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 if(lex->saved_text.length <= 20) {
87 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
88 "%s near '%s'", text, saved_text);
91 snprintf(error->text, JSON_ERROR_TEXT_LENGTH, "%s", text);
95 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
96 "%s near end of file", text);
102 snprintf(error->text, JSON_ERROR_TEXT_LENGTH, "%s", text);
107 /*** lexical analyzer ***/
109 void stream_init(stream_t *stream, get_func get, eof_func eof, void *data)
114 stream->stream_pos = 0;
115 stream->buffer[0] = '\0';
116 stream->buffer_pos = 0;
119 static char stream_get(stream_t *stream, json_error_t *error)
123 if(!stream->buffer[stream->buffer_pos])
125 stream->buffer[0] = stream->get(stream->data);
126 stream->buffer_pos = 0;
128 c = stream->buffer[0];
130 if(c == EOF && stream->eof(stream->data))
135 /* multi-byte UTF-8 sequence */
138 count = utf8_check_first(c);
144 for(i = 1; i < count; i++)
145 stream->buffer[i] = stream->get(stream->data);
147 if(!utf8_check_full(stream->buffer, count))
150 stream->stream_pos += count;
151 stream->buffer[count] = '\0';
154 stream->buffer[1] = '\0';
155 stream->stream_pos++;
159 return (char)stream->buffer[stream->buffer_pos++];
162 error_set(error, NULL, "unable to decode byte 0x%x at position %d",
163 (unsigned char)c, stream->stream_pos);
167 static void stream_unget(stream_t *stream, char c)
169 assert(stream->buffer_pos > 0);
170 stream->buffer_pos--;
171 assert(stream->buffer[stream->buffer_pos] == (unsigned char)c);
175 static int lex_get(lex_t *lex, json_error_t *error)
177 return stream_get(&lex->stream, error);
180 static int lex_eof(lex_t *lex)
182 return lex->stream.eof(lex->stream.data);
185 static void lex_save(lex_t *lex, char c)
187 strbuffer_append_byte(&lex->saved_text, c);
190 static int lex_get_save(lex_t *lex, json_error_t *error)
192 char c = stream_get(&lex->stream, error);
197 static void lex_unget_unsave(lex_t *lex, char c)
201 stream_unget(&lex->stream, c);
202 d = strbuffer_pop(&lex->saved_text);
206 static void lex_save_cached(lex_t *lex)
208 while(lex->stream.buffer[lex->stream.buffer_pos] != '\0')
210 lex_save(lex, lex->stream.buffer[lex->stream.buffer_pos]);
211 lex->stream.buffer_pos++;
215 /* assumes that str points to 'u' plus at least 4 valid hex digits */
216 static int decode_unicode_escape(const char *str)
221 assert(str[0] == 'u');
223 for(i = 1; i <= 4; i++) {
229 value += c - 'a' + 10;
231 value += c - 'A' + 10;
239 static void lex_scan_string(lex_t *lex, json_error_t *error)
246 lex->token = TOKEN_INVALID;
249 c = lex_get_save(lex, error);
254 error_set(error, lex, "premature end of input");
258 else if(0 <= c && c <= 0x1F) {
259 /* control character */
260 lex_unget_unsave(lex, c);
262 error_set(error, lex, "unexpected newline", c);
264 error_set(error, lex, "control character 0x%x", c);
269 c = lex_get_save(lex, error);
271 c = lex_get_save(lex, error);
272 for(i = 0; i < 4; i++) {
274 lex_unget_unsave(lex, c);
275 error_set(error, lex, "invalid escape");
278 c = lex_get_save(lex, error);
281 else if(c == '"' || c == '\\' || c == '/' || c == 'b' ||
282 c == 'f' || c == 'n' || c == 'r' || c == 't')
283 c = lex_get_save(lex, error);
285 lex_unget_unsave(lex, c);
286 error_set(error, lex, "invalid escape");
291 c = lex_get_save(lex, error);
294 /* the actual value is at most of the same length as the source
