Repository navigation
Expand file tree
/
Copy pathconverter_api.py
More file actions
965 lines (838 loc) · 36.9 KB
/
Copy pathconverter_api.py
File metadata and controls
965 lines (838 loc) · 36.9 KB
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
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
from fastapi import FastAPI, Form, HTTPException, Response
from typing import Optional
import typing as t
import uvicorn
import re
import os
import time
import json
# Default to the pure-Python tokenizer; set SQLGLOTRS_TOKENIZER=1 to enable the faster Rust
# tokenizer (sqlglotrs). Must run BEFORE `import sqlglot`, which reads this at import time.
os.environ.setdefault("SQLGLOTRS_TOKENIZER", "0")
import sqlglot
import logging
from datetime import datetime
from log_collector import setup_logger, log_records
from sqlglot.optimizer.qualify_columns import quote_identifiers
from sqlglot import parse_one
from sqlglot.dialects.snowflake_backticks import SnowflakeBackticks
from apis.utils.multidialect import split_pg_outer, split_custom_sql_alias, _splice
from guardrail.main import StorageServiceClient
from guardrail.main import extract_sql_components_per_table_with_alias, get_table_infos
from guardrail.rules_validator import validate_queries
from apis.utils.helpers import (
strip_comment,
sanitize_comments,
unsupported_functionality_identifiers,
extract_functions_from_query,
categorize_functions,
add_comment_to_query,
replace_struct_in_query,
ensure_select_from_values,
extract_udfs,
load_supported_functions,
extract_db_and_Table_names,
extract_joins_from_query,
extract_cte_n_subquery_list,
normalize_unicode_spaces,
transform_table_part,
transform_catalog_schema_only,
set_cte_names_case_sensitively,
apply_e6_ast_transforms,
fix_quote_escapes,
restore_quote_escapes,
extract_large_in_clauses,
restore_large_in_clauses,
)
from formatting_utils import preserve_formatting
if t.TYPE_CHECKING:
from sqlglot._typing import E
setup_logger()
ENABLE_GUARDRAIL = os.getenv("ENABLE_GUARDRAIL", "False")
STORAGE_ENGINE_URL = os.getenv("STORAGE_ENGINE_URL", "localhost") # cops-beta1-storage-storage-blue
STORAGE_ENGINE_PORT = os.getenv("STORAGE_ENGINE_PORT", 9005)
SKIP_COMMENT = os.getenv("SKIP_COMMENT", "True") # Always strip multi-line comments
FIX_QUOTE_ESCAPES = os.getenv("FIX_QUOTE_ESCAPES", "False") # Fix '' inside single-quoted strings
E6_EXECUTOR_TYPE = os.getenv(
"E6_EXECUTOR_TYPE", "java"
) # "java" divides TO_UNIX_TIMESTAMP by 1000; "native" does not
# Samsung Tableau dashboards: detect the native subquery by its "Custom SQL Query" alias
# (multidialect split_custom_sql_alias) instead of the backtick scan / parse-error fallback.
SAMSUNG_TABLEAU_DASHBOARD = os.getenv("SAMSUNG_TABLEAU_DASHBOARD", "False")
# Rust tokenizer flag ("0" default -> pure-Python; "1" -> sqlglotrs). Defaulted before the sqlglot
# import above; read here only for visibility/logging.
SQLGLOTRS_TOKENIZER = os.getenv("SQLGLOTRS_TOKENIZER", "0")
