def de_vectorize(self,
p_cpu,
out_vocab,
input_ptr_values,
schema=None,
table_po=None,
field_po=None,
return_tokens=False):
# convert output prediction vector into a human readable string
if self.model_id in [SEQ2SEQ_PG]:
return vec.de_vectorize_ptr(p_cpu,
out_vocab,
memory=input_ptr_values,
post_process=self.output_post_process,
return_tokens=return_tokens)
elif self.model_id in [BRIDGE]:
return vec.de_vectorize_field_ptr(
p_cpu,
out_vocab,
memory=input_ptr_values,
schema=schema,
table_po=table_po,
field_po=field_po,
post_process=self.output_post_process,
return_tokens=return_tokens)
elif self.model_id == SEQ2SEQ:
return vec.de_vectorize(p_cpu,
out_vocab,
post_process=self.output_post_process,
return_tokens=return_tokens)
else:
raise NotImplementedError
def eval_example(self,
example_pred,
example,
verbose=False,
example_id=None,
schema_graph=None):
example_pred = self.pack_sql(example_pred)
gt_ast = example.program_ast
s_correct, g_correct, o_correct, f_correct, w_correct, h_correct, l_correct = self.match(
example_pred, gt_ast)
sql_correct = s_correct and g_correct and o_correct and f_correct and w_correct and h_correct and l_correct
if not sql_correct and verbose:
assert (schema_graph is not None)
text = example.text
text_tokens = vec.de_vectorize(example.text_ids,
return_tokens=True)
text_ptr_values = example.text_ptr_values
print('Example {}'.format(example_id))
print('NL:\t{}'.format(text.encode('utf-8')))
print('NL tokens:\t{}'.format(encode_str_list(
text_tokens, 'utf-8')))
print('NL tokens (original):\t{}'.format(
encode_str_list(text_ptr_values, 'utf-8')))
print('NL tokens (recovered): {}'.format(vec.de_vectorize_ptr(
example.text_ptr_value_ids, self.value_vocab,
text_ptr_values).encode('utf-8'),
return_tokens=True))
for i, program in enumerate(example.program_list):
print('Target {}'.format(i))
print('- string: {}'.format(program.encode('utf-8')))
badges = [
get_badge(sql_correct),
get_badge(s_correct),
get_badge(g_correct),
get_badge(o_correct),
get_badge(f_correct),
get_badge(w_correct),
get_badge(h_correct),
get_badge(l_correct)
]
print(' '.join(badges))
serializer = SQLSerializer(schema_graph, self.field_vocab,
self.aggregation_ops,
self.arithmetic_ops, self.condition_ops,
self.logical_ops, self.value_vocab)
print('select clause: {}'.format(
serializer.serialize_select(example_pred['select'])))
print('group by clause: {}'.format(
serializer.serialize_group_by(example_pred['groupBy'])))
print('order by clause: {}'.format(
serializer.serialize_order_by(example_pred['orderBy'])))
return sql_correct, s_correct, g_correct, o_correct, f_correct, w_correct, h_correct, l_correct
def preprocess_example(split,
example,
args,
parsed_programs,
text_tokenize,
program_tokenize,
post_process,
trans_utils,
schema_graph,
vocabs,
verbose=False):
tu = trans_utils
text_vocab = vocabs['text']
program_vocab = vocabs['program']
def get_memory_values(features, raw_text, args):
if args.pretrained_transformer.startswith(
'bert-') and args.pretrained_transformer.endswith('-uncased'):
return utils.restore_feature_case(features, raw_text, tu)
else:
return features
def get_text_schema_adjacency_matrix(text_features, s_M):
schema_size = s_M.shape[0]
text_size = len(text_features)
full_size = schema_size + text_size
M = ssp.lil_matrix((full_size, full_size), dtype=np.int)
M[-schema_size:, -schema_size:] = s_M
return M
# sanity check
############################
query_oov = False
denormalized = False
schema_truncated = False
token_restored = True
############################
# Text feature extraction and set program ground truth list
if isinstance(example, Text2SQLExample):
if args.pretrained_transformer:
text_features = text_tokenize(example.text)
text_tokens, token_starts, token_ends = get_memory_values(
text_features, example.text, args)
if not token_starts:
token_restored = False
else:
text_tokens = text_tokenize(example.text, functional_tokens)
text_features = [t.lower() for t in text_tokens]
example.text_tokens = text_features
example.text_ptr_values = text_tokens
example.text_token_starts = token_starts
example.text_token_ends = token_ends
example.text_ids = vec.vectorize(text_features, text_vocab)
example.text_ptr_input_ids = vec.vectorize(text_features, text_vocab)
program_list = example.program_list
else:
text_tokens = example.example.text_ptr_values
text_features = example.example.text_tokens
program_list = example.example.program_list
# Schema feature extraction
if args.model_id in [BRIDGE]:
