clickzetta-semantic-model-generator 1.0.1__py3-none-any.whl → 1.0.3__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (22) hide show
  1. {clickzetta_semantic_model_generator-1.0.1.dist-info → clickzetta_semantic_model_generator-1.0.3.dist-info}/METADATA +5 -5
  2. {clickzetta_semantic_model_generator-1.0.1.dist-info → clickzetta_semantic_model_generator-1.0.3.dist-info}/RECORD +22 -19
  3. semantic_model_generator/clickzetta_utils/clickzetta_connector.py +91 -33
  4. semantic_model_generator/clickzetta_utils/env_vars.py +7 -2
  5. semantic_model_generator/data_processing/cte_utils.py +1 -1
  6. semantic_model_generator/generate_model.py +588 -224
  7. semantic_model_generator/llm/dashscope_client.py +4 -2
  8. semantic_model_generator/llm/enrichment.py +144 -57
  9. semantic_model_generator/llm/progress_tracker.py +16 -15
  10. semantic_model_generator/relationships/__init__.py +15 -0
  11. semantic_model_generator/relationships/discovery.py +202 -0
  12. semantic_model_generator/tests/clickzetta_connector_test.py +3 -7
  13. semantic_model_generator/tests/cte_utils_test.py +1 -1
  14. semantic_model_generator/tests/generate_model_classification_test.py +12 -2
  15. semantic_model_generator/tests/llm_enrichment_test.py +152 -46
  16. semantic_model_generator/tests/relationship_discovery_test.py +114 -0
  17. semantic_model_generator/tests/relationships_filters_test.py +166 -30
  18. semantic_model_generator/tests/utils_test.py +1 -1
  19. semantic_model_generator/validate/keywords.py +453 -53
  20. semantic_model_generator/validate/schema.py +4 -2
  21. {clickzetta_semantic_model_generator-1.0.1.dist-info → clickzetta_semantic_model_generator-1.0.3.dist-info}/LICENSE +0 -0
  22. {clickzetta_semantic_model_generator-1.0.1.dist-info → clickzetta_semantic_model_generator-1.0.3.dist-info}/WHEEL +0 -0
semantic_model_generator/validate/keywords.py CHANGED
@@ -1,57 +1,457 @@
1
1
  # ClickZetta SQL reserved words mirrored from clickzetta.zettapark._internal.reserved_words.
2
2
  # Reference: https://yunqi.tech/documents/sql-reference
3
3
  CZ_RESERVED_WORDS = {
4
- "ABS", "ABSENT", "ABSOLUTE", "ACOS", "ACTION", "ADD", "ALL", "ALLOCATE", "ALTER", "AND", "ANY",
5
- "ARE", "ARRAY", "ARRAY_AGG", "ARRAY_MAX_CARDINALITY", "AS", "ASC", "ASENSITIVE", "ASIN", "ASSERTION",
6
