def main(processes=1):
ptb = Constants().ptb
fileids = list(ptb.fileids())
params = []
for fileid in fileids[:10]:
for sent_num, parse_tree in enumerate(ptb.parsed_sents(fileid)):
params.append((fileid, sent_num, parse_tree))
if processes > 1:
p = Pool(processes)
p.starmap(score, sorted(params, key=lambda x: (x[0], x[1])))
else:
for param in params:
score(*param)
def trees_from_propbank(
self,
word_pos,
single_anchor=True,
verb_anchor=True,
arg_num_match=True,
gerund_match=True,
by_phrase_match=True,
):
tree_families = self.tree_families(word_pos)
instance = Constants().propbank.get_instance(
word_pos.wsj_entry.file_num, word_pos.wsj_entry.sent_num, word_pos.word
)
trees = set()
for tf in tree_families:
for t in Constants().grammar.get_trees_from_tree_family(tf):
# Only want to handle single anchors
if single_anchor and len(t.anchor_positions()) > 1:
continue
# Only want to handle verbal anchors
if verb_anchor and t.anchor_positions()[0].prefix() != "V":
continue
# Require same number of args
if arg_num_match and len(t.subst_nodes()) != len(instance.numbered_args()):
continue
# Skip when one is Gerund and other isn't
if gerund_match and word_pos.pos == "VBN" and "G" not in t.tree_name and "D" not in t.tree_name:
continue
if gerund_match and word_pos.pos != "VBN" and ("G" in t.tree_name or "D" in t.tree_name):
continue
# Skip by-phrases if no "by" in sentence
if by_phrase_match and not word_pos.wsj_entry.by_in_word_list() and "by" in t.tree_name:
continue
trees.add(("V", t.tree_name))
return trees