296 - shortcut escapes (e.g. "\t") (length 2) are converted to 1 byte
297 - a single \uXXXX escape (length 6) is converted to at most 3 bytes
298 - two \uXXXX escapes (length 12) forming an UTF-16 surrogate pair
299 are converted to 4 bytes
301 lex->value.string = malloc(lex->saved_text.length + 1);
302 if(!lex->value.string) {
303 /* this is not very nice, since TOKEN_INVALID is returned */
308 t = lex->value.string;
310 /* + 1 to skip the " */
311 p = strbuffer_value(&lex->saved_text) + 1;
321 value = decode_unicode_escape(p);
324 if(0xD800 <= value && value <= 0xDBFF) {
326 if(*p == '\\' && *(p + 1) == 'u') {
327 int value2 = decode_unicode_escape(++p);
330 if(0xDC00 <= value2 && value2 <= 0xDFFF) {
331 /* valid second surrogate */
332 value = ((value - 0xD800) << 10) +
337 /* invalid second surrogate */
338 error_set(error, lex,
339 "invalid Unicode '\\u%04X\\u%04X'",
345 /* no second surrogate */
346 error_set(error, lex, "invalid Unicode '\\u%04X'",
351 else if(0xDC00 <= value && value <= 0xDFFF) {
352 error_set(error, lex, "invalid Unicode '\\u%04X'", value);
357 error_set(error, lex, "\\u0000 is not allowed");
361 if(utf8_encode(value, buffer, &length))
364 memcpy(t, buffer, length);
369 case '"': case '\\': case '/':
371 case 'b': *t = '\b'; break;
372 case 'f': *t = '\f'; break;
373 case 'n': *t = '\n'; break;
374 case 'r': *t = '\r'; break;
375 case 't': *t = '\t'; break;
386 lex->token = TOKEN_STRING;
392 static void lex_scan_number(lex_t *lex, char c, json_error_t *error)
394 const char *saved_text;
397 lex->token = TOKEN_INVALID;
400 c = lex_get_save(lex, error);
403 c = lex_get_save(lex, error);
405 lex_unget_unsave(lex, c);
409 else /* c != '0' */ {
410 c = lex_get_save(lex, error);
412 c = lex_get_save(lex, error);
415 if(c != '.' && c != 'E' && c != 'e') {
416 lex_unget_unsave(lex, c);
417 lex->token = TOKEN_INTEGER;
419 saved_text = strbuffer_value(&lex->saved_text);
420 lex->value.integer = strtol(saved_text, &end, 10);
421 assert(end == saved_text + lex->saved_text.length);
427 c = lex_get(lex, error);
432 c = lex_get_save(lex, error);
434 c = lex_get_save(lex, error);
437 if(c == 'E' || c == 'e') {
438 c = lex_get_save(lex, error);
439 if(c == '+' || c == '-')
440 c = lex_get_save(lex, error);
443 lex_unget_unsave(lex, c);
447 c = lex_get_save(lex, error);
449 c = lex_get_save(lex, error);
452 lex_unget_unsave(lex, c);
453 lex->token = TOKEN_REAL;
455 saved_text = strbuffer_value(&lex->saved_text);
456 lex->value.real = strtod(saved_text, &end);
457 assert(end == saved_text + lex->saved_text.length);
463 static int lex_scan(lex_t *lex, json_error_t *error)
467 strbuffer_clear(&lex->saved_text);
469 if(lex->token == TOKEN_STRING) {
470 free(lex->value.string);
471 lex->value.string = NULL;
474 c = lex_get(lex, error);
475 while(c == ' ' || c == '\t' || c == '\n' || c == '\r')
480 c = lex_get(lex, error);
485 lex->token = TOKEN_EOF;
487 lex->token = TOKEN_INVALID;
493 if(c == '{' || c == '}' || c == '[' || c == ']' || c == ':' || c == ',')
497 lex_scan_string(lex, error);
499 else if(isdigit(c) || c == '-')
500 lex_scan_number(lex, c, error);
502 else if(isupper(c) || islower(c)) {
503 /* eat up the whole identifier for clearer error messages */
504 const char *saved_text;
506 c = lex_get_save(lex, error);
507 while(isupper(c) || islower(c))
508 c = lex_get_save(lex, error);
509 lex_unget_unsave(lex, c);
