def check_puncs(forest, pos2):
'''returns idx_mapping and the modified forest.
CAUTION: can not modify the original forest.
CAUTION: delete redundant layers'''
newforest = forest.copy()
assert len(newforest) == len(pos2), "different sentence lengths!\n%s\n%s" % (str(test_tree), str(gold_tree))
idx_mapping = {}
j = 0
last_is_punc = True
for i, a in enumerate(pos2):
if not last_is_punc:
j += 1
idx_mapping [i] = j
# delete the non-consistent tags of this word in the newforest
for node in newforest.cells[(i, i+1)]:
if node.is_terminal() or node.sp_terminal():
node.deleted = is_punc(node.label) ^ is_punc(a)
if node.deleted:
print >> logs, node, "deleted"
last_is_punc = is_punc(a)
if not last_is_punc:
j += 1
idx_mapping[i+1] = j
# TODO: CLEAN UP THIS PART!
newforest.nodeorder = [node for node in newforest if not node.is_terminal() or not node.deleted]
newforest.nodes = {}
for node in newforest:
newforest.nodes[node.iden] = node
newforest.rehash()
for node in newforest:
if not node.is_terminal():
mapped_span = node.mapped_span(idx_mapping)
newedges = []
for edge in node.edges:
if edge.unary_cycle():
print >> logs, edge, "deleted (cycle)"
else:
for sub in edge.subs:
if sub.is_terminal() and sub.deleted:
print >> logs, edge, "deleted (punc)"
break
if not node.is_root() and sub.label == node.label \
and sub.mapped_span(idx_mapping) == mapped_span:
## make sure no induced unary cycle
print >> logs, edge, "deleted (induced cycle)"
break
else:
newedges.append(edge)
node.edges = newedges
return lambda x:idx_mapping[x], newforest
def prepare_stuff(self, label, wrd=None, sym=True):
self._coordination = None ## to be evaluated once called (same as C++'s const)
self._str = None
## heads-info
self.headinfo = {
heads.SEM: heads.HeadInfo(),
heads.SYN: heads.HeadInfo()
}
if wrd is not None:
self.word = symbol(wrd) if sym else wrd
self._terminal = True
self._punctuation = is_punc(self.label)
self._conjunction = is_conj(self.label)
self.word_seq = [self.word]
self.tag_seq = [label]
else:
self._terminal = False
self._punctuation = False
self._conjunction = False
self.word_seq = []
self.tag_seq = []
def prepare_stuff(self, label, wrd=None, sym=True):
self._coordination = None ## to be evaluated once called (same as C++'s const)
self._str = None
## heads-info
self.headinfo = { heads.SEM: heads.HeadInfo(), heads.SYN: heads.HeadInfo()}
if wrd is not None:
self.word = symbol(wrd) if sym else wrd
self._terminal = True
self._punctuation = is_punc(self.label)
self._conjunction = is_conj(self.label)
self.word_seq = [self.word]
self.tag_seq = [label]
else:
self._terminal = False
self._punctuation = False
self._conjunction = False
self.word_seq = []
self.tag_seq = []
def is_punctuation(self):
return is_punc(self.label)
def is_punctuation(self): return is_punc(self.label)