open(tokenized_dir + token_file,
encoding='utf-8'))
if id not in companion_data.keys():
continue
else:
spans = ' '.join(
list(companion_data[id]["spans"].keys()))
tokens = companion_data[id]['tokenization']
edges = get_mrp_edges(mrp_dict,
get_remote=False)
edges = eliminate_h(edges)
labels = get_id2lex(mrp_dict)
compressed_edges = compress_c_edge(edges)
compressed_labels = update_id_labels(
compressed_edges, labels)
irtg_format_compressed = edge2irtg(
compressed_edges, labels)
node_tokens = node_to_token_index(
companion_data, mrp_dict,
compressed_labels, id)
aligned = percolate(compressed_edges,
priority_queue,
compressed_labels)
alignments = ''
for alignment in aligned.keys():
for node in aligned[alignment]:
if type(node) == str:
if '<root>' in node:
node = node[:-6]
alignments += str(node) + '|'
alignments += str(
alignment) + '!' + '||' + str(
import sys
import json
import collections
import os
import random
from tqdm import tqdm
from edge_to_irtg import edge2irtg
from get_edges_from_mrp import get_id2lex, get_mrp_edges
from convert_irtg_to_mrp import get_edges, get_mrp_edges, get_nodes, get_tops, irtg2mrp
from eliminate_h_top import eliminate_h
from a_star_mrp import *
mrp_in = sys.argv[1]
with open(mrp_in) as infile:
for line in infile:
mrp = json.loads(line)
print(mrp['id'])
labels = get_id2lex(mrp)
edges = get_mrp_edges(mrp, get_remote = False)
irtg = edge2irtg(edges, labels)
print(irtg)
print('_'*40)
from edge_to_irtg import edge2irtg
from process_c import *
from move_edges import lower_edge, raise_edge
from test_head_percolation import update_id_labels
from utils import number_edges
from a_star_mrp import get_roots
mrp_in = sys.argv[1]
with open(mrp_in) as infile:
for line in infile:
mrp = json.loads(line)
id = mrp['id']
#print(id)
#print('original')
labels = get_id2lex(mrp)
edges = get_mrp_edges(mrp, get_remote=False)
irtg_original = edge2irtg(edges, labels)
#print(irtg_original)
lowered = lower_edge(edges)
decompressed = decompress_c(edges, labels)
labels = update_id_labels(lowered, labels)
irtg_decompressed = edge2irtg(decompressed, labels)
#print('Lowered')
#print(irtg_decompressed)
if len(get_roots(decompressed)) == 0:
print(get_roots(decompressed))
print(id)
print(irtg_decompressed)
#print('_'*40)
from test_head_percolation import update_id_labels
from utils import number_edges
from a_star_mrp import get_roots
mrp_in = sys.argv[1]
with open(mrp_in) as infile:
for line in infile:
mrp = json.loads(line)
id = mrp['id']
print(id)
labels = get_id2lex(mrp)
edges = get_mrp_edges(mrp, get_remote=True)
edges = number_edges(edges, 'A')
print('original')
irtg_original = edge2irtg(edges, labels)
print(irtg_original)
compressed = compress_c_edge(edges)
labels = update_id_labels(compressed, labels)
print('COMPRESSED')
irtg_compressed = edge2irtg(edges, labels)
print(irtg_compressed)
print('RAISED U')
raised_u = raise_edge(compressed,
'U', ['L', 'H', 'P', 'S', 'A', 'D'],
label_dict=labels)
labels = update_id_labels(raised_u, labels)
irtg_raised_u = edge2irtg(raised_u, labels)
print(irtg_raised_u)
print('RAISED F')
raised_f = raise_edge(
raised_t = raise_edge(
raised_r,
'T',
['L', 'H', 'P', 'S', 'Q', 'E', 'F', 'D'],
label_dict=labels)
labels = update_id_labels(raised_t, labels)
raised_q = raise_edge(
raised_t,
'Q', [
'L', 'H', 'P', 'S', 'T', 'Q', 'E', 'F',
'D'
],
label_dict=labels)
updated_id_labels = update_id_labels(
raised_q, labels)
irtg_format_raised = edge2irtg(
raised_q, updated_id_labels)
#node_tokens = node_to_token_index(companion_data, mrp_dict, updated_id_labels, id)
node_tokens = node_to_token_index_mod(
companion_data, mrp_dict,
updated_id_labels, id)
#print(node_tokens)
aligned = percolate(raised_q, priority_queue,
updated_id_labels)
alignments = ''
for alignment in aligned.keys():
for node in aligned[alignment]:
if type(node) == str:
if '<root>' in node:
node = node[:-6]
alignments += str(node) + '|'
alignments += str(