- "ASYMMETRIC", "AT", "ATAN", "ATOMIC", "AUTHORIZATION", "AVG", "BEGIN", "BEGIN_FRAME",
7
- "BEGIN_PARTITION", "BETWEEN", "BIGINT", "BINARY", "BIT", "BIT_LENGTH", "BLOB", "BOOLEAN", "BOTH",
8
- "BY", "CALL", "CALLED", "CARDINALITY", "CASCADE", "CASCADED", "CASE", "CAST", "CATALOG", "CEIL",
9
- "CEILING", "CHAR", "CHARACTER", "CHARACTER_LENGTH", "CHAR_LENGTH", "CHECK", "CLASSIFIER", "CLOB",
10
- "CLOSE", "COALESCE", "COLLATE", "COLLECT", "COLUMN", "COMMIT", "CONDITION", "CONNECT", "CONNECTION",
11
- "CONSTRAINT", "CONTAINS", "CONVERT", "COPY", "CORR", "CORRESPONDING", "COS", "COSH", "COUNT",
12
- "COVAR_POP", "COVAR_SAMP", "CREATE", "CROSS", "CUBE", "CUME_DIST", "CURRENT", "CURRENT_CATALOG",
13
- "CURRENT_DATE", "CURRENT_DEFAULT_TRANSFORM_GROUP", "CURRENT_PATH", "CURRENT_ROLE", "CURRENT_ROW",
14
- "CURRENT_SCHEMA", "CURRENT_TIME", "CURRENT_TIMESTAMP", "CURRENT_TRANSFORM_GROUP_FOR_TYPE",
15
- "CURRENT_USER", "CURSOR", "CYCLE", "DATALINK", "DATE", "DAY", "DEALLOCATE", "DEC", "DECFLOAT",
16
- "DECIMAL", "DECLARE", "DEFAULT", "DEFINE", "DELETE", "DENSE_RANK", "DEREF", "DESCRIBE",
17
- "DETERMINISTIC", "DIAGNOSTICS", "DISCONNECT", "DISTINCT", "DLNEWCOPY", "DLPREVIOUSCOPY",
18
- "DLURLCOMPLETE", "DLURLCOMPLETEONLY", "DLURLCOMPLETEWRITE", "DLURLPATH", "DLURLPATHONLY",
19
- "DLURLPATHWRITE", "DLURLSCHEME", "DLURLSERVER", "DLVALUE", "DOMAIN", "DOUBLE", "DROP", "DYNAMIC",
20
- "EACH", "ELEMENT", "ELSE", "EMPTY", "END", "END-EXEC", "END_FRAME", "END_PARTITION", "EQUALS",
21
- "ESCAPE", "EVERY", "EXCEPT", "EXCEPTION", "EXEC", "EXECUTE", "EXISTS", "EXP", "EXTERNAL", "EXTRACT",
22
- "FALSE", "FETCH", "FILTER", "FIRST", "FIRST_VALUE", "FLOAT", "FLOOR", "FOR", "FOREIGN", "FRAME_ROW",
23
- "FREE", "FROM", "FULL", "FUNCTION", "FUSION", "GET", "GLOBAL", "GO", "GOTO", "GRANT", "GROUP",
24
- "GROUPING", "GROUPS", "HAVING", "HOLD", "HOUR", "IDENTITY", "IMMEDIATE", "IMPORT", "IN",
25
- "INCREMENTAL_DELETE_DELTA", "INDICATOR", "INITIAL", "INITIALLY", "INNER", "INOUT", "INPUT",
26
- "INSENSITIVE", "INSERT", "INT", "INTEGER", "INTERSECT", "INTERSECTION", "INTERVAL", "INTO", "IS",
27
- "ISOLATION", "JOIN", "JSON_ARRAY", "JSON_ARRAYAGG", "JSON_EXISTS", "JSON_OBJECT", "JSON_OBJECTAGG",
28
- "JSON_QUERY", "JSON_TABLE", "JSON_TABLE_PRIMITIVE", "JSON_VALUE", "KEY", "LAG", "LANGUAGE", "LARGE",
29