# Opt-in hybrid multidialect path. When on, the multidialect branch rewrites Databricks
# backtick identifiers to Postgres double-quoted identifiers up front (so the whole query
# parses as Postgres), and the e6 generator reparses any Anonymous-node function it finds on
# the postgres path in the Databricks dialect (see e6.anonymous_sql). Read here and in e6.py.
HYBRID_MULTIDIALECT = os.getenv("HYBRID_MULTIDIALECT", "False").lower() == "true"
storage_service_client = None
app = FastAPI()
logger = logging.getLogger(__name__)
if ENABLE_GUARDRAIL.lower() == "true":
logger.info("Storage Engine URL: ", STORAGE_ENGINE_URL)
logger.info("Storage Engine Port: ", STORAGE_ENGINE_PORT)
storage_service_client = StorageServiceClient(host=STORAGE_ENGINE_URL, port=STORAGE_ENGINE_PORT)
logger.info("Storage Service Client is created")
logger.info(
"Environment flags — ENABLE_GUARDRAIL: %s, SKIP_COMMENT: %s, FIX_QUOTE_ESCAPES: %s, "
"E6_EXECUTOR_TYPE: %s, SAMSUNG_TABLEAU_DASHBOARD: %s, STORAGE_ENGINE_URL: %s, "
"STORAGE_ENGINE_PORT: %s",
ENABLE_GUARDRAIL,
SKIP_COMMENT,
FIX_QUOTE_ESCAPES,
E6_EXECUTOR_TYPE,
SAMSUNG_TABLEAU_DASHBOARD,
STORAGE_ENGINE_URL,
STORAGE_ENGINE_PORT,
)
logger.info(
"Tokenizer — SQLGLOTRS_TOKENIZER: %s, Rust tokenizer active: %s",
SQLGLOTRS_TOKENIZER,
sqlglot.tokens.USE_RS_TOKENIZER,
)
def escape_unicode(s: str) -> str:
"""
Turn every non-ASCII (including all Unicode spaces) into \\uXXXX,
so even “invisible” characters become visible in logs.
"""
return s.encode("unicode_escape").decode("ascii")
def _region_to_e6(region_sql: str, from_sql: str, pretty: bool) -> str:
"""Transpile ONE region of a multi-dialect BI-tool query to e6.
This deliberately runs the SAME steps as the main /convert-query pipeline
(normalize -> strip comments -> parse -> quote identifiers -> ensure SELECT FROM
VALUES -> case-sensitive CTE names -> .sql -> replace STRUCT), so a region
produces output identical to any other query going through the converter.
The one thing that varies is ``from_sql`` -- the dialect is "rewired" per region:
- "databricks" for the primary path (the merged pg->dbr intermediary) and for each
inner subquery, and
- "postgres" for the fallback OUTER, so e6 applies its Postgres-specific rules
(e.g. dropping a 1-arg numeric ``TRUNC`` that Databricks would mis-read as a
date truncation).
"""
# Same input cleanup the main path does before parsing.
region_sql = normalize_unicode_spaces(region_sql)
if SKIP_COMMENT.lower() == "true":
region_sql, _ = strip_comment(region_sql)
# Large IN-clause optimization: pull out oversized literal lists before
# parsing so sqlglot doesn't build/traverse thousands of AST nodes.
region_sql, in_replacements = extract_large_in_clauses(region_sql)
# Parse with the region's own source dialect, then run the standard e6 steps.
_t = time.perf_counter()
tree = sqlglot.parse_one(region_sql, read=from_sql, error_level=None)
_parse_ms = (time.perf_counter() - _t) * 1000
# AST transforms folded into one combined walk (sanitize comments + VALUES/CTE fixes +
# case-sensitive CTE table renaming). Identifier quoting is no longer a separate walk here:
# the e6 Generator quotes reserved-keyword identifiers at emit time.
_t = time.perf_counter()
tree = apply_e6_ast_transforms(tree)
_transform_ms = (time.perf_counter() - _t) * 1000
# from_dialect=from_sql is what lets e6 honor the source dialect's semantics.
# quote_reserved_keywords=True folds the old quote_identifiers walk into generation.
_t = time.perf_counter()
out = tree.sql(dialect="e6", from_dialect=from_sql, pretty=pretty, quote_reserved_keywords=True)
_generate_ms = (time.perf_counter() - _t) * 1000
out = replace_struct_in_query(out)
