sqlglot.dialects.presto
1from __future__ import annotations 2 3import typing as t 4 5from sqlglot import exp, generator, parser, tokens, transforms 6from sqlglot.dialects.dialect import ( 7 Dialect, 8 NormalizationStrategy, 9 binary_from_function, 10 bool_xor_sql, 11 date_trunc_to_time, 12 datestrtodate_sql, 13 encode_decode_sql, 14 build_formatted_time, 15 if_sql, 16 left_to_substring_sql, 17 no_ilike_sql, 18 no_pivot_sql, 19 no_safe_divide_sql, 20 no_timestamp_sql, 21 regexp_extract_sql, 22 rename_func, 23 right_to_substring_sql, 24 struct_extract_sql, 25 timestamptrunc_sql, 26 timestrtotime_sql, 27 ts_or_ds_add_cast, 28) 29from sqlglot.dialects.hive import Hive 30from sqlglot.dialects.mysql import MySQL 31from sqlglot.helper import apply_index_offset, seq_get 32from sqlglot.tokens import TokenType 33 34 35def _explode_to_unnest_sql(self: Presto.Generator, expression: exp.Lateral) -> str: 36 if isinstance(expression.this, exp.Explode): 37 return self.sql( 38 exp.Join( 39 this=exp.Unnest( 40 expressions=[expression.this.this], 41 alias=expression.args.get("alias"), 42 offset=isinstance(expression.this, exp.Posexplode), 43 ), 44 kind="cross", 45 ) 46 ) 47 return self.lateral_sql(expression) 48 49 50def _initcap_sql(self: Presto.Generator, expression: exp.Initcap) -> str: 51 regex = r"(\w)(\w*)" 52 return f"REGEXP_REPLACE({self.sql(expression, 'this')}, '{regex}', x -> UPPER(x[1]) || LOWER(x[2]))" 53 54 55def _no_sort_array(self: Presto.Generator, expression: exp.SortArray) -> str: 56 if expression.args.get("asc") == exp.false(): 57 comparator = "(a, b) -> CASE WHEN a < b THEN 1 WHEN a > b THEN -1 ELSE 0 END" 58 else: 59 comparator = None 60 return self.func("ARRAY_SORT", expression.this, comparator) 61 62 63def _schema_sql(self: Presto.Generator, expression: exp.Schema) -> str: 64 if isinstance(expression.parent, exp.Property): 65 columns = ", ".join(f"'{c.name}'" for c in expression.expressions) 66 return f"ARRAY[{columns}]" 67 68 if expression.parent: 69 for schema in expression.parent.find_all(exp.Schema): 70 column_defs = schema.find_all(exp.ColumnDef) 71 if column_defs and isinstance(schema.parent, exp.Property): 72 expression.expressions.extend(column_defs) 73 74 return self.schema_sql(expression) 75 76 77def _quantile_sql(self: Presto.Generator, expression: exp.Quantile) -> str: 78 self.unsupported("Presto does not support exact quantiles") 79 return self.func("APPROX_PERCENTILE", expression.this, expression.args.get("quantile")) 80 81 82def _str_to_time_sql( 83 self: Presto.Generator, expression: exp.StrToDate | exp.StrToTime | exp.TsOrDsToDate 84) -> str: 85 return self.func("DATE_PARSE", expression.this, self.format_time(expression)) 86 87 88def _ts_or_ds_to_date_sql(self: Presto.Generator, expression: exp.TsOrDsToDate) -> str: 89 time_format = self.format_time(expression) 90 if time_format and time_format not in (Presto.TIME_FORMAT, Presto.DATE_FORMAT): 91 return self.sql(exp.cast(_str_to_time_sql(self, expression), "DATE")) 92 return self.sql(exp.cast(exp.cast(expression.this, "TIMESTAMP"), "DATE")) 93 94 95def _ts_or_ds_add_sql(self: Presto.Generator, expression: exp.TsOrDsAdd) -> str: 96 expression = ts_or_ds_add_cast(expression) 97 unit = exp.Literal.string(expression.text("unit") or "DAY") 98 return self.func("DATE_ADD", unit, expression.expression, expression.this) 99 100 101def _ts_or_ds_diff_sql(self: Presto.Generator, expression: exp.TsOrDsDiff) -> str: 102 this = exp.cast(expression.this, "TIMESTAMP") 103 expr = exp.cast(expression.expression, "TIMESTAMP") 104 unit = exp.Literal.string(expression.text("unit") or "DAY") 105 return self.func("DATE_DIFF", unit, expr, this) 106 107 108def _build_approx_percentile(args: t.List) -> exp.Expression: 109 if len(args) == 4: 110 return exp.ApproxQuantile( 111 this=seq_get(args, 0), 112 weight=seq_get(args, 1), 113 quantile=seq_get(args, 2), 114 accuracy=seq_get(args, 3), 115 ) 116 if len(args) == 3: 117 return exp.ApproxQuantile( 118 this=seq_get(args, 0), quantile=seq_get(args, 1), accuracy=seq_get(args, 2) 119 ) 120 return exp.ApproxQuantile.from_arg_list(args) 121 122 123def _build_from_unixtime(args: t.List) -> exp.Expression: 124 if len(args) == 3: 125 return exp.UnixToTime( 126 this=seq_get(args, 0), 127 hours=seq_get(args, 1), 128 minutes=seq_get(args, 2), 129 ) 130 if len(args) == 2: 131 return exp.UnixToTime(this=seq_get(args, 0), zone=seq_get(args, 1)) 132 133 return exp.UnixToTime.from_arg_list(args) 134 135 136def _unnest_sequence(expression: exp.Expression) -> exp.Expression: 137 if isinstance(expression, exp.Table): 138 if isinstance(expression.this, exp.GenerateSeries): 139 unnest = exp.Unnest(expressions=[expression.this]) 140 141 if expression.alias: 142 return exp.alias_(unnest, alias="_u", table=[expression.alias], copy=False) 143 return unnest 144 return expression 145 146 147def _first_last_sql(self: Presto.Generator, expression: exp.Func) -> str: 148 """ 149 Trino doesn't support FIRST / LAST as functions, but they're valid in the context 150 of MATCH_RECOGNIZE, so we need to preserve them in that case. In all other cases 151 they're converted into an ARBITRARY call. 152 153 Reference: https://trino.io/docs/current/sql/match-recognize.html#logical-navigation-functions 154 """ 155 if isinstance(expression.find_ancestor(exp.MatchRecognize, exp.Select), exp.MatchRecognize): 156 return self.function_fallback_sql(expression) 157 158 return rename_func("ARBITRARY")(self, expression) 159 160 161def _unix_to_time_sql(self: Presto.Generator, expression: exp.UnixToTime) -> str: 162 scale = expression.args.get("scale") 163 timestamp = self.sql(expression, "this") 164 if scale in (None, exp.UnixToTime.SECONDS): 165 return rename_func("FROM_UNIXTIME")(self, expression) 166 167 return f"FROM_UNIXTIME(CAST({timestamp} AS DOUBLE) / POW(10, {scale}))" 168 169 170def _to_int(expression: exp.Expression) -> exp.Expression: 171 if not expression.type: 172 from sqlglot.optimizer.annotate_types import annotate_types 173 174 annotate_types(expression) 175 if expression.type and expression.type.this not in exp.DataType.INTEGER_TYPES: 176 return exp.cast(expression, to=exp.DataType.Type.BIGINT) 177 return expression 178 179 180def _build_to_char(args: t.List) -> exp.TimeToStr: 181 fmt = seq_get(args, 1) 182 if isinstance(fmt, exp.Literal): 183 # We uppercase this to match Teradata's format mapping keys 184 fmt.set("this", fmt.this.upper()) 185 186 # We use "teradata" on purpose here, because the time formats are different in Presto. 187 # See https://prestodb.io/docs/current/functions/teradata.html?highlight=to_char#to_char 188 return build_formatted_time(exp.TimeToStr, "teradata")(args) 189 190 191class Presto(Dialect): 192 INDEX_OFFSET = 1 193 NULL_ORDERING = "nulls_are_last" 194 TIME_FORMAT = MySQL.TIME_FORMAT 195 TIME_MAPPING = MySQL.TIME_MAPPING 196 STRICT_STRING_CONCAT = True 197 SUPPORTS_SEMI_ANTI_JOIN = False 198 TYPED_DIVISION = True 199 TABLESAMPLE_SIZE_IS_PERCENT = True 200 LOG_BASE_FIRST: t.Optional[bool] = None 201 202 # https://github.com/trinodb/trino/issues/17 203 # https://github.com/trinodb/trino/issues/12289 204 # https://github.com/prestodb/presto/issues/2863 205 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 206 207 class Tokenizer(tokens.Tokenizer): 208 UNICODE_STRINGS = [ 209 (prefix + q, q) 210 for q in t.cast(t.List[str], tokens.Tokenizer.QUOTES) 211 for prefix in ("U&", "u&") 212 ] 213 214 KEYWORDS = { 215 **tokens.Tokenizer.KEYWORDS, 216 "START": TokenType.BEGIN, 217 "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE, 218 "ROW": TokenType.STRUCT, 219 "IPADDRESS": TokenType.IPADDRESS, 220 "IPPREFIX": TokenType.IPPREFIX, 221 } 222 223 class Parser(parser.Parser): 224 VALUES_FOLLOWED_BY_PAREN = False 225 226 FUNCTIONS = { 227 **parser.Parser.FUNCTIONS, 228 "ARBITRARY": exp.AnyValue.from_arg_list, 229 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 230 "APPROX_PERCENTILE": _build_approx_percentile, 231 "BITWISE_AND": binary_from_function(exp.BitwiseAnd), 232 "BITWISE_NOT": lambda args: exp.BitwiseNot(this=seq_get(args, 0)), 233 "BITWISE_OR": binary_from_function(exp.BitwiseOr), 234 "BITWISE_XOR": binary_from_function(exp.BitwiseXor), 235 "CARDINALITY": exp.ArraySize.from_arg_list, 236 "CONTAINS": exp.ArrayContains.from_arg_list, 237 "DATE_ADD": lambda args: exp.DateAdd( 238 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 239 ), 240 "DATE_DIFF": lambda args: exp.DateDiff( 241 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 242 ), 243 "DATE_FORMAT": build_formatted_time(exp.TimeToStr, "presto"), 