12 #include "jansson_private.h"
13 #include "strbuffer.h"
16 #define TOKEN_INVALID -1
18 #define TOKEN_STRING 256
19 #define TOKEN_INTEGER 257
20 #define TOKEN_REAL 258
21 #define TOKEN_TRUE 259
22 #define TOKEN_FALSE 260
23 #define TOKEN_NULL 261
25 /* read one byte from stream, return EOF on end of file */
26 typedef int (*get_func)(void *data);
28 /* return non-zero if end of file has been reached */
29 typedef int (*eof_func)(void *data);
43 strbuffer_t saved_text;
54 /*** error reporting ***/
56 static void error_init(json_error_t *error)
60 error->text[0] = '\0';
65 static void error_set(json_error_t *error, const lex_t *lex,
69 char text[JSON_ERROR_TEXT_LENGTH];
71 if(!error || error->text[0] != '\0') {
72 /* error already set */
77 vsnprintf(text, JSON_ERROR_TEXT_LENGTH, msg, ap);
82 const char *saved_text = strbuffer_value(&lex->saved_text);
83 error->line = lex->line;
84 if(saved_text && saved_text[0])
86 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
87 "%s near '%s'", text, saved_text);
91 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
92 "%s near end of file", text);
98 snprintf(error->text, JSON_ERROR_TEXT_LENGTH, "%s", text);
103 /*** lexical analyzer ***/
105 void stream_init(stream_t *stream, get_func get, eof_func eof, void *data)
110 stream->stream_pos = 0;
111 stream->buffer[0] = '\0';
112 stream->buffer_pos = 0;
115 static char stream_get(stream_t *stream, json_error_t *error)
119 if(!stream->buffer[stream->buffer_pos])
121 stream->buffer[0] = stream->get(stream->data);
122 stream->buffer_pos = 0;
124 c = stream->buffer[0];
126 if(c == EOF && stream->eof(stream->data))
131 /* multi-byte UTF-8 sequence */
134 count = utf8_check_first(c);
140 for(i = 1; i < count; i++)
141 stream->buffer[i] = stream->get(stream->data);
143 if(!utf8_check_full(stream->buffer, count))
146 stream->stream_pos += count;
147 stream->buffer[count] = '\0';
150 stream->buffer[1] = '\0';
151 stream->stream_pos++;
155 return (char)stream->buffer[stream->buffer_pos++];
158 error_set(error, NULL, "unable to decode byte 0x%x at position %d",
159 (unsigned char)c, stream->stream_pos);
163 static void stream_unget(stream_t *stream, char c)
165 assert(stream->buffer_pos > 0);
166 stream->buffer_pos--;
167 assert(stream->buffer[stream->buffer_pos] == (unsigned char)c);
171 static int lex_get(lex_t *lex, json_error_t *error)
173 return stream_get(&lex->stream, error);
176 static int lex_eof(lex_t *lex)
178 return lex->stream.eof(lex->stream.data);
181 static void lex_save(lex_t *lex, char c)
183 strbuffer_append_byte(&lex->saved_text, c);
186 static int lex_get_save(lex_t *lex, json_error_t *error)
188 char c = stream_get(&lex->stream, error);
193 static void lex_unget_unsave(lex_t *lex, char c)
197 stream_unget(&lex->stream, c);
198 d = strbuffer_pop(&lex->saved_text);
202 static void lex_scan_string(lex_t *lex, json_error_t *error)
208 lex->token = TOKEN_INVALID;
211 c = lex_get_save(lex, error);
216 error_set(error, lex, "premature end of input");
220 else if(0 <= c && c <= 0x1F) {
221 /* control character */
222 lex_unget_unsave(lex, c);
224 error_set(error, lex, "unexpected newline", c);
226 error_set(error, lex, "control character 0x%x", c);
231 c = lex_get_save(lex, error);
233 c = lex_get_save(lex, error);
