sqlglot.tokens
1from __future__ import annotations 2 3import typing as t 4from enum import auto 5 6from sqlglot.errors import TokenError 7from sqlglot.helper import AutoName 8from sqlglot.trie import TrieResult, in_trie, new_trie 9 10 11class TokenType(AutoName): 12 L_PAREN = auto() 13 R_PAREN = auto() 14 L_BRACKET = auto() 15 R_BRACKET = auto() 16 L_BRACE = auto() 17 R_BRACE = auto() 18 COMMA = auto() 19 DOT = auto() 20 DASH = auto() 21 PLUS = auto() 22 COLON = auto() 23 DCOLON = auto() 24 DQMARK = auto() 25 SEMICOLON = auto() 26 STAR = auto() 27 BACKSLASH = auto() 28 SLASH = auto() 29 LT = auto() 30 LTE = auto() 31 GT = auto() 32 GTE = auto() 33 NOT = auto() 34 EQ = auto() 35 NEQ = auto() 36 NULLSAFE_EQ = auto() 37 AND = auto() 38 OR = auto() 39 AMP = auto() 40 DPIPE = auto() 41 PIPE = auto() 42 CARET = auto() 43 TILDA = auto() 44 ARROW = auto() 45 DARROW = auto() 46 FARROW = auto() 47 HASH = auto() 48 HASH_ARROW = auto() 49 DHASH_ARROW = auto() 50 LR_ARROW = auto() 51 DAT = auto() 52 LT_AT = auto() 53 AT_GT = auto() 54 DOLLAR = auto() 55 PARAMETER = auto() 56 SESSION_PARAMETER = auto() 57 DAMP = auto() 58 XOR = auto() 59 60 BLOCK_START = auto() 61 BLOCK_END = auto() 62 63 SPACE = auto() 64 BREAK = auto() 65 66 STRING = auto() 67 NUMBER = auto() 68 IDENTIFIER = auto() 69 DATABASE = auto() 70 COLUMN = auto() 71 COLUMN_DEF = auto() 72 SCHEMA = auto() 73 TABLE = auto() 74 VAR = auto() 75 BIT_STRING = auto() 76 HEX_STRING = auto() 77 BYTE_STRING = auto() 78 NATIONAL_STRING = auto() 79 RAW_STRING = auto() 80 81 # types 82 BIT = auto() 83 BOOLEAN = auto() 84 TINYINT = auto() 85 UTINYINT = auto() 86 SMALLINT = auto() 87 USMALLINT = auto() 88 MEDIUMINT = auto() 89 UMEDIUMINT = auto() 90 INT = auto() 91 UINT = auto() 92 BIGINT = auto() 93 UBIGINT = auto() 94 INT128 = auto() 95 UINT128 = auto() 96 INT256 = auto() 97 UINT256 = auto() 98 FLOAT = auto() 99 DOUBLE = auto() 100 DECIMAL = auto() 101 BIGDECIMAL = auto() 102 CHAR = auto() 103 NCHAR = auto() 104 VARCHAR = auto() 105 NVARCHAR = auto() 106 TEXT = auto() 107 MEDIUMTEXT = auto() 108 LONGTEXT = auto() 109 MEDIUMBLOB = auto() 110 LONGBLOB = auto() 111 TINYBLOB = auto() 112 TINYTEXT = auto() 113 BINARY = auto() 114 VARBINARY = auto() 115 JSON = auto() 116 JSONB = auto() 117 TIME = auto() 118 TIMETZ = auto() 119 TIMESTAMP = auto() 120 TIMESTAMPTZ = auto() 121 TIMESTAMPLTZ = auto() 122 DATETIME = auto() 123 DATETIME64 = auto() 124 DATE = auto() 125 INT4RANGE = auto() 126 INT4MULTIRANGE = auto() 127 INT8RANGE = auto() 128 INT8MULTIRANGE = auto() 129 NUMRANGE = auto() 130 NUMMULTIRANGE = auto() 131 TSRANGE = auto() 132 TSMULTIRANGE = auto() 133 TSTZRANGE = auto() 134 TSTZMULTIRANGE = auto() 135 DATERANGE = auto() 136 DATEMULTIRANGE = auto() 137 UUID = auto() 138 GEOGRAPHY = auto() 139 NULLABLE = auto() 140 GEOMETRY = auto() 141 HLLSKETCH = auto() 142 HSTORE = auto() 143 SUPER = auto() 144 SERIAL = auto() 145 SMALLSERIAL = auto() 146 BIGSERIAL = auto() 147 XML = auto() 148 YEAR = auto() 149 UNIQUEIDENTIFIER = auto() 150 USERDEFINED = auto() 151 MONEY = auto() 152 SMALLMONEY = auto() 153 ROWVERSION = auto() 154 IMAGE = auto() 155 VARIANT = auto() 156 OBJECT = auto() 157 INET = auto() 158 IPADDRESS = auto() 159 IPPREFIX = auto() 160 ENUM = auto() 161 ENUM8 = auto() 162 ENUM16 = auto() 163 FIXEDSTRING = auto() 164 LOWCARDINALITY = auto() 165 NESTED = auto() 166 UNKNOWN = auto() 167 168 # keywords 169 ALIAS = auto() 170 ALTER = auto() 171 ALWAYS = auto() 172 ALL = auto() 173 ANTI = auto() 174 ANY = auto() 175 APPLY = auto() 176 ARRAY = auto() 177 ASC = auto() 178 ASOF = auto() 179 AUTO_INCREMENT = auto() 180 BEGIN = auto() 181 BETWEEN = auto() 182 CACHE = auto() 183 CASE = auto() 184 CHARACTER_SET = auto() 185 CLUSTER_BY = auto() 186 COLLATE = auto() 187 COMMAND = auto() 188 COMMENT = auto() 189 COMMIT = auto() 190 CONNECT_BY = auto() 191 CONSTRAINT = auto() 192 CREATE = auto() 193 CROSS = auto() 194 CUBE = auto() 195 CURRENT_DATE = auto() 196 CURRENT_DATETIME = auto() 197 CURRENT_TIME = auto() 198 CURRENT_TIMESTAMP = auto() 199 CURRENT_USER = auto() 200 DEFAULT = auto() 201 DELETE = auto() 202 DESC = auto() 203 DESCRIBE = auto() 204 DICTIONARY = auto() 205 DISTINCT = auto() 206 DISTRIBUTE_BY = auto() 207 DIV = auto() 208 DROP = auto() 209 ELSE = auto() 210 END = auto() 211 ESCAPE = auto() 212 EXCEPT = auto() 213 EXECUTE = auto() 214 EXISTS = auto() 215 FALSE = auto() 216 FETCH = auto() 217 FILTER = auto() 218 FINAL = auto() 219 FIRST = auto() 220 FOR = auto() 221 FORCE = auto() 222 FOREIGN_KEY = auto() 223 FORMAT = auto() 224 FROM = auto() 225 FULL = auto() 226 FUNCTION = auto() 227 GLOB = auto() 228 GLOBAL = auto() 229 GROUP_BY = auto() 230 GROUPING_SETS = auto() 231 HAVING = auto() 232 HINT = auto() 233 IGNORE = auto() 234 ILIKE = auto() 235 ILIKE_ANY = auto() 236 IN = auto() 237 INDEX = auto() 238 INNER = auto() 239 INSERT = auto() 240 INTERSECT = auto() 241 INTERVAL = auto() 242 INTO = auto() 243 INTRODUCER = auto() 244 IRLIKE = auto() 245 IS = auto() 246 ISNULL = auto() 247 JOIN = auto() 248 JOIN_MARKER = auto() 249 KEEP = auto() 250 KILL = auto() 251 LANGUAGE = auto() 252 LATERAL = auto() 253 LEFT = auto() 254 LIKE = auto() 255 LIKE_ANY = auto() 256 LIMIT = auto() 257 LOAD = auto() 258 LOCK = auto() 259 MAP = auto() 260 MATCH_RECOGNIZE = auto() 261 MEMBER_OF = auto() 262 MERGE = auto() 263 MOD = auto() 264 NATURAL = auto() 265 NEXT = auto() 266 NOTNULL = auto() 267 NULL = auto() 268 OBJECT_IDENTIFIER = auto() 269 OFFSET = auto() 270 ON = auto() 271 ORDER_BY = auto() 272 ORDERED = auto() 273 ORDINALITY = auto() 274 OUTER = auto() 275 OVER = auto() 276 OVERLAPS = auto() 277 OVERWRITE = auto() 278 PARTITION = auto() 279 PARTITION_BY = auto() 280 PERCENT = auto() 281 PIVOT = auto() 282 PLACEHOLDER = auto() 283 PRAGMA = auto() 284 PRIMARY_KEY = auto() 285 PROCEDURE = auto() 286 PROPERTIES = auto() 287 PSEUDO_TYPE = auto() 288 QUALIFY = auto() 289 QUOTE = auto() 290 RANGE = auto() 291 RECURSIVE = auto() 292 REPLACE = auto() 293 RETURNING = auto() 294 REFERENCES = auto() 295 RIGHT = auto() 296 RLIKE = auto() 297 ROLLBACK = auto() 298 ROLLUP = auto() 299 ROW = auto() 300 ROWS = auto() 301 SELECT = auto() 302 SEMI = auto() 303 SEPARATOR = auto() 304 SERDE_PROPERTIES = auto() 305 SET = auto() 306 SETTINGS = auto() 307 SHOW = auto() 308 SIMILAR_TO = auto() 309 SOME = auto() 310 SORT_BY = auto() 311 START_WITH = auto() 312 STRUCT = auto() 313 TABLE_SAMPLE = auto() 314 TEMPORARY = auto() 315 TOP = auto() 316 THEN = auto() 317 TRUE = auto() 318 UNCACHE = auto() 319 UNION = auto() 320 UNNEST = auto() 321 UNPIVOT = auto() 322 UPDATE = auto() 323 USE = auto() 324 USING = auto() 325 VALUES = auto() 326 VIEW = auto() 327 VOLATILE = auto() 328 WHEN = auto() 329 WHERE = auto() 330 WINDOW = auto() 331 WITH = auto() 332 UNIQUE = auto() 333 VERSION_SNAPSHOT = auto() 334 TIMESTAMP_SNAPSHOT = auto() 335 336 337class Token: 338 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 339 340 @classmethod 341 def number(cls, number: int) -> Token: 342 """Returns a NUMBER token with `number` as its text.""" 343 return cls(TokenType.NUMBER, str(number)) 344 345 @classmethod 346 def string(cls, string: str) -> Token: 347 """Returns a STRING token with `string` as its text.""" 348 return cls(TokenType.STRING, string) 349 350 @classmethod 351 def identifier(cls, identifier: str) -> Token: 352 """Returns an IDENTIFIER token with `identifier` as its text.""" 353 return cls(TokenType.IDENTIFIER, identifier) 354 355 @classmethod 356 def var(cls, var: str) -> Token: 357 """Returns an VAR token with `var` as its text.""" 358 return cls(TokenType.VAR, var) 359 360 def __init__( 361 self, 362 token_type: TokenType, 363 text: str, 364 line: int = 1, 365 col: int = 1, 366 start: int = 0, 367 end: int = 0, 368 comments: t.List[str] = [], 369 ) -> None: 370 """Token initializer. 371 372 Args: 373 token_type: The TokenType Enum. 374 text: The text of the token. 375 line: The line that the token ends on. 376 col: The column that the token ends on. 377 start: The start index of the token. 378 end: The ending index of the token. 379 comments: The comments to attach to the token. 380 """ 381 self.token_type = token_type 382 self.text = text 383 self.line = line 384 self.col = col 385 self.start = start 386 self.end = end 387 self.comments = comments 388 389 def __repr__(self) -> str: 390 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 391 return f"<Token {attributes}>" 392 393 394class _Tokenizer(type): 395 def __new__(cls, clsname, bases, attrs): 396 klass = super().__new__(cls, clsname, bases, attrs) 397 398 def _convert_quotes(arr: t.List[str | t.Tuple[str, str]]) -> t.Dict[str, str]: 399 return dict( 400 (item, item) if isinstance(item, str) else (item[0], item[1]) for item in arr 401 ) 402 403 def _quotes_to_format( 404 token_type: TokenType, arr: t.List[str | t.Tuple[str, str]] 405 ) -> t.Dict[str, t.Tuple[str, TokenType]]: 406 return {k: (v, token_type) for k, v in _convert_quotes(arr).items()} 407 408 klass._QUOTES = _convert_quotes(klass.QUOTES) 409 klass._IDENTIFIERS = _convert_quotes(klass.IDENTIFIERS) 410 411 klass._FORMAT_STRINGS = { 412 **{ 413 p + s: (e, TokenType.NATIONAL_STRING) 414 for s, e in klass._QUOTES.items() 415 for p in ("n", "N") 416 }, 417 **_quotes_to_format(TokenType.BIT_STRING, klass.BIT_STRINGS), 418 **_quotes_to_format(TokenType.BYTE_STRING, klass.BYTE_STRINGS), 419 **_quotes_to_format(TokenType.HEX_STRING, klass.HEX_STRINGS), 420 **_quotes_to_format(TokenType.RAW_STRING, klass.RAW_STRINGS), 421 } 422 423 klass._STRING_ESCAPES = set(klass.STRING_ESCAPES) 424 klass._IDENTIFIER_ESCAPES = set(klass.IDENTIFIER_ESCAPES) 425 klass._COMMENTS = { 426 **dict( 427 (comment, None) if isinstance(comment, str) else (comment[0], comment[1]) 428 for comment in klass.COMMENTS 429 ), 430 "{#": "#}", # Ensure Jinja comments are tokenized correctly in all dialects 431 } 432 433 klass._KEYWORD_TRIE = new_trie( 434 key.upper() 435 for key in ( 436 *klass.KEYWORDS, 437 *klass._COMMENTS, 438 *klass._QUOTES, 439 *klass._FORMAT_STRINGS, 440 ) 441 if " " in key or any(single in key for single in klass.SINGLE_TOKENS) 442 ) 443 444 return klass 445 446 447class Tokenizer(metaclass=_Tokenizer): 448 SINGLE_TOKENS = { 449 "(": TokenType.L_PAREN, 450 ")": TokenType.R_PAREN, 451 "[": TokenType.L_BRACKET, 452 "]": TokenType.R_BRACKET, 453 "{": TokenType.L_BRACE, 454 "}": TokenType.R_BRACE, 455 "&": TokenType.AMP, 456 "^": TokenType.CARET, 457 ":": TokenType.COLON, 458 ",": TokenType.COMMA, 459 ".": TokenType.DOT, 460 "-": TokenType.DASH, 461 "=": TokenType.EQ, 462 ">": TokenType.GT, 463 "<": TokenType.LT, 464 "%": TokenType.MOD, 465 "!": TokenType.NOT, 466 "|": TokenType.PIPE, 467 "+": TokenType.PLUS, 468 ";": TokenType.SEMICOLON, 469 "/": TokenType.SLASH, 470 "\\": TokenType.BACKSLASH, 471 "*": TokenType.STAR, 472 "~": TokenType.TILDA, 473 "?": TokenType.PLACEHOLDER, 474 "@": TokenType.PARAMETER, 475 # used for breaking a var like x'y' but nothing else 476 # the token type doesn't matter 477 "'": TokenType.QUOTE, 478 "`": TokenType.IDENTIFIER, 479 '"': TokenType.IDENTIFIER, 480 "#": TokenType.HASH, 481 } 482 483 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 484 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 485 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 486 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 487 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 488 IDENTIFIER_ESCAPES = ['"'] 489 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 490 STRING_ESCAPES = ["'"] 491 VAR_SINGLE_TOKENS: t.Set[str] = set() 492 493 # Autofilled 494 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 495 496 _COMMENTS: t.Dict[str, str] = {} 497 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 498 _IDENTIFIERS: t.Dict[str, str] = {} 499 _IDENTIFIER_ESCAPES: t.Set[str] = set() 500 _QUOTES: t.Dict[str, str] = {} 501 _STRING_ESCAPES: t.Set[str] = set() 502 _KEYWORD_TRIE: t.Dict = {} 503 504 KEYWORDS: t.Dict[str, TokenType] = { 505 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 506 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 507 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 508 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 509 "/*+": TokenType.HINT, 510 "==": TokenType.EQ, 511 "::": TokenType.DCOLON, 512 "||": TokenType.DPIPE, 513 ">=": TokenType.GTE, 514 "<=": TokenType.LTE, 515 "<>": TokenType.NEQ, 516 "!