- "LAST", "LAST_VALUE", "LATERAL", "LEAD", "LEADING", "LEFT", "LEVEL", "LIKE", "LIKE_REGEX", "LISTAGG",
30
- "LN", "LOCAL", "LOCALTIME", "LOCALTIMESTAMP", "LOG", "LOG10", "LOWER", "MATCH", "MATCHES",
31
- "MATCH_NUMBER", "MATCH_RECOGNIZE", "MAX", "MEASURES", "MEMBER", "MERGE", "METHOD", "MIN", "MINUTE",
32
- "MOD", "MODIFIES", "MODULE", "MONTH", "MULTISET", "NAMES", "NATIONAL", "NATURAL", "NCHAR", "NCLOB",
33
- "NEW", "NEXT", "NO", "NONE", "NORMALIZE", "NOT", "NTH_VALUE", "NTILE", "NULL", "NULLIF", "NUMERIC",
34
- "OCCURRENCES_REGEX", "OCTET_LENGTH", "OF", "OFFSET", "OLD", "OMIT", "ON", "ONE", "ONLY", "OPEN",
35
- "OPTION", "OR", "ORDER", "OUT", "OUTER", "OUTPUT", "OVER", "OVERLAPS", "OVERLAY", "PAD", "PARAMETER",
36
- "PARTIAL", "PARTITION", "PATTERN", "PER", "PERCENT", "PERCENTILE_CONT", "PERCENTILE_DISC",
37
- "PERCENT_RANK", "PERIOD", "PERMUTE", "PORTION", "POSITION", "POSITION_REGEX", "POWER", "PRECEDES",
38
- "PRECISION", "PREPARE", "PRESERVE", "PRIMARY", "PRIOR", "PRIVILEGES", "PROCEDURE", "PTF", "PUBLIC",
39
- "RANGE", "RANK", "READ", "READS", "REAL", "RECURSIVE", "REF", "REFERENCES", "REFERENCING",
40
- "REGR_AVGX", "REGR_AVGY", "REGR_COUNT", "REGR_INTERCEPT", "REGR_R2", "REGR_SLOPE", "REGR_SXX",
41
- "REGR_SXY", "REGR_SYY", "RELATIVE", "RELEASE", "RESTRICT", "RESULT", "RETURN", "RETURNS", "REVOKE",
42
- "RIGHT", "ROLLBACK", "ROLLUP", "ROW", "ROWS", "ROW_NUMBER", "RUNNING", "SAVEPOINT", "SCHEMA",
43
- "SCOPE", "SCROLL", "SEARCH", "SECOND", "SECTION", "SEEK", "SELECT", "SENSITIVE", "SESSION",
44
- "SESSION_USER", "SET", "SHOW", "SIMILAR", "SIN", "SINH", "SIZE", "SKIP", "SMALLINT", "SOME", "SPACE",
45
- "SPECIFIC", "SPECIFICTYPE", "SQL", "SQLCODE", "SQLERROR", "SQLEXCEPTION", "SQLSTATE", "SQLWARNING",
46
- "SQRT", "START", "STATIC", "STDDEV_POP", "STDDEV_SAMP", "STREAM", "SUBMULTISET", "SUBSET",
47
- "SUBSTRING", "SUBSTRING_REGEX", "SUCCEEDS", "SUM", "SYMMETRIC", "SYSTEM", "SYSTEM_TIME",
48
- "SYSTEM_USER", "TABLE", "TABLESAMPLE", "TAN", "TANH", "TEMPORARY", "THEN", "TIME", "TIMESTAMP",
49
- "TIMEZONE_HOUR", "TIMEZONE_MINUTE", "TO", "TRAILING", "TRANSACTION", "TRANSLATE", "TRANSLATE_REGEX",
50
- "TRANSLATION", "TREAT", "TRIGGER", "TRIM", "TRIM_ARRAY", "TRUE", "TRUNCATE", "UESCAPE", "UNION",
51
- "UNIQUE", "UNKNOWN", "UNMATCHED", "UNNEST", "UPDATE", "UPPER", "USAGE", "USER", "USING", "VALUE",
