def test_save(self):
trieobj = trie.trie()
trieobj["foo"] = 1
k = trieobj.keys()
self.assertEqual(k, ["foo"])
v = trieobj.values()
self.assertEqual(v, [1])
self.assertEqual(trieobj.get("bar", 99), 99)
trieobj["hello"] = "55a"
self.assertEqual(trieobj.get_approximate("foo", 0), [("foo", 1, 0)])
self.assertEqual(trieobj.get_approximate("foo", 1), [("foo", 1, 0)])
self.assertEqual(trieobj.get_approximate("foa", 0), [])
self.assertEqual(trieobj.get_approximate("foa", 1), [("foo", 1, 1)])
x = sorted(trieobj.get_approximate("foa", 2))
self.assertEqual(x, [("foo", 1, 1), ("foo", 1, 2), ("foo", 1, 2)])
# foo foo- foo-
# foa f-oa fo-a
# mismatch a->o
# insertion after f, deletion of o
# insertion after o, deletion of o
x = trieobj.get_approximate("foo", 4)
y = {}
for z in x:
y[z] = y.get(z, 0) + 1
x = sorted(y.items())
self.assertEqual(x, [(("foo", 1, 0), 1), (("hello", "55a", 4), 6)])
h = StringIO()
trie.save(h, trieobj)
h.seek(0)
trieobj = trie.load(h)
k = trieobj.keys()
self.assertTrue("foo" in k)
self.assertTrue("hello" in k)
self.assertEqual(repr(trieobj["foo"]), "1")
self.assertEqual(repr(trieobj["hello"]), "'55a'")
def test_large_save_load(self):
"""Generate random key/val pairs in three length categories.
100 items in each category. Insert them into a trie and into a reference dict.
Write the trie to a temp file and read it back, verify that trie entries match
the reference dict.
"""
cmp_dict = {}
trieobj = trie.trie()
self.assertEqual(trieobj.get("foobar"), None)
for max_str_len in [100, 1000, 10000]:
cmp_dict = {}
for i in range(1000):
key = ''.join([
random.choice(ascii_lowercase) for _ in range(max_str_len)
])
val = ''.join([
random.choice(ascii_lowercase) for _ in range(max_str_len)
])
trieobj[key] = val
cmp_dict[key] = val
for key in cmp_dict:
self.assertEqual(trieobj[key], cmp_dict[key])
with tempfile.TemporaryFile(mode='w+b') as f:
trie.save(f, trieobj)
f.seek(0)
trieobj = trie.load(f)
for key in cmp_dict:
self.assertEqual(trieobj[key], cmp_dict[key])
def test_save(self):
trieobj = trie.trie()
trieobj["foo"] = 1
k = list(trieobj.keys())
self.assertEqual(k, ["foo"])
v = list(trieobj.values())
self.assertEqual(v, [1])
self.assertEqual(trieobj.get("bar", 99), 99)
trieobj["hello"] = '55a'
self.assertEqual(trieobj.get_approximate("foo", 0), [("foo", 1, 0)])
self.assertEqual(trieobj.get_approximate("foo", 1), [("foo", 1, 0)])
self.assertEqual(trieobj.get_approximate("foa", 0), [])
self.assertEqual(trieobj.get_approximate("foa", 1), [("foo", 1, 1)])
x = sorted(trieobj.get_approximate("foa", 2))
self.assertEqual(x, [("foo", 1, 1), ("foo", 1, 2), ("foo", 1, 2)])
# foo foo- foo-
# foa f-oa fo-a
# mismatch a->o
# insertion after f, deletion of o
# insertion after o, deletion of o
x = trieobj.get_approximate("foo", 4)
y = {}
for z in x:
y[z] = y.get(z, 0) + 1
x = sorted(y.items())
self.assertEqual(x, [(('foo', 1, 0), 1), (('hello', '55a', 4), 6)])
h = BytesIO()
trie.save(h, trieobj)
h.seek(0)
trieobj = trie.load(h)
k = list(trieobj.keys())
self.assertTrue("foo" in k)
self.assertTrue("hello" in k)
self.assertEqual(repr(trieobj["foo"]), '1')
self.assertEqual(repr(trieobj["hello"]), "'55a'")
def test_large_save_load(self):
"""Generate random key/val pairs in three length categories.
