTrue(Word("sen") in trie.getWordsWithPrefix("sana"))
self.assertTrue(Word("metin") in trie.getWordsWithPrefix("metni"))
self.assertTrue(Word("ağız") in trie.getWordsWithPrefix("ağzı"))
self.assertTrue(Word("ayır") in trie.getWordsWithPrefix("ayrıldı"))
self.assertTrue(Word("buyur") in trie.getWordsWithPrefix("buyruldu"))
self.assertTrue(Word("ahit") in trie.getWordsWithPrefix("ahdi"))
self.assertTrue(Word("kayıp") in trie.getWordsWithPrefix("kaybı"))
self.assertTrue(Word("kutup") in trie.getWordsWithPrefix("kutbu"))
self.assertTrue(
Word("ademelması") in trie.getWordsWithPrefix("ademelmaları"))
self.assertTrue(
Word("ağaçküpesi") in trie.getWordsWithPrefix("ağaçküpeleri"))
self.assertTrue(Word("ağaçlık") in trie.getWordsWithPrefix("ağaçlığı"))
self.assertTrue(Word("sumak") in trie.getWordsWithPrefix("sumağı"))
self.assertTrue(
Word("deveboynu") in trie.getWordsWithPrefix("deveboyunları"))
self.assertTrue(
Word("gökcismi") in trie.getWordsWithPrefix("gökcisimleri"))
self.assertTrue(
Word("gökkuşağı") in trie.getWordsWithPrefix("gökkuşakları"))
self.assertTrue(
Word("hintarmudu") in trie.getWordsWithPrefix("hintarmutları"))
self.assertTrue(
Word("hintpirinci") in trie.getWordsWithPrefix("hintpirinçleri"))
self.assertTrue(
Word("sudolabı") in trie.getWordsWithPrefix("sudolapları"))
self.assertTrue(Word("ye") in trie.getWordsWithPrefix("yiyor"))
self.assertTrue(Word("de") in trie.getWordsWithPrefix("diyor"))
self.assertTrue(Word("depola") in trie.getWordsWithPrefix("depoluyor"))
self.assertTrue(Word("dışla") in trie.getWordsWithPrefix("dışlıyor"))
self.assertTrue(Word("fiyonk") in trie.getWordsWithPrefix("fiyongu"))
self.assertTrue(Word("gonk") in trie.getWordsWithPrefix("gongu"))
class TxtDictionaryTest(unittest.TestCase):
dictionary: TxtDictionary
def setUp(self) -> None:
self.dictionary = TxtDictionary("../../turkish_dictionary.txt",
"../../turkish_misspellings.txt")
def test_GetCorrectForm(self):
for i in range(self.dictionary.size()):
self.assertTrue(
len(
self.dictionary.getCorrectForm(
self.dictionary.getWordWithIndex(i).getName())) == 0)
def test_PrepareTrie(self):
trie = self.dictionary.prepareTrie()
self.assertTrue(Word("ben") in trie.getWordsWithPrefix("bana"))
self.assertTrue(Word("metin") in trie.getWordsWithPrefix("metni"))
self.assertTrue(Word("ağız") in trie.getWordsWithPrefix("ağzı"))
self.assertTrue(Word("ayır") in trie.getWordsWithPrefix("ayrıldı"))
self.assertTrue(Word("buyur") in trie.getWordsWithPrefix("buyruldu"))
self.assertTrue(Word("ahit") in trie.getWordsWithPrefix("ahdi"))
self.assertTrue(Word("kayıp") in trie.getWordsWithPrefix("kaybı"))
self.assertTrue(Word("kutup") in trie.getWordsWithPrefix("kutbu"))
self.assertTrue(
Word("ademelması") in trie.getWordsWithPrefix("ademelmaları"))
self.assertTrue(
Word("ağaçküpesi") in trie.getWordsWithPrefix("ağaçküpeleri"))
self.assertTrue(Word("ağaçlık") in trie.getWordsWithPrefix("ağaçlığı"))
self.assertTrue(Word("sumak") in trie.getWordsWithPrefix("sumağı"))
self.assertTrue(
Word("deveboynu") in trie.getWordsWithPrefix("deveboyunları"))
self.assertTrue(
Word("gökcismi") in trie.getWordsWithPrefix("gökcisimleri"))
self.assertTrue(
Word("gökkuşağı") in trie.getWordsWithPrefix("gökkuşakları"))
self.assertTrue(
Word("hintarmudu") in trie.getWordsWithPrefix("hintarmutları"))
self.assertTrue(
Word("hintpirinci") in trie.getWordsWithPrefix("hintpirinçleri"))
self.assertTrue(
Word("sudolabı") in trie.getWordsWithPrefix("sudolapları"))
self.assertTrue(Word("ye") in trie.getWordsWithPrefix("yiyor"))
self.assertTrue(Word("de") in trie.getWordsWithPrefix("diyor"))
self.assertTrue(Word("depola") in trie.getWordsWithPrefix("depoluyor"))
self.assertTrue(Word("dışla") in trie.getWordsWithPrefix("dışlıyor"))
self.assertTrue(Word("fiyonk") in trie.getWordsWithPrefix("fiyongu"))
self.assertTrue(Word("gonk") in trie.getWordsWithPrefix("gongu"))
def setUp(self) -> None:
self.dictionary = TxtDictionary("../../turkish_dictionary.txt",
"../../turkish_misspellings.txt")
self.lowerCaseDictionary = TxtDictionary("../../lowercase.txt")