52
- "VALUES", "VALUE_OF", "VARBINARY", "VARCHAR", "VARYING", "VAR_POP", "VAR_SAMP", "VERSIONING", "VIEW",
53
- "WHEN", "WHENEVER", "WHERE", "WIDTH_BUCKET", "WINDOW", "WITH", "WITHIN", "WITHOUT", "WORK", "WRITE",
54
- "XML", "XMLAGG", "XMLATTRIBUTES", "XMLBINARY", "XMLCAST", "XMLCOMMENT", "XMLCONCAT", "XMLDOCUMENT",
55
- "XMLELEMENT", "XMLEXISTS", "XMLFOREST", "XMLITERATE", "XMLNAMESPACES", "XMLPARSE", "XMLPI",
56
- "XMLQUERY", "XMLSERIALIZE", "XMLTABLE", "XMLTEXT", "XMLVALIDATE", "YEAR", "ZONE",
4
+ "ABS",
5
+ "ABSENT",
6
+ "ABSOLUTE",
7
+ "ACOS",
8
+ "ACTION",
9
+ "ADD",
10
+ "ALL",
11
+ "ALLOCATE",
12
+ "ALTER",
13
+ "AND",
14
+ "ANY",
15
+ "ARE",
16
+ "ARRAY",
17
+ "ARRAY_AGG",
18
+ "ARRAY_MAX_CARDINALITY",
19
+ "AS",
20
+ "ASC",
21
+ "ASENSITIVE",
22
+ "ASIN",
23
+ "ASSERTION",
24
+ "ASYMMETRIC",
25
+ "AT",
26
+ "ATAN",
27
+ "ATOMIC",
28
+ "AUTHORIZATION",
29
+ "AVG",
30
+ "BEGIN",
31
+ "BEGIN_FRAME",
32
+ "BEGIN_PARTITION",
33
+ "BETWEEN",
34
+ "BIGINT",
35
+ "BINARY",
36
+ "BIT",
37
+ "BIT_LENGTH",
38
+ "BLOB",
39
+ "BOOLEAN",
40
+ "BOTH",
41
+ "BY",
42
+ "CALL",
43
+ "CALLED",
44
+ "CARDINALITY",
45
+ "CASCADE",
46
+ "CASCADED",
47
+ "CASE",
48
+ "CAST",
49
+ "CATALOG",
50
+ "CEIL",
51
+ "CEILING",
52
+ "CHAR",
53
+ "CHARACTER",
54
+ "CHARACTER_LENGTH",
55
+ "CHAR_LENGTH",
56
+ "CHECK",
57
+ "CLASSIFIER",
58
+ "CLOB",
59
+ "CLOSE",
60
+ "COALESCE",
61
+ "COLLATE",
62
+ "COLLECT",
63
+ "COLUMN",
64
+ "COMMIT",
65
+ "CONDITION",
66
+ "CONNECT",
67
+ "CONNECTION",
68
+ "CONSTRAINT",
69
+ "CONTAINS",
70
+ "CONVERT",
71
+ "COPY",
72
+ "CORR",
73
+ "CORRESPONDING",
74
+ "COS",
75
+ "COSH",
76
+ "COUNT",
77
+ "COVAR_POP",
78
+ "COVAR_SAMP",
79
+ "CREATE",
80
+ "CROSS",
81
+ "CUBE",
82
+ "CUME_DIST",
83
+ "CURRENT",
84
+ "CURRENT_CATALOG",
85
+ "CURRENT_DATE",
86
+ "CURRENT_DEFAULT_TRANSFORM_GROUP",
87
+ "CURRENT_PATH",
88
+ "CURRENT_ROLE",
89
+ "CURRENT_ROW",
90
+ "CURRENT_SCHEMA",
91
+ "CURRENT_TIME",
92
+ "CURRENT_TIMESTAMP",
93
+ "CURRENT_TRANSFORM_GROUP_FOR_TYPE",
94
+ "CURRENT_USER",
95
+ "CURSOR",
96
+ "CYCLE",
97
+ "DATALINK",
98
+ "DATE",
99
+ "DAY",
100
+ "DEALLOCATE",
101
+ "DEC",
102
+ "DECFLOAT",
103
+ "DECIMAL",
104
+ "DECLARE",
105
+ "DEFAULT",
106
+ "DEFINE",
107
+ "DELETE",
108
+ "DENSE_RANK",
109
+ "DEREF",
110
+ "DESCRIBE",