100 items in each category. Insert them into a trie and into a reference dict.
Write the trie to a temp file and read it back, verify that trie entries match
the reference dict.
"""
cmp_dict = {}
trieobj = trie.trie()
self.assertEqual(trieobj.get("foobar"), None)
for max_str_len in [100, 1000, 10000]:
cmp_dict = {}
for i in range(1000):
key = ''.join([random.choice(ascii_lowercase) for _ in range(max_str_len)])
val = ''.join([random.choice(ascii_lowercase) for _ in range(max_str_len)])
trieobj[key] = val
cmp_dict[key] = val
for key in cmp_dict:
self.assertEqual(trieobj[key], cmp_dict[key])
with tempfile.TemporaryFile(mode='w+b') as f:
trie.save(f, trieobj)
f.seek(0)
trieobj = trie.load(f)
for key in cmp_dict:
self.assertEqual(trieobj[key], cmp_dict[key])
def loadIndexFromDisk ():
fp = open (indexFileName,'rb')
index = trie.load (fp)
#index = pickle.load(fp)
fp.close()
fp = open(docDictionary, 'rb')
docDict = pickle.load (fp)
fp.close()
return index, docDict[0], docDict[1]
# mismatch a->o
# insertion after f, deletion of o
# insertion after o, deletion of o
x = trieobj2.get_approximate("foo", 4)
y = {}
for z in x:
y[z] = y.get(z, 0) + 1
x = y.items()
x.sort()
print x # [(('foo', 1, 0), 1), (('hello', '55a', 4), 6)]
h = StringIO.StringIO()
trie.save(h, trieobj2)
h.seek(0)
trieobj3 = trie.load(h)
k = trieobj3.keys()
k.sort()
for m in k: # foo 1
print m, repr(trieobj3[m]) # hello '55a'
# Found bug, doesn't handle insertions and deletions at end properly.
trieobj = trie.trie()
trieobj["hello"] = 1
print trieobj.get_approximate('he', 2) # []
print trieobj.get_approximate('he', 3) # [('hello', 1, 3)]
print trieobj.get_approximate('hello me!', 3) # []
print trieobj.get_approximate('hello me!', 4) # [('hello', 1, 4)]
print trieobj.get_approximate('hello me!', 5) # [('hello', 1, 4)]
def loadFromFile (self, indexFile):
# TODO assert that trie should be empty right now,
# otherwise you'll loose the index
fp = open(indexFile, 'r')
self.index = trie.load (fp)
self.modified = False
# mismatch a->o
# insertion after f, deletion of o
# insertion after o, deletion of o
x = trieobj2.get_approximate("foo", 4)
y = {}
for z in x:
y[z] = y.get(z, 0) + 1
x = y.items()
x.sort()
print x # [(('foo', 1, 0), 1), (('hello', '55a', 4), 6)]
h = StringIO.StringIO()
trie.save(h, trieobj2)
h.seek(0)
trieobj3 = trie.load(h)
k = trieobj3.keys()
k.sort()
for m in k: # foo 1
print m, repr(trieobj3[m]) # hello '55a'
# Found bug, doesn't handle insertions and deletions at end properly.
trieobj = trie.trie()
trieobj["hello"] = 1
print trieobj.get_approximate('he', 2) # []
print trieobj.get_approximate('he', 3) # [('hello', 1, 3)]
print trieobj.get_approximate('hello me!', 3) # []
print trieobj.get_approximate('hello me!', 4) # [('hello', 1, 4)]
print trieobj.get_approximate('hello me!', 5) # [('hello', 1, 4)]