=": TokenType.NEQ, 517 "<=>": TokenType.NULLSAFE_EQ, 518 "->": TokenType.ARROW, 519 "->>": TokenType.DARROW, 520 "=>": TokenType.FARROW, 521 "#>": TokenType.HASH_ARROW, 522 "#>>": TokenType.DHASH_ARROW, 523 "<->": TokenType.LR_ARROW, 524 "&&": TokenType.DAMP, 525 "??": TokenType.DQMARK, 526 "ALL": TokenType.ALL, 527 "ALWAYS": TokenType.ALWAYS, 528 "AND": TokenType.AND, 529 "ANTI": TokenType.ANTI, 530 "ANY": TokenType.ANY, 531 "ASC": TokenType.ASC, 532 "AS": TokenType.ALIAS, 533 "ASOF": TokenType.ASOF, 534 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 535 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 536 "BEGIN": TokenType.BEGIN, 537 "BETWEEN": TokenType.BETWEEN, 538 "CACHE": TokenType.CACHE, 539 "UNCACHE": TokenType.UNCACHE, 540 "CASE": TokenType.CASE, 541 "CHARACTER SET": TokenType.CHARACTER_SET, 542 "CLUSTER BY": TokenType.CLUSTER_BY, 543 "COLLATE": TokenType.COLLATE, 544 "COLUMN": TokenType.COLUMN, 545 "COMMIT": TokenType.COMMIT, 546 "CONNECT BY": TokenType.CONNECT_BY, 547 "CONSTRAINT": TokenType.CONSTRAINT, 548 "CREATE": TokenType.CREATE, 549 "CROSS": TokenType.CROSS, 550 "CUBE": TokenType.CUBE, 551 "CURRENT_DATE": TokenType.CURRENT_DATE, 552 "CURRENT_TIME": TokenType.CURRENT_TIME, 553 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 554 "CURRENT_USER": TokenType.CURRENT_USER, 555 "DATABASE": TokenType.DATABASE, 556 "DEFAULT": TokenType.DEFAULT, 557 "DELETE": TokenType.DELETE, 558 "DESC": TokenType.DESC, 559 "DESCRIBE": TokenType.DESCRIBE, 560 "DISTINCT": TokenType.DISTINCT, 561 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 562 "DIV": TokenType.DIV, 563 "DROP": TokenType.DROP, 564 "ELSE": TokenType.ELSE, 565 "END": TokenType.END, 566 "ESCAPE": TokenType.ESCAPE, 567 "EXCEPT": TokenType.EXCEPT, 568 "EXECUTE": TokenType.EXECUTE, 569 "EXISTS": TokenType.EXISTS, 570 "FALSE": TokenType.FALSE, 571 "FETCH": TokenType.FETCH, 572 "FILTER": TokenType.FILTER, 573 "FIRST": TokenType.FIRST, 574 "FULL": TokenType.FULL, 575 "FUNCTION": TokenType.FUNCTION, 576 "FOR": TokenType.FOR, 577 "FOREIGN KEY": TokenType.FOREIGN_KEY, 578 "FORMAT": TokenType.FORMAT, 579 "FROM": TokenType.FROM, 580 "GEOGRAPHY": TokenType.GEOGRAPHY, 581 "GEOMETRY": TokenType.GEOMETRY, 582 "GLOB": TokenType.GLOB, 583 "GROUP BY": TokenType.GROUP_BY, 584 "GROUPING SETS": TokenType.GROUPING_SETS, 585 "HAVING": TokenType.HAVING, 586 "ILIKE": TokenType.ILIKE, 587 "IN": TokenType.IN, 588 "INDEX": TokenType.INDEX, 589 "INET": TokenType.INET, 590 "INNER": TokenType.INNER, 591 "INSERT": TokenType.INSERT, 592 "INTERVAL": TokenType.INTERVAL, 593 "INTERSECT": TokenType.INTERSECT, 594 "INTO": TokenType.INTO, 595 "IS": TokenType.IS, 596 "ISNULL": TokenType.ISNULL, 597 "JOIN": TokenType.JOIN, 598 "KEEP": TokenType.KEEP, 599 "KILL": TokenType.KILL, 600 "LATERAL": TokenType.LATERAL, 601 "LEFT": TokenType.LEFT, 602 "LIKE": TokenType.LIKE, 603 "LIMIT": TokenType.LIMIT, 604 "LOAD": TokenType.LOAD, 605 "LOCK": TokenType.LOCK, 606 "MERGE": TokenType.MERGE, 607 "NATURAL": TokenType.NATURAL, 608 "NEXT": TokenType.NEXT, 609 "NOT": TokenType.NOT, 610 "NOTNULL": TokenType.NOTNULL, 611 "NULL": TokenType.NULL, 612 "OBJECT": TokenType.OBJECT, 613 "OFFSET": TokenType.OFFSET, 614 "ON": TokenType.ON, 615 "OR": TokenType.OR, 616 "XOR": TokenType.XOR, 617 "ORDER BY": TokenType.ORDER_BY, 618 "ORDINALITY": TokenType.ORDINALITY, 619 "OUTER": TokenType.OUTER, 620 "OVER": TokenType.OVER, 621 "OVERLAPS": TokenType.OVERLAPS, 622 "OVERWRITE": TokenType.OVERWRITE, 623 "PARTITION": TokenType.PARTITION, 624 "PARTITION BY": TokenType.PARTITION_BY, 625 "PARTITIONED BY": TokenType.PARTITION_BY, 626 "PARTITIONED_BY": TokenType.PARTITION_BY, 627 "PERCENT": TokenType.PERCENT, 628 "PIVOT": TokenType.PIVOT, 629 "PRAGMA": TokenType.PRAGMA, 630 "PRIMARY KEY": TokenType.PRIMARY_KEY, 631 "PROCEDURE": TokenType.PROCEDURE, 632 "QUALIFY": TokenType.QUALIFY, 633 "RANGE": TokenType.RANGE, 634 "RECURSIVE": TokenType.RECURSIVE, 635 "REGEXP": TokenType.RLIKE, 636 "REPLACE": TokenType.REPLACE, 637 "RETURNING": TokenType.RETURNING, 638 "REFERENCES": TokenType.REFERENCES, 639 "RIGHT": TokenType.RIGHT, 640 "RLIKE": TokenType.RLIKE, 641 "ROLLBACK": TokenType.ROLLBACK, 642 "ROLLUP": TokenType.ROLLUP, 643 "ROW": TokenType.ROW, 644 "ROWS": TokenType.ROWS, 645 "SCHEMA": TokenType.SCHEMA, 646 "SELECT": TokenType.SELECT, 647 "SEMI": TokenType.SEMI, 648 "SET": TokenType.SET, 649 "SETTINGS": TokenType.SETTINGS, 650 "SHOW": TokenType.SHOW, 651 "SIMILAR TO": TokenType.SIMILAR_TO, 652 "SOME": TokenType.SOME, 653 "SORT BY": TokenType.SORT_BY, 654 "START WITH": TokenType.START_WITH, 655 "TABLE": TokenType.TABLE, 656 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 657 "TEMP": TokenType.TEMPORARY, 658 "TEMPORARY": TokenType.TEMPORARY, 659 "THEN": TokenType.THEN, 660 "TRUE": TokenType.TRUE, 661 "UNION": TokenType.UNION, 662 "UNKNOWN": TokenType.UNKNOWN, 663 "UNNEST": TokenType.UNNEST, 664 "UNPIVOT": TokenType.UNPIVOT, 665 "UPDATE": TokenType.UPDATE, 666 "USE": TokenType.USE, 667 "USING": TokenType.USING, 668 "UUID": TokenType.UUID, 669 "VALUES": TokenType.VALUES, 670 "VIEW": TokenType.VIEW, 671 "VOLATILE": TokenType.VOLATILE, 672 "WHEN": TokenType.WHEN, 673 "WHERE": TokenType.WHERE, 674 "WINDOW": TokenType.WINDOW, 675 "WITH": TokenType.WITH, 676 "APPLY": TokenType.APPLY, 677 "ARRAY": TokenType.ARRAY, 678 "BIT": TokenType.BIT, 679 "BOOL": TokenType.BOOLEAN, 680 "BOOLEAN": TokenType.BOOLEAN, 681 "BYTE": TokenType.TINYINT, 682 "MEDIUMINT": TokenType.MEDIUMINT, 683 "TINYINT": TokenType.TINYINT, 684 "SHORT": TokenType.SMALLINT, 685 "SMALLINT": TokenType.SMALLINT, 686 "INT128": TokenType.INT128, 687 "INT2": TokenType.SMALLINT, 688 "INTEGER": TokenType.INT, 689 "INT": TokenType.INT, 690 "INT4": TokenType.INT, 691 "LONG": TokenType.BIGINT, 692 "BIGINT": TokenType.BIGINT, 693 "INT8": TokenType.BIGINT, 694 "DEC": TokenType.DECIMAL, 695 "DECIMAL": TokenType.DECIMAL, 696 "BIGDECIMAL": TokenType.BIGDECIMAL, 697 "BIGNUMERIC": TokenType.BIGDECIMAL, 698 "MAP": TokenType.MAP, 699 "NULLABLE": TokenType.NULLABLE, 700 "NUMBER": TokenType.DECIMAL, 701 "NUMERIC": TokenType.DECIMAL, 702 "FIXED": TokenType.DECIMAL, 703 "REAL": TokenType.FLOAT, 704 "FLOAT": TokenType.FLOAT, 705 "FLOAT4": TokenType.FLOAT, 706 "FLOAT8": TokenType.DOUBLE, 707 "DOUBLE": TokenType.DOUBLE, 708 "DOUBLE PRECISION": TokenType.DOUBLE, 709 "JSON": TokenType.JSON, 710 "CHAR": TokenType.CHAR, 711 "CHARACTER": TokenType.CHAR, 712 "NCHAR": TokenType.NCHAR, 713 "VARCHAR": TokenType.VARCHAR, 714 "VARCHAR2": TokenType.VARCHAR, 715 "NVARCHAR": TokenType.NVARCHAR, 716 "NVARCHAR2": TokenType.NVARCHAR, 717 "STR": TokenType.TEXT, 718 "STRING": TokenType.TEXT, 719 "TEXT": TokenType.TEXT, 720 "LONGTEXT": TokenType.LONGTEXT, 721 "MEDIUMTEXT": TokenType.MEDIUMTEXT, 722 "TINYTEXT": TokenType.TINYTEXT, 723 "CLOB": TokenType.TEXT, 724 "LONGVARCHAR": TokenType.TEXT, 725 "BINARY": TokenType.BINARY, 726 "BLOB": TokenType.VARBINARY, 727 "LONGBLOB": TokenType.LONGBLOB, 728 "MEDIUMBLOB": TokenType.MEDIUMBLOB, 729 "TINYBLOB": TokenType.TINYBLOB, 730 "BYTEA": TokenType.VARBINARY, 731 "VARBINARY": TokenType.VARBINARY, 732 "TIME": TokenType.TIME, 733 "TIMETZ": TokenType.TIMETZ, 734 "TIMESTAMP": TokenType.TIMESTAMP, 735 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 736 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 737 "DATE": TokenType.DATE, 738 "DATETIME": TokenType.DATETIME, 739 "INT4RANGE": TokenType.INT4RANGE, 740 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 741 "INT8RANGE": TokenType.INT8RANGE, 742 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 743 "NUMRANGE": TokenType.NUMRANGE, 744 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 745 "TSRANGE": TokenType.TSRANGE, 746 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 747 "TSTZRANGE": TokenType.TSTZRANGE, 748 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 749 "DATERANGE": TokenType.DATERANGE, 750 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 751 "UNIQUE": TokenType.UNIQUE, 752 "STRUCT": TokenType.STRUCT, 753 "VARIANT": TokenType.VARIANT, 754 "ALTER": TokenType.ALTER, 755 "ANALYZE": TokenType.COMMAND, 756 "CALL": TokenType.COMMAND, 757 "COMMENT": TokenType.COMMENT, 758 "COPY": TokenType.COMMAND, 759 "EXPLAIN": TokenType.COMMAND, 760 "GRANT": TokenType.COMMAND, 761 "OPTIMIZE": TokenType.COMMAND, 762 "PREPARE": TokenType.COMMAND, 763 "TRUNCATE": TokenType.COMMAND, 764 "VACUUM": TokenType.COMMAND, 765 "USER-DEFINED": TokenType.USERDEFINED, 766 "FOR VERSION": TokenType.VERSION_SNAPSHOT, 767 "FOR TIMESTAMP": TokenType.TIMESTAMP_SNAPSHOT, 768 } 769 770 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 771 " ": TokenType.SPACE, 772 "\t": TokenType.SPACE, 773 "\n": TokenType.BREAK, 774 "\r": TokenType.BREAK, 775 "\r\n": TokenType.BREAK, 776 } 777 778 COMMANDS = { 779 TokenType.COMMAND, 780 TokenType.EXECUTE, 781 TokenType.FETCH, 782 TokenType.SHOW, 783 } 784 785 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 786 787 # handle numeric literals like in hive (3L = BIGINT) 788 NUMERIC_LITERALS: t.Dict[str, str] = {} 789 ENCODE: t.Optional[str] = None 790 791 COMMENTS = ["--", ("/*", "*/")] 792 793 __slots__ = ( 794 "sql", 795 "size", 796 "tokens", 797 "_start", 798 "_current", 799 "_line", 800 "_col", 801 "_comments", 802 "_char", 803 "_end", 804 "_peek", 805 "_prev_token_line", 806 ) 807 808 def __init__(self) -> None: 809 self.reset() 810 811 def reset(self) -> None: 812 self.sql = "" 813 self.size = 0 814 self.tokens: t.List[Token] = [] 815 self._start = 0 816 self._current = 0 817 self._line = 1 818 self._col = 0 819 self._comments: t.List[str] = [] 820 821 self._char = "" 822 self._end = False 823 self._peek = "" 824 self._prev_token_line = -1 825 826 def tokenize(self, sql: str) -> t.List[Token]: 827 """Returns a list of tokens corresponding to the SQL string `sql`.""" 828 self.reset() 829 self.sql = sql 830 self.size = len(sql) 831 832 try: 833 self._scan() 834 except Exception as e: 835 start = max(self._current - 50, 0) 836 end = min(self._current + 50, self.size - 1) 837 context = self.sql[start:end] 838 raise TokenError(f"Error tokenizing '{context}'") from e 839 840 return self.tokens 841 842 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 843 while self.size and not self._end: 844 self._start = self._current 845 self._advance() 846 847 if self._char is None: 848 break 849 850 if self._char not in self.WHITE_SPACE: 851 if self._char.isdigit(): 852 self._scan_number() 853 elif self._char in self._IDENTIFIERS: 854 self._scan_identifier(self._IDENTIFIERS[self._char]) 855 else: 856 self._scan_keywords() 857 858 if until and until(): 859 break 860 861 if self.tokens and self._comments: 862 self.tokens[-1].comments.extend(self._comments) 863 864 def _chars(self, size: int) -> str: 865 if size == 1: 866 return self._char 867 868 start = self._current - 1 869 end = start + size 870 871 return self.sql[start:end] if end <= self.size else "" 872 873 def _advance(self, i: int = 1, alnum: bool = False) -> None: 874 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 875 self._col = 1 876 self._line += 1 877 else: 878 self._col += i 879 880 self._current += i 881 self._end = self._current >= self.size 882 self._char = self.sql[self._current - 1] 883 self._peek = "" if self._end else self.sql[self._current] 884 885 if alnum and self._char.isalnum(): 886 # Here we use local variables instead of attributes for better performance 887 _col = self._col 888 _current = self._current 889 _end = self._end 890 _peek = self._peek 891 892 while _peek.isalnum(): 893 _col += 1 894 _current += 1 895 _end = _current >= self.size 896 _peek = "" if _end else self.sql[_current] 897 898 self._col = _col 899 self._current = _current 900 self._end = _end 901 self._peek = _peek 902 self._char = self.sql[_current - 1] 903 904 @property 905 def _text(self) -> str: 906 return self.sql[self._start : self._current] 907 908 def peek(self, i: int = 0) -> str: 909 i = self._current + i 910 if i < self.size: 911 return self.sql[i] 912 return "" 913 914 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 915 self._prev_token_line = self._line 916 917 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 918 self.tokens[-1].comments.extend(self._comments) 919 self._comments = [] 920 921 self.tokens.append( 922 Token( 923 token_type, 924 text=self._text if text is None else text, 925 line=self._line, 926 col=self._col, 927 start=self._start, 928 end=self._current - 1, 929 comments=self._comments, 930 ) 931 ) 932 self._comments = [] 933 934 # If we have either a semicolon or a begin token before the command's token, we'll parse 935 # whatever follows the command's token as a string 936 if ( 937 token_type in self.COMMANDS 938 and self._peek != ";" 939 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 940 ): 941 start = self._current 942 tokens = len(self.tokens) 943 self._scan(lambda: self._peek == ";") 944 self.tokens = self.tokens[:tokens] 945 text = self.sql[start : self._current].strip() 946 if text: 947 self._add(TokenType.STRING, text) 948 949 def _scan_keywords(self) -> None: 950 size = 0 951 word = None 952 chars = self._text 953 char = chars 954 prev_space = False 955 skip = False 956 trie = self._KEYWORD_TRIE 957 single_token = char in self.SINGLE_TOKENS 958 959 while chars: 960 if skip: 961 result = TrieResult.PREFIX 962 else: 963 result, trie = in_trie(trie, char.upper()) 964 965 if result == TrieResult.FAILED: 966 break 967 if result == TrieResult.EXISTS: 968 word = chars 969 970 end = self._current + size 971 size += 1 972 973 if end < self.size: 974 char = self.sql[end] 975 single_token = single_token or