244 "DATE_PARSE": build_formatted_time(exp.StrToTime, "presto"), 245 "DATE_TRUNC": date_trunc_to_time, 246 "ELEMENT_AT": lambda args: exp.Bracket( 247 this=seq_get(args, 0), expressions=[seq_get(args, 1)], offset=1, safe=True 248 ), 249 "FROM_HEX": exp.Unhex.from_arg_list, 250 "FROM_UNIXTIME": _build_from_unixtime, 251 "FROM_UTF8": lambda args: exp.Decode( 252 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 253 ), 254 "NOW": exp.CurrentTimestamp.from_arg_list, 255 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 256 this=seq_get(args, 0), expression=seq_get(args, 1), group=seq_get(args, 2) 257 ), 258 "REGEXP_REPLACE": lambda args: exp.RegexpReplace( 259 this=seq_get(args, 0), 260 expression=seq_get(args, 1), 261 replacement=seq_get(args, 2) or exp.Literal.string(""), 262 ), 263 "ROW": exp.Struct.from_arg_list, 264 "SEQUENCE": exp.GenerateSeries.from_arg_list, 265 "SET_AGG": exp.ArrayUniqueAgg.from_arg_list, 266 "SPLIT_TO_MAP": exp.StrToMap.from_arg_list, 267 "STRPOS": lambda args: exp.StrPosition( 268 this=seq_get(args, 0), substr=seq_get(args, 1), instance=seq_get(args, 2) 269 ), 270 "TO_CHAR": _build_to_char, 271 "TO_HEX": exp.Hex.from_arg_list, 272 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 273 "TO_UTF8": lambda args: exp.Encode( 274 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 275 ), 276 } 277 278 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 279 FUNCTION_PARSERS.pop("TRIM") 280 281 class Generator(generator.Generator): 282 INTERVAL_ALLOWS_PLURAL_FORM = False 283 JOIN_HINTS = False 284 TABLE_HINTS = False 285 QUERY_HINTS = False 286 IS_BOOL_ALLOWED = False 287 TZ_TO_WITH_TIME_ZONE = True 288 NVL2_SUPPORTED = False 289 STRUCT_DELIMITER = ("(", ")") 290 LIMIT_ONLY_LITERALS = True 291 SUPPORTS_SINGLE_ARG_CONCAT = False 292 LIKE_PROPERTY_INSIDE_SCHEMA = True 293 MULTI_ARG_DISTINCT = False 294 SUPPORTS_TO_NUMBER = False 295 296 PROPERTIES_LOCATION = { 297 **generator.Generator.PROPERTIES_LOCATION, 298 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 299 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 300 } 301 302 TYPE_MAPPING = { 303 **generator.Generator.TYPE_MAPPING, 304 exp.DataType.Type.INT: "INTEGER", 305 exp.DataType.Type.FLOAT: "REAL", 306 exp.DataType.Type.BINARY: "VARBINARY", 307 exp.DataType.Type.TEXT: "VARCHAR", 308 exp.DataType.Type.TIMETZ: "TIME", 309 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 310 exp.DataType.Type.STRUCT: "ROW", 311 exp.DataType.Type.DATETIME: "TIMESTAMP", 312 exp.DataType.Type.DATETIME64: "TIMESTAMP", 313 } 314 315 TRANSFORMS = { 316 **generator.Generator.TRANSFORMS, 317 exp.AnyValue: rename_func("ARBITRARY"), 318 exp.ApproxDistinct: lambda self, e: self.func( 319 "APPROX_DISTINCT", e.this, e.args.get("accuracy") 320 ), 321 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 322 exp.ArgMax: rename_func("MAX_BY"), 323 exp.ArgMin: rename_func("MIN_BY"), 324 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 325 exp.ArrayAny: rename_func("ANY_MATCH"), 326 exp.ArrayConcat: rename_func("CONCAT"), 327 exp.ArrayContains: rename_func("CONTAINS"), 328 exp.ArraySize: rename_func("CARDINALITY"), 329 exp.ArrayToString: rename_func("ARRAY_JOIN"), 330 exp.ArrayUniqueAgg: rename_func("SET_AGG"), 331 exp.AtTimeZone: rename_func("AT_TIMEZONE"), 332 exp.BitwiseAnd: lambda self, e: self.func("BITWISE_AND", e.this, e.expression), 333 exp.BitwiseLeftShift: lambda self, e: self.func( 334 "BITWISE_ARITHMETIC_SHIFT_LEFT", e.this, e.expression 335 ), 336 exp.BitwiseNot: lambda self, e: self.func("BITWISE_NOT", e.this), 337 exp.BitwiseOr: lambda self, e: self.func("BITWISE_OR", e.this, e.expression), 338 exp.BitwiseRightShift: lambda self, e: self.func( 339 "BITWISE_ARITHMETIC_SHIFT_RIGHT", e.this, e.expression 340 ), 341 exp.BitwiseXor: lambda self, e: self.func("BITWISE_XOR", e.this, e.expression), 342 exp.Cast: transforms.preprocess([transforms.epoch_cast_to_ts]), 343 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 344 exp.DateAdd: lambda self, e: self.func( 345 "DATE_ADD", 346 exp.Literal.string(e.text("unit") or "DAY"), 347 _to_int(e.expression), 348 e.this, 349 ), 350 exp.DateDiff: lambda self, e: self.func( 351 "DATE_DIFF", exp.Literal.string(e.text("unit") or "DAY"), e.expression, e.this 352 ), 353 exp.DateStrToDate: datestrtodate_sql, 354 exp.DateToDi: lambda self, 355 e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.DATEINT_FORMAT}) AS INT)", 356 exp.DateSub: lambda self, e: self.func( 357 "DATE_ADD", 358 exp.Literal.string(e.text("unit") or "DAY"), 359 _to_int(e.expression * -1), 360 e.this, 361 ), 362 exp.Decode: lambda self, e: encode_decode_sql(self, e, "FROM_UTF8"), 363 exp.DiToDate: lambda self, 364 e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.DATEINT_FORMAT}) AS DATE)", 365 exp.Encode: lambda self, e: encode_decode_sql(self, e, "TO_UTF8"), 366 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 367 exp.First: _first_last_sql, 368 exp.FirstValue: _first_last_sql, 369 exp.FromTimeZone: lambda self, 370 e: f"WITH_TIMEZONE({self.sql(e, 'this')}, {self.sql(e, 'zone')}) AT TIME ZONE 'UTC'", 371 exp.Group: transforms.preprocess([transforms.unalias_group]), 372 exp.GroupConcat: lambda self, e: self.func( 373 "ARRAY_JOIN", self.func("ARRAY_AGG", e.this), e.args.get("separator") 374 ), 375 exp.Hex: rename_func("TO_HEX"), 376 exp.If: if_sql(), 377 exp.ILike: no_ilike_sql, 378 exp.Initcap: _initcap_sql, 379 exp.ParseJSON: rename_func("JSON_PARSE"), 380 exp.Last: _first_last_sql, 381 exp.LastValue: _first_last_sql, 382 exp.LastDay: lambda self, e: self.func("LAST_DAY_OF_MONTH", e.this), 383 exp.Lateral: _explode_to_unnest_sql, 384 exp.Left: left_to_substring_sql, 385 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 386 exp.LogicalAnd: rename_func("BOOL_AND"), 387 exp.LogicalOr: rename_func("BOOL_OR"), 388 exp.Pivot: no_pivot_sql, 389 exp.Quantile: _quantile_sql, 390 exp.RegexpExtract: regexp_extract_sql, 391 exp.Right: right_to_substring_sql, 392 exp.SafeDivide: no_safe_divide_sql, 393 exp.Schema: _schema_sql, 394 exp.SchemaCommentProperty: lambda self, e: self.naked_property(e), 395 exp.Select: transforms.preprocess( 396 [ 397 transforms.eliminate_qualify, 398 transforms.eliminate_distinct_on, 399 transforms.explode_to_unnest(1), 400 transforms.eliminate_semi_and_anti_joins, 401 ] 402 ), 403 exp.SortArray: _no_sort_array, 404 exp.StrPosition: rename_func("STRPOS"), 405 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 406 exp.StrToMap: rename_func("SPLIT_TO_MAP"), 407 exp.StrToTime: _str_to_time_sql, 408 exp.StructExtract: struct_extract_sql, 409 exp.Table: transforms.preprocess([_unnest_sequence]), 410 exp.Timestamp: no_timestamp_sql, 411 exp.TimestampTrunc: timestamptrunc_sql, 412 exp.TimeStrToDate: timestrtotime_sql, 413 exp.TimeStrToTime: timestrtotime_sql, 414 exp.TimeStrToUnix: lambda self, e: self.func( 415 "TO_UNIXTIME", self.func("DATE_PARSE", e.this, Presto.TIME_FORMAT) 416 ), 417 exp.TimeToStr: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 418 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 419 exp.ToChar: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 420 exp.TryCast: transforms.preprocess([transforms.epoch_cast_to_ts]), 421 exp.TsOrDiToDi: lambda self, 422 e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 423 exp.TsOrDsAdd: _ts_or_ds_add_sql, 424 exp.TsOrDsDiff: _ts_or_ds_diff_sql, 425 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 426 exp.Unhex: rename_func("FROM_HEX"), 427 exp.UnixToStr: lambda self, 428 e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 429 exp.UnixToTime: _unix_to_time_sql, 430 exp.UnixToTimeStr: lambda self, 431 e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 432 exp.VariancePop: rename_func("VAR_POP"), 433 exp.With: transforms.preprocess([transforms.add_recursive_cte_column_names]), 434 exp.WithinGroup: transforms.preprocess( 435 [transforms.remove_within_group_for_percentiles] 436 ), 437 exp.Xor: bool_xor_sql, 438 } 439 440 def strtounix_sql(self, expression: exp.StrToUnix) -> str: 441 # Since `TO_UNIXTIME` requires a `TIMESTAMP`, we need to parse the argument into one. 442 # To do this, we first try to `DATE_PARSE` it, but since this can fail when there's a 443 # timezone involved, we wrap it in a `TRY` call and use `PARSE_DATETIME` as a fallback, 444 # which seems to be using the same time mapping as Hive, as per: 445 # https://joda-time.sourceforge.net/apidocs/org/joda/time/format/DateTimeFormat.html 446 value_as_text = exp.cast(expression.this, "text") 447 parse_without_tz = self.func("DATE_PARSE", value_as_text, self.format_time(expression)) 448 parse_with_tz = self.func( 449 "PARSE_DATETIME", 450 value_as_text, 451 self.format_time(expression, Hive.INVERSE_TIME_MAPPING, Hive.INVERSE_TIME_TRIE), 452 ) 453 coalesced = self.func("COALESCE", self.func("TRY", parse_without_tz), parse_with_tz) 454 return self.func("TO_UNIXTIME", coalesced) 455 456 def bracket_sql(self, expression: exp.Bracket) -> str: 457 if expression.args.get("safe"): 458 return self.func( 459 "ELEMENT_AT", 460 expression.this, 461 seq_get( 462 apply_index_offset( 463 expression.this, 464 expression.expressions, 465 1 - expression.args.get("offset", 0), 466 ), 467 0, 468 ), 469 ) 470 return super().bracket_sql(expression) 471 472 def struct_sql(self, expression: exp.Struct) -> str: 473 from sqlglot.optimizer.annotate_types import annotate_types 474 475 expression = annotate_types(expression) 476 values: t.List[str] = [] 477 schema: t.List[str] = [] 478 unknown_type = False 479 480 for e in expression.expressions: 481 if isinstance(e, exp.PropertyEQ): 482 if e.type and e.type.is_type(exp.DataType.Type.UNKNOWN): 483 unknown_type = True 484 else: 485 schema.append(f"{self.sql(e, 'this')} {self.sql(e.type)}") 486 values.append(self.sql(e, "expression")) 487 else: 488 values.append(self.sql(e)) 489 490 size = len(expression.expressions) 491 492 if not size or len(schema) != size: 493 if unknown_type: 494 self.unsupported( 495 "Cannot convert untyped key-value definitions (try annotate_types)." 496 ) 497 return self.func("ROW", *values) 498 return f"CAST(ROW({', '.join(values)}) AS ROW({', '.join(schema)}))" 499 500 def interval_sql(self, expression: exp.Interval) -> str: 501 unit = self.sql(expression, "unit") 502 if expression.this and unit.startswith("WEEK"): 503 return f"({expression.this.name} * INTERVAL '7' DAY)" 504 return super().interval_sql(expression) 505 506 def transaction_sql(self, expression: exp.Transaction) -> str: 507 modes = expression.args.get("modes") 508 modes = f" {', '.join(modes)}" if modes else "" 509 return f"START TRANSACTION{modes}" 510 511 def generateseries_sql(self, expression: exp.GenerateSeries) -> str: 512 start = expression.args["start"] 513 end = expression.args["end"] 514 step = expression.args.get("step") 515 516 if isinstance(start, exp.Cast): 517 target_type = start.to 518 elif isinstance(end, exp.Cast): 519 target_type = end.to 520 else: 521 target_type = None 522 523 if target_type and target_type.is_type("timestamp"): 524 if target_type is start.to: 525 end = exp.cast(end, target_type) 526 else: 527 start = exp.cast(start, target_type) 528 529 return self.func("SEQUENCE", start, end, step) 530 531 def offset_limit_modifiers( 532 self, expression: exp.Expression, fetch: bool, limit: t.Optional[exp.Fetch | exp.Limit] 533 ) -> t.List[str]: 534 return [ 535 self.sql(expression, "offset"), 536 self.sql(limit), 537 ] 538 539 def create_sql(self, expression: exp.Create) -> str: 540 """ 541 Presto doesn't support CREATE VIEW with expressions (ex: `CREATE VIEW x (cola)` then `(cola)` is the expression), 542 so we need to remove them 543 """ 544 kind = expression.args["kind"] 545 schema = expression.this 546 if kind == "VIEW" and schema.expressions: 547 expression.this.set("expressions", None) 548 return super().create_sql(expression)
192class Presto(Dialect): 193 INDEX_OFFSET = 1 194 NULL_ORDERING = "nulls_are_last" 195 TIME_FORMAT = MySQL.TIME_FORMAT 196 TIME_MAPPING = MySQL.TIME_MAPPING 197 STRICT_STRING_CONCAT = True 198 SUPPORTS_SEMI_ANTI_JOIN = False 199 TYPED_DIVISION = True 200 TABLESAMPLE_SIZE_IS_PERCENT = True 201 LOG_BASE_FIRST: t.Optional[bool] = None 202 203 # https://github.com/trinodb/trino/issues/17 204 # https://github.com/trinodb/trino/issues/12289 205 # https://github.com/prestodb/presto/issues/2863 206 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 207 208 class Tokenizer(tokens.Tokenizer): 209 UNICODE_STRINGS = [ 210 (prefix + q, q) 211 for q in t.cast(t.List[str], tokens.Tokenizer.QUOTES) 212 for prefix in ("U&", "u&") 213 ] 214 215 KEYWORDS = { 216 **tokens.Tokenizer.KEYWORDS, 217 "START": TokenType.BEGIN, 218 "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE, 219 "ROW": TokenType.STRUCT, 220 "IPADDRESS": TokenType.IPADDRESS, 221 "IPPREFIX": TokenType.IPPREFIX, 222 } 223 224 class Parser(parser.Parser): 225 VALUES_FOLLOWED_BY_PAREN = False 226 227 FUNCTIONS = { 228 **parser.Parser.FUNCTIONS, 229 "ARBITRARY": exp.AnyValue.from_arg_list, 230 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 231 "APPROX_PERCENTILE": _build_approx_percentile, 232 "BITWISE_AND": binary_from_function(exp.BitwiseAnd), 233 "BITWISE_NOT": lambda args: exp.BitwiseNot(this=seq_get(args, 0)), 234 "BITWISE_OR": binary_from_function(exp.BitwiseOr), 235 "BITWISE_XOR": binary_from_function(exp.BitwiseXor), 236 "CARDINALITY": exp.ArraySize.from_arg_list, 237 "CONTAINS": exp.ArrayContains.from_arg_list, 238 "DATE_ADD": lambda args: exp.DateAdd( 239 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 240 ), 241 "DATE_DIFF": lambda args: exp.DateDiff( 242 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 243 ), 244 "DATE_FORMAT": build_formatted_time(exp.TimeToStr, "presto"), 245 "DATE_PARSE": build_formatted_time(exp.StrToTime, "presto"), 246 "DATE_TRUNC": date_trunc_to_time, 247 "ELEMENT_AT": lambda args: exp.Bracket( 248 this=seq_get(args, 0), expressions=[seq_get(args, 1)], offset=1, safe=True 249 ), 250 "FROM_HEX": exp.Unhex.from_arg_list, 251 "FROM_UNIXTIME": _build_from_unixtime, 252 "FROM_UTF8": lambda args: exp.Decode( 253 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 254 ), 255 "NOW": exp.CurrentTimestamp.from_arg_list, 256 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 257 this=seq_get(args, 0), expression=seq_get(args, 1), group=seq_get(args, 2) 258 ), 259 "REGEXP_REPLACE": lambda args: exp.RegexpReplace( 260 this=seq_get(args, 0), 261 expression=seq_get(args, 1), 262 replacement=seq_get(args, 2) or exp.Literal.string(""), 263 ), 264 "ROW": exp.Struct.from_arg_list, 265 "SEQUENCE": exp.GenerateSeries.from_arg_list, 266 "SET_AGG": exp.ArrayUniqueAgg.from_arg_list, 267 "SPLIT_TO_MAP": exp.StrToMap.from_arg_list, 268 "STRPOS": lambda args: exp.StrPosition( 269 this=seq_get(args, 0), substr=seq_get(args, 1), instance=seq_get(args, 2) 270 ), 271 "TO_CHAR": _build_to_char, 272 "TO_HEX": exp.Hex.from_arg_list, 273 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 274 "TO_UTF8": lambda args: exp.Encode( 275 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 276 ), 277 } 278 279 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 280 FUNCTION_PARSERS.pop("TRIM") 281 282 class Generator(generator.Generator): 283 INTERVAL_ALLOWS_PLURAL_FORM = False 284 JOIN_HINTS = False 285 TABLE_HINTS = False 286 QUERY_HINTS = False 287 IS_BOOL_ALLOWED = False 288 TZ_TO_WITH_TIME_ZONE = True 289 NVL2_SUPPORTED = False 290 STRUCT_DELIMITER = ("(", ")") 291 LIMIT_ONLY_LITERALS = True 292 SUPPORTS_SINGLE_ARG_CONCAT = False 293 LIKE_PROPERTY_INSIDE_SCHEMA = True 294 MULTI_ARG_DISTINCT = False 295 SUPPORTS_TO_NUMBER = False 296 297 PROPERTIES_LOCATION = { 298 **generator.Generator.PROPERTIES_LOCATION, 299 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 300 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 301 } 302 303 TYPE_MAPPING = { 304 **generator.Generator.TYPE_MAPPING, 305 exp.DataType.Type.INT: "INTEGER", 306 exp.DataType.Type.FLOAT: "REAL", 307 exp.DataType.Type.BINARY: "VARBINARY", 308 exp.DataType.Type.TEXT: "VARCHAR", 309 exp.DataType.Type.TIMETZ: "TIME", 310 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 311 exp.DataType.Type.STRUCT: "ROW", 312 exp.DataType.Type.DATETIME: "TIMESTAMP", 313 exp.DataType.Type.DATETIME64: "TIMESTAMP", 314 } 315 316 TRANSFORMS = { 317 **generator.Generator.TRANSFORMS, 318 exp.AnyValue: rename_func("ARBITRARY"), 319 exp.ApproxDistinct: lambda self, e: self.func( 320 "APPROX_DISTINCT", e.this, e.args.get("accuracy") 321 ), 322 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 323 exp.ArgMax: rename_func("MAX_BY"), 324 exp.ArgMin: rename_func("MIN_BY"), 325 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 326 exp.ArrayAny: rename_func("ANY_MATCH"), 327 exp.ArrayConcat: rename_func("CONCAT"), 328 exp.ArrayContains: rename_func("CONTAINS"), 329 exp.ArraySize: rename_func("CARDINALITY"), 330 exp.ArrayToString: rename_func("ARRAY_JOIN"), 331 exp.ArrayUniqueAgg: rename_func("SET_AGG"), 332 exp.AtTimeZone: rename_func("AT_TIMEZONE"), 333 exp.BitwiseAnd: lambda self, e: self.func("BITWISE_AND", e.this, e.expression), 