def build_node(self, node, srcnode):
# if self.m_bRemoveUnary:
# while srcnode.type == "constituent" and len(srcnode.children) == 1:
# srcnode = srcnode.children[0]
if not self.m_bEmptyKeep and srcnode.name == '-NONE-':
return False
node.type = srcnode.type
# if srcnode.name != '-NONE-':
if not (srcnode.name[0]=='-' and srcnode.name[-1]=='-'):
node.name = self.escape(srcnode.name.split("-")[0])
else:
# node.name = '-NONE-'
node.name = srcnode.name
node.start_index = srcnode.start_index
node.end_index = srcnode.end_index
if node.type == 'token':
node.token = self.escape(srcnode.token)
if srcnode.type == "constituent":
node.children = []
for srcchildnode in srcnode.children:
childnode = fidtree.CTreeNode()
if self.build_node(childnode, srcchildnode):
node.children.append(childnode)
if node.children == []:
return False
if self.m_bRemoveUnary and len(node.children)==1:
node.copy(node.children[0])
return True
def process_noroot(self, sSentence, wfile):
# don't process empty sentences
if sSentence.strip() == "":
return
# find the cfg node
head = fidtree.parse_object(sSentence)
if head.name == "ROOT":
head = head.children[0]
if type(head) == list:
lHead = head
else:
lHead = [head]
# output the dep node
for head in lHead:
# print head
if self.m_bBinarize:
outh = CBinarizedTreeNode()
# print outh
self.build_binarized_node(outh, head)
wfile.write(outh.utf8print()+"\n")
#print outh.utf8print()
else:
outh = fidtree.CTreeNode()
self.build_node(outh, head)
wfile.write(outh.utf8print()+"\n")
def cmpbasenp(src, tgt):
if src.type == 'token':
tgt.type = 'token'
tgt.token = src.token
tgt.name = src.name
return False, [src.token]
else:
hasnp = False
children = []
stg = []
for srcchild in src.children:
tgtchild = fidtree.CTreeNode()
subnp, sub = cmpbasenp(srcchild, tgtchild)
hasnp |= subnp
stg.extend(sub)
children.append(tgtchild)
if src.name == 'NP':
if not hasnp:
tgt.type = 'token'
tgt.name = 'NP'
tgt.token = '__' + '_'.join(stg) + '__'
return True, []
hasnp = True
tgt.type = 'constituent'
tgt.children = children
tgt.name = src.name
return hasnp, stg
def build_node(self, srcnode):
if srcnode.type == "token":
node = fidtree.CTreeNode()
node.name = self.escape(srcnode.name)
node.type = srcnode.type
node.token = self.escape(srcnode.token)
return [node]
else:
assert srcnode.type == "constituent"
if srcnode.temporary:
lNode = []
lNode.extend(self.build_node(srcnode.left_child))
if srcnode.head_child != "s":
lNode.extend(self.build_node(srcnode.right_child))
return lNode
else:
node = fidtree.CTreeNode()
node.name = self.escape(srcnode.name)
node.type = srcnode.type
node.children = []
node.children.extend(self.build_node(srcnode.left_child))
if srcnode.head_child != "s":
node.children.extend(self.build_node(srcnode.right_child))
return [node]
def process(self, sSentence):
# don't process empty sentences
if sSentence.strip() == "":
return
# find the cfg node
head = fidtree.parse_object(sSentence)
if type(head) == list:
lHead = head
else:
lHead = [head]
# output the dep node
for head in lHead:
# print head
if self.m_bBinarize:
outh = CBinarizedTreeNode()
# print outh
self.build_binarized_node(outh, head)
print outh
else:
outh = fidtree.CTreeNode()
self.build_node(outh, head)
print outh