111
+ "DETERMINISTIC",
112
+ "DIAGNOSTICS",
113
+ "DISCONNECT",
114
+ "DISTINCT",
115
+ "DLNEWCOPY",
116
+ "DLPREVIOUSCOPY",
117
+ "DLURLCOMPLETE",
118
+ "DLURLCOMPLETEONLY",
119
+ "DLURLCOMPLETEWRITE",
120
+ "DLURLPATH",
121
+ "DLURLPATHONLY",
122
+ "DLURLPATHWRITE",
123
+ "DLURLSCHEME",
124
+ "DLURLSERVER",
125
+ "DLVALUE",
126
+ "DOMAIN",
127
+ "DOUBLE",
128
+ "DROP",
129
+ "DYNAMIC",
130
+ "EACH",
131
+ "ELEMENT",
132
+ "ELSE",
133
+ "EMPTY",
134
+ "END",
135
+ "END-EXEC",
136
+ "END_FRAME",
137
+ "END_PARTITION",
138
+ "EQUALS",
139
+ "ESCAPE",
140
+ "EVERY",
141
+ "EXCEPT",
142
+ "EXCEPTION",
143
+ "EXEC",
144
+ "EXECUTE",
145
+ "EXISTS",
146
+ "EXP",
147
+ "EXTERNAL",
148
+ "EXTRACT",
149
+ "FALSE",
150
+ "FETCH",
151
+ "FILTER",
152
+ "FIRST",
153
+ "FIRST_VALUE",
154
+ "FLOAT",
155
+ "FLOOR",
156
+ "FOR",
157
+ "FOREIGN",
158
+ "FRAME_ROW",
159
+ "FREE",
160
+ "FROM",
161
+ "FULL",
162
+ "FUNCTION",
163
+ "FUSION",
164
+ "GET",
165
+ "GLOBAL",
166
+ "GO",
167
+ "GOTO",
168
+ "GRANT",
169
+ "GROUP",
170
+ "GROUPING",
171
+ "GROUPS",
172
+ "HAVING",
173
+ "HOLD",
174
+ "HOUR",
175
+ "IDENTITY",
176
+ "IMMEDIATE",
177
+ "IMPORT",
178
+ "IN",
179
+ "INCREMENTAL_DELETE_DELTA",
180
+ "INDICATOR",
181
+ "INITIAL",
182
+ "INITIALLY",
183
+ "INNER",
184
+ "INOUT",
185
+ "INPUT",
186
+ "INSENSITIVE",
187
+ "INSERT",
188
+ "INT",
189
+ "INTEGER",
190
+ "INTERSECT",
191
+ "INTERSECTION",
192
+ "INTERVAL",
193
+ "INTO",
194
+ "IS",
195
+ "ISOLATION",
196
+ "JOIN",
197
+ "JSON_ARRAY",
198
+ "JSON_ARRAYAGG",
199
+ "JSON_EXISTS",
200
+ "JSON_OBJECT",
201
+ "JSON_OBJECTAGG",
202
+ "JSON_QUERY",
203
+ "JSON_TABLE",
204
+ "JSON_TABLE_PRIMITIVE",
205
+ "JSON_VALUE",
206
+ "KEY",
207
+ "LAG",
208
+ "LANGUAGE",
209
+ "LARGE",
210
+ "LAST",
211
+ "LAST_VALUE",
212
+ "LATERAL",
213
+ "LEAD",
214
+ "LEADING",
215
+ "LEFT",
216
+ "LEVEL",
217
+ "LIKE",
218
+ "LIKE_REGEX",
219
+ "LISTAGG",
220
+ "LN",
221
+ "LOCAL",
222
+ "LOCALTIME",
223
+ "LOCALTIMESTAMP",
224
+ "LOG",
225
+ "LOG10",
226
+ "LOWER",
227
+ "MATCH",
228
+ "MATCHES",
229
+ "MATCH_NUMBER",
230
+ "MATCH_RECOGNIZE",
231
+ "MAX",
232
+ "MEASURES",
233
+ "MEMBER",
234
+ "MERGE",
235
+ "METHOD",
236
+ "MIN",
237
+ "MINUTE",
238
+ "MOD",
239
+ "MODIFIES",
240
+ "MODULE",
241
+ "MONTH",
242
+ "MULTISET",
243
+ "NAMES",
244