char in self.SINGLE_TOKENS 976 is_space = char in self.WHITE_SPACE 977 978 if not is_space or not prev_space: 979 if is_space: 980 char = " " 981 chars += char 982 prev_space = is_space 983 skip = False 984 else: 985 skip = True 986 else: 987 char = "" 988 chars = " " 989 990 if word: 991 if self._scan_string(word): 992 return 993 if self._scan_comment(word): 994 return 995 if prev_space or single_token or not char: 996 self._advance(size - 1) 997 word = word.upper() 998 self._add(self.KEYWORDS[word], text=word) 999 return 1000 if self._char in self.SINGLE_TOKENS: 1001 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 1002 return 1003 self._scan_var() 1004 1005 def _scan_comment(self, comment_start: str) -> bool: 1006 if comment_start not in self._COMMENTS: 1007 return False 1008 1009 comment_start_line = self._line 1010 comment_start_size = len(comment_start) 1011 comment_end = self._COMMENTS[comment_start] 1012 1013 if comment_end: 1014 # Skip the comment's start delimiter 1015 self._advance(comment_start_size) 1016 1017 comment_end_size = len(comment_end) 1018 while not self._end and self._chars(comment_end_size) != comment_end: 1019 self._advance(alnum=True) 1020 1021 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1022 self._advance(comment_end_size - 1) 1023 else: 1024 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1025 self._advance(alnum=True) 1026 self._comments.append(self._text[comment_start_size:]) 1027 1028 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1029 # Multiple consecutive comments are preserved by appending them to the current comments list. 1030 if comment_start_line == self._prev_token_line: 1031 self.tokens[-1].comments.extend(self._comments) 1032 self._comments = [] 1033 self._prev_token_line = self._line 1034 1035 return True 1036 1037 def _scan_number(self) -> None: 1038 if self._char == "0": 1039 peek = self._peek.upper() 1040 if peek == "B": 1041 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1042 elif peek == "X": 1043 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1044 1045 decimal = False 1046 scientific = 0 1047 1048 while True: 1049 if self._peek.isdigit(): 1050 self._advance() 1051 elif self._peek == "." and not decimal: 1052 after = self.peek(1) 1053 if after.isdigit() or not after.isalpha(): 1054 decimal = True 1055 self._advance() 1056 else: 1057 return self._add(TokenType.VAR) 1058 elif self._peek in ("-", "+") and scientific == 1: 1059 scientific += 1 1060 self._advance() 1061 elif self._peek.upper() == "E" and not scientific: 1062 scientific += 1 1063 self._advance() 1064 elif self._peek.isidentifier(): 1065 number_text = self._text 1066 literal = "" 1067 1068 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1069 literal += self._peek.upper() 1070 self._advance() 1071 1072 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1073 1074 if token_type: 1075 self._add(TokenType.NUMBER, number_text) 1076 self._add(TokenType.DCOLON, "::") 1077 return self._add(token_type, literal) 1078 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1079 return self._add(TokenType.VAR) 1080 1081 self._advance(-len(literal)) 1082 return self._add(TokenType.NUMBER, number_text) 1083 else: 1084 return self._add(TokenType.NUMBER) 1085 1086 def _scan_bits(self) -> None: 1087 self._advance() 1088 value = self._extract_value() 1089 try: 1090 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1091 int(value, 2) 1092 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1093 except ValueError: 1094 self._add(TokenType.IDENTIFIER) 1095 1096 def _scan_hex(self) -> None: 1097 self._advance() 1098 value = self._extract_value() 1099 try: 1100 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1101 int(value, 16) 1102 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1103 except ValueError: 1104 self._add(TokenType.IDENTIFIER) 1105 1106 def _extract_value(self) -> str: 1107 while True: 1108 char = self._peek.strip() 1109 if char and char not in self.SINGLE_TOKENS: 1110 self._advance(alnum=True) 1111 else: 1112 break 1113 1114 return self._text 1115 1116 def _scan_string(self, start: str) -> bool: 1117 base = None 1118 token_type = TokenType.STRING 1119 1120 if start in self._QUOTES: 1121 end = self._QUOTES[start] 1122 elif start in self._FORMAT_STRINGS: 1123 end, token_type = self._FORMAT_STRINGS[start] 1124 1125 if token_type == TokenType.HEX_STRING: 1126 base = 16 1127 elif token_type == TokenType.BIT_STRING: 1128 base = 2 1129 else: 1130 return False 1131 1132 self._advance(len(start)) 1133 text = self._extract_string(end) 1134 1135 if base: 1136 try: 1137 int(text, base) 1138 except: 1139 raise TokenError( 1140 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1141 ) 1142 else: 1143 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1144 1145 self._add(token_type, text) 1146 return True 1147 1148 def _scan_identifier(self, identifier_end: str) -> None: 1149 self._advance() 1150 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1151 self._add(TokenType.IDENTIFIER, text) 1152 1153 def _scan_var(self) -> None: 1154 while True: 1155 char = self._peek.strip() 1156 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1157 self._advance(alnum=True) 1158 else: 1159 break 1160 1161 self._add( 1162 TokenType.VAR 1163 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1164 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1165 ) 1166 1167 def _extract_string(self, delimiter: str, escapes=None) -> str: 1168 text = "" 1169 delim_size = len(delimiter) 1170 escapes = self._STRING_ESCAPES if escapes is None else escapes 1171 1172 while True: 1173 if ( 1174 self._char in escapes 1175 and (self._peek == delimiter or self._peek in escapes) 1176 and (self._char not in self._QUOTES or self._char == self._peek) 1177 ): 1178 if self._peek == delimiter: 1179 text += self._peek 1180 else: 1181 text += self._char + self._peek 1182 1183 if self._current + 1 < self.size: 1184 self._advance(2) 1185 else: 1186 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1187 else: 1188 if self._chars(delim_size) == delimiter: 1189 if delim_size > 1: 1190 self._advance(delim_size - 1) 1191 break 1192 1193 if self._end: 1194 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1195 1196 current = self._current - 1 1197 self._advance(alnum=True) 1198 text += self.sql[current : self._current - 1] 1199 1200 return text
12class TokenType(AutoName): 13 L_PAREN = auto() 14 R_PAREN = auto() 15 L_BRACKET = auto() 16 R_BRACKET = auto() 17 L_BRACE = auto() 18 R_BRACE = auto() 19 COMMA = auto() 20 DOT = auto() 21 DASH = auto() 22 PLUS = auto() 23 COLON = auto() 24 DCOLON = auto() 25 DQMARK = auto() 26 SEMICOLON = auto() 27 STAR = auto() 28 BACKSLASH = auto() 29 SLASH = auto() 30 LT = auto() 31 LTE = auto() 32 GT = auto() 33 GTE = auto() 34 NOT = auto() 35 EQ = auto() 36 NEQ = auto() 37 NULLSAFE_EQ = auto() 38 AND = auto() 39 OR = auto() 40 AMP = auto() 41 DPIPE = auto() 42 PIPE = auto() 43 CARET = auto() 44 TILDA = auto() 45 ARROW = auto() 46 DARROW = auto() 47 FARROW = auto() 48 HASH = auto() 49 HASH_ARROW = auto() 50 DHASH_ARROW = auto() 51 LR_ARROW = auto() 52 DAT = auto() 53 LT_AT = auto() 54 AT_GT = auto() 55 DOLLAR = auto() 56 PARAMETER = auto() 57 SESSION_PARAMETER = auto() 58 DAMP = auto() 59 XOR = auto() 60 61 BLOCK_START = auto() 62 BLOCK_END = auto() 63 64 SPACE = auto() 65 BREAK = auto() 66 67 STRING = auto() 68 NUMBER = auto() 69 IDENTIFIER = auto() 70 DATABASE = auto() 71 COLUMN = auto() 72 COLUMN_DEF = auto() 73 SCHEMA = auto() 74 TABLE = auto() 75 VAR = auto() 76 BIT_STRING = auto() 77 HEX_STRING = auto() 78 BYTE_STRING = auto() 79 NATIONAL_STRING = auto() 80 RAW_STRING = auto() 81 82 # types 83 BIT = auto() 84 BOOLEAN = auto() 85 TINYINT = auto() 86 UTINYINT = auto() 87 SMALLINT = auto() 88 USMALLINT = auto() 89 MEDIUMINT = auto() 90 UMEDIUMINT = auto() 91 INT = auto() 92 UINT = auto() 93 BIGINT = auto() 94 UBIGINT = auto() 95 INT128 = auto() 96 UINT128 = auto() 97 INT256 = auto() 98 UINT256 = auto() 99 FLOAT = auto() 100 DOUBLE = auto() 101 DECIMAL = auto() 102 BIGDECIMAL = auto() 103 CHAR = auto() 104 NCHAR = auto() 105 VARCHAR = auto() 106 NVARCHAR = auto() 107 TEXT = auto() 108 MEDIUMTEXT = auto() 109 LONGTEXT = auto() 110 MEDIUMBLOB = auto() 111 LONGBLOB = auto() 112 TINYBLOB = auto() 113 TINYTEXT = auto() 114 BINARY = auto() 115 VARBINARY = auto() 116 JSON = auto() 117 JSONB = auto() 118 TIME = auto() 119 TIMETZ = auto() 120 TIMESTAMP = auto() 121 TIMESTAMPTZ = auto() 122 TIMESTAMPLTZ = auto() 123 DATETIME = auto() 124 DATETIME64 = auto() 125 DATE = auto() 126 INT4RANGE = auto() 127 INT4MULTIRANGE = auto() 128 INT8RANGE = auto() 129 INT8MULTIRANGE = auto() 130 NUMRANGE = auto() 131 NUMMULTIRANGE = auto() 132 TSRANGE = auto() 133 TSMULTIRANGE = auto() 134 TSTZRANGE = auto() 135 TSTZMULTIRANGE = auto() 136 DATERANGE = auto() 137 DATEMULTIRANGE = auto() 138 UUID = auto() 139 GEOGRAPHY = auto() 140 NULLABLE = auto() 141 GEOMETRY = auto() 142 HLLSKETCH = auto() 143 HSTORE = auto() 144 SUPER = auto() 145 SERIAL = auto() 146 SMALLSERIAL = auto() 147 BIGSERIAL = auto() 148 XML = auto() 149 YEAR = auto() 150 UNIQUEIDENTIFIER = auto() 151 USERDEFINED = auto() 152 MONEY = auto() 153 SMALLMONEY = auto() 154 ROWVERSION = auto() 155 IMAGE = auto() 156 VARIANT = auto() 157 OBJECT = auto() 158 INET = auto() 159 IPADDRESS = auto() 160 IPPREFIX = auto() 161 ENUM = auto() 162 ENUM8 = auto() 163 ENUM16 = auto() 164 FIXEDSTRING = auto() 165 LOWCARDINALITY = auto() 166 NESTED = auto() 167 UNKNOWN = auto() 168 169 # keywords 170 ALIAS = auto() 171 ALTER = auto() 172 ALWAYS = auto() 173 ALL = auto() 174 ANTI = auto() 175 ANY = auto() 176 APPLY = auto() 177 ARRAY = auto() 178 ASC = auto() 179 ASOF = auto() 180 AUTO_INCREMENT = auto() 181 BEGIN = auto() 182 BETWEEN = auto() 183 CACHE = auto() 184 CASE = auto() 185 CHARACTER_SET = auto() 186 CLUSTER_BY = auto() 187 COLLATE = auto() 188 COMMAND = auto() 189 COMMENT = auto() 190 COMMIT = auto() 191 CONNECT_BY = auto() 192 CONSTRAINT = auto() 193 CREATE = auto() 194 CROSS = auto() 195 CUBE = auto() 196 CURRENT_DATE = auto() 197 CURRENT_DATETIME = auto() 198 CURRENT_TIME = auto() 199 CURRENT_TIMESTAMP = auto() 200 CURRENT_USER = auto() 201 DEFAULT = auto() 202 DELETE = auto() 203 DESC = auto() 204 DESCRIBE = auto() 205 DICTIONARY = auto() 206 DISTINCT = auto() 207 DISTRIBUTE_BY = auto() 208 DIV = auto() 209 DROP = auto() 210 ELSE = auto() 211 END = auto() 212 ESCAPE = auto() 213 EXCEPT = auto() 214 EXECUTE = auto() 215 EXISTS = auto() 216 FALSE = auto() 217 FETCH = auto() 218 FILTER = auto() 219 FINAL = auto() 220 FIRST = auto() 221 FOR = auto() 222 FORCE = auto() 223 FOREIGN_KEY = auto() 224 FORMAT = auto() 225 FROM = auto() 226 FULL = auto() 227 FUNCTION = auto() 228 GLOB = auto() 229 GLOBAL = auto() 230 GROUP_BY = auto() 231 GROUPING_SETS = auto() 232 HAVING = auto() 233 HINT = auto() 234 IGNORE = auto() 235 ILIKE = auto() 236 ILIKE_ANY = auto() 237 IN = auto() 238 INDEX = auto() 239 INNER = auto() 240 INSERT = auto() 241 INTERSECT = auto() 242 INTERVAL = auto() 243 INTO = auto() 244 INTRODUCER = auto() 245 IRLIKE = auto() 246 IS = auto() 247 ISNULL = auto() 248 JOIN = auto() 249 JOIN_MARKER = auto() 250 KEEP = auto() 251 KILL = auto() 252 LANGUAGE = auto() 253 LATERAL = auto() 254 LEFT = auto() 255 LIKE = auto() 256 LIKE_ANY = auto() 257 LIMIT = auto() 258 LOAD = auto() 259 LOCK = auto() 260 MAP = auto() 261 MATCH_RECOGNIZE = auto() 262 MEMBER_OF = auto() 263 MERGE = auto() 264 MOD = auto() 265 NATURAL = auto() 266 NEXT = auto() 267 NOTNULL = auto() 268 NULL = auto() 269 OBJECT_IDENTIFIER = auto() 270 OFFSET = auto() 271 ON = auto() 272 ORDER_BY = auto() 273 ORDERED = auto() 274 ORDINALITY = auto() 275 OUTER = auto() 276 OVER = auto() 277 OVERLAPS = auto() 278 OVERWRITE = auto() 279 PARTITION = auto() 280 PARTITION_BY = auto() 281 PERCENT = auto() 282 PIVOT = auto() 283 PLACEHOLDER = auto() 284 PRAGMA = auto() 285 PRIMARY_KEY = auto() 286 PROCEDURE = auto() 287 PROPERTIES = auto() 288 PSEUDO_TYPE = auto() 289 QUALIFY = auto() 290 QUOTE = auto() 291 RANGE = auto() 292 RECURSIVE = auto() 293 REPLACE = auto() 294 RETURNING = auto() 295 REFERENCES = auto() 296 RIGHT = auto() 297 RLIKE = auto() 298 ROLLBACK = auto() 299 ROLLUP = auto() 300 ROW = auto() 301 ROWS = auto() 302 SELECT = auto() 303 SEMI = auto() 304 SEPARATOR = auto() 305 SERDE_PROPERTIES = auto() 306 SET = auto() 307 SETTINGS = auto() 308 SHOW = auto() 309 SIMILAR_TO = auto() 310 SOME = auto() 311 SORT_BY = auto() 312 START_WITH = auto() 313 STRUCT = auto() 314 TABLE_SAMPLE = auto() 315 TEMPORARY = auto() 316 TOP = auto() 317 THEN = auto() 318 TRUE = auto() 319 UNCACHE = auto() 320 UNION = auto() 321 UNNEST = auto() 322 UNPIVOT = auto() 323 UPDATE = auto() 324 USE = auto() 325 USING = auto() 326 VALUES = auto() 327 VIEW = auto() 328 VOLATILE = auto() 329 WHEN = auto() 330 WHERE = auto() 331 WINDOW = auto() 332 WITH = auto() 333 UNIQUE = auto() 334 VERSION_SNAPSHOT = auto() 335 TIMESTAMP_SNAPSHOT = auto()
An enumeration.