234 for(int i = 0; i < 4; i++) {
236 lex_unget_unsave(lex, c);
237 error_set(error, lex, "invalid escape");
240 c = lex_get_save(lex, error);
243 else if(c == '"' || c == '\\' || c == '/' || c == 'b' ||
244 c == 'f' || c == 'n' || c == 'r' || c == 't')
245 c = lex_get_save(lex, error);
247 lex_unget_unsave(lex, c);
248 error_set(error, lex, "invalid escape");
253 c = lex_get_save(lex, error);
256 /* the actual value is at most of the same length as the source
258 - shortcut escapes (e.g. "\t") (length 2) are converted to 1 byte
259 - a single \uXXXX escape (length 6) is converted to at most 3 bytes
260 - two \uXXXX escapes (length 12) forming an UTF-16 surrogate pair
261 are converted to 4 bytes
263 lex->value.string = malloc(lex->saved_text.length + 1);
264 if(!lex->value.string) {
265 /* this is not very nice, since TOKEN_INVALID is returned */
270 t = lex->value.string;
272 /* + 1 to skip the " */
273 p = strbuffer_value(&lex->saved_text) + 1;
280 error_set(error, lex, "\\u escapes are not yet supported");
281 free(lex->value.string);
282 lex->value.string = NULL;
286 case '"': case '\\': case '/':
288 case 'b': *t = '\b'; break;
289 case 'f': *t = '\f'; break;
290 case 'n': *t = '\n'; break;
291 case 'r': *t = '\r'; break;
292 case 't': *t = '\t'; break;
304 lex->token = TOKEN_STRING;
310 static void lex_scan_number(lex_t *lex, char c, json_error_t *error)
312 const char *saved_text;
315 lex->token = TOKEN_INVALID;
318 c = lex_get_save(lex, error);
321 c = lex_get_save(lex, error);
323 lex_unget_unsave(lex, c);
327 else /* c != '0' */ {
328 c = lex_get_save(lex, error);
330 c = lex_get_save(lex, error);
333 if(c != '.' && c != 'E' && c != 'e') {
334 lex_unget_unsave(lex, c);
335 lex->token = TOKEN_INTEGER;
337 saved_text = strbuffer_value(&lex->saved_text);
338 lex->value.integer = strtol(saved_text, &end, 10);
339 assert(end == saved_text + lex->saved_text.length);
345 c = lex_get(lex, error);
350 c = lex_get_save(lex, error);
352 c = lex_get_save(lex, error);
355 if(c == 'E' || c == 'e') {
356 c = lex_get_save(lex, error);
357 if(c == '+' || c == '-')
358 c = lex_get_save(lex, error);
361 lex_unget_unsave(lex, c);
365 c = lex_get_save(lex, error);
367 c = lex_get_save(lex, error);
370 lex_unget_unsave(lex, c);
371 lex->token = TOKEN_REAL;
373 saved_text = strbuffer_value(&lex->saved_text);
374 lex->value.real = strtod(saved_text, &end);
375 assert(end == saved_text + lex->saved_text.length);
381 static int lex_scan(lex_t *lex, json_error_t *error)
385 strbuffer_clear(&lex->saved_text);
387 if(lex->token == TOKEN_STRING) {
388 free(lex->value.string);
389 lex->value.string = NULL;
392 c = lex_get(lex, error);
393 while(c == ' ' || c == '\t' || c == '\n' || c == '\r')
398 c = lex_get(lex, error);
403 lex->token = TOKEN_EOF;
405 lex->token = TOKEN_INVALID;
411 if(c == '{' || c == '}' || c == '[' || c == ']' || c == ':' || c == ',')
415 lex_scan_string(lex, error);
417 else if(isdigit(c) || c == '-')
418 lex_scan_number(lex, c, error);
420 else if(isupper(c) || islower(c)) {
421 /* eat up the whole identifier for clearer error messages */
422 const char *saved_text;
424 c = lex_get_save(lex, error);
425 while(isupper(c) || islower(c))
426 c = lex_get_save(lex, error);
427 lex_unget_unsave(lex, c);