question_encoding = example.text if args.use_picklist else None
tables = sorted([schema_graph.get_table_id(t_name) for t_name in example.gt_table_names]) \
if args.use_oracle_tables else None
table_po, field_po = schema_graph.get_schema_perceived_order(tables)
schema_features, matched_values = schema_graph.get_serialization(
tu,
flatten_features=True,
table_po=table_po,
field_po=field_po,
use_typed_field_markers=args.use_typed_field_markers,
use_graph_encoding=args.use_graph_encoding,
question_encoding=question_encoding,
top_k_matches=args.top_k_picklist_matches,
match_threshold=args.anchor_text_match_threshold,
num_values_per_field=args.num_values_per_field,
no_anchor_text=args.no_anchor_text)
example.matched_values = matched_values
example.input_tokens, example.input_ptr_values, num_excluded_tables, num_excluded_fields = \
get_table_aware_transformer_encoder_inputs(text_tokens, text_features, schema_features, trans_utils)
schema_truncated = (num_excluded_fields > 0)
num_included_nodes = schema_graph.get_num_perceived_nodes(
table_po) + 1 - num_excluded_tables - num_excluded_fields
example.ptr_input_ids = vec.vectorize(example.input_tokens, text_vocab)
if args.read_picklist:
example.transformer_output_value_mask, value_features, value_tokens = \
get_transformer_output_value_mask(example.input_tokens, matched_values, tu)
example.primary_key_ids = schema_graph.get_primary_key_ids(
num_included_nodes, table_po=table_po, field_po=field_po)
example.foreign_key_ids = schema_graph.get_foreign_key_ids(
num_included_nodes, table_po=table_po, field_po=field_po)
example.field_type_ids = schema_graph.get_field_type_ids(
num_included_nodes, table_po=table_po, field_po=field_po)
example.table_masks = schema_graph.get_table_masks(num_included_nodes,
table_po=table_po,
field_po=field_po)
example.field_table_pos = schema_graph.get_field_table_pos(
num_included_nodes, table_po=table_po, field_po=field_po)
example.schema_M = schema_graph.adj_matrix
example.M = get_text_schema_adjacency_matrix(text_features,
example.schema_M)
else:
num_included_nodes = schema_graph.num_nodes
# Value copy feature extraction
if args.read_picklist:
constant_memory_features = text_features + value_features
constant_memory = text_tokens + value_tokens
example.text_ptr_values = constant_memory
else:
constant_memory_features = text_features
constant_ptr_value_ids, constant_unique_input_ids = vec.vectorize_ptr_in(
constant_memory_features, program_vocab)
if isinstance(example, Text2SQLExample):
example.text_ptr_value_ids = constant_ptr_value_ids
example.ptr_value_ids = constant_ptr_value_ids + [
program_vocab.size + len(constant_memory_features) + x
for x in range(num_included_nodes)
]
if not args.leaderboard_submission:
for j, program in enumerate(program_list):
if isinstance(example, Text2SQLExample):
ast, denormalized = get_ast(program, parsed_programs,
args.denormalize_sql, schema_graph)
if ast:
example.program_ast_list.append(ast)
program_tokens = program_tokenize(
ast,
schema=schema_graph,
omit_from_clause=args.omit_from_clause,
no_join_condition=args.no_join_condition,
in_execution_order=args.process_sql_in_execution_order)
assert (len(program_tokens) > 0)
else:
program_tokens = ['from']
program_tokens = [START_TOKEN] + program_tokens + [EOS_TOKEN]
program_input_ids = vec.vectorize(program_tokens,
program_vocab)
example.program_input_ids_list.append(program_input_ids)
if ast:
example.values = extract_values(ast, schema_graph)
else:
example.values = []
# Model I. Vanilla pointer-generator output
if args.model_id in [SEQ2SEQ_PG]:
program_text_ptr_value_ids = vec.vectorize_ptr_out(
program_tokens, program_vocab,
constant_unique_input_ids)
example.program_text_ptr_value_ids_list.append(
program_text_ptr_value_ids)
# sanity check
# NL pointer output contains tokens that does not belong to any of the following categories
# - reserved tokens
# - tokens in the NL input
# - tokens from environment variables (e.g. table schema)
############################
if program_vocab.unk_id in program_text_ptr_value_ids:
# unk_indices = [i for i, x in enumerate(program_text_ptr_value_ids) if x == program_vocab.unk_id]
# print('OOV I: {}'.format(' '.join([program_tokens[i] for i in unk_indices])))
# example.pretty_print(schema=schema_graph,
# de_vectorize_ptr=vec.de_vectorize_ptr,
# de_vectorize_field_ptr=vec.de_vectorize_field_ptr,
# rev_vocab=program_vocab,
# post_process=post_process)
query_oov = True
############################