+ "NATIONAL",
245
+ "NATURAL",
246
+ "NCHAR",
247
+ "NCLOB",
248
+ "NEW",
249
+ "NEXT",
250
+ "NO",
251
+ "NONE",
252
+ "NORMALIZE",
253
+ "NOT",
254
+ "NTH_VALUE",
255
+ "NTILE",
256
+ "NULL",
257
+ "NULLIF",
258
+ "NUMERIC",
259
+ "OCCURRENCES_REGEX",
260
+ "OCTET_LENGTH",
261
+ "OF",
262
+ "OFFSET",
263
+ "OLD",
264
+ "OMIT",
265
+ "ON",
266
+ "ONE",
267
+ "ONLY",
268
+ "OPEN",
269
+ "OPTION",
270
+ "OR",
271
+ "ORDER",
272
+ "OUT",
273
+ "OUTER",
274
+ "OUTPUT",
275
+ "OVER",
276
+ "OVERLAPS",
277
+ "OVERLAY",
278
+ "PAD",
279
+ "PARAMETER",
280
+ "PARTIAL",
281
+ "PARTITION",
282
+ "PATTERN",
283
+ "PER",
284
+ "PERCENT",
285
+ "PERCENTILE_CONT",
286
+ "PERCENTILE_DISC",
287
+ "PERCENT_RANK",
288
+ "PERIOD",
289
+ "PERMUTE",
290
+ "PORTION",
291
+ "POSITION",
292
+ "POSITION_REGEX",
293
+ "POWER",
294
+ "PRECEDES",
295
+ "PRECISION",
296
+ "PREPARE",
297
+ "PRESERVE",
298
+ "PRIMARY",
299
+ "PRIOR",
300
+ "PRIVILEGES",
301
+ "PROCEDURE",
302
+ "PTF",
303
+ "PUBLIC",
304
+ "RANGE",
305
+ "RANK",
306
+ "READ",
307
+ "READS",
308
+ "REAL",
309
+ "RECURSIVE",
310
+ "REF",
311
+ "REFERENCES",
312
+ "REFERENCING",
313
+ "REGR_AVGX",
314
+ "REGR_AVGY",
315
+ "REGR_COUNT",
316
+ "REGR_INTERCEPT",
317
+ "REGR_R2",
318
+ "REGR_SLOPE",
319
+ "REGR_SXX",
320
+ "REGR_SXY",
321
+ "REGR_SYY",
322
+ "RELATIVE",
323
+ "RELEASE",
324
+ "RESTRICT",
325
+ "RESULT",
326
+ "RETURN",
327
+ "RETURNS",
328
+ "REVOKE",
329
+ "RIGHT",
330
+ "ROLLBACK",
331
+ "ROLLUP",
332
+ "ROW",
333
+ "ROWS",
334
+ "ROW_NUMBER",
335
+ "RUNNING",
336
+ "SAVEPOINT",
337
+ "SCHEMA",
338
+ "SCOPE",
339
+ "SCROLL",
340
+ "SEARCH",
341
+ "SECOND",
342
+ "SECTION",
343
+ "SEEK",
344
+ "SELECT",
345
+ "SENSITIVE",
346
+ "SESSION",
347
+ "SESSION_USER",
348
+ "SET",
349
+ "SHOW",
350
+ "SIMILAR",
351
+ "SIN",
352
+ "SINH",
353
+ "SIZE",
354
+ "SKIP",
355
+ "SMALLINT",
356
+ "SOME",
357
+ "SPACE",
358
+ "SPECIFIC",
359
+ "SPECIFICTYPE",
360
+ "SQL",
361
+ "SQLCODE",
362
+ "SQLERROR",
363
+ "SQLEXCEPTION",
364
+ "SQLSTATE",
365
+ "SQLWARNING",
366
+ "SQRT",
367
+ "START",
368
+ "STATIC",
369
+ "STDDEV_POP",
370
+ "STDDEV_SAMP",
371
+ "STREAM",
372
+ "SUBMULTISET",
373
+ "SUBSET",
374
+ "SUBSTRING",
375
+ "SUBSTRING_REGEX",
376
+ "SUCCEEDS",
377
+ "SUM",
378
+ "SYMMETRIC",
379
+ "SYSTEM",
380
+ "SYSTEM_TIME",
381