self.mixedCaseDictionary = TxtDictionary(
"../../mixedcase.txt", "../../turkish_misspellings.txt",
TxtDictionary.turkishIgnoreCaseComparator)
def setUp(self) -> None:
self.simpleTrie = Trie()
self.simpleTrie.addWord("azı", Word("azı"))
self.simpleTrie.addWord("az", Word("az"))
self.simpleTrie.addWord("ad", Word("ad"))
self.simpleTrie.addWord("adi", Word("adi"))
self.simpleTrie.addWord("adil", Word("adil"))
self.simpleTrie.addWord("a", Word("a"))
self.simpleTrie.addWord("adilane", Word("adilane"))
self.simpleTrie.addWord("ısı", Word("ısı"))
self.simpleTrie.addWord("ısıtıcı", Word("ısıtıcı"))
self.simpleTrie.addWord("ölü", Word("ölü"))
self.simpleTrie.addWord("ölüm", Word("ölüm"))
self.simpleTrie.addWord("ören", Word("ören"))
self.simpleTrie.addWord("örgü", Word("örgü"))
self.complexTrie = Trie()
dictionary = TxtDictionary("../../../turkish_dictionary.txt",
"../../../turkish_misspellings.txt")
for i in range(dictionary.size()):
self.complexTrie.addWord(
dictionary.getWordWithIndex(i).getName(),
dictionary.getWordWithIndex(i))
class DictionaryTest(unittest.TestCase):
dictionary: TxtDictionary
lowerCaseDictionary: TxtDictionary
mixedCaseDictionary: TxtDictionary
def setUp(self) -> None:
self.dictionary = TxtDictionary("../../turkish_dictionary.txt",
"../../turkish_misspellings.txt")
self.lowerCaseDictionary = TxtDictionary("../../lowercase.txt")
self.mixedCaseDictionary = TxtDictionary(
"../../mixedcase.txt", "../../turkish_misspellings.txt",
TxtDictionary.turkishIgnoreCaseComparator)
def test_GetWordIndex(self):
self.assertEqual(0, self.lowerCaseDictionary.getWordIndex("a"))
self.assertEqual(3, self.lowerCaseDictionary.getWordIndex("ç"))
self.assertEqual(8, self.lowerCaseDictionary.getWordIndex("ğ"))
self.assertEqual(10, self.lowerCaseDictionary.getWordIndex("ı"))
self.assertEqual(18, self.lowerCaseDictionary.getWordIndex("ö"))
self.assertEqual(22, self.lowerCaseDictionary.getWordIndex("ş"))
self.assertEqual(25, self.lowerCaseDictionary.getWordIndex("ü"))
self.assertEqual(28, self.lowerCaseDictionary.getWordIndex("z"))
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("A") == 0
or self.mixedCaseDictionary.getWordIndex("A") == 1)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("Ç") == 6
or self.mixedCaseDictionary.getWordIndex("Ç") == 7)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("Ğ") == 16
or self.mixedCaseDictionary.getWordIndex("Ğ") == 17)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("I") == 20
or self.mixedCaseDictionary.getWordIndex("I") == 21)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("İ") == 22
or self.mixedCaseDictionary.getWordIndex("İ") == 23)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("Ş") == 44
or self.mixedCaseDictionary.getWordIndex("Ş") == 45)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("Ü") == 50
or self.mixedCaseDictionary.getWordIndex("Ü") == 51)
self.assertTrue(
self.mixedCaseDictionary.getWordIndex("Z") == 56
or self.mixedCaseDictionary.getWordIndex("Z") == 57)
def test_Size(self):
self.assertEqual(29, self.lowerCaseDictionary.size())
self.assertEqual(58, self.mixedCaseDictionary.size())
self.assertEqual(62112, self.dictionary.size())
def test_GetWord(self):
self.assertEqual("a",
self.lowerCaseDictionary.getWordWithIndex(0).name)
self.assertEqual("ç",
self.lowerCaseDictionary.getWordWithIndex(3).name)
self.assertEqual("ğ",
self.lowerCaseDictionary.getWordWithIndex(8).name)
self.assertEqual("ı",
self.lowerCaseDictionary.getWordWithIndex(10).name)
self.assertEqual("ö",
self.lowerCaseDictionary.getWordWithIndex(18).name)
self.assertEqual("ş",
self.lowerCaseDictionary.getWordWithIndex(22).name)
self.assertEqual("ü",
self.lowerCaseDictionary.getWordWithIndex(25).name)
self.assertEqual("z",
self.lowerCaseDictionary.getWordWithIndex(28).name)
for i in range(self.dictionary.size()):
self.assertIsNotNone(self.dictionary.getWordWithIndex(i))
def test_LongestWordSize(self):
self.assertEqual(1, self.lowerCaseDictionary.longestWordSize())
self.assertEqual(1, self.mixedCaseDictionary.longestWordSize())