429 saved_text = strbuffer_value(&lex->saved_text);
431 if(strcmp(saved_text, "true") == 0)
432 lex->token = TOKEN_TRUE;
433 else if(strcmp(saved_text, "false") == 0)
434 lex->token = TOKEN_FALSE;
435 else if(strcmp(saved_text, "null") == 0)
436 lex->token = TOKEN_NULL;
438 lex->token = TOKEN_INVALID;
442 lex->token = TOKEN_INVALID;
448 static int lex_init(lex_t *lex, get_func get, eof_func eof, void *data)
450 stream_init(&lex->stream, get, eof, data);
451 if(strbuffer_init(&lex->saved_text))
454 lex->token = TOKEN_INVALID;
460 static void lex_close(lex_t *lex)
462 if(lex->token == TOKEN_STRING)
463 free(lex->value.string);
469 static json_t *parse_value(lex_t *lex, json_error_t *error);
471 static json_t *parse_object(lex_t *lex, json_error_t *error)
473 json_t *object = json_object();
477 lex_scan(lex, error);
478 if(lex->token == '}')
485 if(lex->token != TOKEN_STRING) {
486 error_set(error, lex, "string or '}' expected");
490 key = strdup(lex->value.string);
494 lex_scan(lex, error);
495 if(lex->token != ':') {
497 error_set(error, lex, "':' expected");
501 lex_scan(lex, error);
502 value = parse_value(lex, error);
508 if(json_object_set_nocheck(object, key, value)) {
517 lex_scan(lex, error);
518 if(lex->token != ',')
521 lex_scan(lex, error);
524 if(lex->token != '}') {
525 error_set(error, lex, "'}' expected");
536 static json_t *parse_array(lex_t *lex, json_error_t *error)
538 json_t *array = json_array();
542 lex_scan(lex, error);
543 if(lex->token == ']')
547 json_t *elem = parse_value(lex, error);
551 if(json_array_append(array, elem)) {
557 lex_scan(lex, error);
558 if(lex->token != ',')
561 lex_scan(lex, error);
564 if(lex->token != ']') {
565 error_set(error, lex, "']' expected");
576 static json_t *parse_value(lex_t *lex, json_error_t *error)
582 json = json_string_nocheck(lex->value.string);
586 case TOKEN_INTEGER: {
587 json = json_integer(lex->value.integer);
592 json = json_real(lex->value.real);
609 json = parse_object(lex, error);
613 json = parse_array(lex, error);
617 error_set(error, lex, "invalid token");
621 error_set(error, lex, "unexpected token");
631 json_t *parse_json(lex_t *lex, json_error_t *error)
635 lex_scan(lex, error);
636 if(lex->token != '[' && lex->token != '{') {
637 error_set(error, lex, "'[' or '{' expected");
641 return parse_value(lex, error);
644 json_t *json_load(const char *path, json_error_t *error)
649 fp = fopen(path, "r");
652 error_set(error, NULL, "unable to open %s: %s",
653 path, strerror(errno));
657 result = json_loadf(fp, error);
669 static int string_get(void *data)
672 string_data_t *stream = (string_data_t *)data;
673 c = stream->data[stream->pos++];
680 static int string_eof(void *data)
682 string_data_t *stream = (string_data_t *)data;
683 return (stream->data[stream->pos] == '\0');
686 json_t *json_loads(const char *string, json_error_t *error)
691 string_data_t stream_data = {
696 if(lex_init(&lex, string_get, string_eof, (void *)&stream_data))
699 result = parse_json(&lex, error);
703 lex_scan(&lex, error);
704 if(lex.token != TOKEN_EOF) {
705 error_set(error, &lex, "end of file expected");
715 json_t *json_loadf(FILE *input, json_error_t *error)
720 if(lex_init(&lex, (get_func)fgetc, (eof_func)feof, input))
723 result = parse_json(&lex, error);