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