Token.h 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. * Copyright (c) 2021, Jan de Visser <jan@de-visser.net>
  4. *
  5. * SPDX-License-Identifier: BSD-2-Clause
  6. */
  7. #pragma once
  8. #include <AK/HashMap.h>
  9. #include <AK/String.h>
  10. #include <AK/StringView.h>
  11. namespace SQL::AST {
  12. // https://sqlite.org/lang_keywords.html
  13. #define ENUMERATE_SQL_TOKENS \
  14. __ENUMERATE_SQL_TOKEN("ABORT", Abort, Keyword) \
  15. __ENUMERATE_SQL_TOKEN("ACTION", Action, Keyword) \
  16. __ENUMERATE_SQL_TOKEN("ADD", Add, Keyword) \
  17. __ENUMERATE_SQL_TOKEN("AFTER", After, Keyword) \
  18. __ENUMERATE_SQL_TOKEN("ALL", All, Keyword) \
  19. __ENUMERATE_SQL_TOKEN("ALTER", Alter, Keyword) \
  20. __ENUMERATE_SQL_TOKEN("ALWAYS", Always, Keyword) \
  21. __ENUMERATE_SQL_TOKEN("ANALYZE", Analyze, Keyword) \
  22. __ENUMERATE_SQL_TOKEN("AND", And, Keyword) \
  23. __ENUMERATE_SQL_TOKEN("AS", As, Keyword) \
  24. __ENUMERATE_SQL_TOKEN("ASC", Asc, Keyword) \
  25. __ENUMERATE_SQL_TOKEN("ATTACH", Attach, Keyword) \
  26. __ENUMERATE_SQL_TOKEN("AUTOINCREMENT", Autoincrement, Keyword) \
  27. __ENUMERATE_SQL_TOKEN("BEFORE", Before, Keyword) \
  28. __ENUMERATE_SQL_TOKEN("BEGIN", Begin, Keyword) \
  29. __ENUMERATE_SQL_TOKEN("BETWEEN", Between, Keyword) \
  30. __ENUMERATE_SQL_TOKEN("BY", By, Keyword) \
  31. __ENUMERATE_SQL_TOKEN("CASCADE", Cascade, Keyword) \
  32. __ENUMERATE_SQL_TOKEN("CASE", Case, Keyword) \
  33. __ENUMERATE_SQL_TOKEN("CAST", Cast, Keyword) \
  34. __ENUMERATE_SQL_TOKEN("CHECK", Check, Keyword) \
  35. __ENUMERATE_SQL_TOKEN("COLLATE", Collate, Keyword) \
  36. __ENUMERATE_SQL_TOKEN("COLUMN", Column, Keyword) \
  37. __ENUMERATE_SQL_TOKEN("COMMIT", Commit, Keyword) \
  38. __ENUMERATE_SQL_TOKEN("CONFLICT", Conflict, Keyword) \
  39. __ENUMERATE_SQL_TOKEN("CONSTRAINT", Constraint, Keyword) \
  40. __ENUMERATE_SQL_TOKEN("CREATE", Create, Keyword) \
  41. __ENUMERATE_SQL_TOKEN("CROSS", Cross, Keyword) \
  42. __ENUMERATE_SQL_TOKEN("CURRENT", Current, Keyword) \
  43. __ENUMERATE_SQL_TOKEN("CURRENT_DATE", CurrentDate, Keyword) \
  44. __ENUMERATE_SQL_TOKEN("CURRENT_TIME", CurrentTime, Keyword) \
  45. __ENUMERATE_SQL_TOKEN("CURRENT_TIMESTAMP", CurrentTimestamp, Keyword) \
  46. __ENUMERATE_SQL_TOKEN("DATABASE", Database, Keyword) \
  47. __ENUMERATE_SQL_TOKEN("DEFAULT", Default, Keyword) \
  48. __ENUMERATE_SQL_TOKEN("DEFERRABLE", Deferrable, Keyword) \
  49. __ENUMERATE_SQL_TOKEN("DEFERRED", Deferred, Keyword) \
  50. __ENUMERATE_SQL_TOKEN("DELETE", Delete, Keyword) \
  51. __ENUMERATE_SQL_TOKEN("DESC", Desc, Keyword) \
  52. __ENUMERATE_SQL_TOKEN("DETACH", Detach, Keyword) \
  53. __ENUMERATE_SQL_TOKEN("DISTINCT", Distinct, Keyword) \
  54. __ENUMERATE_SQL_TOKEN("DO", Do, Keyword) \
  55. __ENUMERATE_SQL_TOKEN("DROP", Drop, Keyword) \
  56. __ENUMERATE_SQL_TOKEN("EACH", Each, Keyword) \
  57. __ENUMERATE_SQL_TOKEN("ELSE", Else, Keyword) \
  58. __ENUMERATE_SQL_TOKEN("END", End, Keyword) \
  59. __ENUMERATE_SQL_TOKEN("ESCAPE", Escape, Keyword) \