+ "SYSTEM_USER",
382
+ "TABLE",
383
+ "TABLESAMPLE",
384
+ "TAN",
385
+ "TANH",
386
+ "TEMPORARY",
387
+ "THEN",
388
+ "TIME",
389
+ "TIMESTAMP",
390
+ "TIMEZONE_HOUR",
391
+ "TIMEZONE_MINUTE",
392
+ "TO",
393
+ "TRAILING",
394
+ "TRANSACTION",
395
+ "TRANSLATE",
396
+ "TRANSLATE_REGEX",
397
+ "TRANSLATION",
398
+ "TREAT",
399
+ "TRIGGER",
400
+ "TRIM",
401
+ "TRIM_ARRAY",
402
+ "TRUE",
403
+ "TRUNCATE",
404
+ "UESCAPE",
405
+ "UNION",
406
+ "UNIQUE",
407
+ "UNKNOWN",
408
+ "UNMATCHED",
409
+ "UNNEST",
410
+ "UPDATE",
411
+ "UPPER",
412
+ "USAGE",
413
+ "USER",
414
+ "USING",
415
+ "VALUE",
416
+ "VALUES",
417
+ "VALUE_OF",
418
+ "VARBINARY",
419
+ "VARCHAR",
420
+ "VARYING",
421
+ "VAR_POP",
422
+ "VAR_SAMP",
423
+ "VERSIONING",
424
+ "VIEW",
425
+ "WHEN",
426
+ "WHENEVER",
427
+ "WHERE",
428
+ "WIDTH_BUCKET",
429
+ "WINDOW",
430
+ "WITH",
431
+ "WITHIN",
432
+ "WITHOUT",
433
+ "WORK",
434
+ "WRITE",
435
+ "XML",
436
+ "XMLAGG",
437
+ "XMLATTRIBUTES",
438
+ "XMLBINARY",
439
+ "XMLCAST",
440
+ "XMLCOMMENT",
441
+ "XMLCONCAT",
442
+ "XMLDOCUMENT",
443
+ "XMLELEMENT",
444
+ "XMLEXISTS",
445
+ "XMLFOREST",
446
+ "XMLITERATE",
447
+ "XMLNAMESPACES",
448
+ "XMLPARSE",
449
+ "XMLPI",
450
+ "XMLQUERY",
451
+ "XMLSERIALIZE",
452
+ "XMLTABLE",
453
+ "XMLTEXT",
454
+ "XMLVALIDATE",
455
+ "YEAR",
456
+ "ZONE",
57
457
  }
semantic_model_generator/validate/schema.py CHANGED
@@ -22,9 +22,9 @@ from strictyaml import (
22
22
  YAMLValidationError,
23
23
  )
24
24
 
25
+ from semantic_model_generator.data_processing.cte_utils import ClickzettaDialect
25
26
  from semantic_model_generator.protos import semantic_model_pb2
26
27
  from semantic_model_generator.validate.keywords import CZ_RESERVED_WORDS
27
- from semantic_model_generator.data_processing.cte_utils import ClickzettaDialect
28
28
 
29
29
  scalar_type_map = {
30
30
  FieldDescriptor.TYPE_BOOL: Bool,
@@ -53,7 +53,9 @@ class IdField(Str): # type: ignore
53
53
  "name can only contain letters, underscores, decimal digits (0-9), and dollar signs ($).",
54
54
  )
55
55
  if chunk.contents.upper() in CZ_RESERVED_WORDS:
56
- chunk.expecting_but_found("", "name cannot be a ClickZetta reserved keyword")
56
+ chunk.expecting_but_found(
57
+ "", "name cannot be a ClickZetta reserved keyword"
58
+ )
57
59
  return chunk.contents
58
60
 
59
61