-
Notifications
You must be signed in to change notification settings - Fork 23
/
parse_sparql_cwq.py
686 lines (587 loc) · 30.2 KB
/
parse_sparql_cwq.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
import re
import os
from components.utils import *
from executor.sparql_executor import execute_query_with_odbc
from executor.logic_form_util_cwq import lisp_to_sparql
class ParseError(Exception):
pass
class Parser:
def __init__(self):
pass
def parse_query_cwq(self, query, mid_list):
"""parse a sparql query into a s-expression
@param query: sparql query
@param mid_list: all mids appeared in the sparql query
"""
# print('QUERY', query)
lines = query.split('\n')
lines = [x for x in lines if x]
assert lines[0] != '#MANUAL SPARQL'
prefix_stmts = []
line_num = 0
while True:
l = lines[line_num]
if l.startswith('PREFIX'):
prefix_stmts.append(l)
else:
break
line_num = line_num + 1
next_line = lines[line_num]
assert next_line.startswith('SELECT DISTINCT ?x')
line_num = line_num + 1
next_line = lines[line_num]
assert next_line == 'WHERE {'
if re.match(r'ORDER BY .*\?\w*.* LIMIT 1', lines[-1]):
lines[-1] = lines[-1].replace('LIMIT 1', '').strip()
lines.append('LIMIT 1')
if re.match(r'LIMIT \d*', lines[-1]): # //TODO LIMIT n
lines[-1]='LIMIT 1' # transform to LIMIT 1, temporally
assert lines[-1] in ['}', 'LIMIT 1']
lines = lines[line_num:]
filter_string_flag = not all(['FILTER (str' not in x for x in lines])
# assert all(['FILTER (str' not in x for x in lines])
# normalize body lines
body_lines, spec_condition, filter_lines = self.normalize_body_lines(
lines, filter_string_flag)
body_lines = [x.strip() for x in body_lines] # strip spaces
# assert all([x.startswith('?') or x.startswith('ns') or x.startswith('FILTER') for x in body_lines])
# we only parse query following this format
if body_lines[0].startswith('FILTER'):
predefined_filter0 = body_lines[0]
predefined_filter1 = body_lines[1]
# filter_0_line validation
filter_0_valid = (predefined_filter0 == f'FILTER (?x != ?c)')
if not filter_0_valid:
for mid in mid_list:
filter_0_valid = filter_0_valid or (
predefined_filter0 == f'FILTER (?x != {mid})')
assert filter_0_valid
# filter_1_line validation
assert predefined_filter1 == "FILTER (!isLiteral(?x) OR lang(?x) = '' OR langMatches(lang(?x), 'en'))"
# if predefined_filter0 != f'FILTER (?x != ns:{topic_mid})':
# print('QUERY', query)
# print('First Filter')
# if predefined_filter1 != "FILTER (!isLiteral(?x) OR lang(?x) = '' OR langMatches(lang(?x), 'en'))":
# print('QUERY', query)
# print('Second Filter')
# if any([not (x.startswith('?') or x.startswith('ns:')) for x in body_lines]):
# print('Unprincipled Filter')
# print('QUERY', query)
body_lines = body_lines[2:]
# body line form assertion
assert all([(x.startswith('?') or x.startswith('ns:'))
for x in body_lines])
# print(body_lines)
var_dep_list = self.parse_naive_body(
body_lines, filter_lines, '?x', spec_condition)
s_expr = self.dep_graph_to_s_expr(var_dep_list, '?x', spec_condition)
return s_expr
def normalize_body_lines(self, lines, filter_string_flag=False):
"""return normalized body lines of sparql, specially return filter lines starting with `FILTER (str(`
@param lines: sparql lines list
@param filter_string_flag: flag indicates existence of filter lines
@return: (body_lines,
spec_condition,
# [
# ['SUPERLATIVE', argmax/argmin, arg_var, arg_r],
# ['COMPARATIVE', gt/lt/ge/le, compare_var, compare_value, compare_rel],
# ['RANGE', range_relation, range_var, range_year],
# ]
filter_lines
)
"""
spec_condition = []