self.assertEqual(21, self.dictionary.longestWordSize())
def test_GetWordStartingWith(self):
self.assertEqual(
20, self.lowerCaseDictionary.getWordStartingWith("pırasa"))
self.assertEqual(27,
self.lowerCaseDictionary.getWordStartingWith("veli"))
self.assertEqual(
40, self.mixedCaseDictionary.getWordStartingWith("Pırasa"))
self.assertEqual(54,
self.mixedCaseDictionary.getWordStartingWith("Veli"))
def setUp(self) -> None:
self.dictionary = TxtDictionary("../../turkish_dictionary.txt",
"../../turkish_misspellings.txt")
class TxtWordTest(unittest.TestCase):
dictionary: TxtDictionary
def setUp(self) -> None:
self.dictionary = TxtDictionary("../../turkish_dictionary.txt",
"../../turkish_misspellings.txt")
def test_VerbType(self):
verbs = {}
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
verbType = word.verbType()
if verbType in verbs:
verbs[verbType] = verbs[verbType] + 1
else:
verbs[verbType] = 1
self.assertEqual(5, verbs["F2P1-NO-REF"])
self.assertEqual(1, verbs["F3P1-NO-REF"])
self.assertEqual(1, verbs["F4P1-NO-REF"])
self.assertEqual(14, verbs["F4PR-NO-REF"])
self.assertEqual(2, verbs["F4PL-NO-REF"])
self.assertEqual(67, verbs["F4PW-NO-REF"])
self.assertEqual(10, verbs["F5PL-NO-REF"])
self.assertEqual(111, verbs["F5PR-NO-REF"])
self.assertEqual(1, verbs["F5PW-NO-REF"])
self.assertEqual(2, verbs["F1P1"])
self.assertEqual(11, verbs["F2P1"])
self.assertEqual(4, verbs["F3P1"])
self.assertEqual(1, verbs["F4P1"])
self.assertEqual(1, verbs["F5P1"])
self.assertEqual(7, verbs["F6P1"])
self.assertEqual(2, verbs["F2PL"])
self.assertEqual(49, verbs["F4PL"])
self.assertEqual(18, verbs["F5PL"])
self.assertEqual(173, verbs["F4PR"])
self.assertEqual(808, verbs["F5PR"])
self.assertEqual(1396, verbs["F4PW"])
self.assertEqual(13, verbs["F5PW"])
def test_IsNominal(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isNominal():
count = count + 1
self.assertEqual(30593, count)
def test_IsPassive(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPassive():
count = count + 1
self.assertEqual(10, count)
def test_IsAbbreviation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isAbbreviation():
count = count + 1
self.assertEqual(102, count)
def test_IsInterjection(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isInterjection():
count = count + 1
self.assertEqual(104, count)
def test_IsDuplicate(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isInterjection():
count = count + 1
self.assertEqual(104, count)
def test_IsAdjective(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isAdjective():
count = count + 1
self.assertEqual(9671, count)
def test_IsPronoun(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPronoun():
count = count + 1
self.assertEqual(48, count)
def test_IsQuestion(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isQuestion():
count = count + 1
self.assertEqual(4, count)
def test_IsVerb(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isVerb():
count = count + 1
self.assertEqual(5041, count)
def test_IsPortmanteau(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPortmanteau():
count = count + 1
self.assertEqual(1294, count)
def test_IsDeterminer(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isDeterminer():
count = count + 1
self.assertEqual(13, count)
def test_IsConjunction(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isConjunction():
count = count + 1
self.assertEqual(48, count)
def test_IsAdverb(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isAdverb():
count = count + 1
self.assertEqual(1845, count)
def test_IsPostP(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPostP():
count = count + 1
self.assertEqual(47, count)
def test_IsPortmanteauEndingWithSI(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPortmanteauEndingWithSI():
count = count + 1
self.assertEqual(178, count)
def test_IsPortmanteauFacedVowelEllipsis(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPortmanteauFacedVowelEllipsis():
count = count + 1
self.assertEqual(25, count)
def test_IsPortmanteauFacedSoftening(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPortmanteauFacedSoftening():
