def test_writebyte(self):
output = "testWrite"
with TemporaryDirectory() as td:
tempFileName = os.path.join(td, output)
with codecs.open(tempFileName, 'w', 'utf-8') as testOutput:
writer = bit.Writer(testOutput)
writer.writebyte('a')
writer.close()
with codecs.open(tempFileName, 'r', 'utf-8') as testOutput:
self.assertEqual(testOutput.read(), 'a' + chr(0b00000000))
def test_writeNode(self):
output = "testNode"
with TemporaryDirectory() as td:
tempFileName = os.path.join(td, output)
with codecs.open(tempFileName, 'w', 'utf-8') as testOutput:
writer = bit.Writer(testOutput)
huffman.writeNode(testTree, writer)
writer.close()
with codecs.open(tempFileName, 'r', 'utf-8') as testOutput:
s = getFileBinary(testOutput)
self.assertEqual(s, testTreeBinary)
def test_writesinglebit(self):
output = "testWrite"
with TemporaryDirectory() as td:
tempFileName = os.path.join(td, output)
with codecs.open(tempFileName, 'w', 'utf-8') as testOutput:
writer = bit.Writer(testOutput)
writer.write(1)
writer.close()
with codecs.open(tempFileName, 'r', 'utf-8') as testOutput:
r = testOutput.read()
self.assertEqual(r, chr(0b10000000) + chr(0b00000111))
def test_readTree(self):
output = "testNode"
with TemporaryDirectory() as td:
tempFileName = os.path.join(td, output)
with codecs.open(tempFileName, 'w', 'utf-8') as testOutput:
writer = bit.Writer(testOutput)
for b in testTreeBinary:
writer.write(int(b))
writer.close()
with codecs.open(tempFileName, 'r', 'utf-8') as testOutput:
reader = bit.Reader(testOutput)
t = huffman.readTree(reader)
self.assertEqual(str(t), str(testTree))
def test_writeCodified(self):
output = "testEncoded"
with TemporaryDirectory() as td:
tempFileName = os.path.join(td, 'testFile')
with open(tempFileName, 'w') as testFile:
testFile.write(testString)
with open(tempFileName, 'r') as testFile:
tempFileName = os.path.join(td, output)
with codecs.open(tempFileName, 'w', 'utf-8') as testOutput:
writer = bit.Writer(testOutput)
huffman.writeCodified(testFile, testCodeDict, writer)
with codecs.open(tempFileName, 'r', 'utf-8') as testOutput:
s = getFileBinary(testOutput)
self.assertEqual(s, testEncoded)
def test_decodeFile(self):
output = "testDecoded"
input_ = "testEncoded"
with TemporaryDirectory() as td:
outputName = os.path.join(td, output)
tempFileName = os.path.join(td, input_)
with codecs.open(tempFileName, 'w', 'utf-8') as testInput:
writer = bit.Writer(testInput)
for b in testEncoded:
writer.write(int(b))
writer.close()
with codecs.open(tempFileName, 'r', 'utf-8') as testInput:
reader = bit.Reader(testInput)
huffman.decodeFile(reader, outputName, testTree)
with codecs.open(outputName, 'r', 'utf-8') as testDecoded:
r = testDecoded.read()
self.assertEqual(r, testString)
def Compress(filename, outputName):
with codecs.open(filename, 'r', 'utf-8') as file:
# gerar a arvore a partit da frequencia dos caracteres no texto
root = Harvest(GetMap(file))
# gerar o dicionario
dict = {}
# Caso só tenha uma letra
if(root.isLeaf()):
dict[root.Value] = 0
else:
createDict(root, dict)
# Resetar cursor
file.seek(0, 0)
# Escrever arvora
with codecs.open(outputName, 'w', 'utf-8') as outputFile:
writer = bit.Writer(outputFile)
writeNode(root, writer)
# Codificar
writeCodified(file, dict, writer)