# 1. get literal filter_lines
# ?x ns:base.biblioness.bibs_location.loc_type ?sk0 .
# FILTER (str(?sk0) = "Country")
if filter_string_flag:
filter_lines = [x.strip() for x in lines if 'FILTER (str' in x]
lines = [x.strip() for x in lines if 'FILTER (str' not in x]
else:
lines = [x.strip() for x in lines]
filter_lines = None
# 2. get compare lines
# 2.1 FILTER (?num > "2009-01-02"^^xsd:dateTime) .
# 2.2 FILTER (xsd:integer(?num) < 33351310952) .
if re.match(r'FILTER \(\?\w* (>|<|>=|<=) .*',lines[-2]) \
or re.match(r'FILTER \(xsd:integer\(\?\w*\) (>|<|>=|<=) .*',lines[-2]):
compare_line = lines.pop(-2)
compare_var = re.findall(r'\?\w*',compare_line)[0]
compare_operator = re.findall(r'(>|>=|<|<=)',compare_line)[0]
operator_mapper = {'<':'lt','<=':'le','>':'gt',">=":"ge"}
if "^^xsd:dateTime" in compare_line: # dateTime
compare_value = re.findall(r'".*"\^\^xsd:dateTime',compare_line)[0]
else: # number
compare_value = compare_line.replace(") .","").split(" ")[-1]
compare_value = compare_value.replace('"','') # remove \" in compare value
# print(variable,compare_operator,compare_value)
compare_condition = ['COMPARATIVE', operator_mapper[compare_operator],compare_var,compare_value]
spec_condition.append(compare_condition)
# 3. get range lines, move to the end of where clause
# WHERE {
# ns:m.04f_xd8 ns:government.government_office_or_title.office_holders ?y .
# ?y ns:government.government_position_held.office_holder ?x .
# FILTER(NOT EXISTS {?y ns:government.government_position_held.from ?sk0} ||
# EXISTS {?y ns:government.government_position_held.from ?sk1 .
# FILTER(xsd:datetime(?sk1) <= "2009-12-31"^^xsd:dateTime) })
# FILTER(NOT EXISTS {?y ns:government.government_position_held.to ?sk2} ||
# EXISTS {?y ns:government.government_position_held.to ?sk3 .
# FILTER(xsd:datetime(?sk3) >= "2009-01-01"^^xsd:dateTime) })
# }
start_line = -1
right_parantheses_line = -1
not_exists_num = 0
for i, line in enumerate(lines):
if line.startswith("FILTER(NOT EXISTS"):
not_exists_num +=1
if start_line == -1:
start_line = i
# if line.startswith("FILTER(") and "2015-08-10" in line and start_line != -1:
# meaningless_time_flag = True
if line == '}':
right_parantheses_line = i
if start_line != -1:
if not_exists_num==4: # redundant range filters
end_line = start_line+12
else:
end_line = start_line+6
assert end_line <= right_parantheses_line
if end_line==start_line+12: # discard redundant range filters
lines = lines[:start_line]+lines[end_line:right_parantheses_line] + \
lines[start_line:end_line-6]+lines[right_parantheses_line:]
else:
lines = lines[:start_line]+lines[end_line:right_parantheses_line] + \
lines[start_line:end_line]+lines[right_parantheses_line:]
# 4. get SUPERLATIVE lines
body_lines = []
if lines[-1] == 'LIMIT 1':
# spec_condition = argmax
# who did jackie robinson first play for?
# WHERE {
# ns:m.0443c ns:sports.pro_athlete.teams ?y .
# ?y ns:sports.sports_team_roster.team ?x .
# ?y ns:sports.sports_team_roster.from ?sk0 .
# }
# ORDER BY DESC(xsd:datetime(?sk0))
# LIMIT 1
order_line = lines[-2]
direction = 'argmax' if 'DESC(' in order_line else 'argmin'
compare_var = re.findall(r'\?\w*', order_line)[0]
# assert ('?sk0' in order_line) # variable in order_line
assert(compare_var in order_line)
_tmp_body_lines = lines[1:-3]
hit = False
for l in _tmp_body_lines:
if compare_var in l:
# self.parse_assert(l.endswith('?sk0 .') and not hit)
self.parse_assert(l.endswith(compare_var+" .")
