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