# Restore original IN-clause values after transpilation.
out = restore_large_in_clauses(out, in_replacements)
logger.info(
"[TRANSPILE-TIMING] region=%s: parse=%.1f ms, transforms=%.1f ms, generate=%.1f ms",
from_sql,
_parse_ms,
_transform_ms,
_generate_ms,
)
return out
@app.post("/convert-query")
async def convert_query(
query: str = Form(...),
query_id: Optional[str] = Form("NO_ID_MENTIONED"),
from_sql: str = Form(...),
to_sql: Optional[str] = Form("e6"),
feature_flags: Optional[str] = Form(None),
):
timestamp = datetime.now().isoformat()
to_sql = to_sql.lower()
flags_dict = {}
if feature_flags:
try:
flags_dict = json.loads(feature_flags)
except json.JSONDecodeError as je:
return HTTPException(status_code=500, detail=str(je))
samsung = SAMSUNG_TABLEAU_DASHBOARD.lower() == "true"
if flags_dict.get("MULTIDIALECT", False) or samsung:
# Multi-dialect BI-tool query: a Postgres outer wrapper around inner subqueries
# written in another dialect (INNER_DIALECT, default "databricks"). Rule:
# - a subquery that FAILS the Postgres parse is inner-dialect -> INNER_DIALECT -> e6
# - everything else is the Postgres wrapper -> postgres -> e6
# then splice the pieces back. We never merge into the inner dialect first (that
# mis-reads Postgres constructs, e.g. numeric TRUNC -> date truncation).
inner_dialect = flags_dict.get("INNER_DIALECT", "databricks").lower()
pretty = flags_dict.get("PRETTY_PRINT", True)
# --- Stage-by-stage transpile timing ---
_t_arrival = time.perf_counter()
logger.info(
"[TRANSPILE-TIMING] %s — query arrived at transpiler (%d chars)",
query_id,
len(query),
)
# HYBRID_MULTIDIALECT: rewrite Databricks backtick identifiers to Postgres double-quoted
# identifiers up front, so the whole query (outer identifiers included) parses as
# Postgres. Databricks functions Postgres can't resolve surface as Anonymous nodes and
# are reparsed as Databricks in the e6 generator (see e6.anonymous_sql).
#
# Token-aware (not a blind regex): only the backticks the Postgres tokenizer emits as
# UNKNOWN are identifier delimiters -- a backtick inside a '...' string is part of one
# STRING token, so it is left untouched. `` ` `` and ``"`` are both one char, so swapping
# in place preserves every offset.
if HYBRID_MULTIDIALECT:
_toks = sqlglot.tokenize(query, dialect="postgres")
_bt = [
tok.start
for tok in _toks
if tok.token_type == sqlglot.tokens.TokenType.UNKNOWN and tok.text == "`"
]
if _bt:
_chars = list(query)
for _pos in _bt:
_chars[_pos] = '"'
query = "".join(_chars)
# Samsung Tableau dashboards always alias the native subquery "Custom SQL Query", so we
# detect it by that alias alone (no backtick scan / parse-error fallback).
_t = time.perf_counter()
if samsung:
outer, inner_subqueries = split_custom_sql_alias(query)
else:
outer, inner_subqueries = split_pg_outer(query)
_tokenize_ms = (time.perf_counter() - _t) * 1000
logger.info(
"[TRANSPILE-TIMING] %s — tokenize+subquery-detection=%.1f ms (%d inner subquery(s))",
query_id,
_tokenize_ms,
len(inner_subqueries),
)
# Outer wrapper: postgres -> e6 (logs its own parse/generate times).
converted_query = _region_to_e6(outer, "postgres", pretty)
# Each inner subquery: <inner_dialect> -> e6, then splice it back into the outer.
_splice_ms = 0.0
for marker, subquery in inner_subqueries.items():
inner_e6 = _region_to_e6(subquery, inner_dialect, pretty)
_t = time.perf_counter()
converted_query = _splice(converted_query, marker, inner_e6)
_splice_ms += (time.perf_counter() - _t) * 1000
logger.info("[TRANSPILE-TIMING] %s — reconstruct/splice=%.1f ms", query_id, _splice_ms)
_total_ms = (time.perf_counter() - _t_arrival) * 1000
logger.info(
"[TRANSPILE-TIMING] %s — TOTAL=%.1f ms — sending transpiled query away",
query_id,
_total_ms,
)
logger.info(
"%s AT %s — MULTIDIALECT: %d inner subquery(s) via %s, outer via postgres:\n%s",
query_id,
timestamp,
len(inner_subqueries),
inner_dialect,
converted_query,
)
return {"converted_query": converted_query}
elif flags_dict.get("POWERBI_SF_TO_DBR", False):