L_PAREN =
<TokenType.L_PAREN: 'L_PAREN'>
R_PAREN =
<TokenType.R_PAREN: 'R_PAREN'>
L_BRACKET =
<TokenType.L_BRACKET: 'L_BRACKET'>
R_BRACKET =
<TokenType.R_BRACKET: 'R_BRACKET'>
L_BRACE =
<TokenType.L_BRACE: 'L_BRACE'>
R_BRACE =
<TokenType.R_BRACE: 'R_BRACE'>
COMMA =
<TokenType.COMMA: 'COMMA'>
DOT =
<TokenType.DOT: 'DOT'>
DASH =
<TokenType.DASH: 'DASH'>
PLUS =
<TokenType.PLUS: 'PLUS'>
COLON =
<TokenType.COLON: 'COLON'>
DCOLON =
<TokenType.DCOLON: 'DCOLON'>
DQMARK =
<TokenType.DQMARK: 'DQMARK'>
SEMICOLON =
<TokenType.SEMICOLON: 'SEMICOLON'>
STAR =
<TokenType.STAR: 'STAR'>
BACKSLASH =
<TokenType.BACKSLASH: 'BACKSLASH'>
SLASH =
<TokenType.SLASH: 'SLASH'>
LT =
<TokenType.LT: 'LT'>
LTE =
<TokenType.LTE: 'LTE'>
GT =
<TokenType.GT: 'GT'>
GTE =
<TokenType.GTE: 'GTE'>
NOT =
<TokenType.NOT: 'NOT'>
EQ =
<TokenType.EQ: 'EQ'>
NEQ =
<TokenType.NEQ: 'NEQ'>
NULLSAFE_EQ =
<TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>
AND =
<TokenType.AND: 'AND'>
OR =
<TokenType.OR: 'OR'>
AMP =
<TokenType.AMP: 'AMP'>
DPIPE =
<TokenType.DPIPE: 'DPIPE'>
PIPE =
<TokenType.PIPE: 'PIPE'>
CARET =
<TokenType.CARET: 'CARET'>
TILDA =
<TokenType.TILDA: 'TILDA'>
ARROW =
<TokenType.ARROW: 'ARROW'>
DARROW =
<TokenType.DARROW: 'DARROW'>
FARROW =
<TokenType.FARROW: 'FARROW'>
HASH =
<TokenType.HASH: 'HASH'>
HASH_ARROW =
<TokenType.HASH_ARROW: 'HASH_ARROW'>
DHASH_ARROW =
<TokenType.DHASH_ARROW: 'DHASH_ARROW'>
LR_ARROW =
<TokenType.LR_ARROW: 'LR_ARROW'>
DAT =
<TokenType.DAT: 'DAT'>
LT_AT =
<TokenType.LT_AT: 'LT_AT'>
AT_GT =
<TokenType.AT_GT: 'AT_GT'>
DOLLAR =
<TokenType.DOLLAR: 'DOLLAR'>
PARAMETER =
<TokenType.PARAMETER: 'PARAMETER'>
SESSION_PARAMETER =
<TokenType.SESSION_PARAMETER: 'SESSION_PARAMETER'>
DAMP =
<TokenType.DAMP: 'DAMP'>
XOR =
<TokenType.XOR: 'XOR'>
BLOCK_START =
<TokenType.BLOCK_START: 'BLOCK_START'>
BLOCK_END =
<TokenType.BLOCK_END: 'BLOCK_END'>
SPACE =
<TokenType.SPACE: 'SPACE'>
BREAK =
<TokenType.BREAK: 'BREAK'>
STRING =
<TokenType.STRING: 'STRING'>
NUMBER =
<TokenType.NUMBER: 'NUMBER'>
IDENTIFIER =
<TokenType.IDENTIFIER: 'IDENTIFIER'>
DATABASE =
<TokenType.DATABASE: 'DATABASE'>
COLUMN =
<TokenType.COLUMN: 'COLUMN'>
COLUMN_DEF =
<TokenType.COLUMN_DEF: 'COLUMN_DEF'>
SCHEMA =
<TokenType.SCHEMA: 'SCHEMA'>
TABLE =
<TokenType.TABLE: 'TABLE'>
VAR =
<TokenType.VAR: 'VAR'>
BIT_STRING =
<TokenType.BIT_STRING: 'BIT_STRING'>
HEX_STRING =
<TokenType.HEX_STRING: 'HEX_STRING'>
BYTE_STRING =
<TokenType.BYTE_STRING: 'BYTE_STRING'>
NATIONAL_STRING =
<TokenType.NATIONAL_STRING: 'NATIONAL_STRING'>
RAW_STRING =
<TokenType.RAW_STRING: 'RAW_STRING'>
BIT =
<TokenType.BIT: 'BIT'>
BOOLEAN =
<TokenType.BOOLEAN: 'BOOLEAN'>
TINYINT =
<TokenType.TINYINT: 'TINYINT'>
UTINYINT =
<TokenType.UTINYINT: 'UTINYINT'>
SMALLINT =
<TokenType.SMALLINT: 'SMALLINT'>
USMALLINT =
<TokenType.USMALLINT: 'USMALLINT'>
MEDIUMINT =
<TokenType.MEDIUMINT: 'MEDIUMINT'>
UMEDIUMINT =
<TokenType.UMEDIUMINT: 'UMEDIUMINT'>
INT =
<TokenType.INT: 'INT'>
UINT =
<TokenType.UINT: 'UINT'>
BIGINT =
<TokenType.BIGINT: 'BIGINT'>
UBIGINT =
<TokenType.UBIGINT: 'UBIGINT'>
INT128 =
<TokenType.INT128: 'INT128'>
UINT128 =
<TokenType.UINT128: 'UINT128'>
INT256 =
<TokenType.INT256: 'INT256'>
UINT256 =
<TokenType.UINT256: 'UINT256'>
FLOAT =
<TokenType.FLOAT: 'FLOAT'>
DOUBLE =
<TokenType.DOUBLE: 'DOUBLE'>
DECIMAL =
<TokenType.DECIMAL: 'DECIMAL'>
BIGDECIMAL =
<TokenType.BIGDECIMAL: 'BIGDECIMAL'>
CHAR =
<TokenType.CHAR: 'CHAR'>
NCHAR =
<TokenType.NCHAR: 'NCHAR'>
VARCHAR =
<TokenType.VARCHAR: 'VARCHAR'>
NVARCHAR =
<TokenType.NVARCHAR: 'NVARCHAR'>
TEXT =
<TokenType.TEXT: 'TEXT'>
MEDIUMTEXT =
<TokenType.MEDIUMTEXT: 'MEDIUMTEXT'>
LONGTEXT =
<TokenType.LONGTEXT: 'LONGTEXT'>
MEDIUMBLOB =
<TokenType.MEDIUMBLOB: 'MEDIUMBLOB'>
LONGBLOB =
<TokenType.LONGBLOB: 'LONGBLOB'>
TINYBLOB =
<TokenType.TINYBLOB: 'TINYBLOB'>
TINYTEXT =
<TokenType.TINYTEXT: 'TINYTEXT'>
BINARY =
<TokenType.BINARY: 'BINARY'>
VARBINARY =
<TokenType.VARBINARY: 'VARBINARY'>
JSON =
<TokenType.JSON: 'JSON'>
JSONB =
<TokenType.JSONB: 'JSONB'>
TIME =
<TokenType.TIME: 'TIME'>
TIMETZ =
<TokenType.TIMETZ: 'TIMETZ'>
TIMESTAMP =
<TokenType.TIMESTAMP: 'TIMESTAMP'>
TIMESTAMPTZ =
<TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>
TIMESTAMPLTZ =
<TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>
DATETIME =
<TokenType.DATETIME: 'DATETIME'>
DATETIME64 =
<TokenType.DATETIME64: 'DATETIME64'>
DATE =
<TokenType.DATE: 'DATE'>
INT4RANGE =
<TokenType.INT4RANGE: 'INT4RANGE'>
INT4MULTIRANGE =
<TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>
INT8RANGE =
<TokenType.INT8RANGE: 'INT8RANGE'>
INT8MULTIRANGE =
<TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>
NUMRANGE =
<TokenType.NUMRANGE: 'NUMRANGE'>
NUMMULTIRANGE =
<TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>
TSRANGE =
<TokenType.TSRANGE: 'TSRANGE'>
TSMULTIRANGE =
<TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>
TSTZRANGE =
<TokenType.TSTZRANGE: 'TSTZRANGE'>
TSTZMULTIRANGE =
<TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>
DATERANGE =
<TokenType.DATERANGE: 'DATERANGE'>
DATEMULTIRANGE =
<TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>
UUID =
<TokenType.UUID: 'UUID'>
GEOGRAPHY =
<TokenType.GEOGRAPHY: 'GEOGRAPHY'>
NULLABLE =
<TokenType.NULLABLE: 'NULLABLE'>
GEOMETRY =
<TokenType.GEOMETRY: 'GEOMETRY'>
HLLSKETCH =
<TokenType.HLLSKETCH: 'HLLSKETCH'>
HSTORE =
<TokenType.HSTORE: 'HSTORE'>
SUPER =
<TokenType.SUPER: 'SUPER'>
SERIAL =
<TokenType.SERIAL: 'SERIAL'>
SMALLSERIAL =
<TokenType.SMALLSERIAL: 'SMALLSERIAL'>
BIGSERIAL =
<TokenType.BIGSERIAL: 'BIGSERIAL'>
XML =
<TokenType.XML: 'XML'>
YEAR =
<TokenType.YEAR: 'YEAR'>
UNIQUEIDENTIFIER =
<TokenType.UNIQUEIDENTIFIER: 'UNIQUEIDENTIFIER'>
USERDEFINED =
<TokenType.USERDEFINED: 'USERDEFINED'>
MONEY =
<TokenType.MONEY: 'MONEY'>
SMALLMONEY =
<TokenType.SMALLMONEY: 'SMALLMONEY'>
ROWVERSION =
<TokenType.ROWVERSION: 'ROWVERSION'>
IMAGE =
<TokenType.IMAGE: 'IMAGE'>
VARIANT =
<TokenType.VARIANT: 'VARIANT'>
OBJECT =
<TokenType.OBJECT: 'OBJECT'>
INET =
<TokenType.INET: 'INET'>
IPADDRESS =
<TokenType.IPADDRESS: 'IPADDRESS'>
IPPREFIX =
<TokenType.IPPREFIX: 'IPPREFIX'>
ENUM =
<TokenType.ENUM: 'ENUM'>
ENUM8 =
<TokenType.ENUM8: 'ENUM8'>
ENUM16 =
<TokenType.ENUM16: 'ENUM16'>
FIXEDSTRING =
<TokenType.FIXEDSTRING: 'FIXEDSTRING'>
LOWCARDINALITY =
<TokenType.LOWCARDINALITY: 'LOWCARDINALITY'>
NESTED =
<TokenType.NESTED: 'NESTED'>
UNKNOWN =
<TokenType.UNKNOWN: 'UNKNOWN'>
ALIAS =
<TokenType.ALIAS: 'ALIAS'>
ALTER =
<TokenType.ALTER: 'ALTER'>
ALWAYS =
<TokenType.ALWAYS: 'ALWAYS'>
ALL =
<TokenType.ALL: 'ALL'>
ANTI =
<TokenType.ANTI: 'ANTI'>
ANY =
<TokenType.ANY: 'ANY'>
APPLY =
<TokenType.APPLY: 'APPLY'>
ARRAY =
<TokenType.ARRAY: 'ARRAY'>
ASC =
<TokenType.ASC: 'ASC'>
ASOF =
<TokenType.ASOF: 'ASOF'>
AUTO_INCREMENT =
<TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>
BEGIN =
<TokenType.BEGIN: 'BEGIN'>
BETWEEN =
<TokenType.BETWEEN: 'BETWEEN'>
CACHE =
<TokenType.CACHE: 'CACHE'>
CASE =
<TokenType.CASE: 'CASE'>
CHARACTER_SET =
<TokenType.CHARACTER_SET: 'CHARACTER_SET'>
CLUSTER_BY =
<TokenType.CLUSTER_BY: 'CLUSTER_BY'>
COLLATE =
<TokenType.COLLATE: 'COLLATE'>
COMMAND =
<TokenType.COMMAND: 'COMMAND'>
COMMENT =
<TokenType.COMMENT: 'COMMENT'>
COMMIT =
<TokenType.COMMIT: 'COMMIT'>
CONNECT_BY =
<TokenType.CONNECT_BY: 'CONNECT_BY'>
CONSTRAINT =
<TokenType.CONSTRAINT: 'CONSTRAINT'>
CREATE =
<TokenType.CREATE: 'CREATE'>
CROSS =
<TokenType.CROSS: 'CROSS'>
CUBE =
<TokenType.CUBE: 'CUBE'>
CURRENT_DATE =
<TokenType.CURRENT_DATE: 'CURRENT_DATE'>
CURRENT_DATETIME =
<TokenType.CURRENT_DATETIME: 'CURRENT_DATETIME'>
CURRENT_TIME =
<TokenType.CURRENT_TIME: 'CURRENT_TIME'>
CURRENT_TIMESTAMP =
<TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>
CURRENT_USER =
<TokenType.CURRENT_USER: 'CURRENT_USER'>
DEFAULT =
<TokenType.DEFAULT: 'DEFAULT'>
DELETE =
<TokenType.DELETE: 'DELETE'>
DESC =
<TokenType.DESC: 'DESC'>
DESCRIBE =
<TokenType.DESCRIBE: 'DESCRIBE'>
DICTIONARY =
<TokenType.DICTIONARY: 'DICTIONARY'>
DISTINCT =
<TokenType.DISTINCT: 'DISTINCT'>
DISTRIBUTE_BY =
<TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>
DIV =
<TokenType.DIV: 'DIV'>
DROP =
<TokenType.DROP: 'DROP'>
ELSE =
<TokenType.ELSE: 'ELSE'>
END =
<TokenType.END: 'END'>
ESCAPE =
<TokenType.ESCAPE: 'ESCAPE'>
