plum
git clone https://git.pyrossh.dev/plum
A statically typed, imperative programming language inspired by rust, python
plum-tooling/tree-sitter-plum/src/tree_sitter/parser.h
| 4bd2c1c | 1 | #ifndef TREE_SITTER_PARSER_H_ |
| 4bd2c1c | 2 | #define TREE_SITTER_PARSER_H_ |
| 4bd2c1c | 3 | |
| 4bd2c1c | 4 | #ifdef __cplusplus |
| 4bd2c1c | 5 | extern "C" { |
| 4bd2c1c | 6 | #endif |
| 4bd2c1c | 7 | |
| 4bd2c1c | 8 | #include <stdbool.h> |
| 4bd2c1c | 9 | #include <stdint.h> |
| 4bd2c1c | 10 | #include <stdlib.h> |
| 4bd2c1c | 11 | |
| 4bd2c1c | 12 | #define ts_builtin_sym_error ((TSSymbol)-1) |
| 4bd2c1c | 13 | #define ts_builtin_sym_end 0 |
| 4bd2c1c | 14 | #define TREE_SITTER_SERIALIZATION_BUFFER_SIZE 1024 |
| 4bd2c1c | 15 | |
| 4bd2c1c | 16 | #ifndef TREE_SITTER_API_H_ |
| 5d12d71 | 17 | typedef uint16_t TSStateId; |
| 4bd2c1c | 18 | typedef uint16_t TSSymbol; |
| 4bd2c1c | 19 | typedef uint16_t TSFieldId; |
| 4bd2c1c | 20 | typedef struct TSLanguage TSLanguage; |
| f502d22 | 21 | typedef struct TSLanguageMetadata { |
| f502d22 | 22 | uint8_t major_version; |
| f502d22 | 23 | uint8_t minor_version; |
| f502d22 | 24 | uint8_t patch_version; |
| f502d22 | 25 | } TSLanguageMetadata; |
| 4bd2c1c | 26 | #endif |
| 4bd2c1c | 27 | |
| 4bd2c1c | 28 | typedef struct { |
| 4bd2c1c | 29 | TSFieldId field_id; |
| 4bd2c1c | 30 | uint8_t child_index; |
| 4bd2c1c | 31 | bool inherited; |
| 4bd2c1c | 32 | } TSFieldMapEntry; |
| 4bd2c1c | 33 | |
| f502d22 | 34 | // Used to index the field and supertype maps. |
| 4bd2c1c | 35 | typedef struct { |
| 4bd2c1c | 36 | uint16_t index; |
| 4bd2c1c | 37 | uint16_t length; |
| f502d22 | 38 | } TSMapSlice; |
| 4bd2c1c | 39 | |
| 4bd2c1c | 40 | typedef struct { |
| 4bd2c1c | 41 | bool visible; |
| 4bd2c1c | 42 | bool named; |
| 4bd2c1c | 43 | bool supertype; |
| 4bd2c1c | 44 | } TSSymbolMetadata; |
| 4bd2c1c | 45 | |
| 4bd2c1c | 46 | typedef struct TSLexer TSLexer; |
| 4bd2c1c | 47 | |
| 4bd2c1c | 48 | struct TSLexer { |
| 4bd2c1c | 49 | int32_t lookahead; |
| 4bd2c1c | 50 | TSSymbol result_symbol; |
| 4bd2c1c | 51 | void (*advance)(TSLexer *, bool); |
| 4bd2c1c | 52 | void (*mark_end)(TSLexer *); |
| 4bd2c1c | 53 | uint32_t (*get_column)(TSLexer *); |
| 4bd2c1c | 54 | bool (*is_at_included_range_start)(const TSLexer *); |
| 4bd2c1c | 55 | bool (*eof)(const TSLexer *); |
| 0d43379 | 56 | void (*log)(const TSLexer *, const char *, ...); |
| 4bd2c1c | 57 | }; |
| 4bd2c1c | 58 | |
| 4bd2c1c | 59 | typedef enum { |
| 4bd2c1c | 60 | TSParseActionTypeShift, |
| 4bd2c1c | 61 | TSParseActionTypeReduce, |
| 4bd2c1c | 62 | TSParseActionTypeAccept, |
| 4bd2c1c | 63 | TSParseActionTypeRecover, |
| 4bd2c1c | 64 | } TSParseActionType; |
| 4bd2c1c | 65 | |
| 4bd2c1c | 66 | typedef union { |
| 4bd2c1c | 67 | struct { |
| 4bd2c1c | 68 | uint8_t type; |
| 4bd2c1c | 69 | TSStateId state; |
| 4bd2c1c | 70 | bool extra; |
