class TamilDTriesForward(unittest.TestCase):
def setUp(self):
rhymie = [
(u"மாங்குயில்", u"ல்யிகுங்மா"),
(u"பூங்குயில்", u"ல்யிகுங்பூ"),
(u"அல்லவா", u"வாலல்அ"),
(u"செல்வாயா", u"யாவால்செ"),
(u"சொல்வாயா", u"யாவால்சொ"),
]
self.obj = DTrie()
self.count = len(rhymie)
for k, v in rhymie:
self.obj.add(v)
return
def test_all_words(self):
self.assertTrue(len(self.obj.getAllWords()), self.count)
def word_n_prefix_test(self, pfx, wordlist, no):
itr = 0
for word in self.obj.getAllWordsPrefix(pfx):
itr = itr + 1
self.assertTrue(word in wordlist)
self.assertEqual(itr, no)
return
def test_fwd_dictionaries(self):
kuyils = [u"ல்யிகுங்பூ", u"ல்யிகுங்மா"]
pfx = u"ல்யிகு"
self.word_n_prefix_test(pfx, kuyils, 2)
def test_fwd_dictionaries_two(self):
verbie = [u"யாவால்செ", u"யாவால்சொ"]
pfx = u"யாவா"
self.word_n_prefix_test(pfx, verbie, 2)
class TamilDTriesForward(unittest.TestCase):
def setUp(self):
rhymie = [(u"மாங்குயில்",u"ல்யிகுங்மா"),(u"பூங்குயில்",u"ல்யிகுங்பூ"),(u"அல்லவா",u"வாலல்அ"),\
(u"செல்வாயா",u"யாவால்செ"),(u"சொல்வாயா",u"யாவால்சொ")]
self.obj = DTrie()
self.count = len(rhymie)
for k,v in rhymie:
self.obj.add(v)
return
def test_all_words(self):
self.assertTrue(len(self.obj.getAllWords()),self.count)
def word_n_prefix_test(self,pfx,wordlist,no):
itr = 0
for word in self.obj.getAllWordsPrefix(pfx):
itr = itr + 1
self.assertTrue(word in wordlist)
self.assertEqual(itr,no)
return
def test_fwd_dictionaries(self):
kuyils = [u"ல்யிகுங்பூ",u"ல்யிகுங்மா"]
pfx = u"ல்யிகு"
self.word_n_prefix_test(pfx,kuyils,2)
def test_fwd_dictionaries_two(self):
verbie = [u"யாவால்செ",u"யாவால்சொ"]
pfx = u"யாவா"
self.word_n_prefix_test(pfx,verbie,2)
def test_load_dictionary(self):
obj = DTrie()
obj.loadWordFile(DICTIONARY_DATA_FILES['tamilvu'])
self.assertEqual(len(obj.getAllWords()),63896)
count = 0
for word in obj.getAllWordsIterable():
count = count + 1
self.assertEqual(count,63896)
words = obj.getAllWordsPrefix(u'பெரு')
print(len(words))
#for w in words:
# print(w)
self.assertEqual( len(words), 215 )
def test_load_dictionary(self):
obj = DTrie()
obj.loadWordFile(DICTIONARY_DATA_FILES['tamilvu'])
self.assertEqual(len(obj.getAllWords()),63896)
count = 0
for word in obj.getAllWordsIterable():
count = count + 1
self.assertEqual(count,63896)
words = obj.getAllWordsPrefix(u'பெரு')
print(len(words))
#for w in words:
# print(w)
self.assertEqual( len(words), 215 )
def test_trie_counts_and_prefix(self):
obj = DTrie()
actual_words = ['a','ab','abc','abc','bbc']
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue(obj.isWord(w))
self.assertEqual(len(obj.getAllWords()),4)
self.assertEqual( obj.getAllWordsPrefix('ab'),['ab','abc'] )
self.assertEqual(obj.getWordCount('abc'),2)
obj = DTrie()
list(map(obj.add,['foo','bar','bar','baz']))
self.assertEqual((obj.getWordCount('bar'),\
obj.getWordCount('baz'),\
obj.getWordCount('foo')),(2,1,1))
def test_stuff_3letter(self):
obj = DTrie()
self.assertFalse( obj.isWord('apple') )
try:
obj.add('')
except AssertionError as exp:
pass
actual_words = ['a','ab','abc','bbc']
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue( obj.isWord(w) )
self.assertEqual( sorted(obj.getAllWords()),sorted(actual_words))
self.assertEqual( obj.getAllWordsPrefix('ab'), ['ab','abc'] )
return
def test_trie_counts_and_prefix(self):
obj = DTrie()
actual_words = ['a','ab','abc','abc','bbc']
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue(obj.isWord(w))
self.assertEqual(len(obj.getAllWords()),4)
self.assertEqual( obj.getAllWordsPrefix('ab'),['ab','abc'] )
self.assertEqual(obj.getWordCount('abc'),2)
obj = DTrie()
list(map(obj.add,['foo','bar','bar','baz']))
self.assertEqual((obj.getWordCount('bar'),\
obj.getWordCount('baz'),\
obj.getWordCount('foo')),(2,1,1))
def test_stuff_3letter(self):
obj = DTrie()
self.assertFalse( obj.isWord('apple') )
try:
obj.add('')
except AssertionError as exp:
pass
actual_words = ['a','ab','abc','bbc']
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue( obj.isWord(w) )
self.assertEqual( sorted(obj.getAllWords()),sorted(actual_words))
self.assertEqual( obj.getAllWordsPrefix('ab'), ['ab','abc'] )
return
def test_stuff_3letter(self):
obj = DTrie()
self.assertFalse(obj.isWord("apple"))
try:
obj.add("")
except AssertionError as exp:
pass
actual_words = ["a", "ab", "abc", "bbc"]
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue(obj.isWord(w))
self.assertEqual(sorted(obj.getAllWords()), sorted(actual_words))
self.assertEqual(obj.getAllWordsPrefix("ab"), ["ab", "abc"])
return
def test_trie_counts_and_prefix(self):
obj = DTrie()
actual_words = ["a", "ab", "abc", "abc", "bbc"]
[obj.add(w) for w in actual_words]
for w in actual_words:
self.assertTrue(obj.isWord(w))
self.assertEqual(len(obj.getAllWords()), 4)
self.assertEqual(obj.getAllWordsPrefix("ab"), ["ab", "abc"])
self.assertEqual(obj.getWordCount("abc"), 2)
obj = DTrie()
list(map(obj.add, ["foo", "bar", "bar", "baz"]))
self.assertEqual(
(obj.getWordCount("bar"), obj.getWordCount("baz"), obj.getWordCount("foo")),
(2, 1, 1),
)
def test_trie_neg(self):
obj = DTrie()
self.assertEqual( obj.getAllWords(), [] )
self.assertEqual( obj.getAllWordsPrefix('none'), [] )
self.assertFalse( obj.isWord('fubar',True)[0] )
self.assertTrue( obj.isWord('fubar',True)[1] )
def test_trie_neg(self):
obj = DTrie()
self.assertEqual( obj.getAllWords(), [] )
self.assertEqual( obj.getAllWordsPrefix('none'), [] )
self.assertFalse( obj.isWord('fubar',True)[0] )
self.assertTrue( obj.isWord('fubar',True)[1] )