EXCEPT =
<TokenType.EXCEPT: 'EXCEPT'>
EXECUTE =
<TokenType.EXECUTE: 'EXECUTE'>
EXISTS =
<TokenType.EXISTS: 'EXISTS'>
FALSE =
<TokenType.FALSE: 'FALSE'>
FETCH =
<TokenType.FETCH: 'FETCH'>
FILTER =
<TokenType.FILTER: 'FILTER'>
FINAL =
<TokenType.FINAL: 'FINAL'>
FIRST =
<TokenType.FIRST: 'FIRST'>
FOR =
<TokenType.FOR: 'FOR'>
FORCE =
<TokenType.FORCE: 'FORCE'>
FOREIGN_KEY =
<TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>
FORMAT =
<TokenType.FORMAT: 'FORMAT'>
FROM =
<TokenType.FROM: 'FROM'>
FULL =
<TokenType.FULL: 'FULL'>
FUNCTION =
<TokenType.FUNCTION: 'FUNCTION'>
GLOB =
<TokenType.GLOB: 'GLOB'>
GLOBAL =
<TokenType.GLOBAL: 'GLOBAL'>
GROUP_BY =
<TokenType.GROUP_BY: 'GROUP_BY'>
GROUPING_SETS =
<TokenType.GROUPING_SETS: 'GROUPING_SETS'>
HAVING =
<TokenType.HAVING: 'HAVING'>
HINT =
<TokenType.HINT: 'HINT'>
IGNORE =
<TokenType.IGNORE: 'IGNORE'>
ILIKE =
<TokenType.ILIKE: 'ILIKE'>
ILIKE_ANY =
<TokenType.ILIKE_ANY: 'ILIKE_ANY'>
IN =
<TokenType.IN: 'IN'>
INDEX =
<TokenType.INDEX: 'INDEX'>
INNER =
<TokenType.INNER: 'INNER'>
INSERT =
<TokenType.INSERT: 'INSERT'>
INTERSECT =
<TokenType.INTERSECT: 'INTERSECT'>
INTERVAL =
<TokenType.INTERVAL: 'INTERVAL'>
INTO =
<TokenType.INTO: 'INTO'>
INTRODUCER =
<TokenType.INTRODUCER: 'INTRODUCER'>
IRLIKE =
<TokenType.IRLIKE: 'IRLIKE'>
IS =
<TokenType.IS: 'IS'>
ISNULL =
<TokenType.ISNULL: 'ISNULL'>
JOIN =
<TokenType.JOIN: 'JOIN'>
JOIN_MARKER =
<TokenType.JOIN_MARKER: 'JOIN_MARKER'>
KEEP =
<TokenType.KEEP: 'KEEP'>
KILL =
<TokenType.KILL: 'KILL'>
LANGUAGE =
<TokenType.LANGUAGE: 'LANGUAGE'>
LATERAL =
<TokenType.LATERAL: 'LATERAL'>
LEFT =
<TokenType.LEFT: 'LEFT'>
LIKE =
<TokenType.LIKE: 'LIKE'>
LIKE_ANY =
<TokenType.LIKE_ANY: 'LIKE_ANY'>
LIMIT =
<TokenType.LIMIT: 'LIMIT'>
LOAD =
<TokenType.LOAD: 'LOAD'>
LOCK =
<TokenType.LOCK: 'LOCK'>
MAP =
<TokenType.MAP: 'MAP'>
MATCH_RECOGNIZE =
<TokenType.MATCH_RECOGNIZE: 'MATCH_RECOGNIZE'>
MEMBER_OF =
<TokenType.MEMBER_OF: 'MEMBER_OF'>
MERGE =
<TokenType.MERGE: 'MERGE'>
MOD =
<TokenType.MOD: 'MOD'>
NATURAL =
<TokenType.NATURAL: 'NATURAL'>
NEXT =
<TokenType.NEXT: 'NEXT'>
NOTNULL =
<TokenType.NOTNULL: 'NOTNULL'>
NULL =
<TokenType.NULL: 'NULL'>
OBJECT_IDENTIFIER =
<TokenType.OBJECT_IDENTIFIER: 'OBJECT_IDENTIFIER'>
OFFSET =
<TokenType.OFFSET: 'OFFSET'>
ON =
<TokenType.ON: 'ON'>
ORDER_BY =
<TokenType.ORDER_BY: 'ORDER_BY'>
ORDERED =
<TokenType.ORDERED: 'ORDERED'>
ORDINALITY =
<TokenType.ORDINALITY: 'ORDINALITY'>
OUTER =
<TokenType.OUTER: 'OUTER'>
OVER =
<TokenType.OVER: 'OVER'>
OVERLAPS =
<TokenType.OVERLAPS: 'OVERLAPS'>
OVERWRITE =
<TokenType.OVERWRITE: 'OVERWRITE'>
PARTITION =
<TokenType.PARTITION: 'PARTITION'>
PARTITION_BY =
<TokenType.PARTITION_BY: 'PARTITION_BY'>
PERCENT =
<TokenType.PERCENT: 'PERCENT'>
PIVOT =
<TokenType.PIVOT: 'PIVOT'>
PLACEHOLDER =
<TokenType.PLACEHOLDER: 'PLACEHOLDER'>
PRAGMA =
<TokenType.PRAGMA: 'PRAGMA'>
PRIMARY_KEY =
<TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>
PROCEDURE =
<TokenType.PROCEDURE: 'PROCEDURE'>
PROPERTIES =
<TokenType.PROPERTIES: 'PROPERTIES'>
PSEUDO_TYPE =
<TokenType.PSEUDO_TYPE: 'PSEUDO_TYPE'>
QUALIFY =
<TokenType.QUALIFY: 'QUALIFY'>
QUOTE =
<TokenType.QUOTE: 'QUOTE'>
RANGE =
<TokenType.RANGE: 'RANGE'>
RECURSIVE =
<TokenType.RECURSIVE: 'RECURSIVE'>
REPLACE =
<TokenType.REPLACE: 'REPLACE'>
RETURNING =
<TokenType.RETURNING: 'RETURNING'>
REFERENCES =
<TokenType.REFERENCES: 'REFERENCES'>
RIGHT =
<TokenType.RIGHT: 'RIGHT'>
RLIKE =
<TokenType.RLIKE: 'RLIKE'>
ROLLBACK =
<TokenType.ROLLBACK: 'ROLLBACK'>
ROLLUP =
<TokenType.ROLLUP: 'ROLLUP'>
ROW =
<TokenType.ROW: 'ROW'>
ROWS =
<TokenType.ROWS: 'ROWS'>
SELECT =
<TokenType.SELECT: 'SELECT'>
SEMI =
<TokenType.SEMI: 'SEMI'>
SEPARATOR =
<TokenType.SEPARATOR: 'SEPARATOR'>
SERDE_PROPERTIES =
<TokenType.SERDE_PROPERTIES: 'SERDE_PROPERTIES'>
SET =
<TokenType.SET: 'SET'>
SETTINGS =
<TokenType.SETTINGS: 'SETTINGS'>
SHOW =
<TokenType.SHOW: 'SHOW'>
SIMILAR_TO =
<TokenType.SIMILAR_TO: 'SIMILAR_TO'>
SOME =
<TokenType.SOME: 'SOME'>
SORT_BY =
<TokenType.SORT_BY: 'SORT_BY'>
START_WITH =
<TokenType.START_WITH: 'START_WITH'>
STRUCT =
<TokenType.STRUCT: 'STRUCT'>
TABLE_SAMPLE =
<TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>
TEMPORARY =
<TokenType.TEMPORARY: 'TEMPORARY'>
TOP =
<TokenType.TOP: 'TOP'>
THEN =
<TokenType.THEN: 'THEN'>
TRUE =
<TokenType.TRUE: 'TRUE'>
UNCACHE =
<TokenType.UNCACHE: 'UNCACHE'>
UNION =
<TokenType.UNION: 'UNION'>
UNNEST =
<TokenType.UNNEST: 'UNNEST'>
UNPIVOT =
<TokenType.UNPIVOT: 'UNPIVOT'>
UPDATE =
<TokenType.UPDATE: 'UPDATE'>
USE =
<TokenType.USE: 'USE'>
USING =
<TokenType.USING: 'USING'>
VALUES =
<TokenType.VALUES: 'VALUES'>
VIEW =
<TokenType.VIEW: 'VIEW'>
VOLATILE =
<TokenType.VOLATILE: 'VOLATILE'>
WHEN =
<TokenType.WHEN: 'WHEN'>
WHERE =
<TokenType.WHERE: 'WHERE'>
WINDOW =
<TokenType.WINDOW: 'WINDOW'>
WITH =
<TokenType.WITH: 'WITH'>
UNIQUE =
<TokenType.UNIQUE: 'UNIQUE'>
VERSION_SNAPSHOT =
<TokenType.VERSION_SNAPSHOT: 'VERSION_SNAPSHOT'>
TIMESTAMP_SNAPSHOT =
<TokenType.TIMESTAMP_SNAPSHOT: 'TIMESTAMP_SNAPSHOT'>
Inherited Members
- enum.Enum
- name
- value
class
Token:
338class Token: 339 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 340 341 @classmethod 342 def number(cls, number: int) -> Token: 343 """Returns a NUMBER token with `number` as its text.""" 344 return cls(TokenType.NUMBER, str(number)) 345 346 @classmethod 347 def string(cls, string: str) -> Token: 348 """Returns a STRING token with `string` as its text.""" 349 return cls(TokenType.STRING, string) 350 351 @classmethod 352 def identifier(cls, identifier: str) -> Token: 353 """Returns an IDENTIFIER token with `identifier` as its text.""" 354 return cls(TokenType.IDENTIFIER, identifier) 355 356 @classmethod 357 def var(cls, var: str) -> Token: 358 """Returns an VAR token with `var` as its text.""" 359 return cls(TokenType.VAR, var) 360 361 def __init__( 362 self, 363 token_type: TokenType, 364 text: str, 365 line: int = 1, 366 col: int = 1, 367 start: int = 0, 368 end: int = 0, 369 comments: t.List[str] = [], 370 ) -> None: 371 """Token initializer. 372 373 Args: 374 token_type: The TokenType Enum. 375 text: The text of the token. 376 line: The line that the token ends on. 377 col: The column that the token ends on. 378 start: The start index of the token. 379 end: The ending index of the token. 380 comments: The comments to attach to the token. 381 """ 382 self.token_type = token_type 383 self.text = text 384 self.line = line 385 self.col = col 386 self.start = start 387 self.end = end 388 self.comments = comments 389 390 def __repr__(self) -> str: 391 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 392 return f"<Token {attributes}>"
Token( token_type: TokenType, text: str, line: int = 1, col: int = 1, start: int = 0, end: int = 0, comments: List[str] = [])
361 def __init__( 362 self, 363 token_type: TokenType, 364 text: str, 365 line: int = 1, 366 col: int = 1, 367 start: int = 0, 368 end: int = 0, 369 comments: t.List[str] = [], 370 ) -> None: 371 """Token initializer. 372 373 Args: 374 token_type: The TokenType Enum. 375 text: The text of the token. 376 line: The line that the token ends on. 377 col: The column that the token ends on. 378 start: The start index of the token. 379 end: The ending index of the token. 380 comments: The comments to attach to the token. 381 """ 382 self.token_type = token_type 383 self.text = text 384 self.line = line 385 self.col = col 386 self.start = start 387 self.end = end 388 self.comments = comments
Token initializer.
Arguments:
- token_type: The TokenType Enum.
- text: The text of the token.
- line: The line that the token ends on.
- col: The column that the token ends on.
- start: The start index of the token.
- end: The ending index of the token.
- comments: The comments to attach to the token.
341 @classmethod 342 def number(cls, number: int) -> Token: 343 """Returns a NUMBER token with `number` as its text.""" 344 return cls(TokenType.NUMBER, str(number))
Returns a NUMBER token with number
as its text.
346 @classmethod 347 def string(cls, string: str) -> Token: 348 """Returns a STRING token with `string` as its text.""" 349 return cls(TokenType.STRING, string)
Returns a STRING token with string
as its text.