| 4bd2c1c | 71 | bool repetition; |
| 4bd2c1c | 72 | } shift; |
| 4bd2c1c | 73 | struct { |
| 4bd2c1c | 74 | uint8_t type; |
| 4bd2c1c | 75 | uint8_t child_count; |
| 4bd2c1c | 76 | TSSymbol symbol; |
| 4bd2c1c | 77 | int16_t dynamic_precedence; |
| 4bd2c1c | 78 | uint16_t production_id; |
| 4bd2c1c | 79 | } reduce; |
| 4bd2c1c | 80 | uint8_t type; |
| 4bd2c1c | 81 | } TSParseAction; |
| 4bd2c1c | 82 | |
| 4bd2c1c | 83 | typedef struct { |
| 4bd2c1c | 84 | uint16_t lex_state; |
| 4bd2c1c | 85 | uint16_t external_lex_state; |
| 4bd2c1c | 86 | } TSLexMode; |
| 4bd2c1c | 87 | |
| f502d22 | 88 | typedef struct { |
| f502d22 | 89 | uint16_t lex_state; |
| f502d22 | 90 | uint16_t external_lex_state; |
| f502d22 | 91 | uint16_t reserved_word_set_id; |
| f502d22 | 92 | } TSLexerMode; |
| f502d22 | 93 | |
| 4bd2c1c | 94 | typedef union { |
| 4bd2c1c | 95 | TSParseAction action; |
| 4bd2c1c | 96 | struct { |
| 4bd2c1c | 97 | uint8_t count; |
| 4bd2c1c | 98 | bool reusable; |
| 4bd2c1c | 99 | } entry; |
| 4bd2c1c | 100 | } TSParseActionEntry; |
| 4bd2c1c | 101 | |
| 5d12d71 | 102 | typedef struct { |
| 5d12d71 | 103 | int32_t start; |
| 5d12d71 | 104 | int32_t end; |
| 5d12d71 | 105 | } TSCharacterRange; |
| 5d12d71 | 106 | |
| 4bd2c1c | 107 | struct TSLanguage { |
| f502d22 | 108 | uint32_t abi_version; |
| 4bd2c1c | 109 | uint32_t symbol_count; |
| 4bd2c1c | 110 | uint32_t alias_count; |
| 4bd2c1c | 111 | uint32_t token_count; |
| 4bd2c1c | 112 | uint32_t external_token_count; |
| 4bd2c1c | 113 | uint32_t state_count; |
| 4bd2c1c | 114 | uint32_t large_state_count; |
| 4bd2c1c | 115 | uint32_t production_id_count; |
| 4bd2c1c | 116 | uint32_t field_count; |
| 4bd2c1c | 117 | uint16_t max_alias_sequence_length; |
| 4bd2c1c | 118 | const uint16_t *parse_table; |
| 4bd2c1c | 119 | const uint16_t *small_parse_table; |
| 4bd2c1c | 120 | const uint32_t *small_parse_table_map; |
| 4bd2c1c | 121 | const TSParseActionEntry *parse_actions; |
| 4bd2c1c | 122 | const char * const *symbol_names; |
| 4bd2c1c | 123 | const char * const *field_names; |
| f502d22 | 124 | const TSMapSlice *field_map_slices; |
| 4bd2c1c | 125 | const TSFieldMapEntry *field_map_entries; |
| 4bd2c1c | 126 | const TSSymbolMetadata *symbol_metadata; |
| 4bd2c1c | 127 | const TSSymbol *public_symbol_map; |
| 4bd2c1c | 128 | const uint16_t *alias_map; |
| 4bd2c1c | 129 | const TSSymbol *alias_sequences; |
| f502d22 | 130 | const TSLexerMode *lex_modes; |
| 4bd2c1c | 131 | bool (*lex_fn)(TSLexer *, TSStateId); |
| 4bd2c1c | 132 | bool (*keyword_lex_fn)(TSLexer *, TSStateId); |
| 4bd2c1c | 133 | TSSymbol keyword_capture_token; |
| 4bd2c1c | 134 | struct { |
| 4bd2c1c | 135 | const bool *states; |
| 4bd2c1c | 136 | const TSSymbol *symbol_map; |
| 4bd2c1c | 137 | void *(*create)(void); |
| 4bd2c1c | 138 | void (*destroy)(void *); |
| 4bd2c1c | 139 | bool (*scan)(void *, TSLexer *, const bool *symbol_whitelist); |
