def main():
authorkey,combinedkey,readerkey = data.trainingKeys()
i = 0
for path,f,doc in data.walkTrainDataDoc():
print doc
print i
i += 1
def answerhist():
trainingkeylist = map(lambda (x,y,key): key, data.trainingKeys())
hist = []
for key in trainingkeylist:
for docname,answers in key.items():
for a in answers:
hist.append(filter(lambda x: ord(x) < 128, a))
return hist
def assembleAllPages():
"""
Assembles all training documents as marked-up HTML
"""
authorkey,combinedkey,readerkey = data.trainingKeys()
keydict = {'authorkey': authorkey, 'combinedkey': combinedkey,
'readerkey': readerkey}
for path,f,doc in data.walkTrainDataDoc():
page = assemblePage(path, f, doc, keydict)