count = count + 1
self.assertEqual(348, count)
def test_IsSuffix(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isSuffix():
count = count + 1
self.assertEqual(3, count)
def test_IsProperNoun(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isProperNoun():
count = count + 1
self.assertEqual(19011, count)
def test_IsPlural(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isPlural():
count = count + 1
self.assertEqual(398, count)
def test_IsNumeral(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isNumeral():
count = count + 1
self.assertEqual(33, count)
def test_NotObeysVowelHarmonyDuringAgglutination(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.notObeysVowelHarmonyDuringAgglutination():
count = count + 1
self.assertEqual(315, count)
def test_ObeysAndNotObeysVowelHarmonyDuringAgglutination(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.obeysAndNotObeysVowelHarmonyDuringAgglutination():
count = count + 1
self.assertEqual(5, count)
def test_RootSoftenDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.rootSoftenDuringSuffixation():
count = count + 1
self.assertEqual(5529, count)
def test_RootSoftenAndNotSoftenDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.rootSoftenAndNotSoftenDuringSuffixation():
count = count + 1
self.assertEqual(14, count)
def test_VerbSoftenDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.verbSoftenDuringSuffixation():
count = count + 1
self.assertEqual(87, count)
def test_NounSoftenDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.nounSoftenDuringSuffixation():
count = count + 1
self.assertEqual(5443, count)
def test_EndingKChangesIntoG(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.endingKChangesIntoG():
count = count + 1
self.assertEqual(26, count)
def test_IsExceptional(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.isExceptional():
count = count + 1
self.assertEqual(31, count)
def test_DuplicatesDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.duplicatesDuringSuffixation():
count = count + 1
self.assertEqual(36, count)
def test_DuplicatesAndNotDuplicatesDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.duplicatesAndNotDuplicatesDuringSuffixation():
count = count + 1
self.assertEqual(4, count)
def test_LastIdropsDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.lastIdropsDuringSuffixation():
count = count + 1
self.assertEqual(167, count)
def test_LastIDropsAndNotDropDuringSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.lastIDropsAndNotDropDuringSuffixation():
count = count + 1
self.assertEqual(4, count)
def test_TakesRelativeSuffixKi(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.takesRelativeSuffixKi():
count = count + 1
self.assertEqual(16, count)
def test_TakesRelativeSuffixKu(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.takesRelativeSuffixKu():
count = count + 1
self.assertEqual(4, count)
def test_LastIdropsDuringPassiveSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.lastIdropsDuringPassiveSuffixation():
count = count + 1
self.assertEqual(11, count)
def test_VowelAChangesToIDuringYSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.vowelAChangesToIDuringYSuffixation():
count = count + 1
self.assertEqual(1299, count)
def test_VowelEChangesToIDuringYSuffixation(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.vowelEChangesToIDuringYSuffixation():
count = count + 1
self.assertEqual(2, count)
def test_TakesSuffixIRAsAorist(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if not word.takesSuffixIRAsAorist():
count = count + 1
self.assertEqual(51, count)
def test_TakesSuffixDIRAsFactitive(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if not word.takesSuffixDIRAsFactitive():
count = count + 1
self.assertEqual(197, count)
def test_ShowsSuRegularities(self):
count = 0
for i in range(self.dictionary.size()):
word = self.dictionary.getWordWithIndex(i)
if word.showsSuRegularities():
count = count + 1
self.assertEqual(5, count)
def setUp(self) -> None:
self.dictionary = TxtDictionary()