def encodeDataFromFile(inputFile,
outputFile,
markovInputFile,
textFileFormat,
wordsPerState=1):
initTime = time.time()
f = open(markovInputFile, 'r')
jsonData = f.read()
f.close()
markovData = json.JSONDecoder().decode(jsonData)
if (wordsPerState == 1 and type(markovData[0][0]) != str
and type(markovData[0][0]) != unicode) or (
wordsPerState == 2 and type(markovData[0][0]) != list):
raise RuntimeError(
"error; markov chain structure doesn't match wordsPerState value")
inputData = []
f = open(inputFile, 'rb')
char = None
while char != "":
char = f.read(1)
if char != "": inputData.append(ord(char))
f.close()
initTimeCode = time.time()
encodedData = encodeDataToWordList(inputData, 4, markovData, wordsPerState)
# save
if textFileFormat:
outputData = utils.wordListToText(encodedData)
else:
outputData = json.JSONEncoder().encode(encodedData)
endTimeCode = time.time()
f = open(outputFile, 'w')
f.write(outputData)
f.close()
print "wrote " + repr(len(inputData) * 8) + " bits"
print "elapsed time: " + repr(time.time() - initTime) + " seconds"
print " - encoding time: " + repr(endTimeCode - initTimeCode) + " seconds"
def encodeDataFromFile(inputFile, outputFile, markovInputFile, textFileFormat, wordsPerState = 1):
initTime = time.time()
f = open(markovInputFile, 'r')
jsonData = f.read()
f.close()
markovData = json.JSONDecoder().decode(jsonData)
if (wordsPerState == 1 and type(markovData[0][0]) != str and type(markovData[0][0]) != unicode) or (wordsPerState == 2 and type(markovData[0][0]) != list):
raise RuntimeError("error; markov chain structure doesn't match wordsPerState value")
inputData = []
f = open(inputFile, 'rb')
char = None
while char != "":
char = f.read(1)
if char != "": inputData.append(ord(char))
f.close()
initTimeCode = time.time()
encodedData = encodeDataToWordList(inputData, 4, markovData, wordsPerState)
# save
if textFileFormat:
outputData = utils.wordListToText(encodedData)
else:
outputData = json.JSONEncoder().encode(encodedData)
endTimeCode = time.time()
f = open(outputFile, 'w')
f.write(outputData)
f.close()
print "wrote " + repr(len(inputData) * 8) + " bits"
print "elapsed time: " + repr(time.time() - initTime) + " seconds"
print " - encoding time: " + repr(endTimeCode - initTimeCode) + " seconds"
if wordsPerState == None: wordsPerState = 1
if mode == "createMarkov":
print "creating markov chain"
print "using wordsPerState = " + repr(wordsPerState)
markov.createMarkovChainFromFile(inputFile, outputFile, wordsPerState)
print "done"
elif mode == "testMarkov":
print "testing markov chain"
markov.testMarkovChain(inputFile)
print "done"
elif mode == "genTextWithMarkov":
print "generating text using markov chain"
print "using wordsPerState = " + repr(wordsPerState)
print(
utils.wordListToText(
markov.generateTextUsingMarkovChain(inputFile, wordsPerState)))
print "done"
elif mode == "encode":
markovInputFile = args.markovInput
print "encoding file (number to text) using markov chain, saving as json"
variableSizeCode.encodeDataFromFile(inputFile, outputFile, markovInputFile,
False, wordsPerState)
print "done"
elif mode == "decode":
markovInputFile = args.markovInput
print "decoding file (json text to number) using markov chain"
variableSizeCode.decodeDataFromFile(inputFile, outputFile, markovInputFile,
False, wordsPerState)
print "done"
elif mode == "encodeFullText":
markovInputFile = args.markovInput
if wordsPerState == "1" or wordsPerState == "2": wordsPerState = int(wordsPerState) if wordsPerState == None: wordsPerState = 1 if mode == "createMarkov": print "creating markov chain" print "using wordsPerState = " + repr(wordsPerState) markov.createMarkovChainFromFile(inputFile, outputFile, wordsPerState) print "done" elif mode == "testMarkov": print "testing markov chain" markov.testMarkovChain(inputFile) print "done" elif mode == "genTextWithMarkov": print "generating text using markov chain" print "using wordsPerState = " + repr(wordsPerState) print(utils.wordListToText(markov.generateTextUsingMarkovChain(inputFile, wordsPerState))) print "done" elif mode == "encode": markovInputFile = args.markovInput print "encoding file (number to text) using markov chain, saving as json" variableSizeCode.encodeDataFromFile(inputFile, outputFile, markovInputFile, False, wordsPerState) print "done" elif mode == "decode": markovInputFile = args.markovInput print "decoding file (json text to number) using markov chain" variableSizeCode.decodeDataFromFile(inputFile, outputFile, markovInputFile, False, wordsPerState) print "done" elif mode == "encodeFullText": markovInputFile = args.markovInput print "encoding file (number to text) using markov chain, saving as txt" variableSizeCode.encodeDataFromFile(inputFile, outputFile, markovInputFile, True, wordsPerState)