  60. __ENUMERATE_SQL_TOKEN("EXCEPT", Except, Keyword) \
  61. __ENUMERATE_SQL_TOKEN("EXCLUDE", Exclude, Keyword) \
  62. __ENUMERATE_SQL_TOKEN("EXCLUSIVE", Exclusive, Keyword) \
  63. __ENUMERATE_SQL_TOKEN("EXISTS", Exists, Keyword) \
  64. __ENUMERATE_SQL_TOKEN("EXPLAIN", Explain, Keyword) \
  65. __ENUMERATE_SQL_TOKEN("FAIL", Fail, Keyword) \
  66. __ENUMERATE_SQL_TOKEN("FILTER", Filter, Keyword) \
  67. __ENUMERATE_SQL_TOKEN("FIRST", First, Keyword) \
  68. __ENUMERATE_SQL_TOKEN("FOLLOWING", Following, Keyword) \
  69. __ENUMERATE_SQL_TOKEN("FOR", For, Keyword) \
  70. __ENUMERATE_SQL_TOKEN("FOREIGN", Foreign, Keyword) \
  71. __ENUMERATE_SQL_TOKEN("FROM", From, Keyword) \
  72. __ENUMERATE_SQL_TOKEN("FULL", Full, Keyword) \
  73. __ENUMERATE_SQL_TOKEN("GENERATED", Generated, Keyword) \
  74. __ENUMERATE_SQL_TOKEN("GLOB", Glob, Keyword) \
  75. __ENUMERATE_SQL_TOKEN("GROUP", Group, Keyword) \
  76. __ENUMERATE_SQL_TOKEN("GROUPS", Groups, Keyword) \
  77. __ENUMERATE_SQL_TOKEN("HAVING", Having, Keyword) \
  78. __ENUMERATE_SQL_TOKEN("IF", If, Keyword) \
  79. __ENUMERATE_SQL_TOKEN("IGNORE", Ignore, Keyword) \
  80. __ENUMERATE_SQL_TOKEN("IMMEDIATE", Immediate, Keyword) \
  81. __ENUMERATE_SQL_TOKEN("IN", In, Keyword) \
  82. __ENUMERATE_SQL_TOKEN("INDEX", Index, Keyword) \
  83. __ENUMERATE_SQL_TOKEN("INDEXED", Indexed, Keyword) \
  84. __ENUMERATE_SQL_TOKEN("INITIALLY", Initially, Keyword) \
  85. __ENUMERATE_SQL_TOKEN("INNER", Inner, Keyword) \
  86. __ENUMERATE_SQL_TOKEN("INSERT", Insert, Keyword) \
  87. __ENUMERATE_SQL_TOKEN("INSTEAD", Instead, Keyword) \
  88. __ENUMERATE_SQL_TOKEN("INTERSECT", Intersect, Keyword) \
  89. __ENUMERATE_SQL_TOKEN("INTO", Into, Keyword) \
  90. __ENUMERATE_SQL_TOKEN("IS", Is, Keyword) \
  91. __ENUMERATE_SQL_TOKEN("ISNULL", Isnull, Keyword) \
  92. __ENUMERATE_SQL_TOKEN("JOIN", Join, Keyword) \
  93. __ENUMERATE_SQL_TOKEN("KEY", Key, Keyword) \
  94. __ENUMERATE_SQL_TOKEN("LAST", Last, Keyword) \
  95. __ENUMERATE_SQL_TOKEN("LEFT", Left, Keyword) \
  96. __ENUMERATE_SQL_TOKEN("LIKE", Like, Keyword) \
  97. __ENUMERATE_SQL_TOKEN("LIMIT", Limit, Keyword) \
  98. __ENUMERATE_SQL_TOKEN("MATCH", Match, Keyword) \
  99. __ENUMERATE_SQL_TOKEN("MATERIALIZED", Materialized, Keyword) \
  100. __ENUMERATE_SQL_TOKEN("NATURAL", Natural, Keyword) \
  101. __ENUMERATE_SQL_TOKEN("NO", No, Keyword) \
  102. __ENUMERATE_SQL_TOKEN("NOT", Not, Keyword) \
  103. __ENUMERATE_SQL_TOKEN("NOTHING", Nothing, Keyword) \
  104. __ENUMERATE_SQL_TOKEN("NOTNULL", Notnull, Keyword) \
  105. __ENUMERATE_SQL_TOKEN("NULL", Null, Keyword) \
  106. __ENUMERATE_SQL_TOKEN("NULLS", Nulls, Keyword) \
  107. __ENUMERATE_SQL_TOKEN("OF", Of, Keyword) \
  108. __ENUMERATE_SQL_TOKEN("OFFSET", Offset, Keyword) \
  109. __ENUMERATE_SQL_TOKEN("ON", On, Keyword) \
  110. __ENUMERATE_SQL_TOKEN("OR", Or, Keyword) \
  111. __ENUMERATE_SQL_TOKEN("ORDER", Order, Keyword) \
  112. __ENUMERATE_SQL_TOKEN("OTHERS", Others, Keyword) \
