def create_tree(self,zone=None):
root = Node()
root.ord = 0
root._aux['descendants'] = []
root.set_zone(zone)
self.trees.append(root)
return root
def load(self,args):
fh = None
try:
fh = args['filehandle']
except:
filename = args['filename']
fh = open(filename, 'r')
fh = codecs.getreader('utf8')(fh)
nodes = []
comment = ''
for line in fh:
if re.search('^#',line):
comment = comment + line
elif re.search('^\d+\-',line): # HACK: multiword tokens temporarily avoided
pass
elif line.strip():
if not nodes:
bundle = Bundle()
self.bundles.append(bundle)
root = Root() # TODO: nahradit bundle.create_tree, az bude odladene
root._aux['comment'] = comment # TODO: ulozit nekam poradne
nodes = [root]
bundle.trees.append(root)
columns = line.strip().split('\t')
node = Node()
nodes.append(node)
for index in xrange(0,len(Document.attrnames)):
setattr( node, Document.attrnames[index], columns[index] )
try: # TODO: kde se v tomhle sloupecku berou podtrzitka
node.head = int(node.head)
except ValueError:
node.head = 0
try: # TODO: poresit multitokeny
node.ord = int(node.ord)
except ValueError:
node.ord = 0
else: # an empty line is guaranteed even after the last sentence in a conll-u file
nodes[0]._aux['descendants'] = nodes[1:]
for node in nodes[1:]:
node.set_parent( nodes[node.head] )
nodes = []
comment = ''