| 4bd2c1c | 140 | unsigned (*serialize)(void *, char *); |
| 4bd2c1c | 141 | void (*deserialize)(void *, const char *, unsigned); |
| 4bd2c1c | 142 | } external_scanner; |
| 4bd2c1c | 143 | const TSStateId *primary_state_ids; |
| f502d22 | 144 | const char *name; |
| f502d22 | 145 | const TSSymbol *reserved_words; |
| f502d22 | 146 | uint16_t max_reserved_word_set_size; |
| f502d22 | 147 | uint32_t supertype_count; |
| f502d22 | 148 | const TSSymbol *supertype_symbols; |
| f502d22 | 149 | const TSMapSlice *supertype_map_slices; |
| f502d22 | 150 | const TSSymbol *supertype_map_entries; |
| f502d22 | 151 | TSLanguageMetadata metadata; |
| 4bd2c1c | 152 | }; |
| 4bd2c1c | 153 | |
| f502d22 | 154 | static inline bool set_contains(const TSCharacterRange *ranges, uint32_t len, int32_t lookahead) { |
| 5d12d71 | 155 | uint32_t index = 0; |
| 5d12d71 | 156 | uint32_t size = len - index; |
| 5d12d71 | 157 | while (size > 1) { |
| 5d12d71 | 158 | uint32_t half_size = size / 2; |
| 5d12d71 | 159 | uint32_t mid_index = index + half_size; |
| f502d22 | 160 | const TSCharacterRange *range = &ranges[mid_index]; |
| 5d12d71 | 161 | if (lookahead >= range->start && lookahead <= range->end) { |
| 5d12d71 | 162 | return true; |
| 5d12d71 | 163 | } else if (lookahead > range->end) { |
| 5d12d71 | 164 | index = mid_index; |
| 5d12d71 | 165 | } |
| 5d12d71 | 166 | size -= half_size; |
| 5d12d71 | 167 | } |
| f502d22 | 168 | const TSCharacterRange *range = &ranges[index]; |
| 5d12d71 | 169 | return (lookahead >= range->start && lookahead <= range->end); |
| 5d12d71 | 170 | } |
| 5d12d71 | 171 | |
| 4bd2c1c | 172 | /* |
| 4bd2c1c | 173 | * Lexer Macros |
| 4bd2c1c | 174 | */ |
| 4bd2c1c | 175 | |
| 5d12d71 | 176 | #ifdef _MSC_VER |
| 5d12d71 | 177 | #define UNUSED __pragma(warning(suppress : 4101)) |
| 5d12d71 | 178 | #else |
| 5d12d71 | 179 | #define UNUSED __attribute__((unused)) |
| 5d12d71 | 180 | #endif |
| 5d12d71 | 181 | |
| 4bd2c1c | 182 | #define START_LEXER() \ |
| 4bd2c1c | 183 | bool result = false; \ |
| 4bd2c1c | 184 | bool skip = false; \ |
| 5d12d71 | 185 | UNUSED \ |
| 4bd2c1c | 186 | bool eof = false; \ |
| 4bd2c1c | 187 | int32_t lookahead; \ |
| 4bd2c1c | 188 | goto start; \ |
| 4bd2c1c | 189 | next_state: \ |
| 4bd2c1c | 190 | lexer->advance(lexer, skip); \ |
| 4bd2c1c | 191 | start: \ |
| 4bd2c1c | 192 | skip = false; \ |
| 4bd2c1c | 193 | lookahead = lexer->lookahead; |
| 4bd2c1c | 194 | |
| 4bd2c1c | 195 | #define ADVANCE(state_value) \ |
| 4bd2c1c | 196 | { \ |
| 4bd2c1c | 197 | state = state_value; \ |
| 4bd2c1c | 198 | goto next_state; \ |
| 4bd2c1c | 199 | } |
| 4bd2c1c | 200 | |
| 5d12d71 | 201 | #define ADVANCE_MAP(...) \ |
| 5d12d71 | 202 | { \ |
| 5d12d71 | 203 | static const uint16_t map[] = { __VA_ARGS__ }; \ |
| 5d12d71 | 204 | for (uint32_t i = 0; i < sizeof(map) / sizeof(map[0]); i += 2) { \ |