# Intermediary vanilla Snowflake -> Databricks transpile, run
# unconditionally whenever the flag is set -- the caller's `from_sql`
# is intentionally ignored so the planner can opt a query into the
# SF -> DBR step without first having to assert its dialect.
# - on success: `query` is now Databricks-shaped (SF identifiers
# turned into backticks, function renames, etc.).
# - on failure: keep the original query unchanged; assume it was
# already Databricks-shaped.
# In either case the query going into the downstream pipeline is
# Databricks-shaped, so override `from_sql` to "databricks" so the
# rest of the handler parses it with the right dialect.
logger.info(
"%s AT %s — POWERBI_SF_TO_DBR: intermediary Snowflake -> Databricks transpile (from_sql=%s ignored)",
query_id,
timestamp,
from_sql,
)
try:
# Parse as Snowflake but tolerate Databricks backtick identifiers:
# Power BI queries mix ANSI double-quoted identifiers (the outer
# wrapper) with backtick identifiers (inner CTEs). Plain Snowflake
# chokes on the backticks; SnowflakeBackticks treats both " and `
# as identifiers, so every identifier is correctly carried over to
# Databricks backticks while single-quoted strings stay literals.
query = sqlglot.transpile(
query,
read=SnowflakeBackticks,
write="databricks",
identify=False,
)[0]
logger.info(
"%s AT %s — Intermediary (SF -> DBR) result:\n%s",
query_id,
timestamp,
query,
)
except Exception as e:
logger.warning(
"%s AT %s — Intermediary SF -> DBR failed (%s); forwarding original query as Databricks",
query_id,
timestamp,
e,
)
from_sql = "databricks"
if not query or not query.strip():
logger.info(
"%s AT %s FROM %s — Empty query received, returning empty result",
query_id,
timestamp,
from_sql.upper(),
)
return {"converted_query": ""}
try:
logger.info(
"%s AT %s FROM %s — Original:\n%s",
query_id,
timestamp,
from_sql.upper(),
escape_unicode(query),
)
query = normalize_unicode_spaces(query)
logger.info(
"%s AT %s FROM %s — Normalized (escaped):\n%s",
query_id,
timestamp,
from_sql.upper(),
escape_unicode(query),
)
if SKIP_COMMENT.lower() == "true":
query, comment = strip_comment(query)
logger.info("%s — SKIP_COMMENT: stripped comments", query_id)
# Large IN-clause optimization: extract oversized literal-only value
# lists before parsing so sqlglot doesn't build/traverse thousands of
# AST nodes for values that need no dialect transformation.
query, in_replacements = extract_large_in_clauses(query)
if in_replacements:
logger.info(
"%s — Large IN-clause optimization: extracted %d clause(s)",
query_id,
len(in_replacements),
)
tree = sqlglot.parse_one(query, read=from_sql, error_level=None)
tree = sanitize_comments(tree)
if flags_dict.get("USE_TWO_PHASE_QUALIFICATION_SCHEME", False):
logger.info("%s — USE_TWO_PHASE_QUALIFICATION_SCHEME: enabled", query_id)
# Check if we should only transform catalog.schema without full transpilation
if flags_dict.get("SKIP_E6_TRANSPILATION", False):
logger.info("%s — SKIP_E6_TRANSPILATION: enabled", query_id)
transformed_query = transform_catalog_schema_only(query, from_sql)
# transformed_query = add_comment_to_query(transformed_query, comment)
logger.info(
"%s AT %s FROM %s — Catalog.Schema Transformed Query:\n%s",
query_id,
timestamp,
from_sql.upper(),
transformed_query,
)
return {"converted_query": transformed_query}
tree = transform_table_part(tree)
tree2 = quote_identifiers(tree, dialect=to_sql)
values_ensured_ast = ensure_select_from_values(tree2)
cte_names_equivalence_checked_ast = set_cte_names_case_sensitively(values_ensured_ast)
double_quotes_added_query = cte_names_equivalence_checked_ast.sql(
dialect=to_sql,
from_dialect=from_sql,
pretty=flags_dict.get("PRETTY_PRINT", True),
)
double_quotes_added_query = replace_struct_in_query(double_quotes_added_query)