334 exp.BitwiseLeftShift: lambda self, e: self.func( 335 "BITWISE_ARITHMETIC_SHIFT_LEFT", e.this, e.expression 336 ), 337 exp.BitwiseNot: lambda self, e: self.func("BITWISE_NOT", e.this), 338 exp.BitwiseOr: lambda self, e: self.func("BITWISE_OR", e.this, e.expression), 339 exp.BitwiseRightShift: lambda self, e: self.func( 340 "BITWISE_ARITHMETIC_SHIFT_RIGHT", e.this, e.expression 341 ), 342 exp.BitwiseXor: lambda self, e: self.func("BITWISE_XOR", e.this, e.expression), 343 exp.Cast: transforms.preprocess([transforms.epoch_cast_to_ts]), 344 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 345 exp.DateAdd: lambda self, e: self.func( 346 "DATE_ADD", 347 exp.Literal.string(e.text("unit") or "DAY"), 348 _to_int(e.expression), 349 e.this, 350 ), 351 exp.DateDiff: lambda self, e: self.func( 352 "DATE_DIFF", exp.Literal.string(e.text("unit") or "DAY"), e.expression, e.this 353 ), 354 exp.DateStrToDate: datestrtodate_sql, 355 exp.DateToDi: lambda self, 356 e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.DATEINT_FORMAT}) AS INT)", 357 exp.DateSub: lambda self, e: self.func( 358 "DATE_ADD", 359 exp.Literal.string(e.text("unit") or "DAY"), 360 _to_int(e.expression * -1), 361 e.this, 362 ), 363 exp.Decode: lambda self, e: encode_decode_sql(self, e, "FROM_UTF8"), 364 exp.DiToDate: lambda self, 365 e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.DATEINT_FORMAT}) AS DATE)", 366 exp.Encode: lambda self, e: encode_decode_sql(self, e, "TO_UTF8"), 367 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 368 exp.First: _first_last_sql, 369 exp.FirstValue: _first_last_sql, 370 exp.FromTimeZone: lambda self, 371 e: f"WITH_TIMEZONE({self.sql(e, 'this')}, {self.sql(e, 'zone')}) AT TIME ZONE 'UTC'", 372 exp.Group: transforms.preprocess([transforms.unalias_group]), 373 exp.GroupConcat: lambda self, e: self.func( 374 "ARRAY_JOIN", self.func("ARRAY_AGG", e.this), e.args.get("separator") 375 ), 376 exp.Hex: rename_func("TO_HEX"), 377 exp.If: if_sql(), 378 exp.ILike: no_ilike_sql, 379 exp.Initcap: _initcap_sql, 380 exp.ParseJSON: rename_func("JSON_PARSE"), 381 exp.Last: _first_last_sql, 382 exp.LastValue: _first_last_sql, 383 exp.LastDay: lambda self, e: self.func("LAST_DAY_OF_MONTH", e.this), 384 exp.Lateral: _explode_to_unnest_sql, 385 exp.Left: left_to_substring_sql, 386 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 387 exp.LogicalAnd: rename_func("BOOL_AND"), 388 exp.LogicalOr: rename_func("BOOL_OR"), 389 exp.Pivot: no_pivot_sql, 390 exp.Quantile: _quantile_sql, 391 exp.RegexpExtract: regexp_extract_sql, 392 exp.Right: right_to_substring_sql, 393 exp.SafeDivide: no_safe_divide_sql, 394 exp.Schema: _schema_sql, 395 exp.SchemaCommentProperty: lambda self, e: self.naked_property(e), 396 exp.Select: transforms.preprocess( 397 [ 398 transforms.eliminate_qualify, 399 transforms.eliminate_distinct_on, 400 transforms.explode_to_unnest(1), 401 transforms.eliminate_semi_and_anti_joins, 402 ] 403 ), 404 exp.SortArray: _no_sort_array, 405 exp.StrPosition: rename_func("STRPOS"), 406 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 407 exp.StrToMap: rename_func("SPLIT_TO_MAP"), 408 exp.StrToTime: _str_to_time_sql, 409 exp.StructExtract: struct_extract_sql, 410 exp.Table: transforms.preprocess([_unnest_sequence]), 411 exp.Timestamp: no_timestamp_sql, 412 exp.TimestampTrunc: timestamptrunc_sql, 413 exp.TimeStrToDate: timestrtotime_sql, 414 exp.TimeStrToTime: timestrtotime_sql, 415 exp.TimeStrToUnix: lambda self, e: self.func( 416 "TO_UNIXTIME", self.func("DATE_PARSE", e.this, Presto.TIME_FORMAT) 417 ), 418 exp.TimeToStr: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 419 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 420 exp.ToChar: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 421 exp.TryCast: transforms.preprocess([transforms.epoch_cast_to_ts]), 422 exp.TsOrDiToDi: lambda self, 423 e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 424 exp.TsOrDsAdd: _ts_or_ds_add_sql, 425 exp.TsOrDsDiff: _ts_or_ds_diff_sql, 426 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 427 exp.Unhex: rename_func("FROM_HEX"), 428 exp.UnixToStr: lambda self, 429 e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 430 exp.UnixToTime: _unix_to_time_sql, 431 exp.UnixToTimeStr: lambda self, 432 e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 433 exp.VariancePop: rename_func("VAR_POP"), 434 exp.With: transforms.preprocess([transforms.add_recursive_cte_column_names]), 435 exp.WithinGroup: transforms.preprocess( 436 [transforms.remove_within_group_for_percentiles] 437 ), 438 exp.Xor: bool_xor_sql, 439 } 440 441 def strtounix_sql(self, expression: exp.StrToUnix) -> str: 442 # Since `TO_UNIXTIME` requires a `TIMESTAMP`, we need to parse the argument into one. 443 # To do this, we first try to `DATE_PARSE` it, but since this can fail when there's a 444 # timezone involved, we wrap it in a `TRY` call and use `PARSE_DATETIME` as a fallback, 445 # which seems to be using the same time mapping as Hive, as per: 446 # https://joda-time.sourceforge.net/apidocs/org/joda/time/format/DateTimeFormat.html 447 value_as_text = exp.cast(expression.this, "text") 448 parse_without_tz = self.func("DATE_PARSE", value_as_text, self.format_time(expression)) 449 parse_with_tz = self.func( 450 "PARSE_DATETIME", 451 value_as_text, 452 self.format_time(expression, Hive.INVERSE_TIME_MAPPING, Hive.INVERSE_TIME_TRIE), 453 ) 454 coalesced = self.func("COALESCE", self.func("TRY", parse_without_tz), parse_with_tz) 455 return self.func("TO_UNIXTIME", coalesced) 456 457 def bracket_sql(self, expression: exp.Bracket) -> str: 458 if expression.args.get("safe"): 459 return self.func( 460 "ELEMENT_AT", 461 expression.this, 462 seq_get( 463 apply_index_offset( 464 expression.this, 465 expression.expressions, 466 1 - expression.args.get("offset", 0), 467 ), 468 0, 469 ), 470 ) 471 return super().bracket_sql(expression) 472 473 def struct_sql(self, expression: exp.Struct) -> str: 474 from sqlglot.optimizer.annotate_types import annotate_types 475 476 expression = annotate_types(expression) 477 values: t.List[str] = [] 478 schema: t.List[str] = [] 479 unknown_type = False 480 481 for e in expression.expressions: 482 if isinstance(e, exp.PropertyEQ): 483 if e.type and e.type.is_type(exp.DataType.Type.UNKNOWN): 484 unknown_type = True 485 else: 486 schema.append(f"{self.sql(e, 'this')} {self.sql(e.type)}") 487 values.append(self.sql(e, "expression")) 488 else: 489 values.append(self.sql(e)) 490 491 size = len(expression.expressions) 492 493 if not size or len(schema) != size: 494 if unknown_type: 495 self.unsupported( 496 "Cannot convert untyped key-value definitions (try annotate_types)." 497 ) 498 return self.func("ROW", *values) 499 return f"CAST(ROW({', '.join(values)}) AS ROW({', '.join(schema)}))" 500 501 def interval_sql(self, expression: exp.Interval) -> str: 502 unit = self.sql(expression, "unit") 503 if expression.this and unit.startswith("WEEK"): 504 return f"({expression.this.name} * INTERVAL '7' DAY)" 505 return super().interval_sql(expression) 506 507 def transaction_sql(self, expression: exp.Transaction) -> str: 508 modes = expression.args.get("modes") 509 modes = f" {', '.join(modes)}" if modes else "" 510 return f"START TRANSACTION{modes}" 511 512 def generateseries_sql(self, expression: exp.GenerateSeries) -> str: 513 start = expression.args["start"] 514 end = expression.args["end"] 515 step = expression.args.get("step") 516 517 if isinstance(start, exp.Cast): 518 target_type = start.to 519 elif isinstance(end, exp.Cast): 520 target_type = end.to 521 else: 522 target_type = None 523 524 if target_type and target_type.is_type("timestamp"): 525 if target_type is start.to: 526 end = exp.cast(end, target_type) 527 else: 528 start = exp.cast(start, target_type) 529 530 return self.func("SEQUENCE", start, end, step) 531 532 def offset_limit_modifiers( 533 self, expression: exp.Expression, fetch: bool, limit: t.Optional[exp.Fetch | exp.Limit] 534 ) -> t.List[str]: 535 return [ 536 self.sql(expression, "offset"), 537 self.sql(limit), 538 ] 539 540 def create_sql(self, expression: exp.Create) -> str: 541 """ 542 Presto doesn't support CREATE VIEW with expressions (ex: `CREATE VIEW x (cola)` then `(cola)` is the expression), 543 so we need to remove them 544 """ 545 kind = expression.args["kind"] 546 schema = expression.this 547 if kind == "VIEW" and schema.expressions: 548 expression.this.set("expressions", None) 549 return super().create_sql(expression)
Default NULL ordering method to use if not explicitly set.