351 @classmethod 352 def identifier(cls, identifier: str) -> Token: 353 """Returns an IDENTIFIER token with `identifier` as its text.""" 354 return cls(TokenType.IDENTIFIER, identifier)
Returns an IDENTIFIER token with identifier
as its text.
class
Tokenizer:
448class Tokenizer(metaclass=_Tokenizer): 449 SINGLE_TOKENS = { 450 "(": TokenType.L_PAREN, 451 ")": TokenType.R_PAREN, 452 "[": TokenType.L_BRACKET, 453 "]": TokenType.R_BRACKET, 454 "{": TokenType.L_BRACE, 455 "}": TokenType.R_BRACE, 456 "&": TokenType.AMP, 457 "^": TokenType.CARET, 458 ":": TokenType.COLON, 459 ",": TokenType.COMMA, 460 ".": TokenType.DOT, 461 "-": TokenType.DASH, 462 "=": TokenType.EQ, 463 ">": TokenType.GT, 464 "<": TokenType.LT, 465 "%": TokenType.MOD, 466 "!": TokenType.NOT, 467 "|": TokenType.PIPE, 468 "+": TokenType.PLUS, 469 ";": TokenType.SEMICOLON, 470 "/": TokenType.SLASH, 471 "\\": TokenType.BACKSLASH, 472 "*": TokenType.STAR, 473 "~": TokenType.TILDA, 474 "?": TokenType.PLACEHOLDER, 475 "@": TokenType.PARAMETER, 476 # used for breaking a var like x'y' but nothing else 477 # the token type doesn't matter 478 "'": TokenType.QUOTE, 479 "`": TokenType.IDENTIFIER, 480 '"': TokenType.IDENTIFIER, 481 "#": TokenType.HASH, 482 } 483 484 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 485 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 486 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 487 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 488 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 489 IDENTIFIER_ESCAPES = ['"'] 490 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 491 STRING_ESCAPES = ["'"] 492 VAR_SINGLE_TOKENS: t.Set[str] = set() 493 494 # Autofilled 495 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 496 497 _COMMENTS: t.Dict[str, str] = {} 498 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 499 _IDENTIFIERS: t.Dict[str, str] = {} 500 _IDENTIFIER_ESCAPES: t.Set[str] = set() 501 _QUOTES: t.Dict[str, str] = {} 502 _STRING_ESCAPES: t.Set[str] = set() 503 _KEYWORD_TRIE: t.Dict = {} 504 505 KEYWORDS: t.Dict[str, TokenType] = { 506 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 507 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 508 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 509 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 510 "/*+": TokenType.HINT, 511 "==": TokenType.EQ, 512 "::": TokenType.DCOLON, 513 "||": TokenType.DPIPE, 514 ">=": TokenType.GTE, 515 "<=": TokenType.LTE, 516 "<>": TokenType.NEQ, 517 "!=": TokenType.NEQ, 518 "<=>": TokenType.NULLSAFE_EQ, 519 "->": TokenType.ARROW, 520 "->>": TokenType.DARROW, 521 "=>": TokenType.FARROW, 522 "#>": TokenType.HASH_ARROW, 523 "#>>": TokenType.DHASH_ARROW, 524 "<->": TokenType.LR_ARROW, 525 "&&": TokenType.DAMP, 526 "??": TokenType.DQMARK, 527 "ALL": TokenType.ALL, 528 "ALWAYS": TokenType.ALWAYS, 529 "AND": TokenType.AND, 530 "ANTI": TokenType.ANTI, 531 "ANY": TokenType.ANY, 532 "ASC": TokenType.ASC, 533 "AS": TokenType.ALIAS, 534 "ASOF": TokenType.ASOF, 535 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 536 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 537 "BEGIN": TokenType.BEGIN, 538 "BETWEEN": TokenType.BETWEEN, 539 "CACHE": TokenType.CACHE, 540 "UNCACHE": TokenType.UNCACHE, 541 "CASE": TokenType.CASE, 542 "CHARACTER SET": TokenType.CHARACTER_SET, 543 "CLUSTER BY": TokenType.CLUSTER_BY, 544 "COLLATE": TokenType.COLLATE, 545 "COLUMN": TokenType.COLUMN, 546 "COMMIT": TokenType.COMMIT, 547 "CONNECT BY": TokenType.CONNECT_BY, 548 "CONSTRAINT": TokenType.CONSTRAINT, 549 "CREATE": TokenType.CREATE, 550 "CROSS": TokenType.CROSS, 551 "CUBE": TokenType.CUBE, 552 "CURRENT_DATE": TokenType.CURRENT_DATE, 553 "CURRENT_TIME": TokenType.CURRENT_TIME, 554 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 555 "CURRENT_USER": TokenType.CURRENT_USER, 556 "DATABASE": TokenType.DATABASE, 557 "DEFAULT": TokenType.DEFAULT, 558 "DELETE": TokenType.DELETE, 559 "DESC": TokenType.DESC, 560 "DESCRIBE": TokenType.DESCRIBE, 561 "DISTINCT": TokenType.DISTINCT, 562 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 563 "DIV": TokenType.DIV, 564 "DROP": TokenType.DROP, 565 "ELSE": TokenType.ELSE, 566 "END": TokenType.END, 567 "ESCAPE": TokenType.ESCAPE, 568 "EXCEPT": TokenType.EXCEPT, 569 "EXECUTE": TokenType.EXECUTE, 570 "EXISTS": TokenType.EXISTS, 571 "FALSE": TokenType.FALSE, 572 "FETCH": TokenType.FETCH, 573 "FILTER": TokenType.FILTER, 574 "FIRST": TokenType.FIRST, 575 "FULL": TokenType.FULL, 576 "FUNCTION": TokenType.FUNCTION, 577 "FOR": TokenType.FOR, 578 "FOREIGN KEY": TokenType.FOREIGN_KEY, 579 "FORMAT": TokenType.FORMAT, 580 "FROM": TokenType.FROM, 581 "GEOGRAPHY": TokenType.GEOGRAPHY, 582 "GEOMETRY": TokenType.GEOMETRY, 583 "GLOB": TokenType.GLOB, 584 "GROUP BY": TokenType.GROUP_BY, 585 "GROUPING SETS": TokenType.GROUPING_SETS, 586 "HAVING": TokenType.HAVING, 587 "ILIKE": TokenType.ILIKE, 588 "IN": TokenType.IN, 589 "INDEX": TokenType.INDEX, 590 "INET": TokenType.INET, 591 "INNER": TokenType.INNER, 592 "INSERT": TokenType.INSERT, 593 "INTERVAL": TokenType.INTERVAL, 594 "INTERSECT": TokenType.INTERSECT, 595 "INTO": TokenType.INTO, 596 "IS": TokenType.IS, 597 "ISNULL": TokenType.ISNULL, 598 "JOIN": TokenType.JOIN, 599 "KEEP": TokenType.KEEP, 600 "KILL": TokenType.KILL, 601 "LATERAL": TokenType.LATERAL, 602 "LEFT": TokenType.LEFT, 603 "LIKE": TokenType.LIKE, 604 "LIMIT": TokenType.LIMIT, 605 "LOAD": TokenType.LOAD, 606 "LOCK": TokenType.LOCK, 607 "MERGE": TokenType.MERGE, 608 "NATURAL": TokenType.NATURAL, 609 "NEXT": TokenType.NEXT, 610 "NOT": TokenType.NOT, 611 "NOTNULL": TokenType.NOTNULL, 612 "NULL": TokenType.NULL, 613 "OBJECT": TokenType.OBJECT, 614 "OFFSET": TokenType.OFFSET, 615 "ON": TokenType.ON, 616 "OR": TokenType.OR, 617 "XOR": TokenType.XOR, 618 "ORDER BY": TokenType.ORDER_BY, 619 "ORDINALITY": TokenType.ORDINALITY, 620 "OUTER": TokenType.OUTER, 621 "OVER": TokenType.OVER, 622 "OVERLAPS": TokenType.OVERLAPS, 623 "OVERWRITE": TokenType.OVERWRITE, 624 "PARTITION": TokenType.PARTITION, 625 "PARTITION BY": TokenType.PARTITION_BY, 626 "PARTITIONED BY": TokenType.PARTITION_BY, 627 "PARTITIONED_BY": TokenType.PARTITION_BY, 628 "PERCENT": TokenType.PERCENT, 629 "PIVOT": TokenType.PIVOT, 630 "PRAGMA": TokenType.PRAGMA, 631 "PRIMARY KEY": TokenType.PRIMARY_KEY, 632 "PROCEDURE": TokenType.PROCEDURE, 633 "QUALIFY": TokenType.QUALIFY, 634 "RANGE": TokenType.RANGE, 635 "RECURSIVE": TokenType.RECURSIVE, 636 "REGEXP": TokenType.RLIKE, 637 "REPLACE": TokenType.REPLACE, 638 "RETURNING": TokenType.RETURNING, 639 "REFERENCES": TokenType.REFERENCES, 640 "RIGHT": TokenType.RIGHT, 641 "RLIKE": TokenType.RLIKE, 642 "ROLLBACK": TokenType.ROLLBACK, 643 "ROLLUP": TokenType.ROLLUP, 644 "ROW": TokenType.ROW, 645 "ROWS": TokenType.ROWS, 646 "SCHEMA": TokenType.SCHEMA, 647 "SELECT": TokenType.SELECT, 648 "SEMI": TokenType.SEMI, 649 "SET": TokenType.SET, 650 "SETTINGS": TokenType.SETTINGS, 651 "SHOW": TokenType.SHOW, 652 "SIMILAR TO": TokenType.SIMILAR_TO, 653 "SOME": TokenType.SOME, 654 "SORT BY": TokenType.SORT_BY, 655 "START WITH": TokenType.START_WITH, 656 "TABLE": TokenType.TABLE, 657 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 658 "TEMP": TokenType.TEMPORARY, 659 "TEMPORARY": TokenType.TEMPORARY, 660 "THEN": TokenType.THEN, 661 "TRUE": TokenType.TRUE, 662 "UNION": TokenType.UNION, 663 "UNKNOWN": TokenType.UNKNOWN, 664 "UNNEST": TokenType.UNNEST, 665 "UNPIVOT": TokenType.UNPIVOT, 666 "UPDATE": TokenType.UPDATE, 667 "USE": TokenType.USE, 668 "USING": TokenType.USING, 669 "UUID": TokenType.UUID, 670 "VALUES": TokenType.VALUES, 671 "VIEW": TokenType.VIEW, 672 "VOLATILE": TokenType.VOLATILE, 673 "WHEN": TokenType.WHEN, 674 "WHERE": TokenType.WHERE, 675 "WINDOW": TokenType.WINDOW, 676 "WITH": TokenType.WITH, 677 "APPLY": TokenType.APPLY, 678 "ARRAY": TokenType.ARRAY, 679 "BIT": TokenType.BIT, 680 "BOOL": TokenType.BOOLEAN, 681 "BOOLEAN": TokenType.BOOLEAN, 682 "BYTE": TokenType.TINYINT, 683 "MEDIUMINT": TokenType.MEDIUMINT, 684 "TINYINT": TokenType.TINYINT, 685 "SHORT": TokenType.SMALLINT, 686 "SMALLINT": TokenType.SMALLINT, 687 "INT128": TokenType.INT128, 688 "INT2": TokenType.SMALLINT, 689 "INTEGER": TokenType.INT, 690 "INT": TokenType.INT, 691 "INT4": TokenType.INT, 692 "LONG": TokenType.BIGINT, 693 "BIGINT": TokenType.BIGINT, 694 "INT8": TokenType.BIGINT, 695 "DEC": TokenType.DECIMAL, 696 "DECIMAL": TokenType.DECIMAL, 697 "BIGDECIMAL": TokenType.BIGDECIMAL, 698 "BIGNUMERIC": TokenType.BIGDECIMAL, 699 "MAP": TokenType.MAP, 700 "NULLABLE": TokenType.NULLABLE, 701 "NUMBER": TokenType.DECIMAL, 702 "NUMERIC": TokenType.DECIMAL, 703 "FIXED": TokenType.DECIMAL, 704 "REAL": TokenType.FLOAT, 705 "FLOAT": TokenType.FLOAT, 706 "FLOAT4": TokenType.FLOAT, 707 "FLOAT8": TokenType.DOUBLE, 708 "DOUBLE": TokenType.DOUBLE, 709 "DOUBLE PRECISION": TokenType.DOUBLE, 710 "JSON": TokenType.JSON, 711 "CHAR": TokenType.CHAR, 712 "CHARACTER": TokenType.CHAR, 713 "NCHAR": TokenType.NCHAR, 714 "VARCHAR": TokenType.VARCHAR, 715 "VARCHAR2": TokenType.VARCHAR, 716 "NVARCHAR": TokenType.NVARCHAR, 717 "NVARCHAR2": TokenType.NVARCHAR, 718 "STR": TokenType.TEXT, 719 "STRING": TokenType.TEXT, 720 "TEXT": TokenType.TEXT, 721 "LONGTEXT": TokenType.LONGTEXT, 722 "MEDIUMTEXT": TokenType.MEDIUMTEXT, 723 "TINYTEXT": TokenType.TINYTEXT, 724 "CLOB": TokenType.TEXT, 725 "LONGVARCHAR": TokenType.TEXT, 726 "BINARY": TokenType.BINARY, 727 "BLOB": TokenType.VARBINARY, 728 "LONGBLOB": TokenType.LONGBLOB, 729 "MEDIUMBLOB": TokenType.MEDIUMBLOB, 730 "TINYBLOB": TokenType.TINYBLOB, 731 "BYTEA": TokenType.VARBINARY, 732 "VARBINARY": TokenType.VARBINARY, 733 "TIME": TokenType.TIME, 734 "TIMETZ": TokenType.TIMETZ, 735 "TIMESTAMP": TokenType.TIMESTAMP, 736 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 737 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 738 "DATE": TokenType.DATE, 739 "DATETIME": TokenType.DATETIME, 740 "INT4RANGE": TokenType.INT4RANGE, 741 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 742 "INT8RANGE": TokenType.INT8RANGE, 743 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 744 "NUMRANGE": TokenType.NUMRANGE, 745 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 746 "TSRANGE": TokenType.TSRANGE, 747 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 748 "TSTZRANGE": TokenType.TSTZRANGE, 749 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 750 "DATERANGE": TokenType.DATERANGE, 751 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 752 "UNIQUE": TokenType.UNIQUE, 753 "STRUCT": TokenType.STRUCT, 754 "VARIANT": TokenType.VARIANT, 755 "ALTER": TokenType.ALTER, 756 "ANALYZE": TokenType.COMMAND, 757 "CALL": TokenType.COMMAND, 758 "COMMENT": TokenType.COMMENT, 759 "COPY": TokenType.COMMAND, 760 "EXPLAIN": TokenType.COMMAND, 761 "GRANT": TokenType.COMMAND, 762 "OPTIMIZE": TokenType.COMMAND, 763 "PREPARE": TokenType.COMMAND, 764 "TRUNCATE": TokenType.COMMAND, 765 "VACUUM": TokenType.COMMAND, 766 "USER-DEFINED": TokenType.USERDEFINED, 767 "FOR VERSION": TokenType.VERSION_SNAPSHOT, 768 "FOR TIMESTAMP": TokenType.TIMESTAMP_SNAPSHOT, 769 } 770 771 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 772 " ": TokenType.SPACE, 773 "\t": TokenType.SPACE, 774 "\n": TokenType.BREAK, 775 "\r": TokenType.BREAK, 776 "\r\n": TokenType.BREAK, 777 } 778 779 COMMANDS = { 780 TokenType.COMMAND, 781 TokenType.EXECUTE, 782 TokenType.FETCH, 783 TokenType.SHOW, 784 } 785 786 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 787 788 # handle numeric literals like in hive (3L = BIGINT) 789 NUMERIC_LITERALS: t.Dict[str, str] = {} 790 ENCODE: t.Optional[str] = None 791 792 COMMENTS = ["--", ("/*", "*/")] 793 794 __slots__ = ( 795 "sql", 796 "size", 797 "tokens", 798 "_start", 799 "_current", 800 "_line", 801 "_col", 802 "_comments", 803 "_char", 804 "_end", 805 "_peek", 806 "_prev_token_line", 807 ) 808 809 def __init__(self) -> None: 810 self.reset() 811 812 def reset(self) -> None: 813 self.sql = "" 814 self.size = 0 815 self.tokens: t.List[Token] = [] 816 self._start = 0 817 self._current = 0 818 self._line = 1 819 self._col = 0 820 self._comments: t.List[str] = [] 821 822 self._char = "" 823 self._end = False 824 self._peek = "" 825 self._prev_token_line = -1 826 827 def tokenize(self, sql: str) -> t.List[Token]: 828 """Returns a list of tokens corresponding to the SQL string `sql`.""" 