# Restore original IN-clause values that were extracted before parsing.
double_quotes_added_query = restore_large_in_clauses(
double_quotes_added_query, in_replacements
)
# Preserve original formatting if enabled via feature flag
if flags_dict.get("PRESERVE_FORMATTING", False):
logger.info("%s — PRESERVE_FORMATTING: enabled", query_id)
double_quotes_added_query = preserve_formatting(
query, double_quotes_added_query, from_sql, to_sql
)
# double_quotes_added_query = add_comment_to_query(double_quotes_added_query, comment)
logger.info(
"%s AT %s FROM %s — Transpiled Query:\n%s",
query_id,
timestamp,
from_sql.upper(),
double_quotes_added_query,
)
return {"converted_query": double_quotes_added_query}
except Exception as e:
logger.error(
"%s AT %s FROM %s — Error:\n%s",
query_id,
timestamp,
from_sql.upper(),
str(e),
exc_info=True,
)
raise HTTPException(status_code=500, detail=str(e))
@app.get("/health")
def health_check():
return Response(status_code=200)
@app.post("/guardrail")
async def gaurd(
query: str = Form(...),
schema: str = Form(...),
catalog: str = Form(...),
):
try:
if storage_service_client is not None:
parsed = sqlglot.parse(query, error_level=None)
queries, tables = extract_sql_components_per_table_with_alias(parsed)
# tables = client.get_table_names(catalog_name="hive", db_name="tpcds_1000")
table_map = get_table_infos(tables, storage_service_client, catalog, schema)
logger.info("table info is ", table_map)
violations_found = validate_queries(queries, table_map)
if violations_found:
return {"action": "deny", "violations": violations_found}
else:
return {"action": "allow", "violations": []}
else:
detail = (
"Storage Service Not Initialized. Guardrail service status: " + ENABLE_GUARDRAIL
)
logger.error(detail)
raise HTTPException(status_code=500, detail=detail)
except Exception as e:
logger.error(f"Error in guardrail API: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/transpile-guardrail")
async def Transgaurd(
query: str = Form(...),
schema: str = Form(...),
catalog: str = Form(...),
from_sql: str = Form(...),
to_sql: Optional[str] = Form("e6"),
):
to_sql = to_sql.lower()
try:
if storage_service_client is not None:
# This is the main method will which help in transpiling to our e6data SQL dialects from other dialects
converted_query = sqlglot.transpile(query, read=from_sql, write=to_sql, identify=False)[
0
]
# This is additional steps to replace the STRUCT(STRUCT()) --> {{}}
converted_query = replace_struct_in_query(converted_query)
converted_query_ast = parse_one(converted_query, read=to_sql)
double_quotes_added_query = quote_identifiers(converted_query_ast, dialect=to_sql).sql(
dialect=to_sql
)
# ------------------------#
# GuardRail Application
parsed = sqlglot.parse(double_quotes_added_query, error_level=None)
# now lets validate the query
queries, tables = extract_sql_components_per_table_with_alias(parsed)
table_map = get_table_infos(tables, storage_service_client, catalog, schema)
violations_found = validate_queries(queries, table_map)
if violations_found:
return {"action": "deny", "violations": violations_found}
else:
return {"action": "allow", "violations": []}
else:
detail = (
"Storage Service Not Initialized. Guardrail service status: " + ENABLE_GUARDRAIL
)
raise HTTPException(status_code=500, detail=detail)
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
@app.post("/statistics")
async def stats_api(
query: str = Form(...),
query_id: Optional[str] = Form("NO_ID_MENTIONED"),
from_sql: str = Form(...),
to_sql: Optional[str] = Form("e6"),
feature_flags: Optional[str] = Form(None),
):
"""
API endpoint to extract supported and unsupported SQL functions from a query.
