def test_classify(self):
db = {}
Classifier.train(db, "Ruby", self.fixture("Ruby/foo.rb"))
Classifier.train(db, "Objective-C", self.fixture("Objective-C/Foo.h"))
Classifier.train(db, "Objective-C", self.fixture("Objective-C/Foo.m"))
rs = Classifier.classify(db, self.fixture("Objective-C/hello.m"))
assert "Objective-C" == rs[0][0]
tokens = Tokenizer.tokenize(self.fixture("Objective-C/hello.m"))
rs = Classifier.classify(db, tokens)
assert "Objective-C" == rs[0][0]
def test_classify(self):
db = {}
Classifier.train(db, "Ruby", self.fixture("Ruby/foo.rb"))
Classifier.train(db, "Objective-C", self.fixture("Objective-C/Foo.h"))
Classifier.train(db, "Objective-C", self.fixture("Objective-C/Foo.m"))
rs = Classifier.classify(db, self.fixture("Objective-C/hello.m"))
assert "Objective-C" == rs[0][0]
tokens = Tokenizer.tokenize(self.fixture("Objective-C/hello.m"))
rs = Classifier.classify(db, tokens)
assert "Objective-C" == rs[0][0]
def tokenize(self, data='', is_path=None):
if is_path:
data = open(join(join(ROOT_DIR, "samples"), str(data))).read()
return Tokenizer.tokenize(data)
def tokenize(self, data='', is_path=None):
if is_path:
data = open(join(join(ROOT_DIR, "samples"), str(data))).read()
return Tokenizer.tokenize(data)