Possible values: "nulls_are_small", "nulls_are_large", "nulls_are_last"
Associates this dialect's time formats with their equivalent Python strftime formats.
Whether the behavior of a / b depends on the types of a and b.
False means a / b is always float division.
True means a / b is integer division if both a and b are integers.
Whether the base comes first in the LOG function.
Possible values: True, False, None (two arguments are not supported by LOG)
Specifies the strategy according to which identifiers should be normalized.
Inherited Members
- sqlglot.dialects.dialect.Dialect
- Dialect
- WEEK_OFFSET
- UNNEST_COLUMN_ONLY
- ALIAS_POST_TABLESAMPLE
- IDENTIFIERS_CAN_START_WITH_DIGIT
- DPIPE_IS_STRING_CONCAT
- SUPPORTS_USER_DEFINED_TYPES
- NORMALIZE_FUNCTIONS
- SAFE_DIVISION
- CONCAT_COALESCE
- DATE_FORMAT
- DATEINT_FORMAT
- FORMAT_MAPPING
- ESCAPE_SEQUENCES
- PSEUDOCOLUMNS
- PREFER_CTE_ALIAS_COLUMN
- get_or_raise
- format_time
- normalize_identifier
- case_sensitive
- can_identify
- quote_identifier
- to_json_path
- parse
- parse_into
- generate
- transpile
- tokenize
- tokenizer
- parser
- generator
208 class Tokenizer(tokens.Tokenizer): 209 UNICODE_STRINGS = [ 210 (prefix + q, q) 211 for q in t.cast(t.List[str], tokens.Tokenizer.QUOTES) 212 for prefix in ("U&", "u&") 213 ] 214 215 KEYWORDS = { 216 **tokens.Tokenizer.KEYWORDS, 217 "START": TokenType.BEGIN, 218 "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE, 219 "ROW": TokenType.STRUCT, 220 "IPADDRESS": TokenType.IPADDRESS, 221 "IPPREFIX": TokenType.IPPREFIX, 222 }
Inherited Members
- sqlglot.tokens.Tokenizer
- Tokenizer
- SINGLE_TOKENS
- BIT_STRINGS
- BYTE_STRINGS
- HEX_STRINGS
- RAW_STRINGS
- HEREDOC_STRINGS
- IDENTIFIERS
- IDENTIFIER_ESCAPES
- QUOTES
- STRING_ESCAPES
- VAR_SINGLE_TOKENS
- HEREDOC_TAG_IS_IDENTIFIER
- HEREDOC_STRING_ALTERNATIVE
- WHITE_SPACE
- COMMANDS
- COMMAND_PREFIX_TOKENS
- NUMERIC_LITERALS
- COMMENTS
- dialect
- reset
- tokenize
- tokenize_rs
- size
- sql
- tokens
224 class Parser(parser.Parser): 225 VALUES_FOLLOWED_BY_PAREN = False 226 227 FUNCTIONS = { 228 **parser.Parser.FUNCTIONS, 229 "ARBITRARY": exp.AnyValue.from_arg_list, 230 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 231 "APPROX_PERCENTILE": _build_approx_percentile, 232 "BITWISE_AND": binary_from_function(exp.BitwiseAnd), 233 "BITWISE_NOT": lambda args: exp.BitwiseNot(this=seq_get(args, 0)), 234 "BITWISE_OR": binary_from_function(exp.BitwiseOr), 235 "BITWISE_XOR": binary_from_function(exp.BitwiseXor), 236 "CARDINALITY": exp.ArraySize.from_arg_list, 237 "CONTAINS": exp.ArrayContains.from_arg_list, 238 "DATE_ADD": lambda args: exp.DateAdd( 239 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 240 ), 241 "DATE_DIFF": lambda args: exp.DateDiff( 242 this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0) 243 ), 244 "DATE_FORMAT": build_formatted_time(exp.TimeToStr, "presto"), 245 "DATE_PARSE": build_formatted_time(exp.StrToTime, "presto"), 246 "DATE_TRUNC": date_trunc_to_time, 247 "ELEMENT_AT": lambda args: exp.Bracket( 248 this=seq_get(args, 0), expressions=[seq_get(args, 1)], offset=1, safe=True 249 ), 250 "FROM_HEX": exp.Unhex.from_arg_list, 251 "FROM_UNIXTIME": _build_from_unixtime, 252 "FROM_UTF8": lambda args: exp.Decode( 253 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 254 ), 255 "NOW": exp.CurrentTimestamp.from_arg_list, 256 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 257 this=seq_get(args, 0), expression=seq_get(args, 1), group=seq_get(args, 2) 258 ), 259 "REGEXP_REPLACE": lambda args: exp.RegexpReplace( 260 this=seq_get(args, 0), 261 expression=seq_get(args, 1), 262 replacement=seq_get(args, 2) or exp.Literal.string(""), 263 ), 264 "ROW": exp.Struct.from_arg_list, 265 "SEQUENCE": exp.GenerateSeries.from_arg_list, 266 "SET_AGG": exp.ArrayUniqueAgg.from_arg_list, 267 "SPLIT_TO_MAP": exp.StrToMap.from_arg_list, 268 "STRPOS": lambda args: exp.StrPosition( 269 this=seq_get(args, 0), substr=seq_get(args, 1), instance=seq_get(args, 2) 270 ), 271 "TO_CHAR": _build_to_char, 272 "TO_HEX": exp.Hex.from_arg_list, 273 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 274 "TO_UTF8": lambda args: exp.Encode( 275 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 276 ), 277 } 278 279 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 280 FUNCTION_PARSERS.pop("TRIM")
Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree.