829 self.reset() 830 self.sql = sql 831 self.size = len(sql) 832 833 try: 834 self._scan() 835 except Exception as e: 836 start = max(self._current - 50, 0) 837 end = min(self._current + 50, self.size - 1) 838 context = self.sql[start:end] 839 raise TokenError(f"Error tokenizing '{context}'") from e 840 841 return self.tokens 842 843 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 844 while self.size and not self._end: 845 self._start = self._current 846 self._advance() 847 848 if self._char is None: 849 break 850 851 if self._char not in self.WHITE_SPACE: 852 if self._char.isdigit(): 853 self._scan_number() 854 elif self._char in self._IDENTIFIERS: 855 self._scan_identifier(self._IDENTIFIERS[self._char]) 856 else: 857 self._scan_keywords() 858 859 if until and until(): 860 break 861 862 if self.tokens and self._comments: 863 self.tokens[-1].comments.extend(self._comments) 864 865 def _chars(self, size: int) -> str: 866 if size == 1: 867 return self._char 868 869 start = self._current - 1 870 end = start + size 871 872 return self.sql[start:end] if end <= self.size else "" 873 874 def _advance(self, i: int = 1, alnum: bool = False) -> None: 875 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 876 self._col = 1 877 self._line += 1 878 else: 879 self._col += i 880 881 self._current += i 882 self._end = self._current >= self.size 883 self._char = self.sql[self._current - 1] 884 self._peek = "" if self._end else self.sql[self._current] 885 886 if alnum and self._char.isalnum(): 887 # Here we use local variables instead of attributes for better performance 888 _col = self._col 889 _current = self._current 890 _end = self._end 891 _peek = self._peek 892 893 while _peek.isalnum(): 894 _col += 1 895 _current += 1 896 _end = _current >= self.size 897 _peek = "" if _end else self.sql[_current] 898 899 self._col = _col 900 self._current = _current 901 self._end = _end 902 self._peek = _peek 903 self._char = self.sql[_current - 1] 904 905 @property 906 def _text(self) -> str: 907 return self.sql[self._start : self._current] 908 909 def peek(self, i: int = 0) -> str: 910 i = self._current + i 911 if i < self.size: 912 return self.sql[i] 913 return "" 914 915 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 916 self._prev_token_line = self._line 917 918 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 919 self.tokens[-1].comments.extend(self._comments) 920 self._comments = [] 921 922 self.tokens.append( 923 Token( 924 token_type, 925 text=self._text if text is None else text, 926 line=self._line, 927 col=self._col, 928 start=self._start, 929 end=self._current - 1, 930 comments=self._comments, 931 ) 932 ) 933 self._comments = [] 934 935 # If we have either a semicolon or a begin token before the command's token, we'll parse 936 # whatever follows the command's token as a string 937 if ( 938 token_type in self.COMMANDS 939 and self._peek != ";" 940 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 941 ): 942 start = self._current 943 tokens = len(self.tokens) 944 self._scan(lambda: self._peek == ";") 945 self.tokens = self.tokens[:tokens] 946 text = self.sql[start : self._current].strip() 947 if text: 948 self._add(TokenType.STRING, text) 949 950 def _scan_keywords(self) -> None: 951 size = 0 952 word = None 953 chars = self._text 954 char = chars 955 prev_space = False 956 skip = False 957 trie = self._KEYWORD_TRIE 958 single_token = char in self.SINGLE_TOKENS 959 960 while chars: 961 if skip: 962 result = TrieResult.PREFIX 963 else: 964 result, trie = in_trie(trie, char.upper()) 965 966 if result == TrieResult.FAILED: 967 break 968 if result == TrieResult.EXISTS: 969 word = chars 970 971 end = self._current + size 972 size += 1 973 974 if end < self.size: 975 char = self.sql[end] 976 single_token = single_token or char in self.SINGLE_TOKENS 977 is_space = char in self.WHITE_SPACE 978 979 if not is_space or not prev_space: 980 if is_space: 981 char = " " 982 chars += char 983 prev_space = is_space 984 skip = False 985 else: 986 skip = True 987 else: 988 char = "" 989 chars = " " 990 991 if word: 992 if self._scan_string(word): 993 return 994 if self._scan_comment(word): 995 return 996 if prev_space or single_token or not char: 997 self._advance(size - 1) 998 word = word.upper() 999 self._add(self.KEYWORDS[word], text=word) 1000 return 1001 if self._char in self.SINGLE_TOKENS: 1002 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 1003 return 1004 self._scan_var() 1005 1006 def _scan_comment(self, comment_start: str) -> bool: 1007 if comment_start not in self._COMMENTS: 1008 return False 1009 1010 comment_start_line = self._line 1011 comment_start_size = len(comment_start) 1012 comment_end = self._COMMENTS[comment_start] 1013 1014 if comment_end: 1015 # Skip the comment's start delimiter 1016 self._advance(comment_start_size) 1017 1018 comment_end_size = len(comment_end) 1019 while not self._end and self._chars(comment_end_size) != comment_end: 1020 self._advance(alnum=True) 1021 1022 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1023 self._advance(comment_end_size - 1) 1024 else: 1025 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1026 self._advance(alnum=True) 1027 self._comments.append(self._text[comment_start_size:]) 1028 1029 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1030 # Multiple consecutive comments are preserved by appending them to the current comments list. 1031 if comment_start_line == self._prev_token_line: 1032 self.tokens[-1].comments.extend(self._comments) 1033 self._comments = [] 1034 self._prev_token_line = self._line 1035 1036 return True 1037 1038 def _scan_number(self) -> None: 1039 if self._char == "0": 1040 peek = self._peek.upper() 1041 if peek == "B": 1042 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1043 elif peek == "X": 1044 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1045 1046 decimal = False 1047 scientific = 0 1048 1049 while True: 1050 if self._peek.isdigit(): 1051 self._advance() 1052 elif self._peek == "." and not decimal: 1053 after = self.peek(1) 1054 if after.isdigit() or not after.isalpha(): 1055 decimal = True 1056 self._advance() 1057 else: 1058 return self._add(TokenType.VAR) 1059 elif self._peek in ("-", "+") and scientific == 1: 1060 scientific += 1 1061 self._advance() 1062 elif self._peek.upper() == "E" and not scientific: 1063 scientific += 1 1064 self._advance() 1065 elif self._peek.isidentifier(): 1066 number_text = self._text 1067 literal = "" 1068 1069 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1070 literal += self._peek.upper() 1071 self._advance() 1072 1073 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1074 1075 if token_type: 1076 self._add(TokenType.NUMBER, number_text) 1077 self._add(TokenType.DCOLON, "::") 1078 return self._add(token_type, literal) 1079 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1080 return self._add(TokenType.VAR) 1081 1082 self._advance(-len(literal)) 1083 return self._add(TokenType.NUMBER, number_text) 1084 else: 1085 return self._add(TokenType.NUMBER) 1086 1087 def _scan_bits(self) -> None: 1088 self._advance() 1089 value = self._extract_value() 1090 try: 1091 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1092 int(value, 2) 1093 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1094 except ValueError: 1095 self._add(TokenType.IDENTIFIER) 1096 1097 def _scan_hex(self) -> None: 1098 self._advance() 1099 value = self._extract_value() 1100 try: 1101 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1102 int(value, 16) 1103 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1104 except ValueError: 1105 self._add(TokenType.IDENTIFIER) 1106 1107 def _extract_value(self) -> str: 1108 while True: 1109 char = self._peek.strip() 1110 if char and char not in self.SINGLE_TOKENS: 1111 self._advance(alnum=True) 1112 else: 1113 break 1114 1115 return self._text 1116 1117 def _scan_string(self, start: str) -> bool: 1118 base = None 1119 token_type = TokenType.STRING 1120 1121 if start in self._QUOTES: 1122 end = self._QUOTES[start] 1123 elif start in self._FORMAT_STRINGS: 1124 end, token_type = self._FORMAT_STRINGS[start] 1125 1126 if token_type == TokenType.HEX_STRING: 1127 base = 16 1128 elif token_type == TokenType.BIT_STRING: 1129 base = 2 1130 else: 1131 return False 1132 1133 self._advance(len(start)) 1134 text = self._extract_string(end) 1135 1136 if base: 1137 try: 1138 int(text, base) 1139 except: 1140 raise TokenError( 1141 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1142 ) 1143 else: 1144 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1145 1146 self._add(token_type, text) 1147 return True 1148 1149 def _scan_identifier(self, identifier_end: str) -> None: 1150 self._advance() 1151 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1152 self._add(TokenType.IDENTIFIER, text) 1153 1154 def _scan_var(self) -> None: 1155 while True: 1156 char = self._peek.strip() 1157 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1158 self._advance(alnum=True) 1159 else: 1160 break 1161 1162 self._add( 1163 TokenType.VAR 1164 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1165 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1166 ) 1167 1168 def _extract_string(self, delimiter: str, escapes=None) -> str: 1169 text = "" 1170 delim_size = len(delimiter) 1171 escapes = self._STRING_ESCAPES if escapes is None else escapes 1172 1173 while True: 1174 if ( 1175 self._char in escapes 1176 and (self._peek == delimiter or self._peek in escapes) 1177 and (self._char not in self._QUOTES or self._char == self._peek) 1178 ): 1179 if self._peek == delimiter: 1180 text += self._peek 1181 else: 1182 text += self._char + self._peek 1183 1184 if self._current + 1 < self.size: 1185 self._advance(2) 1186 else: 1187 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1188 else: 1189 if self._chars(delim_size) == delimiter: 1190 if delim_size > 1: 1191 self._advance(delim_size - 1) 1192 break 1193 1194 if self._end: 1195 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1196 1197 current = self._current - 1 1198 self._advance(alnum=True) 1199 text += self.sql[current : self._current - 1] 1200 1201 return text
SINGLE_TOKENS =
{'(': <TokenType.L_PAREN: 'L_PAREN'>, ')': <TokenType.R_PAREN: 'R_PAREN'>, '[': <TokenType.L_BRACKET: 'L_BRACKET'>, ']': <TokenType.R_BRACKET: 'R_BRACKET'>, '{': <TokenType.L_BRACE: 'L_BRACE'>, '}': <TokenType.R_BRACE: 'R_BRACE'>, '&': <TokenType.AMP: 'AMP'>, '^': <TokenType.CARET: 'CARET'>, ':': <TokenType.COLON: 'COLON'>, ',': <TokenType.COMMA: 'COMMA'>, '.': <TokenType.DOT: 'DOT'>, '-': <TokenType.DASH: 'DASH'>, '=': <TokenType.EQ: 'EQ'>, '>': <TokenType.GT: 'GT'>, '<': <TokenType.LT: 'LT'>, '%': <TokenType.MOD: 'MOD'>, '!': <TokenType.NOT: 'NOT'>, '|': <TokenType.PIPE: 'PIPE'>, '+': <TokenType.PLUS: 'PLUS'>, ';': <TokenType.SEMICOLON: 'SEMICOLON'>, '/': <TokenType.SLASH: 'SLASH'>, '\\': <TokenType.BACKSLASH: 'BACKSLASH'>, '*': <TokenType.STAR: 'STAR'>, '~': <TokenType.TILDA: 'TILDA'>, '?': <TokenType.PLACEHOLDER: 'PLACEHOLDER'>, '@': <TokenType.PARAMETER: 'PARAMETER'>, "'": <TokenType.QUOTE: 'QUOTE'>, '`': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '"': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '#': <TokenType.HASH: 'HASH'>}
KEYWORDS: Dict[str, TokenType] =