| 5d12d71 | 205 | if (map[i] == lookahead) { \ |
| 5d12d71 | 206 | state = map[i + 1]; \ |
| 5d12d71 | 207 | goto next_state; \ |
| 5d12d71 | 208 | } \ |
| 5d12d71 | 209 | } \ |
| 5d12d71 | 210 | } |
| 5d12d71 | 211 | |
| 4bd2c1c | 212 | #define SKIP(state_value) \ |
| 4bd2c1c | 213 | { \ |
| 4bd2c1c | 214 | skip = true; \ |
| 4bd2c1c | 215 | state = state_value; \ |
| 4bd2c1c | 216 | goto next_state; \ |
| 4bd2c1c | 217 | } |
| 4bd2c1c | 218 | |
| 4bd2c1c | 219 | #define ACCEPT_TOKEN(symbol_value) \ |
| 4bd2c1c | 220 | result = true; \ |
| 4bd2c1c | 221 | lexer->result_symbol = symbol_value; \ |
| 4bd2c1c | 222 | lexer->mark_end(lexer); |
| 4bd2c1c | 223 | |
| 4bd2c1c | 224 | #define END_STATE() return result; |
| 4bd2c1c | 225 | |
| 4bd2c1c | 226 | /* |
| 4bd2c1c | 227 | * Parse Table Macros |
| 4bd2c1c | 228 | */ |
| 4bd2c1c | 229 | |
| 5d12d71 | 230 | #define SMALL_STATE(id) ((id) - LARGE_STATE_COUNT) |
| 4bd2c1c | 231 | |
| 4bd2c1c | 232 | #define STATE(id) id |
| 4bd2c1c | 233 | |
| 4bd2c1c | 234 | #define ACTIONS(id) id |
| 4bd2c1c | 235 | |
| 4bd2c1c | 236 | #define SHIFT(state_value) \ |
| 4bd2c1c | 237 | {{ \ |
| 4bd2c1c | 238 | .shift = { \ |
| 4bd2c1c | 239 | .type = TSParseActionTypeShift, \ |
| 5d12d71 | 240 | .state = (state_value) \ |
| 4bd2c1c | 241 | } \ |
| 4bd2c1c | 242 | }} |
| 4bd2c1c | 243 | |
| 4bd2c1c | 244 | #define SHIFT_REPEAT(state_value) \ |
| 4bd2c1c | 245 | {{ \ |
| 4bd2c1c | 246 | .shift = { \ |
| 4bd2c1c | 247 | .type = TSParseActionTypeShift, \ |
| 5d12d71 | 248 | .state = (state_value), \ |
| 4bd2c1c | 249 | .repetition = true \ |
| 4bd2c1c | 250 | } \ |
| 4bd2c1c | 251 | }} |
| 4bd2c1c | 252 | |
| 4bd2c1c | 253 | #define SHIFT_EXTRA() \ |
| 4bd2c1c | 254 | {{ \ |
| 4bd2c1c | 255 | .shift = { \ |
| 4bd2c1c | 256 | .type = TSParseActionTypeShift, \ |
| 4bd2c1c | 257 | .extra = true \ |
| 4bd2c1c | 258 | } \ |
| 4bd2c1c | 259 | }} |
| 4bd2c1c | 260 | |
| 5d12d71 | 261 | #define REDUCE(symbol_name, children, precedence, prod_id) \ |
| 5d12d71 | 262 | {{ \ |
| 5d12d71 | 263 | .reduce = { \ |
| 5d12d71 | 264 | .type = TSParseActionTypeReduce, \ |
| 5d12d71 | 265 | .symbol = symbol_name, \ |
| 5d12d71 | 266 | .child_count = children, \ |
| 5d12d71 | 267 | .dynamic_precedence = precedence, \ |
| 5d12d71 | 268 | .production_id = prod_id \ |
| 5d12d71 | 269 | }, \ |
| 4bd2c1c | 270 | }} |
| 4bd2c1c | 271 | |
| 4bd2c1c | 272 | #define RECOVER() \ |
| 4bd2c1c | 273 | {{ \ |
| 4bd2c1c | 274 | .type = TSParseActionTypeRecover \ |
| 4bd2c1c | 275 | }} |
| 4bd2c1c | 276 | |
| 4bd2c1c | 277 | #define ACCEPT_INPUT() \ |
| 4bd2c1c | 278 | {{ \ |
| 4bd2c1c | 279 | .type = TSParseActionTypeAccept \ |
| 4bd2c1c | 280 | }} |
| 4bd2c1c | 281 | |
| 4bd2c1c | 282 | #ifdef __cplusplus |
| 4bd2c1c | 283 | } |
| 4bd2c1c | 284 | #endif |
| 4bd2c1c | 285 | |
| 4bd2c1c | 286 | #endif // TREE_SITTER_PARSER_H_ |