Arguments:
- error_level: The desired error level. Default: ErrorLevel.IMMEDIATE
- error_message_context: The amount of context to capture from a query string when displaying the error message (in number of characters). Default: 100
- max_errors: Maximum number of error messages to include in a raised ParseError. This is only relevant if error_level is ErrorLevel.RAISE. Default: 3
Inherited Members
- sqlglot.parser.Parser
- Parser
- NO_PAREN_FUNCTIONS
- STRUCT_TYPE_TOKENS
- NESTED_TYPE_TOKENS
- ENUM_TYPE_TOKENS
- AGGREGATE_TYPE_TOKENS
- TYPE_TOKENS
- SIGNED_TO_UNSIGNED_TYPE_TOKEN
- SUBQUERY_PREDICATES
- RESERVED_TOKENS
- DB_CREATABLES
- CREATABLES
- ID_VAR_TOKENS
- INTERVAL_VARS
- ALIAS_TOKENS
- COMMENT_TABLE_ALIAS_TOKENS
- UPDATE_ALIAS_TOKENS
- TRIM_TYPES
- FUNC_TOKENS
- CONJUNCTION
- EQUALITY
- COMPARISON
- BITWISE
- TERM
- FACTOR
- EXPONENT
- TIMES
- TIMESTAMPS
- SET_OPERATIONS
- JOIN_METHODS
- JOIN_SIDES
- JOIN_KINDS
- JOIN_HINTS
- LAMBDAS
- COLUMN_OPERATORS
- EXPRESSION_PARSERS
- STATEMENT_PARSERS
- UNARY_PARSERS
- STRING_PARSERS
- NUMERIC_PARSERS
- PRIMARY_PARSERS
- PLACEHOLDER_PARSERS
- RANGE_PARSERS
- PROPERTY_PARSERS
- CONSTRAINT_PARSERS
- ALTER_PARSERS
- SCHEMA_UNNAMED_CONSTRAINTS
- NO_PAREN_FUNCTION_PARSERS
- INVALID_FUNC_NAME_TOKENS
- FUNCTIONS_WITH_ALIASED_ARGS
- KEY_VALUE_DEFINITIONS
- QUERY_MODIFIER_PARSERS
- SET_PARSERS
- SHOW_PARSERS
- TYPE_LITERAL_PARSERS
- DDL_SELECT_TOKENS
- PRE_VOLATILE_TOKENS
- TRANSACTION_KIND
- TRANSACTION_CHARACTERISTICS
- CONFLICT_ACTIONS
- CREATE_SEQUENCE
- ISOLATED_LOADING_OPTIONS
- USABLES
- CAST_ACTIONS
- INSERT_ALTERNATIVES
- CLONE_KEYWORDS
- HISTORICAL_DATA_KIND
- OPCLASS_FOLLOW_KEYWORDS
- OPTYPE_FOLLOW_TOKENS
- TABLE_INDEX_HINT_TOKENS
- VIEW_ATTRIBUTES
- WINDOW_ALIAS_TOKENS
- WINDOW_BEFORE_PAREN_TOKENS
- WINDOW_SIDES
- JSON_KEY_VALUE_SEPARATOR_TOKENS
- FETCH_TOKENS
- ADD_CONSTRAINT_TOKENS
- DISTINCT_TOKENS
- NULL_TOKENS
- UNNEST_OFFSET_ALIAS_TOKENS
- SELECT_START_TOKENS
- STRICT_CAST
- PREFIXED_PIVOT_COLUMNS
- IDENTIFY_PIVOT_STRINGS
- LOG_DEFAULTS_TO_LN
- ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN
- TABLESAMPLE_CSV
- SET_REQUIRES_ASSIGNMENT_DELIMITER
- TRIM_PATTERN_FIRST
- STRING_ALIASES
- MODIFIERS_ATTACHED_TO_UNION
- UNION_MODIFIERS
- NO_PAREN_IF_COMMANDS
- JSON_ARROWS_REQUIRE_JSON_TYPE
- SUPPORTS_IMPLICIT_UNNEST
- error_level
- error_message_context
- max_errors
- dialect
- reset
- parse
- parse_into
- check_errors
- raise_error
- expression
- validate_expression
- errors
- sql
282 class Generator(generator.Generator): 283 INTERVAL_ALLOWS_PLURAL_FORM = False 284 JOIN_HINTS = False 285 TABLE_HINTS = False 286 QUERY_HINTS = False 287 IS_BOOL_ALLOWED = False 288 TZ_TO_WITH_TIME_ZONE = True 289 NVL2_SUPPORTED = False 290 STRUCT_DELIMITER = ("(", ")") 291 LIMIT_ONLY_LITERALS = True 292 SUPPORTS_SINGLE_ARG_CONCAT = False 293 LIKE_PROPERTY_INSIDE_SCHEMA = True 294 MULTI_ARG_DISTINCT = False 295 SUPPORTS_TO_NUMBER = False 296 297 PROPERTIES_LOCATION = { 298 **generator.Generator.PROPERTIES_LOCATION, 299 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 300 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 301 } 302 303 TYPE_MAPPING = { 304 **generator.Generator.TYPE_MAPPING, 305 exp.DataType.Type.INT: "INTEGER", 306 exp.DataType.Type.FLOAT: "REAL", 307 exp.DataType.Type.BINARY: "VARBINARY", 308 exp.DataType.Type.TEXT: "VARCHAR", 309 exp.DataType.Type.TIMETZ: "TIME", 310 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 311 exp.DataType.Type.STRUCT: "ROW", 312 exp.DataType.Type.DATETIME: "TIMESTAMP", 313 exp.DataType.Type.DATETIME64: "TIMESTAMP", 314 } 315 316 TRANSFORMS = { 317 **generator.Generator.TRANSFORMS, 318 exp.AnyValue: rename_func("ARBITRARY"), 319 exp.ApproxDistinct: lambda self, e: self.func( 320 "APPROX_DISTINCT", e.this, e.args.get("accuracy") 321 ), 322 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 323 exp.ArgMax: rename_func("MAX_BY"), 324 exp.ArgMin: rename_func("MIN_BY"), 325 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 326 exp.ArrayAny: rename_func("ANY_MATCH"), 327 exp.ArrayConcat: rename_func("CONCAT"), 328 exp.ArrayContains: rename_func("CONTAINS"), 329 exp.ArraySize: rename_func("CARDINALITY"), 330 exp.ArrayToString: rename_func("ARRAY_JOIN"), 331 exp.ArrayUniqueAgg: rename_func("SET_AGG"), 332 exp.AtTimeZone: rename_func("AT_TIMEZONE"), 333 exp.BitwiseAnd: lambda self, e: self.func("BITWISE_AND", e.this, e.expression), 334 exp.BitwiseLeftShift: lambda self, e: self.func( 335 "BITWISE_ARITHMETIC_SHIFT_LEFT", e.this, e.expression 336 ), 337 exp.BitwiseNot: lambda self, e: self.func("BITWISE_NOT", e.this), 338 exp.BitwiseOr: lambda self, e: self.func("BITWISE_OR", e.this, e.expression), 339 exp.BitwiseRightShift: lambda self, e: self.func( 340 "BITWISE_ARITHMETIC_SHIFT_RIGHT", e.this, e.expression 341 ), 342 exp.BitwiseXor: lambda self, e: self.func("BITWISE_XOR", e.this, e.expression), 343 exp.Cast: transforms.preprocess([transforms.epoch_cast_to_ts]), 344 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 345 exp.DateAdd: lambda self, e: self.func( 346 "DATE_ADD", 347 exp.Literal.string(e.text("unit") or "DAY"), 348 _to_int(e.expression), 349 e.this, 350 ), 351 exp.DateDiff: lambda self, e: self.func( 352 "DATE_DIFF", exp.Literal.string(e.text("unit") or "DAY"), e.expression, e.this 353 ), 354 exp.DateStrToDate: datestrtodate_sql, 355 exp.DateToDi: lambda self, 356 e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.DATEINT_FORMAT}) AS INT)", 357 exp.DateSub: lambda self, e: self.func( 358 "DATE_ADD", 359 exp.Literal.string(e.text("unit") or "DAY"), 360 _to_int(e.expression * -1), 361 e.this, 362 ), 363 exp.Decode: lambda self, e: encode_decode_sql(self, e, "FROM_UTF8"), 364 exp.DiToDate: lambda self, 365 e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.DATEINT_FORMAT}) AS DATE)", 366 exp.Encode: lambda self, e: encode_decode_sql(self, e, "TO_UTF8"), 367 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 368 exp.First: _first_last_sql, 369 exp.FirstValue: _first_last_sql, 370 exp.FromTimeZone: lambda self, 371 e: f"WITH_TIMEZONE({self.sql(e, 'this')}, {self.sql(e, 'zone')}) AT TIME ZONE 'UTC'", 372 exp.Group: transforms.preprocess([transforms.unalias_group]), 373 exp.GroupConcat: lambda self, e: self.func( 374 "ARRAY_JOIN", self.func("ARRAY_AGG", e.this), e.args.get("separator") 375 ), 376 exp.Hex: rename_func("TO_HEX"), 377 exp.If: if_sql(), 378 exp.ILike: no_ilike_sql, 379 exp.Initcap: _initcap_sql, 380 exp.ParseJSON: rename_func("JSON_PARSE"), 381 exp.Last: _first_last_sql, 382 exp.LastValue: _first_last_sql, 383 exp.LastDay: lambda self, e: self.func("LAST_DAY_OF_MONTH", e.this), 384 exp.Lateral: _explode_to_unnest_sql, 385 exp.Left: left_to_substring_sql, 386 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 387 exp.LogicalAnd: rename_func("BOOL_AND"), 388 exp.LogicalOr: rename_func("BOOL_OR"), 389 exp.Pivot: no_pivot_sql, 390 exp.Quantile: _quantile_sql, 391 exp.RegexpExtract: regexp_extract_sql, 392 exp.Right: right_to_substring_sql, 393 exp.SafeDivide: no_safe_divide_sql, 394 exp.Schema: _schema_sql, 395 exp.SchemaCommentProperty: lambda self, e: self.naked_property(e), 396 exp.Select: transforms.preprocess( 397 [ 398 transforms.eliminate_qualify, 399 transforms.eliminate_distinct_on, 400 transforms.explode_to_unnest(1), 401 transforms.eliminate_semi_and_anti_joins, 402 ] 403 ), 404 exp.SortArray: _no_sort_array, 405 exp.StrPosition: rename_func("STRPOS"), 406 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 407 exp.StrToMap: rename_func("SPLIT_TO_MAP"), 408 exp.StrToTime: _str_to_time_sql, 409 exp.StructExtract: struct_extract_sql, 410 exp.Table: transforms.preprocess([_unnest_sequence]), 411 exp.Timestamp: no_timestamp_sql, 412 exp.TimestampTrunc: timestamptrunc_sql, 413 exp.TimeStrToDate: timestrtotime_sql, 414 exp.TimeStrToTime: timestrtotime_sql, 415 exp.TimeStrToUnix: lambda self, e: self.func( 416 "TO_UNIXTIME", self.func("DATE_PARSE", e.this, Presto.TIME_FORMAT) 417 ), 418 exp.TimeToStr: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 419 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 420 exp.ToChar: lambda self, e: self.func("DATE_FORMAT", e.this, self.format_time(e)), 421 