and not hit) # appear only once
hit = True
arg_var, arg_r = l.split(' ')[0], l.split(' ')[1]
arg_r = arg_r[3:] # rm ns:
else:
body_lines.append(l)
superlative_cond = ['SUPERLATIVE',direction,arg_var,arg_r]
spec_condition.append(superlative_cond)
# if not lines[-4].startswith('FILTER(NOT EXISTS {?'):
# if filter_string_flag:
# return body_lines, [direction, arg_var, arg_r], filter_lines
# else:
# return body_lines, [direction, arg_var, arg_r], None
# else:
# # contains range constraints FILTER
# pass
# 4. process range lines
if body_lines: # already processed by superlative extraction
lines = body_lines
range_line_num = -6
else:
range_line_num = -7
if len(lines)>= abs(range_line_num) and lines[range_line_num].startswith('FILTER(NOT EXISTS {?'):
# WHERE {
# ns:m.04f_xd8 ns:government.government_office_or_title.office_holders ?y .
# ?y ns:government.government_position_held.office_holder ?x .
# FILTER(NOT EXISTS {?y ns:government.government_position_held.from ?sk0} ||
# EXISTS {?y ns:government.government_position_held.from ?sk1 .
# FILTER(xsd:datetime(?sk1) <= "2009-12-31"^^xsd:dateTime) })
# FILTER(NOT EXISTS {?y ns:government.government_position_held.to ?sk2} ||
# EXISTS {?y ns:government.government_position_held.to ?sk3 .
# FILTER(xsd:datetime(?sk3) >= "2009-01-01"^^xsd:dateTime) })
# }
if not body_lines:
body_lines = lines[1:-7]
range_lines = lines[-7:-1]
else:
body_lines = lines[:-6]
range_lines = lines[-6:]
range_prompt = range_lines[0]
range_prompt = range_prompt[range_prompt.index(
'{') + 1:range_prompt.index('}')]
range_var = range_prompt.split(' ')[0]
range_relation = range_prompt.split(' ')[1]
# range_relation = '.'.join(
# range_relation.split('.')[:2]) + '.time_macro'
range_relation = range_relation[3:] # rm ns:
range_start_time = re.findall(f'".*"\^\^',range_lines[2])[0].split("^^")[0].strip('"')
if range_start_time =='2015-08-10':
range_start_time = 'NOW'
range_start = range_lines[2].split(' ')[2]
range_start = range_start[1:]
range_start = range_start[:range_start.index('"')]
range_end = range_lines[5].split(' ')[2]
range_end = range_end[1:]
range_end = range_end[:range_end.index('"')]
assert range_start[:4] == range_end[:4]
# to fit parsable
# range_year = range_start[:4] + \
# '^^http://www.w3.org/2001/XMLSchema#dateTime' if range_start_time != 'NOW' else 'NOW'
range_year = range_start[:4] if range_start_time != 'NOW' else 'NOW'
range_start_cond = ['RANGE', range_relation, range_var, range_year]
spec_condition.append(range_start_cond)
# if filter_string_flag:
# return body_lines, ['range', range_var, range_relation, range_year], filter_lines
# else:
# return body_lines, ['range', range_var, range_relation, range_year], None
# body_lines not extracted yet
if not body_lines:
body_lines = lines[1:-1]
# if filter_string_flag:
# return body_lines, None, filter_lines
# else:
# return body_lines, None, None
return body_lines, spec_condition, filter_lines
def dep_graph_to_s_expr(self, var_dep_list, ret_var, spec_condition=None):
"""Convert dependancy graph to s_expression
@param var_dep_list: varialbe dependancy list
@param ret_var: return var
@param spec_condition: special condition
@return s_expression
"""
self.parse_assert(var_dep_list[0][0] == ret_var)
var_dep_list.reverse() # reverse the var_dep_list
parsed_dict = {} # dict for parsed variables
# spec_condition,
# # [
# # ['SUPERLATIVE', argmax/argmin, arg_var, arg_r],
# # ['COMPARATIVE', gt/lt/ge/le, compare_var, compare_value],
# # ['RANGE', range_relation, range_var, range_year],
# # ]
# specical condition var map {spec_var:idx in spec_condition}
spec_var_map = {cond[2]:i for i,cond in enumerate(spec_condition)} if spec_condition else None
# spec_var = spec_condition[1] if spec_condition is not None else None
for var_name, dep_relations in var_dep_list:
# expr = ''
dep_relations[0]
clause = self.triplet_to_clause(
var_name, dep_relations[0], parsed_dict)
for tri in dep_relations[1:]:
n_clause = self.triplet_to_clause(var_name, tri, parsed_dict)
clause = 'AND ({}) ({})'.format(n_clause, clause)
# if var_name == spec_var:
if spec_var_map and var_name in spec_var_map: # spec_condition
cond = spec_condition[spec_var_map[var_name]]
# if cond[0] == 'argmax' or cond[0] == 'argmin': # superlative
if cond[0]=='SUPERLATIVE':
#relation = spec_condition[2]
relation = cond[3]
clause = '{} ({}) {}'.format(
cond[1].upper(), clause, relation)
elif cond[0] == 'RANGE':
relation, time_point = cond[1], cond[3]
clause = 'TC ({}) {} {}'.format(clause, relation, time_point)
# n_clause = 'TC {} {}'.format(relation, time_point)
# clause = 'AND ({}) ({})'.format(n_clause, clause)
elif cond[0] == 'COMPARATIVE':
op = cond[1]
value = cond[3]
rel = cond[4]
n_clause = '{} {} {}'.format(op, rel, value)
clause = 'AND ({}) ({})'.format(n_clause, clause)
# pass
parsed_dict[var_name] = clause
res = '(' + parsed_dict[ret_var] + ')'
res = res.replace('xsd:','http://www.w3.org/2001/XMLSchema#')
return res
def triplet_to_clause(self, tgt_var, triplet, parsed_dict):
"""Convert a triplet to S_expression clause
@param tgt_var: target variable
@param triplet: triplet in sparql
@param parsed_dict: dict for variables already parsed
"""
if triplet[0] == tgt_var:
this = triplet[0]
other = triplet[-1]
if other in parsed_dict:
other = '(' + parsed_dict[other] + ')'
return 'JOIN {} {}'.format(triplet[1], other)
elif triplet[-1] == tgt_var:
this = triplet[-1]
other = triplet[0]
if other in parsed_dict:
other = '(' + parsed_dict[other] + ')'
return 'JOIN (R {}) {}'.format(triplet[1], other)
else:
raise ParseError()
def parse_assert(self, eval):
if not eval:
raise ParseError()
def parse_naive_body(self, body_lines, filter_lines, ret_var, spec_condition=None):
"""Parse body lines
@param body_lines: list of sparql body lines
@param ret_var: return var, default `?x`
@param filter_lines: lines that start with `FILTER (str(?`
@return: variable dependancy list
"""
# ret_variable
# body_lines
assert all([x[-1] == '.' for x in body_lines])
# filter lines assertion
if filter_lines:
assert all(['FILTER (str' in x for x in filter_lines])
triplets = [x.replace('"','') if "^^xsd:" in x else x for x in body_lines]
triplets = [x.split(' ') for x in triplets] # split by '
triplets = [x[:2] + [" ".join(x[2:-1]), x[-1]] if len(x)>4 else x for x in triplets] # avoid error splitting like "2100 Woodward Avenue"@en
triplets = [x[:-1] if x[-1] == '.' else x for x in triplets] # remove '.'
# remove ns
triplets = [[x[3:] if x.startswith(
'ns:') else x for x in tri] for tri in triplets]
# dependancy graph
triplets_pool = triplets
# while True:
# varaible dependancy list, in the form like [(?x,[['?x','ns:aaa.aaa.aaa','?y'],['ns:m.xx','ns:bbb.bbb.bbb','?x''])]
var_dep_list = []
successors = []
# firstly solve the return variable
dep_triplets, triplets_pool = self.resolve_dependancy(
triplets_pool, filter_lines, ret_var, successors)
var_dep_list.append((ret_var, dep_triplets))
# vars_pool = []
# go over un resolved vars
# for tri in triplets_pool:
# if tri[0].startswith('?') and tri[0] not in vars_pool and tri[0] != ret_var:
# vars_pool.append(tri[0])
# if tri[-1].startswith('?') and tri[-1] not in vars_pool and tri[-1] != ret_var:
# vars_pool.append(tri[-1])
# for tgt_var in vars_pool:
# dep_triplets, triplets_pool = self.resolve_dependancy(triplets_pool, tgt_var)
# self.parse_assert(len(dep_triplets) > 0)
# var_dep_list.append((tgt_var, dep_triplets))
# handle all the successor variables
while len(successors):
tgt_var = successors[0]
successors = successors[1:]
dep_triplets, triplets_pool = self.resolve_dependancy(
triplets_pool, filter_lines, tgt_var, successors)
# if (len(dep_triplets)==0):
# # no triplet for tgt_var
# # ?x ns:xxx ?c
# # ?c ns:xxx ?num
# # ORDER BY ?num LIMIT 1
# print(dep_triplets)
# assert len(dep_triplets) > 0 # at least one dependancy triplets
if len(dep_triplets) == 0:
# zero dep_triples, can be a 2-hop constraint
# e.g.
# 'ns:m.0d0x8 ns:government.political_district.representatives ?y .'
# '?y ns:government.government_position_held.office_holder ?x .'
# '?y ns:government.government_position_held.governmental_body ns:m.07t58 .'
# '?x ns:government.politician.government_positions_held ?c .'
if spec_condition and any([tgt_var in x for x in spec_condition]):
cond = []
for x in spec_condition:
if tgt_var in x:
cond = x
break
repeat = True
while repeat:
# tgt_var is a var in spec_condition
for (var, triplets) in var_dep_list:
if any([tgt_var in trip for trip in triplets]):
head_var = var # find the real constrained var
_temp_triplets = triplets[:]
triplets.clear()
for trip in _temp_triplets:
if tgt_var not in trip:
triplets.append(trip)
else:
# find the constraint relation
cons_rel = trip[1]
if trip[0] == head_var:
reversed_direction = False
else:
reversed_direction = True
cons_rel = f'(R {cons_rel})' if reversed_direction else cons_rel
# modify spec_condition
# spec_condition[1] = head_var
if cond[0]=='COMPARATIVE':
cond[2] = head_var
if len(cond)<5:
cond.append(cons_rel)
else:
cond[4] = "(JOIN " + cons_rel+" "+ cond[4]+")"
else: # SUPERLATIVE
cond[2] = head_var
cond[3] = "(JOIN "+ cons_rel+" "+cond[3]+")"
tgt_var = head_var
# check whether need to repeat
remove_idx=-1
for i,(var,triplets) in enumerate(var_dep_list):
if var == head_var:
if len(triplets)==0:
repeat = True
remove_idx = i
else:
repeat = False
break
if remove_idx>=0:
var_dep_list.pop(remove_idx)
else:
repeat=False
else:
# uncovered situation
assert 1 == 2
else:
"""dep_triplets not None"""
self.parse_assert(len(dep_triplets) >
0) # at least dependancy triplets
var_dep_list.append((tgt_var, dep_triplets))
if(len(triplets_pool) != 0):
print(triplets_pool)
self.parse_assert(len(triplets_pool) == 0)
return var_dep_list
def resolve_dependancy(self, triplets, filter_lines, target_var, successors):
"""resolve dependancy of variables
@param triplets: all sparql triplet lines
@param filter_lines: filter lines that start with `Filter (str(`
@param target_var: target variable
@param successors: successor variables of target variable
@return: dependancy triplets of target_var, left triplets (independant of target_var)
"""
dep = []
left = []
if not triplets: # empty triplets, target_var constrained by filter
# ns:m.0f9wd ns:influence.influence_node.influenced ?x .
# ?x ns:government.politician.government_positions_held ?c .
# ?c ns:government.government_position_held.from ?num .
# ORDER BY ?num LIMIT 1
pass
else:
for tri in triplets:
if tri[0] == target_var: # head is target variable
dep.append(tri) # add to dependancy triplets
# tail is variable
if tri[-1].startswith('?') and tri[-1] not in successors:
successor_var = tri[-1]
if filter_lines: # check filter variable `?sk0`
new_filter_lines = []
found_filter_variable = False
for line in filter_lines:
if successor_var in line:
found_filter_variable = True
line = line.replace(
'FILTER (str(', '').replace(')', '')
tuple_list = line.split('=')
var = tuple_list[0].strip()
value = tuple_list[1].strip()
assert successor_var == var
if value.isalpha():
tri[-1] = value+'@en'
else:
tri[-1] = value
# tri[-1] = value+'@en'
else:
new_filter_lines.append(line)
# remove corresponding filter_lines
if not found_filter_variable: # no filter variable found
# add to successor variable
successors.append(successor_var)
filter_lines = new_filter_lines
else:
# add to successor variable
successors.append(successor_var)
elif tri[-1] == target_var: # tail is target variable
dep.append(tri) # add to dependancy triplets
# head is variable
if tri[0].startswith('?') and tri[0] not in successors:
successors.append(tri[0]) # add to successor variable