  113. __ENUMERATE_SQL_TOKEN("OUTER", Outer, Keyword) \
  114. __ENUMERATE_SQL_TOKEN("OVER", Over, Keyword) \
  115. __ENUMERATE_SQL_TOKEN("PARTITION", Partition, Keyword) \
  116. __ENUMERATE_SQL_TOKEN("PLAN", Plan, Keyword) \
  117. __ENUMERATE_SQL_TOKEN("PRAGMA", Pragma, Keyword) \
  118. __ENUMERATE_SQL_TOKEN("PRECEDING", Preceding, Keyword) \
  119. __ENUMERATE_SQL_TOKEN("PRIMARY", Primary, Keyword) \
  120. __ENUMERATE_SQL_TOKEN("QUERY", Query, Keyword) \
  121. __ENUMERATE_SQL_TOKEN("RAISE", Raise, Keyword) \
  122. __ENUMERATE_SQL_TOKEN("RANGE", Range, Keyword) \
  123. __ENUMERATE_SQL_TOKEN("RECURSIVE", Recursive, Keyword) \
  124. __ENUMERATE_SQL_TOKEN("REFERENCES", References, Keyword) \
  125. __ENUMERATE_SQL_TOKEN("REGEXP", Regexp, Keyword) \
  126. __ENUMERATE_SQL_TOKEN("REINDEX", Reindex, Keyword) \
  127. __ENUMERATE_SQL_TOKEN("RELEASE", Release, Keyword) \
  128. __ENUMERATE_SQL_TOKEN("RENAME", Rename, Keyword) \
  129. __ENUMERATE_SQL_TOKEN("REPLACE", Replace, Keyword) \
  130. __ENUMERATE_SQL_TOKEN("RESTRICT", Restrict, Keyword) \
  131. __ENUMERATE_SQL_TOKEN("RETURNING", Returning, Keyword) \
  132. __ENUMERATE_SQL_TOKEN("RIGHT", Right, Keyword) \
  133. __ENUMERATE_SQL_TOKEN("ROLLBACK", Rollback, Keyword) \
  134. __ENUMERATE_SQL_TOKEN("ROW", Row, Keyword) \
  135. __ENUMERATE_SQL_TOKEN("ROWS", Rows, Keyword) \
  136. __ENUMERATE_SQL_TOKEN("SAVEPOINT", Savepoint, Keyword) \
  137. __ENUMERATE_SQL_TOKEN("SCHEMA", Schema, Keyword) \
  138. __ENUMERATE_SQL_TOKEN("SELECT", Select, Keyword) \
  139. __ENUMERATE_SQL_TOKEN("SET", Set, Keyword) \
  140. __ENUMERATE_SQL_TOKEN("TABLE", Table, Keyword) \
  141. __ENUMERATE_SQL_TOKEN("TEMP", Temp, Keyword) \
  142. __ENUMERATE_SQL_TOKEN("TEMPORARY", Temporary, Keyword) \
  143. __ENUMERATE_SQL_TOKEN("THEN", Then, Keyword) \
  144. __ENUMERATE_SQL_TOKEN("TIES", Ties, Keyword) \
  145. __ENUMERATE_SQL_TOKEN("TO", To, Keyword) \
  146. __ENUMERATE_SQL_TOKEN("TRANSACTION", Transaction, Keyword) \
  147. __ENUMERATE_SQL_TOKEN("TRIGGER", Trigger, Keyword) \
  148. __ENUMERATE_SQL_TOKEN("UNBOUNDED", Unbounded, Keyword) \
  149. __ENUMERATE_SQL_TOKEN("UNION", Union, Keyword) \
  150. __ENUMERATE_SQL_TOKEN("UNIQUE", Unique, Keyword) \
  151. __ENUMERATE_SQL_TOKEN("UPDATE", Update, Keyword) \
  152. __ENUMERATE_SQL_TOKEN("USING", Using, Keyword) \
  153. __ENUMERATE_SQL_TOKEN("VACUUM", Vacuum, Keyword) \
  154. __ENUMERATE_SQL_TOKEN("VALUES", Values, Keyword) \
  155. __ENUMERATE_SQL_TOKEN("VIEW", View, Keyword) \
  156. __ENUMERATE_SQL_TOKEN("VIRTUAL", Virtual, Keyword) \
  157. __ENUMERATE_SQL_TOKEN("WHEN", When, Keyword) \
  158. __ENUMERATE_SQL_TOKEN("WHERE", Where, Keyword) \
  159. __ENUMERATE_SQL_TOKEN("WINDOW", Window, Keyword) \
  160. __ENUMERATE_SQL_TOKEN("WITH", With, Keyword) \
  161. __ENUMERATE_SQL_TOKEN("WITHOUT", Without, Keyword) \
  162. __ENUMERATE_SQL_TOKEN("_identifier_", Identifier, Identifier) \
  163. __ENUMERATE_SQL_TOKEN("_numeric_", NumericLiteral, Number) \
  164. __ENUMERATE_SQL_TOKEN("_string_", StringLiteral, String) \