"""
timestamp = datetime.now().isoformat()
to_sql = to_sql.lower()
logger.info(f"{query_id} AT start time: {timestamp} FROM {from_sql.upper()}")
flags_dict = {}
if feature_flags:
try:
flags_dict = json.loads(feature_flags)
except json.JSONDecodeError as je:
return HTTPException(status_code=500, detail=str(je))
try:
supported_functions_in_e6 = load_supported_functions(to_sql)
# Functions treated as keywords (no parentheses required)
functions_as_keywords = [
"LIKE",
"ILIKE",
"RLIKE",
"AT TIME ZONE",
"||",
"DISTINCT",
"QUALIFY",
]
# Exclusion list for words that are followed by '(' but are not functions
exclusion_list = [
"AS",
"AND",
"THEN",
"OR",
"ELSE",
"WHEN",
"WHERE",
"FROM",
"JOIN",
"OVER",
"ON",
"ALL",
"NOT",
"BETWEEN",
"UNION",
"SELECT",
"BY",
"GROUP",
"EXCEPT",
"SETS",
]
# Regex patterns
function_pattern = r"\b([A-Za-z_][A-Za-z0-9_]*)\s*\("
keyword_pattern = (
r"\b(?:" + "|".join([re.escape(func) for func in functions_as_keywords]) + r")\b"
)
if not query.strip():
logger.info("Query is empty or only contains comments!")
return {
"supported_functions": [],
"unsupported_functions": [],
"udf_list": [],
"converted-query": "Query is empty or only contains comments.",
"unsupported_functions_after_transpilation": [],
"executable": "NO",
"error": True,
"log_records": log_records,
}
query, comment = strip_comment(query)
# Extract functions from the query
all_functions = extract_functions_from_query(
query, function_pattern, keyword_pattern, exclusion_list
)
supported, unsupported = categorize_functions(
all_functions, supported_functions_in_e6, functions_as_keywords
)
from_dialect_function_list = load_supported_functions(from_sql)
udf_list, unsupported = extract_udfs(unsupported, from_dialect_function_list)
# --------------------------
# HANDLING PARSING ERRORS
# --------------------------
executable = "YES"
error_flag = False
try:
# ------------------------------
# Step 1: Parse the Original Query
# ------------------------------
original_ast = parse_one(query, read=from_sql)
tables_list = extract_db_and_Table_names(original_ast)
supported, unsupported = unsupported_functionality_identifiers(
original_ast, unsupported, supported
)
values_ensured_ast = ensure_select_from_values(original_ast)
cte_names_equivalence_ast = set_cte_names_case_sensitively(values_ensured_ast)
query = cte_names_equivalence_ast.sql(from_sql)
# ------------------------------
# Step 2: Transpile the Query
# ------------------------------
tree = sqlglot.parse_one(query, read=from_sql, error_level=None)
tree2 = quote_identifiers(tree, dialect=to_sql)
double_quotes_added_query = tree2.sql(
dialect=to_sql,
from_dialect=from_sql,
pretty=flags_dict.get("PRETTY_PRINT", True),
)
double_quotes_added_query = replace_struct_in_query(double_quotes_added_query)
# Preserve original formatting if enabled via feature flag
if flags_dict.get("PRESERVE_FORMATTING", False):
double_quotes_added_query = preserve_formatting(
query, double_quotes_added_query, from_sql, to_sql
)
double_quotes_added_query = add_comment_to_query(double_quotes_added_query, comment)
logger.info("Got the converted query!!!!")
all_functions_converted_query = extract_functions_from_query(
double_quotes_added_query,
function_pattern,
keyword_pattern,
exclusion_list,
)
(
supported_functions_in_converted_query,
unsupported_functions_in_converted_query,
) = categorize_functions(
all_functions_converted_query,
supported_functions_in_e6,
functions_as_keywords,
)
double_quote_ast = parse_one(double_quotes_added_query, read=to_sql)
supported_in_converted, unsupported_in_converted = (
unsupported_functionality_identifiers(
double_quote_ast,
unsupported_functions_in_converted_query,
supported_functions_in_converted_query,
)
)
joins_list = extract_joins_from_query(original_ast)
cte_values_subquery_list = extract_cte_n_subquery_list(original_ast)
if unsupported_in_converted:
executable = "NO"
logger.info(
f"{query_id} executed in {(datetime.now() - datetime.fromisoformat(timestamp)).total_seconds()} seconds FROM {from_sql.upper()}\n"