{'{%': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%-': <TokenType.BLOCK_START: 'BLOCK_START'>, '%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '+%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '{{+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{{-': <TokenType.BLOCK_START: 'BLOCK_START'>, '+}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '/*+': <TokenType.HINT: 'HINT'>, '==': <TokenType.EQ: 'EQ'>, '::': <TokenType.DCOLON: 'DCOLON'>, '||': <TokenType.DPIPE: 'DPIPE'>, '>=': <TokenType.GTE: 'GTE'>, '<=': <TokenType.LTE: 'LTE'>, '<>': <TokenType.NEQ: 'NEQ'>, '!=': <TokenType.NEQ: 'NEQ'>, '<=>': <TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>, '->': <TokenType.ARROW: 'ARROW'>, '->>': <TokenType.DARROW: 'DARROW'>, '=>': <TokenType.FARROW: 'FARROW'>, '#>': <TokenType.HASH_ARROW: 'HASH_ARROW'>, '#>>': <TokenType.DHASH_ARROW: 'DHASH_ARROW'>, '<->': <TokenType.LR_ARROW: 'LR_ARROW'>, '&&': <TokenType.DAMP: 'DAMP'>, '??': <TokenType.DQMARK: 'DQMARK'>, 'ALL': <TokenType.ALL: 'ALL'>, 'ALWAYS': <TokenType.ALWAYS: 'ALWAYS'>, 'AND': <TokenType.AND: 'AND'>, 'ANTI': <TokenType.ANTI: 'ANTI'>, 'ANY': <TokenType.ANY: 'ANY'>, 'ASC': <TokenType.ASC: 'ASC'>, 'AS': <TokenType.ALIAS: 'ALIAS'>, 'ASOF': <TokenType.ASOF: 'ASOF'>, 'AUTOINCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'AUTO_INCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'BEGIN': <TokenType.BEGIN: 'BEGIN'>, 'BETWEEN': <TokenType.BETWEEN: 'BETWEEN'>, 'CACHE': <TokenType.CACHE: 'CACHE'>, 'UNCACHE': <TokenType.UNCACHE: 'UNCACHE'>, 'CASE': <TokenType.CASE: 'CASE'>, 'CHARACTER SET': <TokenType.CHARACTER_SET: 'CHARACTER_SET'>, 'CLUSTER BY': <TokenType.CLUSTER_BY: 'CLUSTER_BY'>, 'COLLATE': <TokenType.COLLATE: 'COLLATE'>, 'COLUMN': <TokenType.COLUMN: 'COLUMN'>, 'COMMIT': <TokenType.COMMIT: 'COMMIT'>, 'CONNECT BY': <TokenType.CONNECT_BY: 'CONNECT_BY'>, 'CONSTRAINT': <TokenType.CONSTRAINT: 'CONSTRAINT'>, 'CREATE': <TokenType.CREATE: 'CREATE'>, 'CROSS': <TokenType.CROSS: 'CROSS'>, 'CUBE': <TokenType.CUBE: 'CUBE'>, 'CURRENT_DATE': <TokenType.CURRENT_DATE: 'CURRENT_DATE'>, 'CURRENT_TIME': <TokenType.CURRENT_TIME: 'CURRENT_TIME'>, 'CURRENT_TIMESTAMP': <TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>, 'CURRENT_USER': <TokenType.CURRENT_USER: 'CURRENT_USER'>, 'DATABASE': <TokenType.DATABASE: 'DATABASE'>, 'DEFAULT': <TokenType.DEFAULT: 'DEFAULT'>, 'DELETE': <TokenType.DELETE: 'DELETE'>, 'DESC': <TokenType.DESC: 'DESC'>, 'DESCRIBE': <TokenType.DESCRIBE: 'DESCRIBE'>, 'DISTINCT': <TokenType.DISTINCT: 'DISTINCT'>, 'DISTRIBUTE BY': <TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>, 'DIV': <TokenType.DIV: 'DIV'>, 'DROP': <TokenType.DROP: 'DROP'>, 'ELSE': <TokenType.ELSE: 'ELSE'>, 'END': <TokenType.END: 'END'>, 'ESCAPE': <TokenType.ESCAPE: 'ESCAPE'>, 'EXCEPT': <TokenType.EXCEPT: 'EXCEPT'>, 'EXECUTE': <TokenType.EXECUTE: 'EXECUTE'>, 'EXISTS': <TokenType.EXISTS: 'EXISTS'>, 'FALSE': <TokenType.FALSE: 'FALSE'>, 'FETCH': <TokenType.FETCH: 'FETCH'>, 'FILTER': <TokenType.FILTER: 'FILTER'>, 'FIRST': <TokenType.FIRST: 'FIRST'>, 'FULL': <TokenType.FULL: 'FULL'>, 'FUNCTION': <TokenType.FUNCTION: 'FUNCTION'>, 'FOR': <TokenType.FOR: 'FOR'>, 'FOREIGN KEY': <TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>, 'FORMAT': <TokenType.FORMAT: 'FORMAT'>, 'FROM': <TokenType.FROM: 'FROM'>, 'GEOGRAPHY': <TokenType.GEOGRAPHY: 'GEOGRAPHY'>, 'GEOMETRY': <TokenType.GEOMETRY: 'GEOMETRY'>, 'GLOB': <TokenType.GLOB: 'GLOB'>, 'GROUP BY': <TokenType.GROUP_BY: 'GROUP_BY'>, 'GROUPING SETS': <TokenType.GROUPING_SETS: 'GROUPING_SETS'>, 'HAVING': <TokenType.HAVING: 'HAVING'>, 'ILIKE': <TokenType.ILIKE: 'ILIKE'>, 'IN': <TokenType.IN: 'IN'>, 'INDEX': <TokenType.INDEX: 'INDEX'>, 'INET': <TokenType.INET: 'INET'>, 'INNER': <TokenType.INNER: 'INNER'>, 'INSERT': <TokenType.INSERT: 'INSERT'>, 'INTERVAL': <TokenType.INTERVAL: 'INTERVAL'>, 'INTERSECT': <TokenType.INTERSECT: 'INTERSECT'>, 'INTO': <TokenType.INTO: 'INTO'>, 'IS': <TokenType.IS: 'IS'>, 'ISNULL': <TokenType.ISNULL: 'ISNULL'>, 'JOIN': <TokenType.JOIN: 'JOIN'>, 'KEEP': <TokenType.KEEP: 'KEEP'>, 'KILL': <TokenType.KILL: 'KILL'>, 'LATERAL': <TokenType.LATERAL: 'LATERAL'>, 'LEFT': <TokenType.LEFT: 'LEFT'>, 'LIKE': <TokenType.LIKE: 'LIKE'>, 'LIMIT': <TokenType.LIMIT: 'LIMIT'>, 'LOAD': <TokenType.LOAD: 'LOAD'>, 'LOCK': <TokenType.LOCK: 'LOCK'>, 'MERGE': <TokenType.MERGE: 'MERGE'>, 'NATURAL': <TokenType.NATURAL: 'NATURAL'>, 'NEXT': <TokenType.NEXT: 'NEXT'>, 'NOT': <TokenType.NOT: 'NOT'>, 'NOTNULL': <TokenType.NOTNULL: 'NOTNULL'>, 'NULL': <TokenType.NULL: 'NULL'>, 'OBJECT': <TokenType.OBJECT: 'OBJECT'>, 'OFFSET': <TokenType.OFFSET: 'OFFSET'>, 'ON': <TokenType.ON: 'ON'>, 'OR': <TokenType.OR: 'OR'>, 'XOR': <TokenType.XOR: 'XOR'>, 'ORDER BY': <TokenType.ORDER_BY: 'ORDER_BY'>, 'ORDINALITY': <TokenType.ORDINALITY: 'ORDINALITY'>, 'OUTER': <TokenType.OUTER: 'OUTER'>, 'OVER': <TokenType.OVER: 'OVER'>, 'OVERLAPS': <TokenType.OVERLAPS: 'OVERLAPS'>, 'OVERWRITE': <TokenType.OVERWRITE: 'OVERWRITE'>, 'PARTITION': <TokenType.PARTITION: 'PARTITION'>, 'PARTITION BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED_BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PERCENT': <TokenType.PERCENT: 'PERCENT'>, 'PIVOT': <TokenType.PIVOT: 'PIVOT'>, 'PRAGMA': <TokenType.PRAGMA: 'PRAGMA'>, 'PRIMARY KEY': <TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>, 'PROCEDURE': <TokenType.PROCEDURE: 'PROCEDURE'>, 'QUALIFY': <TokenType.QUALIFY: 'QUALIFY'>, 'RANGE': <TokenType.RANGE: 'RANGE'>, 'RECURSIVE': <TokenType.RECURSIVE: 'RECURSIVE'>, 'REGEXP': <TokenType.RLIKE: 'RLIKE'>, 'REPLACE': <TokenType.REPLACE: 'REPLACE'>, 'RETURNING': <TokenType.RETURNING: 'RETURNING'>, 'REFERENCES': <TokenType.REFERENCES: 'REFERENCES'>, 'RIGHT': <TokenType.RIGHT: 'RIGHT'>, 'RLIKE': <TokenType.RLIKE: 'RLIKE'>, 'ROLLBACK': <TokenType.ROLLBACK: 'ROLLBACK'>, 'ROLLUP': <TokenType.ROLLUP: 'ROLLUP'>, 'ROW': <TokenType.ROW: 'ROW'>, 'ROWS': <TokenType.ROWS: 'ROWS'>, 'SCHEMA': <TokenType.SCHEMA: 'SCHEMA'>, 'SELECT': <TokenType.SELECT: 'SELECT'>, 'SEMI': <TokenType.SEMI: 'SEMI'>, 'SET': <TokenType.SET: 'SET'>, 'SETTINGS': <TokenType.SETTINGS: 'SETTINGS'>, 'SHOW': <TokenType.SHOW: 'SHOW'>, 'SIMILAR TO': <TokenType.SIMILAR_TO: 'SIMILAR_TO'>, 'SOME': <TokenType.SOME: 'SOME'>, 'SORT BY': <TokenType.SORT_BY: 'SORT_BY'>, 'START WITH': <TokenType.START_WITH: 'START_WITH'>, 'TABLE': <TokenType.TABLE: 'TABLE'>, 'TABLESAMPLE': <TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>, 'TEMP': <TokenType.TEMPORARY: 'TEMPORARY'>, 'TEMPORARY': <TokenType.TEMPORARY: 'TEMPORARY'>, 'THEN': <TokenType.THEN: 'THEN'>, 'TRUE': <TokenType.TRUE: 'TRUE'>, 'UNION': <TokenType.UNION: 'UNION'>, 'UNKNOWN': <TokenType.UNKNOWN: 'UNKNOWN'>, 'UNNEST': <TokenType.UNNEST: 'UNNEST'>, 'UNPIVOT': <TokenType.UNPIVOT: 'UNPIVOT'>, 'UPDATE': <TokenType.UPDATE: 'UPDATE'>, 'USE': <TokenType.USE: 'USE'>, 'USING': <TokenType.USING: 'USING'>, 'UUID': <TokenType.UUID: 'UUID'>, 'VALUES': <TokenType.VALUES: 'VALUES'>, 'VIEW': <TokenType.VIEW: 'VIEW'>, 'VOLATILE': <TokenType.VOLATILE: 'VOLATILE'>, 'WHEN': <TokenType.WHEN: 'WHEN'>, 'WHERE': <TokenType.WHERE: 'WHERE'>, 'WINDOW': <TokenType.WINDOW: 'WINDOW'>, 'WITH': <TokenType.WITH: 'WITH'>, 'APPLY': <TokenType.APPLY: 'APPLY'>, 'ARRAY': <TokenType.ARRAY: 'ARRAY'>, 'BIT': <TokenType.BIT: 'BIT'>, 'BOOL': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BOOLEAN': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BYTE': <TokenType.TINYINT: 'TINYINT'>, 'MEDIUMINT': <TokenType.MEDIUMINT: 'MEDIUMINT'>, 'TINYINT': <TokenType.TINYINT: 'TINYINT'>, 'SHORT': <TokenType.SMALLINT: 'SMALLINT'>, 'SMALLINT': <TokenType.SMALLINT: 'SMALLINT'>, 'INT128': <TokenType.INT128: 'INT128'>, 'INT2': <TokenType.SMALLINT: 'SMALLINT'>, 'INTEGER': <TokenType.INT: 'INT'>, 'INT': <TokenType.INT: 'INT'>, 'INT4': <TokenType.INT: 'INT'>, 'LONG': <TokenType.BIGINT: 'BIGINT'>, 'BIGINT': <TokenType.BIGINT: 'BIGINT'>, 'INT8': <TokenType.BIGINT: 'BIGINT'>, 'DEC': <TokenType.DECIMAL: 'DECIMAL'>, 'DECIMAL': <TokenType.DECIMAL: 'DECIMAL'>, 'BIGDECIMAL': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'BIGNUMERIC': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'MAP': <TokenType.MAP: 'MAP'>, 'NULLABLE': <TokenType.NULLABLE: 'NULLABLE'>, 'NUMBER': <TokenType.DECIMAL: 'DECIMAL'>, 'NUMERIC': <TokenType.DECIMAL: 'DECIMAL'>, 'FIXED': <TokenType.DECIMAL: 'DECIMAL'>, 'REAL': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT4': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT8': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE PRECISION': <TokenType.DOUBLE: 'DOUBLE'>, 'JSON': <TokenType.JSON: 'JSON'>, 'CHAR': <TokenType.CHAR: 'CHAR'>, 'CHARACTER': <TokenType.CHAR: 'CHAR'>, 'NCHAR': <TokenType.NCHAR: 'NCHAR'>, 'VARCHAR': <TokenType.VARCHAR: 'VARCHAR'>, 'VARCHAR2': <TokenType.VARCHAR: 'VARCHAR'>, 'NVARCHAR': <TokenType.NVARCHAR: 'NVARCHAR'>, 'NVARCHAR2': <TokenType.NVARCHAR: 'NVARCHAR'>, 'STR': <TokenType.TEXT: 'TEXT'>, 'STRING': <TokenType.TEXT: 'TEXT'>, 'TEXT': <TokenType.TEXT: 'TEXT'>, 'LONGTEXT': <TokenType.LONGTEXT: 'LONGTEXT'>, 'MEDIUMTEXT': <TokenType.MEDIUMTEXT: 'MEDIUMTEXT'>, 'TINYTEXT': <TokenType.TINYTEXT: 'TINYTEXT'>, 'CLOB': <TokenType.TEXT: 'TEXT'>, 'LONGVARCHAR': <TokenType.TEXT: 'TEXT'>, 'BINARY': <TokenType.BINARY: 'BINARY'>, 'BLOB': <TokenType.VARBINARY: 'VARBINARY'>, 'LONGBLOB': <TokenType.LONGBLOB: 'LONGBLOB'>, 'MEDIUMBLOB': <TokenType.MEDIUMBLOB: 'MEDIUMBLOB'>, 'TINYBLOB': <TokenType.TINYBLOB: 'TINYBLOB'>, 'BYTEA': <TokenType.VARBINARY: 'VARBINARY'>, 'VARBINARY': <TokenType.VARBINARY: 'VARBINARY'>, 'TIME': <TokenType.TIME: 'TIME'>, 'TIMETZ': <TokenType.TIMETZ: 'TIMETZ'>, 'TIMESTAMP': <TokenType.TIMESTAMP: 'TIMESTAMP'>, 'TIMESTAMPTZ': <TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>, 'TIMESTAMPLTZ': <TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>, 'DATE': <TokenType.DATE: 'DATE'>, 'DATETIME': <TokenType.DATETIME: 'DATETIME'>, 'INT4RANGE': <TokenType.INT4RANGE: 'INT4RANGE'>, 'INT4MULTIRANGE': <TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>, 'INT8RANGE': <TokenType.INT8RANGE: 'INT8RANGE'>, 'INT8MULTIRANGE': <TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>, 'NUMRANGE': <TokenType.NUMRANGE: 'NUMRANGE'>, 'NUMMULTIRANGE': <TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>, 'TSRANGE': <TokenType.TSRANGE: 'TSRANGE'>, 'TSMULTIRANGE': <TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>, 'TSTZRANGE': <TokenType.TSTZRANGE: 'TSTZRANGE'>, 'TSTZMULTIRANGE': <TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>, 'DATERANGE': <TokenType.DATERANGE: 'DATERANGE'>, 'DATEMULTIRANGE': <TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>, 'UNIQUE': <TokenType.UNIQUE: 'UNIQUE'>, 'STRUCT': <TokenType.STRUCT: 'STRUCT'>, 'VARIANT': <TokenType.VARIANT: 'VARIANT'>, 'ALTER': <TokenType.ALTER: 'ALTER'>, 'ANALYZE': <TokenType.COMMAND: 'COMMAND'>, 'CALL': <TokenType.COMMAND: 'COMMAND'>, 'COMMENT': <TokenType.COMMENT: 'COMMENT'>, 'COPY': <TokenType.COMMAND: 'COMMAND'>, 'EXPLAIN': <TokenType.COMMAND: 'COMMAND'>, 'GRANT': <TokenType.COMMAND: 'COMMAND'>, 'OPTIMIZE': <TokenType.COMMAND: 'COMMAND'>, 'PREPARE': <TokenType.COMMAND: 'COMMAND'>, 'TRUNCATE': <TokenType.COMMAND: 'COMMAND'>, 'VACUUM': <TokenType.COMMAND: 'COMMAND'>, 'USER-DEFINED': <TokenType.USERDEFINED: 'USERDEFINED'>, 'FOR VERSION': <TokenType.VERSION_SNAPSHOT: 'VERSION_SNAPSHOT'>, 'FOR TIMESTAMP': <TokenType.TIMESTAMP_SNAPSHOT: 'TIMESTAMP_SNAPSHOT'>}
WHITE_SPACE: Dict[Optional[str], TokenType] =
{' ': <TokenType.SPACE: 'SPACE'>, '\t': <TokenType.SPACE: 'SPACE'>, '\n': <TokenType.BREAK: 'BREAK'>, '\r': <TokenType.BREAK: 'BREAK'>, '\r\n': <TokenType.BREAK: 'BREAK'>}
COMMANDS =
{<TokenType.COMMAND: 'COMMAND'>, <TokenType.FETCH: 'FETCH'>, <TokenType.EXECUTE: 'EXECUTE'>, <TokenType.SHOW: 'SHOW'>}
COMMAND_PREFIX_TOKENS =
{<TokenType.BEGIN: 'BEGIN'>, <TokenType.SEMICOLON: 'SEMICOLON'>}
def
reset(self) -> None:
812 def reset(self) -> None: 813 self.sql = "" 814 self.size = 0 815 self.tokens: t.List[Token] = [] 816 self._start = 0 817 self._current = 0 818 self._line = 1 819 self._col = 0 820 self._comments: t.List[str] = [] 821 822 self._char = "" 823 self._end = False 824 self._peek = "" 825 self._prev_token_line = -1
827 def tokenize(self, sql: str) -> t.List[Token]: 828 """Returns a list of tokens corresponding to the SQL string `sql`.""" 829 self.reset() 830 self.sql = sql 831 self.size = len(sql) 832 833 try: 834 self._scan() 835 except Exception as e: 836 start = max(self._current - 50, 0) 837 end = min(self._current + 50, self.size - 1) 838 context = self.sql[start:end] 839 raise TokenError(f"Error tokenizing '{context}'") from e 840 841 return self.tokens
Returns a list of tokens corresponding to the SQL string sql
.