exp.TryCast: transforms.preprocess([transforms.epoch_cast_to_ts]), 422 exp.TsOrDiToDi: lambda self, 423 e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 424 exp.TsOrDsAdd: _ts_or_ds_add_sql, 425 exp.TsOrDsDiff: _ts_or_ds_diff_sql, 426 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 427 exp.Unhex: rename_func("FROM_HEX"), 428 exp.UnixToStr: lambda self, 429 e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 430 exp.UnixToTime: _unix_to_time_sql, 431 exp.UnixToTimeStr: lambda self, 432 e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 433 exp.VariancePop: rename_func("VAR_POP"), 434 exp.With: transforms.preprocess([transforms.add_recursive_cte_column_names]), 435 exp.WithinGroup: transforms.preprocess( 436 [transforms.remove_within_group_for_percentiles] 437 ), 438 exp.Xor: bool_xor_sql, 439 } 440 441 def strtounix_sql(self, expression: exp.StrToUnix) -> str: 442 # Since `TO_UNIXTIME` requires a `TIMESTAMP`, we need to parse the argument into one. 443 # To do this, we first try to `DATE_PARSE` it, but since this can fail when there's a 444 # timezone involved, we wrap it in a `TRY` call and use `PARSE_DATETIME` as a fallback, 445 # which seems to be using the same time mapping as Hive, as per: 446 # https://joda-time.sourceforge.net/apidocs/org/joda/time/format/DateTimeFormat.html 447 value_as_text = exp.cast(expression.this, "text") 448 parse_without_tz = self.func("DATE_PARSE", value_as_text, self.format_time(expression)) 449 parse_with_tz = self.func( 450 "PARSE_DATETIME", 451 value_as_text, 452 self.format_time(expression, Hive.INVERSE_TIME_MAPPING, Hive.INVERSE_TIME_TRIE), 453 ) 454 coalesced = self.func("COALESCE", self.func("TRY", parse_without_tz), parse_with_tz) 455 return self.func("TO_UNIXTIME", coalesced) 456 457 def bracket_sql(self, expression: exp.Bracket) -> str: 458 if expression.args.get("safe"): 459 return self.func( 460 "ELEMENT_AT", 461 expression.this, 462 seq_get( 463 apply_index_offset( 464 expression.this, 465 expression.expressions, 466 1 - expression.args.get("offset", 0), 467 ), 468 0, 469 ), 470 ) 471 return super().bracket_sql(expression) 472 473 def struct_sql(self, expression: exp.Struct) -> str: 474 from sqlglot.optimizer.annotate_types import annotate_types 475 476 expression = annotate_types(expression) 477 values: t.List[str] = [] 478 schema: t.List[str] = [] 479 unknown_type = False 480 481 for e in expression.expressions: 482 if isinstance(e, exp.PropertyEQ): 483 if e.type and e.type.is_type(exp.DataType.Type.UNKNOWN): 484 unknown_type = True 485 else: 486 schema.append(f"{self.sql(e, 'this')} {self.sql(e.type)}") 487 values.append(self.sql(e, "expression")) 488 else: 489 values.append(self.sql(e)) 490 491 size = len(expression.expressions) 492 493 if not size or len(schema) != size: 494 if unknown_type: 495 self.unsupported( 496 "Cannot convert untyped key-value definitions (try annotate_types)." 497 ) 498 return self.func("ROW", *values) 499 return f"CAST(ROW({', '.join(values)}) AS ROW({', '.join(schema)}))" 500 501 def interval_sql(self, expression: exp.Interval) -> str: 502 unit = self.sql(expression, "unit") 503 if expression.this and unit.startswith("WEEK"): 504 return f"({expression.this.name} * INTERVAL '7' DAY)" 505 return super().interval_sql(expression) 506 507 def transaction_sql(self, expression: exp.Transaction) -> str: 508 modes = expression.args.get("modes") 509 modes = f" {', '.join(modes)}" if modes else "" 510 return f"START TRANSACTION{modes}" 511 512 def generateseries_sql(self, expression: exp.GenerateSeries) -> str: 513 start = expression.args["start"] 514 end = expression.args["end"] 515 step = expression.args.get("step") 516 517 if isinstance(start, exp.Cast): 518 target_type = start.to 519 elif isinstance(end, exp.Cast): 520 target_type = end.to 521 else: 522 target_type = None 523 524 if target_type and target_type.is_type("timestamp"): 525 if target_type is start.to: 526 end = exp.cast(end, target_type) 527 else: 528 start = exp.cast(start, target_type) 529 530 return self.func("SEQUENCE", start, end, step) 531 532 def offset_limit_modifiers( 533 self, expression: exp.Expression, fetch: bool, limit: t.Optional[exp.Fetch | exp.Limit] 534 ) -> t.List[str]: 535 return [ 536 self.sql(expression, "offset"), 537 self.sql(limit), 538 ] 539 540 def create_sql(self, expression: exp.Create) -> str: 541 """ 542 Presto doesn't support CREATE VIEW with expressions (ex: `CREATE VIEW x (cola)` then `(cola)` is the expression), 543 so we need to remove them 544 """ 545 kind = expression.args["kind"] 546 schema = expression.this 547 if kind == "VIEW" and schema.expressions: 548 expression.this.set("expressions", None) 549 return super().create_sql(expression)
Generator converts a given syntax tree to the corresponding SQL string.
Arguments:
- pretty: Whether to format the produced SQL string. Default: False.
- identify: Determines when an identifier should be quoted. Possible values are: False (default): Never quote, except in cases where it's mandatory by the dialect. True or 'always': Always quote. 'safe': Only quote identifiers that are case insensitive.
- normalize: Whether to normalize identifiers to lowercase. Default: False.
- pad: The pad size in a formatted string. For example, this affects the indentation of a projection in a query, relative to its nesting level. Default: 2.
- indent: The indentation size in a formatted string. For example, this affects the
indentation of subqueries and filters under a
WHEREclause. Default: 2. - normalize_functions: How to normalize function names. Possible values are: "upper" or True (default): Convert names to uppercase. "lower": Convert names to lowercase. False: Disables function name normalization.
- unsupported_level: Determines the generator's behavior when it encounters unsupported expressions. Default ErrorLevel.WARN.
- max_unsupported: Maximum number of unsupported messages to include in a raised UnsupportedError. This is only relevant if unsupported_level is ErrorLevel.RAISE. Default: 3
- leading_comma: Whether the comma is leading or trailing in select expressions. This is only relevant when generating in pretty mode. Default: False
- max_text_width: The max number of characters in a segment before creating new lines in pretty mode. The default is on the smaller end because the length only represents a segment and not the true line length. Default: 80
- comments: Whether to preserve comments in the output SQL code. Default: True
441 def strtounix_sql(self, expression: exp.StrToUnix) -> str: 442 # Since `TO_UNIXTIME` requires a `TIMESTAMP`, we need to parse the argument into one. 443 # To do this, we first try to `DATE_PARSE` it, but since this can fail when there's a 444 # timezone involved, we wrap it in a `TRY` call and use `PARSE_DATETIME` as a fallback, 445 # which seems to be using the same time mapping as Hive, as per: 446 # https://joda-time.sourceforge.net/apidocs/org/joda/time/format/DateTimeFormat.html 447 value_as_text = exp.cast(expression.this, "text") 448 parse_without_tz = self.func("DATE_PARSE", value_as_text, self.format_time(expression)) 449 parse_with_tz = self.func( 450 "PARSE_DATETIME", 451 value_as_text, 452 self.format_time(expression, Hive.INVERSE_TIME_MAPPING, Hive.INVERSE_TIME_TRIE), 453 ) 454 coalesced = self.func("COALESCE", self.func("TRY", parse_without_tz), parse_with_tz) 455 return self.func("TO_UNIXTIME", coalesced)
457 def bracket_sql(self, expression: exp.Bracket) -> str: 458 if expression.args.get("safe"): 459 return self.func( 460 "ELEMENT_AT", 461 expression.this, 462 seq_get( 463 apply_index_offset( 464 expression.this, 465 expression.expressions, 466 1 - expression.args.get("offset", 0), 467 ), 468 0, 469 ), 470 ) 471 return super().bracket_sql(expression)