"-----------------------\n"
"--- Original query ---\n"
"-----------------------\n"
f"{query}"
"-----------------------\n"
"--- Transpiled query ---\n"
"-----------------------\n"
f"{double_quotes_added_query}"
)
except Exception as e:
logger.info(
f"{query_id} executed in {(datetime.now() - datetime.fromisoformat(timestamp)).total_seconds()} seconds FROM {from_sql.upper()}\n"
"-----------------------\n"
"--- Original query ---\n"
"-----------------------\n"
f"{query}"
"-----------------------\n"
"-------- Error --------\n"
"-----------------------\n"
f"{str(e)}"
)
error_message = f"{str(e)}"
error_flag = True
double_quotes_added_query = error_message
tables_list = []
joins_list = []
cte_values_subquery_list = []
unsupported_in_converted = []
executable = "NO"
return {
"supported_functions": supported,
"unsupported_functions": set(unsupported),
"udf_list": set(udf_list),
"converted-query": double_quotes_added_query, # Will contain error message if error_flag is True
"unsupported_functions_after_transpilation": set(unsupported_in_converted),
"executable": executable,
"tables_list": set(tables_list),
"joins_list": joins_list,
"cte_values_subquery_list": cte_values_subquery_list,
"error": error_flag,
"log_records": log_records,
}
except Exception as e:
logger.error(
f"{query_id} occurred at time {datetime.now().isoformat()} with processing time {(datetime.now() - datetime.fromisoformat(timestamp)).total_seconds()} FROM {from_sql.upper()}\n"
"-----------------------\n"
"--- Original query ---\n"
"-----------------------\n"
f"{query}"
"-----------------------\n"
"-------- Error --------\n"
"-----------------------\n"
f"{str(e)}"
)
return {
"supported_functions": [],
"unsupported_functions": [],
"udf_list": [],
"converted-query": f"Internal Server Error: {str(e)}",
"unsupported_functions_after_transpilation": [],
"executable": "NO",
"tables_list": [],
"joins_list": [],
"cte_values_subquery_list": [],
"error": True,
"log_records": log_records,
}
@app.post("/guardstats")
async def guardstats(
query: str = Form(...),
from_sql: str = Form(...),
to_sql: Optional[str] = Form("e6"),
schema: str = Form(...),
catalog: str = Form(...),
):
to_sql = to_sql.lower()
try:
supported_functions_in_e6 = load_supported_functions(to_sql)
# Functions treated as keywords (no parentheses required)
functions_as_keywords = [
"LIKE",
"ILIKE",
"RLIKE",
"AT TIME ZONE",
"||",
"DISTINCT",
"QUALIFY",
]
# Exclusion list for words that are followed by '(' but are not functions
exclusion_list = [
"AS",
"AND",
"THEN",
"OR",
"ELSE",
"WHEN",
"WHERE",
"FROM",
"JOIN",
"OVER",
"ON",
"ALL",
"NOT",
"BETWEEN",
"UNION",
"SELECT",
"BY",
"GROUP",
"EXCEPT",
]
# Regex patterns
function_pattern = r"\b([A-Za-z_][A-Za-z0-9_]*)\s*\("
keyword_pattern = (
r"\b(?:" + "|".join([re.escape(func) for func in functions_as_keywords]) + r")\b"
)
query, comment = strip_comment(query)
# Extract functions from the query
all_functions = extract_functions_from_query(
query, function_pattern, keyword_pattern, exclusion_list
)
supported, unsupported = categorize_functions(
all_functions, supported_functions_in_e6, functions_as_keywords
)
logger.info(f"supported: {supported}\n\nunsupported: {unsupported}")
original_ast = parse_one(query, read=from_sql)
tables_list = extract_db_and_Table_names(original_ast)
supported, unsupported = unsupported_functionality_identifiers(
original_ast, unsupported, supported
)
values_ensured_ast = ensure_select_from_values(original_ast)
query = values_ensured_ast.sql(dialect=from_sql)
tree = sqlglot.parse_one(query, read=from_sql, error_level=None)
tree2 = quote_identifiers(tree, dialect=to_sql)
double_quotes_added_query = tree2.sql(dialect=to_sql, from_dialect=from_sql)
double_quotes_added_query = replace_struct_in_query(double_quotes_added_query)
# Note: PRESERVE_FORMATTING not available here as no flags_dict
# Can be added if needed by adding feature_flags parameter to this endpoint
double_quotes_added_query = add_comment_to_query(double_quotes_added_query, comment)
