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_save_cached(lex_t *lex)
204 while(lex->stream.buffer[lex->stream.buffer_pos] != '\0')
206 lex_save(lex, lex->stream.buffer[lex->stream.buffer_pos]);
207 lex->stream.buffer_pos++;
211 static void lex_scan_string(lex_t *lex, json_error_t *error)
217 lex->token = TOKEN_INVALID;
220 c = lex_get_save(lex, error);
225 error_set(error, lex, "premature end of input");
229 else if(0 <= c && c <= 0x1F) {
230 /* control character */
231 lex_unget_unsave(lex, c);
233 error_set(error, lex, "unexpected newline", c);
235 error_set(error, lex, "control character 0x%x", c);
240 c = lex_get_save(lex, error);
242 c = lex_get_save(lex, error);
243 for(int i = 0; i < 4; i++) {
245 lex_unget_unsave(lex, c);
246 error_set(error, lex, "invalid escape");
249 c = lex_get_save(lex, error);
252 else if(c == '"' || c == '\\' || c == '/' || c == 'b' ||
253 c == 'f' || c == 'n' || c == 'r' || c == 't')
254 c = lex_get_save(lex, error);
256 lex_unget_unsave(lex, c);
257 error_set(error, lex, "invalid escape");
262 c = lex_get_save(lex, error);
265 /* the actual value is at most of the same length as the source
267 - shortcut escapes (e.g. "\t") (length 2) are converted to 1 byte
268 - a single \uXXXX escape (length 6) is converted to at most 3 bytes
269 - two \uXXXX escapes (length 12) forming an UTF-16 surrogate pair
270 are converted to 4 bytes
272 lex->value.string = malloc(lex->saved_text.length + 1);
273 if(!lex->value.string) {
274 /* this is not very nice, since TOKEN_INVALID is returned */
279 t = lex->value.string;
281 /* + 1 to skip the " */
282 p = strbuffer_value(&lex->saved_text) + 1;
289 error_set(error, lex, "\\u escapes are not yet supported");
290 free(lex->value.string);
291 lex->value.string = NULL;
295 case '"': case '\\': case '/':
297 case 'b': *t = '\b'; break;
298 case 'f': *t = '\f'; break;
299 case 'n': *t = '\n'; break;
300 case 'r': *t = '\r'; break;
301 case 't': *t = '\t'; break;
313 lex->token = TOKEN_STRING;
319 static void lex_scan_number(lex_t *lex, char c, json_error_t *error)
321 const char *saved_text;
324 lex->token = TOKEN_INVALID;
327 c = lex_get_save(lex, error);
330 c = lex_get_save(lex, error);
332 lex_unget_unsave(lex, c);
336 else /* c != '0' */ {
337 c = lex_get_save(lex, error);
339 c = lex_get_save(lex, error);
342 if(c != '.' && c != 'E' && c != 'e') {
343 lex_unget_unsave(lex, c);
344 lex->token = TOKEN_INTEGER;
346 saved_text = strbuffer_value(&lex->saved_text);
347 lex->value.integer = strtol(saved_text, &end, 10);
348 assert(end == saved_text + lex->saved_text.length);
354 c = lex_get(lex, error);
359 c = lex_get_save(lex, error);
361 c = lex_get_save(lex, error);
364 if(c == 'E' || c == 'e') {
365 c = lex_get_save(lex, error);
366 if(c == '+' || c == '-')
367 c = lex_get_save(lex, error);
370 lex_unget_unsave(lex, c);
374 c = lex_get_save(lex, error);
376 c = lex_get_save(lex, error);
379 lex_unget_unsave(lex, c);
380 lex->token = TOKEN_REAL;
382 saved_text = strbuffer_value(&lex->saved_text);
383 lex->value.real = strtod(saved_text, &end);
384 assert(end == saved_text + lex->saved_text.length);
390 static int lex_scan(lex_t *lex, json_error_t *error)
394 strbuffer_clear(&lex->saved_text);
396 if(lex->token == TOKEN_STRING) {
397 free(lex->value.string);
398 lex->value.string = NULL;
401 c = lex_get(lex, error);
402 while(c == ' ' || c == '\t' || c == '\n' || c == '\r')
407 c = lex_get(lex, error);
412 lex->token = TOKEN_EOF;
414 lex->token = TOKEN_INVALID;
420 if(c == '{' || c == '}' || c == '[' || c == ']' || c == ':' || c == ',')
424 lex_scan_string(lex, error);
426 else if(isdigit(c) || c == '-')
427 lex_scan_number(lex, c, error);