473 def struct_sql(self, expression: exp.Struct) -> str: 474 from sqlglot.optimizer.annotate_types import annotate_types 475 476 expression = annotate_types(expression) 477 values: t.List[str] = [] 478 schema: t.List[str] = [] 479 unknown_type = False 480 481 for e in expression.expressions: 482 if isinstance(e, exp.PropertyEQ): 483 if e.type and e.type.is_type(exp.DataType.Type.UNKNOWN): 484 unknown_type = True 485 else: 486 schema.append(f"{self.sql(e, 'this')} {self.sql(e.type)}") 487 values.append(self.sql(e, "expression")) 488 else: 489 values.append(self.sql(e)) 490 491 size = len(expression.expressions) 492 493 if not size or len(schema) != size: 494 if unknown_type: 495 self.unsupported( 496 "Cannot convert untyped key-value definitions (try annotate_types)." 497 ) 498 return self.func("ROW", *values) 499 return f"CAST(ROW({', '.join(values)}) AS ROW({', '.join(schema)}))"
512 def generateseries_sql(self, expression: exp.GenerateSeries) -> str: 513 start = expression.args["start"] 514 end = expression.args["end"] 515 step = expression.args.get("step") 516 517 if isinstance(start, exp.Cast): 518 target_type = start.to 519 elif isinstance(end, exp.Cast): 520 target_type = end.to 521 else: 522 target_type = None 523 524 if target_type and target_type.is_type("timestamp"): 525 if target_type is start.to: 526 end = exp.cast(end, target_type) 527 else: 528 start = exp.cast(start, target_type) 529 530 return self.func("SEQUENCE", start, end, step)
540 def create_sql(self, expression: exp.Create) -> str: 541 """ 542 Presto doesn't support CREATE VIEW with expressions (ex: `CREATE VIEW x (cola)` then `(cola)` is the expression), 543 so we need to remove them 544 """ 545 kind = expression.args["kind"] 546 schema = expression.this 547 if kind == "VIEW" and schema.expressions: 548 expression.this.set("expressions", None) 549 return super().create_sql(expression)
Presto doesn't support CREATE VIEW with expressions (ex: CREATE VIEW x (cola) then (cola) is the expression),
so we need to remove them
Inherited Members
- sqlglot.generator.Generator
- Generator
- NULL_ORDERING_SUPPORTED
- IGNORE_NULLS_IN_FUNC
- LOCKING_READS_SUPPORTED
- EXPLICIT_UNION
- WRAP_DERIVED_VALUES
- CREATE_FUNCTION_RETURN_AS
- MATCHED_BY_SOURCE
- SINGLE_STRING_INTERVAL
- LIMIT_FETCH
- RENAME_TABLE_WITH_DB
- GROUPINGS_SEP
- INDEX_ON
- QUERY_HINT_SEP
- DUPLICATE_KEY_UPDATE_WITH_SET
- LIMIT_IS_TOP
- RETURNING_END
- COLUMN_JOIN_MARKS_SUPPORTED
- EXTRACT_ALLOWS_QUOTES
- VALUES_AS_TABLE
- ALTER_TABLE_INCLUDE_COLUMN_KEYWORD
- UNNEST_WITH_ORDINALITY
- AGGREGATE_FILTER_SUPPORTED
- SEMI_ANTI_JOIN_WITH_SIDE
- COMPUTED_COLUMN_WITH_TYPE
- SUPPORTS_TABLE_COPY
- TABLESAMPLE_REQUIRES_PARENS
- TABLESAMPLE_SIZE_IS_ROWS
- TABLESAMPLE_KEYWORDS
- TABLESAMPLE_WITH_METHOD
- TABLESAMPLE_SEED_KEYWORD
- COLLATE_IS_FUNC
- DATA_TYPE_SPECIFIERS_ALLOWED
- ENSURE_BOOLS
- CTE_RECURSIVE_KEYWORD_REQUIRED
- LAST_DAY_SUPPORTS_DATE_PART
- SUPPORTS_TABLE_ALIAS_COLUMNS
- UNPIVOT_ALIASES_ARE_IDENTIFIERS
- JSON_KEY_VALUE_PAIR_SEP
- INSERT_OVERWRITE
- SUPPORTS_SELECT_INTO
- SUPPORTS_UNLOGGED_TABLES
- SUPPORTS_CREATE_TABLE_LIKE
- JSON_TYPE_REQUIRED_FOR_EXTRACTION
- JSON_PATH_BRACKETED_KEY_SUPPORTED
- JSON_PATH_SINGLE_QUOTE_ESCAPE
- SUPPORTED_JSON_PATH_PARTS
- CAN_IMPLEMENT_ARRAY_ANY
- STAR_MAPPING
- TIME_PART_SINGULARS
- TOKEN_MAPPING
- PARAMETER_TOKEN
- NAMED_PLACEHOLDER_TOKEN
- RESERVED_KEYWORDS
- WITH_SEPARATED_COMMENTS
- EXCLUDE_COMMENTS
- UNWRAPPED_INTERVAL_VALUES
- PARAMETERIZABLE_TEXT_TYPES
- EXPRESSIONS_WITHOUT_NESTED_CTES
- SENTINEL_LINE_BREAK
- pretty
- identify
- normalize
- pad
- unsupported_level
- max_unsupported
- leading_comma
- max_text_width
- comments
- dialect
- normalize_functions
- unsupported_messages
- generate
- preprocess
- unsupported
- sep
- seg
- pad_comment
- maybe_comment
- wrap
- no_identify
- normalize_func
- indent
- sql
- uncache_sql
- cache_sql
- characterset_sql
- column_sql
- columnposition_sql
- columndef_sql
- columnconstraint_sql
- computedcolumnconstraint_sql
- autoincrementcolumnconstraint_sql
- compresscolumnconstraint_sql
- generatedasidentitycolumnconstraint_sql
- generatedasrowcolumnconstraint_sql
- periodforsystemtimeconstraint_sql
- notnullcolumnconstraint_sql
- transformcolumnconstraint_sql
- primarykeycolumnconstraint_sql
- uniquecolumnconstraint_sql
- createable_sql
- sequenceproperties_sql
- clone_sql
- describe_sql
- heredoc_sql
- prepend_ctes
- with_sql
- cte_sql
- tablealias_sql
- bitstring_sql
- hexstring_sql
- bytestring_sql
- unicodestring_sql
- rawstring_sql
- datatypeparam_sql
- datatype_sql
- directory_sql
- delete_sql
- drop_sql
- except_sql
- except_op
- fetch_sql
- filter_sql
- hint_sql
- indexparameters_sql
- index_sql
- identifier_sql
- inputoutputformat_sql
- national_sql
- partition_sql
- properties_sql
- root_properties
- properties
- with_properties
- locate_properties
- property_name
- property_sql
- likeproperty_sql
- fallbackproperty_sql
- journalproperty_sql
- freespaceproperty_sql
- checksumproperty_sql
- mergeblockratioproperty_sql
- datablocksizeproperty_sql
- blockcompressionproperty_sql
- isolatedloadingproperty_sql
- partitionboundspec_sql
- partitionedofproperty_sql
- lockingproperty_sql
- withdataproperty_sql
- withsystemversioningproperty_sql
- insert_sql
- intersect_sql
- intersect_op
- introducer_sql
- kill_sql
- pseudotype_sql
- objectidentifier_sql
- onconflict_sql
- returning_sql
- rowformatdelimitedproperty_sql
- withtablehint_sql
- indextablehint_sql
- historicaldata_sql
- table_parts
- table_sql
- tablesample_sql
- pivot_sql
- version_sql
- tuple_sql
- update_sql
- values_sql
- var_sql
- into_sql
- from_sql
- group_sql
- having_sql
- connect_sql
- prior_sql
- join_sql
- lambda_sql
- lateral_op
- lateral_sql
- limit_sql
- offset_sql
- setitem_sql
- set_sql
- pragma_sql
- lock_sql
- literal_sql
- escape_str
- loaddata_sql
- null_sql
- boolean_sql
- order_sql
- withfill_sql
- cluster_sql
- distribute_sql
- sort_sql
- ordered_sql
- matchrecognize_sql
- query_modifiers
- queryoption_sql
- after_limit_modifiers
- select_sql
- schema_sql
- schema_columns_sql
- star_sql
- parameter_sql
- sessionparameter_sql
- placeholder_sql
- subquery_sql
- qualify_sql
- set_operations
- union_sql
- union_op
- unnest_sql
- prewhere_sql
- where_sql
- window_sql
- partition_by_sql
- windowspec_sql
- withingroup_sql
- between_sql
- all_sql
- any_sql
- exists_sql
- case_sql
- constraint_sql
- nextvaluefor_sql
- extract_sql
- trim_sql
- convert_concat_args
- concat_sql
- concatws_sql
- check_sql
- foreignkey_sql
- primarykey_sql
- if_sql
- matchagainst_sql
- jsonkeyvalue_sql
- jsonpath_sql
- json_path_part
- formatjson_sql
- jsonobject_sql
- jsonobjectagg_sql
- jsonarray_sql
- jsonarrayagg_sql
- jsoncolumndef_sql
- jsonschema_sql
- jsontable_sql
- openjsoncolumndef_sql
- openjson_sql
- in_sql
- in_unnest_op
- return_sql
- reference_sql
- anonymous_sql
- paren_sql
- neg_sql
- not_sql
- alias_sql
- pivotalias_sql
- aliases_sql
- atindex_sql
- attimezone_sql
- fromtimezone_sql
- add_sql
- and_sql
- or_sql
- xor_sql
- connector_sql
- bitwiseand_sql
- bitwiseleftshift_sql
- bitwisenot_sql
- bitwiseor_sql
- bitwiserightshift_sql
- bitwisexor_sql
- cast_sql
- currentdate_sql
- currenttimestamp_sql
- collate_sql
- command_sql
- comment_sql
- mergetreettlaction_sql
- mergetreettl_sql
- commit_sql
- rollback_sql
- altercolumn_sql
- renametable_sql
- renamecolumn_sql
- altertable_sql
- add_column_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- havingmax_sql
- intdiv_sql
- dpipe_sql
- div_sql
- overlaps_sql
- distance_sql
- dot_sql
- eq_sql
- propertyeq_sql
- escape_sql
- glob_sql
- gt_sql
- gte_sql
- ilike_sql
- ilikeany_sql
- is_sql
- like_sql
- likeany_sql
- similarto_sql
- lt_sql
- lte_sql
- mod_sql
- mul_sql
- neq_sql
- nullsafeeq_sql
- nullsafeneq_sql
- slice_sql
- sub_sql
- trycast_sql
- log_sql
- use_sql
- binary
- function_fallback_sql
- func
- format_args
- text_width
- format_time
- expressions
- op_expressions
- naked_property
- tag_sql
- token_sql
- userdefinedfunction_sql
- joinhint_sql
- kwarg_sql
- when_sql
- merge_sql
- tochar_sql
- tonumber_sql
- dictproperty_sql
- dictrange_sql
- dictsubproperty_sql
- oncluster_sql
- clusteredbyproperty_sql
- anyvalue_sql
- querytransform_sql
- indexconstraintoption_sql
- checkcolumnconstraint_sql
- indexcolumnconstraint_sql
- nvl2_sql
- comprehension_sql
- columnprefix_sql
- opclass_sql
- predict_sql
- forin_sql
- refresh_sql
- operator_sql
- toarray_sql
- tsordstotime_sql
- tsordstotimestamp_sql
- tsordstodate_sql
- unixdate_sql
- lastday_sql
- arrayany_sql
- partitionrange_sql
- truncatetable_sql
- convert_sql