# Model II. Bridge output
if ast:
denormalized_ast, _ = denormalize(ast,
schema_graph,
return_parse_tree=True)
example.program_denormalized_ast_list.append(
denormalized_ast)
tokenizer_output = program_tokenize(
denormalized_ast,
return_token_types=True,
schema=schema_graph,
keep_singleton_fields=True,
omit_from_clause=args.omit_from_clause,
no_join_condition=args.no_join_condition,
atomic_value=False,
num_token=NUM_TOKEN,
str_token=STR_TOKEN,
in_execution_order=args.process_sql_in_execution_order)
program_singleton_field_tokens, program_singleton_field_token_types = tokenizer_output[:
2]
else:
program_singleton_field_tokens = ['from']
program_singleton_field_token_types = [RESERVED_TOKEN_TYPE]
program_singleton_field_tokens = [
START_TOKEN
] + program_singleton_field_tokens + [EOS_TOKEN]
program_singleton_field_token_types = \
[RESERVED_TOKEN_TYPE] + program_singleton_field_token_types + [RESERVED_TOKEN_TYPE]
example.program_singleton_field_tokens_list.append(
program_singleton_field_tokens)
example.program_singleton_field_token_types_list.append(
program_singleton_field_token_types)
program_singleton_field_input_ids = vec.vectorize_singleton(
program_singleton_field_tokens,
program_singleton_field_token_types, program_vocab)
example.program_singleton_field_input_ids_list.append(
program_singleton_field_input_ids)
else:
# Model II. Bridge output
example.program_singleton_field_input_ids_list.append(
example.example.program_singleton_field_input_ids_list[j])
program_singleton_field_tokens = example.example.program_singleton_field_tokens_list[
j]
program_singleton_field_token_types = example.example.program_singleton_field_token_types_list[
j]
program_field_ptr_value_ids = vec.vectorize_field_ptr_out(
program_singleton_field_tokens,
program_singleton_field_token_types,
program_vocab,
constant_unique_input_ids,
max_memory_size=len(constant_memory_features),
schema=schema_graph,
num_included_nodes=num_included_nodes)
example.program_text_and_field_ptr_value_ids_list.append(
program_field_ptr_value_ids)
if example.gt_table_names_list:
table_ids = [
schema_graph.get_table_id(table_name)
for table_name in example.gt_table_names_list[j]
]
example.table_ids_list.append(table_ids)
assert ([schema_graph.get_table(x).name
for x in table_ids] == example.gt_table_names)
# sanity check
############################
# NL+Schema pointer output contains tokens that does not belong to any of the following categories
if verbose:
if program_vocab.unk_id in program_field_ptr_value_ids:
unk_indices = [
i for i, x in enumerate(program_field_ptr_value_ids)
if x == program_vocab.unk_id
]
print('OOV II: {}'.format(' '.join([
program_singleton_field_tokens[i] for i in unk_indices
])))
example.pretty_print(
schema=schema_graph,
de_vectorize_ptr=vec.de_vectorize_ptr,
de_vectorize_field_ptr=vec.de_vectorize_field_ptr,
rev_vocab=program_vocab,
post_process=post_process,
use_table_aware_te=(args.model_id in [BRIDGE]))
query_oov = True
if program_vocab.unk_field_id in program_field_ptr_value_ids:
example.pretty_print(
schema=schema_graph,
de_vectorize_ptr=vec.de_vectorize_ptr,
de_vectorize_field_ptr=vec.de_vectorize_field_ptr,
rev_vocab=program_vocab,
post_process=post_process,
use_table_aware_te=(args.model_id in [BRIDGE]))
if program_vocab.unk_table_id in program_field_ptr_value_ids:
example.pretty_print(
schema=schema_graph,
de_vectorize_ptr=vec.de_vectorize_ptr,
de_vectorize_field_ptr=vec.de_vectorize_field_ptr,
rev_vocab=program_vocab,
post_process=post_process,
use_table_aware_te=(args.model_id in [BRIDGE]))
############################
# Store the ground truth queries after preprocessing to run a relaxed evaluation or
# to evaluate with partial queries
if split == 'dev':
input_tokens = text_tokens
if args.model_id in [BRIDGE]:
_p = vec.de_vectorize_field_ptr(
program_field_ptr_value_ids,
program_vocab,
input_tokens,
schema=schema_graph,
post_process=post_process)
elif args.model_id in [SEQ2SEQ_PG]:
_p = vec.de_vectorize_ptr(program_text_ptr_value_ids,
program_vocab,
input_tokens,
post_process=post_process)
else:
_p = program
example.gt_program_list.append(_p)
# sanity check
############################
# try:
# assert(equal_ignoring_trivial_diffs(_p, program.lower(), verbose=True))
# except Exception:
# print('_p:\t\t{}'.format(_p))
# print('program:\t{}'.format(program))
# print()
# import pdb
# pdb.set_trace()
############################
example.run_unit_tests()
return query_oov, denormalized, schema_truncated, token_restored