429 else if(isupper(c) || islower(c)) {
430 /* eat up the whole identifier for clearer error messages */
431 const char *saved_text;
433 c = lex_get_save(lex, error);
434 while(isupper(c) || islower(c))
435 c = lex_get_save(lex, error);
436 lex_unget_unsave(lex, c);
438 saved_text = strbuffer_value(&lex->saved_text);
440 if(strcmp(saved_text, "true") == 0)
441 lex->token = TOKEN_TRUE;
442 else if(strcmp(saved_text, "false") == 0)
443 lex->token = TOKEN_FALSE;
444 else if(strcmp(saved_text, "null") == 0)
445 lex->token = TOKEN_NULL;
447 lex->token = TOKEN_INVALID;
451 /* save the rest of the input UTF-8 sequence to get an error
452 message of valid UTF-8 */
453 lex_save_cached(lex);
454 lex->token = TOKEN_INVALID;
461 static int lex_init(lex_t *lex, get_func get, eof_func eof, void *data)
463 stream_init(&lex->stream, get, eof, data);
464 if(strbuffer_init(&lex->saved_text))
467 lex->token = TOKEN_INVALID;
473 static void lex_close(lex_t *lex)
475 if(lex->token == TOKEN_STRING)
476 free(lex->value.string);
482 static json_t *parse_value(lex_t *lex, json_error_t *error);
484 static json_t *parse_object(lex_t *lex, json_error_t *error)
486 json_t *object = json_object();
490 lex_scan(lex, error);
491 if(lex->token == '}')
498 if(lex->token != TOKEN_STRING) {
499 error_set(error, lex, "string or '}' expected");
503 key = strdup(lex->value.string);
507 lex_scan(lex, error);
508 if(lex->token != ':') {
510 error_set(error, lex, "':' expected");
514 lex_scan(lex, error);
515 value = parse_value(lex, error);
521 if(json_object_set_nocheck(object, key, value)) {
530 lex_scan(lex, error);
531 if(lex->token != ',')
534 lex_scan(lex, error);
537 if(lex->token != '}') {
538 error_set(error, lex, "'}' expected");
549 static json_t *parse_array(lex_t *lex, json_error_t *error)
551 json_t *array = json_array();
555 lex_scan(lex, error);
556 if(lex->token == ']')
560 json_t *elem = parse_value(lex, error);
564 if(json_array_append(array, elem)) {
570 lex_scan(lex, error);
571 if(lex->token != ',')
574 lex_scan(lex, error);
577 if(lex->token != ']') {
578 error_set(error, lex, "']' expected");
589 static json_t *parse_value(lex_t *lex, json_error_t *error)
595 json = json_string_nocheck(lex->value.string);
599 case TOKEN_INTEGER: {
600 json = json_integer(lex->value.integer);
605 json = json_real(lex->value.real);
622 json = parse_object(lex, error);
626 json = parse_array(lex, error);
630 error_set(error, lex, "invalid token");
634 error_set(error, lex, "unexpected token");
644 json_t *parse_json(lex_t *lex, json_error_t *error)
648 lex_scan(lex, error);
649 if(lex->token != '[' && lex->token != '{') {
650 error_set(error, lex, "'[' or '{' expected");
654 return parse_value(lex, error);
657 json_t *json_load(const char *path, json_error_t *error)
662 fp = fopen(path, "r");
665 error_set(error, NULL, "unable to open %s: %s",
666 path, strerror(errno));
670 result = json_loadf(fp, error);
682 static int string_get(void *data)
685 string_data_t *stream = (string_data_t *)data;
686 c = stream->data[stream->pos++];
693 static int string_eof(void *data)
695 string_data_t *stream = (string_data_t *)data;
696 return (stream->data[stream->pos] == '\0');
699 json_t *json_loads(const char *string, json_error_t *error)
704 string_data_t stream_data = {
709 if(lex_init(&lex, string_get, string_eof, (void *)&stream_data))
712 result = parse_json(&lex, error);
716 lex_scan(&lex, error);
717 if(lex.token != TOKEN_EOF) {
718 error_set(error, &lex, "end of file expected");
728 json_t *json_loadf(FILE *input, json_error_t *error)
733 if(lex_init(&lex, (get_func)fgetc, (eof_func)feof, input))
736 result = parse_json(&lex, error);