511 saved_text = strbuffer_value(&lex->saved_text);
513 if(strcmp(saved_text, "true") == 0)
514 lex->token = TOKEN_TRUE;
515 else if(strcmp(saved_text, "false") == 0)
516 lex->token = TOKEN_FALSE;
517 else if(strcmp(saved_text, "null") == 0)
518 lex->token = TOKEN_NULL;
520 lex->token = TOKEN_INVALID;
524 /* save the rest of the input UTF-8 sequence to get an error
525 message of valid UTF-8 */
526 lex_save_cached(lex);
527 lex->token = TOKEN_INVALID;
534 static int lex_init(lex_t *lex, get_func get, eof_func eof, void *data)
536 stream_init(&lex->stream, get, eof, data);
537 if(strbuffer_init(&lex->saved_text))
540 lex->token = TOKEN_INVALID;
546 static void lex_close(lex_t *lex)
548 if(lex->token == TOKEN_STRING)
549 free(lex->value.string);
555 static json_t *parse_value(lex_t *lex, json_error_t *error);
557 static json_t *parse_object(lex_t *lex, json_error_t *error)
559 json_t *object = json_object();
563 lex_scan(lex, error);
564 if(lex->token == '}')
571 if(lex->token != TOKEN_STRING) {
572 error_set(error, lex, "string or '}' expected");
576 key = strdup(lex->value.string);
580 lex_scan(lex, error);
581 if(lex->token != ':') {
583 error_set(error, lex, "':' expected");
587 lex_scan(lex, error);
588 value = parse_value(lex, error);
594 if(json_object_set_nocheck(object, key, value)) {
603 lex_scan(lex, error);
604 if(lex->token != ',')
607 lex_scan(lex, error);
610 if(lex->token != '}') {
611 error_set(error, lex, "'}' expected");
622 static json_t *parse_array(lex_t *lex, json_error_t *error)
624 json_t *array = json_array();
628 lex_scan(lex, error);
629 if(lex->token == ']')
633 json_t *elem = parse_value(lex, error);
637 if(json_array_append(array, elem)) {
643 lex_scan(lex, error);
644 if(lex->token != ',')
647 lex_scan(lex, error);
650 if(lex->token != ']') {
651 error_set(error, lex, "']' expected");
662 static json_t *parse_value(lex_t *lex, json_error_t *error)
668 json = json_string_nocheck(lex->value.string);
672 case TOKEN_INTEGER: {
673 json = json_integer(lex->value.integer);
678 json = json_real(lex->value.real);
695 json = parse_object(lex, error);
699 json = parse_array(lex, error);
703 error_set(error, lex, "invalid token");
707 error_set(error, lex, "unexpected token");
717 json_t *parse_json(lex_t *lex, json_error_t *error)
721 lex_scan(lex, error);
722 if(lex->token != '[' && lex->token != '{') {
723 error_set(error, lex, "'[' or '{' expected");
727 return parse_value(lex, error);
730 json_t *json_load(const char *path, json_error_t *error)
735 fp = fopen(path, "r");
738 error_set(error, NULL, "unable to open %s: %s",
739 path, strerror(errno));
743 result = json_loadf(fp, error);
755 static int string_get(void *data)
758 string_data_t *stream = (string_data_t *)data;
759 c = stream->data[stream->pos++];
766 static int string_eof(void *data)
768 string_data_t *stream = (string_data_t *)data;
769 return (stream->data[stream->pos] == '\0');
772 json_t *json_loads(const char *string, json_error_t *error)
777 string_data_t stream_data = {
782 if(lex_init(&lex, string_get, string_eof, (void *)&stream_data))
785 result = parse_json(&lex, error);
789 lex_scan(&lex, error);
790 if(lex.token != TOKEN_EOF) {
791 error_set(error, &lex, "end of file expected");
801 json_t *json_loadf(FILE *input, json_error_t *error)
806 if(lex_init(&lex, (get_func)fgetc, (eof_func)feof, input))
809 result = parse_json(&lex, error);