  165. __ENUMERATE_SQL_TOKEN("_blob_", BlobLiteral, Blob) \
  166. __ENUMERATE_SQL_TOKEN("_eof_", Eof, Invalid) \
  167. __ENUMERATE_SQL_TOKEN("_invalid_", Invalid, Invalid) \
  168. __ENUMERATE_SQL_TOKEN("&", Ampersand, Operator) \
  169. __ENUMERATE_SQL_TOKEN("*", Asterisk, Operator) \
  170. __ENUMERATE_SQL_TOKEN(",", Comma, Punctuation) \
  171. __ENUMERATE_SQL_TOKEN("/", Divide, Operator) \
  172. __ENUMERATE_SQL_TOKEN("||", DoublePipe, Operator) \
  173. __ENUMERATE_SQL_TOKEN("=", Equals, Operator) \
  174. __ENUMERATE_SQL_TOKEN("==", EqualsEquals, Operator) \
  175. __ENUMERATE_SQL_TOKEN(">", GreaterThan, Operator) \
  176. __ENUMERATE_SQL_TOKEN(">=", GreaterThanEquals, Operator) \
  177. __ENUMERATE_SQL_TOKEN("<", LessThan, Operator) \
  178. __ENUMERATE_SQL_TOKEN("<=", LessThanEquals, Operator) \
  179. __ENUMERATE_SQL_TOKEN("-", Minus, Operator) \
  180. __ENUMERATE_SQL_TOKEN("%", Modulus, Operator) \
  181. __ENUMERATE_SQL_TOKEN("!=", NotEquals1, Operator) \
  182. __ENUMERATE_SQL_TOKEN("<>", NotEquals2, Operator) \
  183. __ENUMERATE_SQL_TOKEN(")", ParenClose, Punctuation) \
  184. __ENUMERATE_SQL_TOKEN("(", ParenOpen, Punctuation) \
  185. __ENUMERATE_SQL_TOKEN(".", Period, Operator) \
  186. __ENUMERATE_SQL_TOKEN("|", Pipe, Operator) \
  187. __ENUMERATE_SQL_TOKEN("+", Plus, Operator) \
  188. __ENUMERATE_SQL_TOKEN(";", SemiColon, Punctuation) \
  189. __ENUMERATE_SQL_TOKEN("<<", ShiftLeft, Operator) \
  190. __ENUMERATE_SQL_TOKEN(">>", ShiftRight, Operator) \
  191. __ENUMERATE_SQL_TOKEN("~", Tilde, Operator)
  192. enum class TokenType {
  193. #define __ENUMERATE_SQL_TOKEN(value, type, category) type,
  194. ENUMERATE_SQL_TOKENS
  195. #undef __ENUMERATE_SQL_TOKEN
  196. _COUNT_OF_TOKENS,
  197. };
  198. enum class TokenCategory {
  199. Invalid,
  200. Keyword,
  201. Identifier,
  202. Number,
  203. String,
  204. Blob,
  205. Operator,
  206. Punctuation,
  207. };
  208. struct SourcePosition {
  209. size_t line { 0 };
  210. size_t column { 0 };
  211. };
  212. class Token {
  213. public:
  214. Token(TokenType type, String value, SourcePosition start_position, SourcePosition end_position)
  215. : m_type(type)
  216. , m_value(move(value))
  217. , m_start_position(start_position)
  218. , m_end_position(end_position)
  219. {
  220. }
  221. static StringView name(TokenType);
  222. static TokenCategory category(TokenType);
  223. StringView name() const { return name(m_type); }
  224. TokenType type() const { return m_type; }
  225. TokenCategory category() const { return category(m_type); }
  226. String const& value() const { return m_value; }
  227. double double_value() const;
  228. SourcePosition const& start_position() const { return m_start_position; }
  229. SourcePosition const& end_position() const { return m_end_position; }
  230. private:
  231. TokenType m_type;
  232. String m_value;
  233. SourcePosition m_start_position;
  234. SourcePosition m_end_position;
  235. };
  236. }