else:
left.append(tri) # left triplets
return dep, left
def augment_with_s_expr_cwq(split, check_execute_accuracy=False):
"""augment original cwq datasets with s-expression"""
if split == 'train':
dataset_train = load_json(f'data/CWQ/origin/ComplexWebQuestions_train.json')
dataset_dev = load_json(f'data/CWQ/origin/ComplexWebQuestions_dev.json')
dataset = dataset_train + dataset_dev
else:
dataset = load_json(f'data/CWQ/origin/ComplexWebQuestions_{split}.json')
total_num = 0
hit_num = 0
execute_hit_num = 0
failed_instances = []
for i,data in enumerate(dataset):
total_num += 1
sparql = data['sparql'] # sparql string
instance, flag_success = convert_cwq_sparql_instance(sparql, data)
if flag_success:
hit_num += 1
if check_execute_accuracy:
execute_right_flag = False
try:
execute_ans = [s.replace('http://rdf.freebase.com/ns/','') if type(s) == str else str(s) for s in execute_query_with_odbc(lisp_to_sparql(instance['SExpr']))]
if split=='test':
gold_ans = [s.replace('http://rdf.freebase.com/ns/','') if type(s) == str else str(s) for s in execute_query_with_odbc(lisp_to_sparql(data['sparql']))]
else:
gold_ans = [x['answer_id'] for x in data['answers']]
if set(execute_ans) == set(gold_ans):
execute_hit_num +=1
execute_right_flag = True
else:
temp = lisp_to_sparql(instance['SExpr'])
temp = [s.replace('http://rdf.freebase.com/ns/','') if type(s) == str else str(s) for s in execute_query_with_odbc(lisp_to_sparql(instance['SExpr']))]
# print(f'{i}: SExpr generation:{flag_success}, Execute right:{execute_right_flag}')
instance['SExpr_execute_right'] = execute_right_flag
except Exception:
temp = [s.replace('http://rdf.freebase.com/ns/','') if type(s) == str else str(s) for s in execute_query_with_odbc(lisp_to_sparql(instance['SExpr']))]
# instance['SExpr_executed_succeed']=False
instance['SExpr_execute_right'] = execute_right_flag
if not execute_right_flag:
pass
# print(f'ID:{instance["ID"]},\nExpected Ansewr:{gold_ans},\nGot Answer:{execute_ans}')
else:
# if check_execute_accuracy:
# instance['SExpr_execute_right'] = False
failed_instances.append(instance)
# print(hit_num, total_num, hit_num/total_num, len(dataset))
if (i+1)%100==0:
print(f'In the First {i+1} questions, S-Expression Gen rate [{split}]: {hit_num}, {total_num}, {hit_num/total_num}, {i+1}')
if check_execute_accuracy:
print(f'In the First {i+1} questions, Execute right rate [{split}]: {execute_hit_num}, {total_num}, {execute_hit_num/total_num}, {i+1}', )
print(f'S-Expression Gen rate [{split}]: {hit_num}, {total_num}, {hit_num/total_num}, {len(dataset)}')
print(f'Execute right rate [{split}]: {execute_hit_num}, {total_num}, {execute_hit_num/total_num}, {len(dataset)}', )
sexpr_dir = 'data/CWQ/sexpr'
if not os.path.exists(sexpr_dir):
os.makedirs(sexpr_dir)
dump_json(dataset, f'{sexpr_dir}/CWQ.{split}.expr.json', indent=4)
#dump_json(failed_instances, f'{sexpr_dir}/CWQ.{split}.expr_failed.json', indent=4)
def convert_cwq_sparql_instance(sparql, origin_data):
"""convert a cwq sparql to a s_expr"""
pattern_str = r'ns:m\.0\w*'
# pattern = re.compile(pattern_str)
mid_list = list(set([mid.strip()
for mid in re.findall(pattern_str, sparql)]))
try:
s_expr = parser.parse_query_cwq(sparql, mid_list)
except AssertionError:
# print(f'Error processing sparql: {sparql}')
s_expr = 'null'
origin_data['SExpr'] = s_expr
return origin_data, s_expr != 'null'
def parse_cwq_sparql(check_execute_accuracy=False):
"""Parse CWQ sparqls into s-expressions"""
augment_with_s_expr_cwq('train',check_execute_accuracy)
# augment_with_s_expr_cwq('dev',check_execute_accuracy)
augment_with_s_expr_cwq('test',check_execute_accuracy)
if __name__ == '__main__':
parser = Parser()
parse_cwq_sparql(check_execute_accuracy=False)