all_functions_converted_query = extract_functions_from_query(
double_quotes_added_query, function_pattern, keyword_pattern, exclusion_list
)
(
supported_functions_in_converted_query,
unsupported_functions_in_converted_query,
) = categorize_functions(
all_functions_converted_query,
supported_functions_in_e6,
functions_as_keywords,
)
double_quote_ast = parse_one(double_quotes_added_query, read=to_sql)
supported_in_converted, unsupported_in_converted = unsupported_functionality_identifiers(
double_quote_ast,
unsupported_functions_in_converted_query,
supported_functions_in_converted_query,
)
from_dialect_func_list = load_supported_functions(from_sql)
udf_list, unsupported = extract_udfs(unsupported, from_dialect_func_list)
executable = "NO" if unsupported_in_converted else "YES"
if storage_service_client is not None:
parsed = sqlglot.parse(double_quotes_added_query, error_level=None)
queries, tables = extract_sql_components_per_table_with_alias(parsed)
# tables = client.get_table_names(catalog_name="hive", db_name="tpcds_1000")
table_map = get_table_infos(tables, storage_service_client, catalog, schema)
logger.info("table info is ", table_map)
violations_found = validate_queries(queries, table_map)
joins_list = extract_joins_from_query(original_ast)
cte_values_subquery_list = extract_cte_n_subquery_list(original_ast)
if violations_found:
return {
"supported_functions": supported,
"unsupported_functions": unsupported,
"udf_list": udf_list,
"converted-query": double_quotes_added_query,
"unsupported_functions_after_transpilation": unsupported_in_converted,
"executable": executable,
"tables_list": tables_list,
"joins_list": joins_list,
"cte_values_subquery_list": cte_values_subquery_list,
"action": "deny",
"violations": violations_found,
"log_records": log_records,
}
else:
return {
"supported_functions": supported,
"unsupported_functions": unsupported,
"converted-query": double_quotes_added_query,
"unsupported_functions_after_transpilation": unsupported_in_converted,
"udf_list": udf_list,
"executable": executable,
"tables_list": tables_list,
"joins_list": joins_list,
"cte_values_subquery_list": cte_values_subquery_list,
"action": "allow",
"violations": [],
"log_records": log_records,
}
else:
detail = (
"Storage Service Not Initialized. Guardrail service status: " + ENABLE_GUARDRAIL
)
raise HTTPException(status_code=500, detail=detail)
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
def _effective_cpus():
"""Cores this process may actually use — the cgroup CPU limit, not the host.
multiprocessing.cpu_count() reports the HOST/node cores, ignoring the
Kubernetes CPU limit (enforced via cgroup cfs_quota). On a pod limited to N
cores it still returns the node's cores, so the worker math over-provisions.
Read the real quota instead.
"""
# cgroup v2 (k8s default): /sys/fs/cgroup/cpu.max -> "<quota> <period>"
try:
quota, period = open("/sys/fs/cgroup/cpu.max").read().split()
if quota != "max":
return max(1.0, int(quota) / int(period))
except (OSError, ValueError):
pass
# cgroup v1: cpu.cfs_quota_us / cpu.cfs_period_us
try:
q = int(open("/sys/fs/cgroup/cpu/cpu.cfs_quota_us").read())
p = int(open("/sys/fs/cgroup/cpu/cpu.cfs_period_us").read())
if q > 0:
return max(1.0, q / p)
except (OSError, ValueError):
pass
# no CPU limit set: fall back to affinity, then host count
try:
return float(len(os.sched_getaffinity(0)))
except AttributeError:
return float(os.cpu_count() or 1)
if __name__ == "__main__":
cpu_cores = _effective_cpus()
# Transpiling is CPU-bound (pins a core), NOT I/O-bound, so run ~1 worker
# per core (+1 to overlap the small per-request I/O). The old (2*cores)+1
# rule is for I/O-bound servers and here just oversubscribes the cores.
optimal_workers = max(2, round(cpu_cores) + 1)
# Allow override via environment variable
workers = int(os.getenv("UVICORN_WORKERS", optimal_workers))
logger.info(f"Detected {cpu_cores:.2f} effective CPU cores, using {workers} workers")
uvicorn.run(
"converter_api:app",
host="0.0.0.0",
port=8100,
proxy_headers=True,
workers=workers,
)