if input_file <> None :
if os.path.isdir(input_file):
treebank = reader.read_dir_xml(input_file)
else:
instream = open(input_file)
treebank = reader.read_xml(instream)
else:
treebank = reader.read_xml(sys.stdin)
#Perform transformations
treebank = transform_treebank(treebank)
#Check for eval mode
if eval_file <> None and not xfold:
if cross:
(treebank,dev,test) = reader.build_eval_data(treebank)
else:
(treebank,dev,test) = reader.build_std_eval_data(treebank)
#Prints the additional dev and test files if eval mode is on
trainposfile= eval_file+'_1.pos'
pstream = open(trainposfile,'w')
printout_pos_corpus(treebank,pstream)
posdevfile = eval_file+'_2.pos'
treedevfile = eval_file+'_2.mrg'
gstream = open(treedevfile,'w')
pstream = open(posdevfile,'w')
printout_ptb_corpus(dev,gstream)
printout_pos_corpus(dev,pstream)
gstream.close()
pstream.close()
treebank = []
test = []
gold = []
reader = XmlReader()
if input_file <> None :
if os.path.isdir(input_file):
treebank = reader.read_dir_xml(input_file)
else:
instream = open(input_file)
treebank = reader.read_xml(instream)
else:
treebank = reader.read_xml(sys.stdin)
#Check for eval mode
if eval_file <> None :
(treebank,test,gold) = reader.build_eval_data(treebank)
#Prints the corpus (or the training treebank)
if raw :
printout_raw_corpus(treebank,sys.stdout)
else:
printout_pos_corpus(treebank,sys.stdout)
#Prints the additional eval files if eval mode is on
if eval_file <> None :
testfile = eval_file+'.tst'
goldfile = eval_file+'.gld'
gstream = open(goldfile,'w')
tstream = open(testfile,'w')
printout_pos_corpus(gold,gstream)
printout_raw_corpus(test,tstream)
