if not options.input:
optparser.error("No input file defined")
if not options.output:
optparser.error("No output file defined")
if not options.dictionary:
optparser.error("No dictionary file defined")
return options, args
if __name__=="__main__":
options, args = getOptions()
f = gzip.GzipFile(options.dictionary, 'r')
dictionary = readDictionaryMapping(f)
f.close()
f = gzip.GzipFile(options.input, 'r')
documents = GraphMatrices.readInstances(f)
instances = GraphMatrices.buildAMFromFullSentences(documents, MatrixBuilders.buildAdjacencyMatrix, settings, options.parser, options.tokenizer)
f.close()
f = gzip.GzipFile(options.output,'w')
datavector = []
identities = []
for document in instances.itervalues():
for sentence in document.itervalues():
for identity, instance in sentence.iteritems():
identities.append(identity)
datavector.append(instance)
outputs = [x[2] for x in datavector]
datavector = [(x[0], x[1]) for x in datavector]
datavector = GraphMatrices.LinearizeGraphs(datavector, dictionary, options.mode)
for id, output, features in zip(identities, outputs, datavector):
keys = features.keys()
