2
* Copyright (c) 2009-2011 Petri Lehtinen <petri@digip.org>
4
* Jansson is free software; you can redistribute it and/or modify
5
* it under the terms of the MIT license. See LICENSE for details.
18
#include "jansson_private.h"
19
#include "strbuffer.h"
22
#define STREAM_STATE_OK 0
23
#define STREAM_STATE_EOF -1
24
#define STREAM_STATE_ERROR -2
26
#define TOKEN_INVALID -1
28
#define TOKEN_STRING 256
29
#define TOKEN_INTEGER 257
30
#define TOKEN_REAL 258
31
#define TOKEN_TRUE 259
32
#define TOKEN_FALSE 260
33
#define TOKEN_NULL 261
35
/* Read one byte from stream, convert to unsigned char, then int, and
36
return. return EOF on end of file. This corresponds to the
37
behaviour of fgetc(). */
38
typedef int (*get_func)(void *data);
47
int column, last_column;
53
strbuffer_t saved_text;
62
#define stream_to_lex(stream) container_of(stream, lex_t, stream)
65
/*** error reporting ***/
67
static void error_set(json_error_t *error, const lex_t *lex,
71
char msg_text[JSON_ERROR_TEXT_LENGTH];
73
int line = -1, col = -1;
75
const char *result = msg_text;
81
vsnprintf(msg_text, JSON_ERROR_TEXT_LENGTH, msg, ap);
86
const char *saved_text = strbuffer_value(&lex->saved_text);
87
char msg_with_context[JSON_ERROR_TEXT_LENGTH];
89
line = lex->stream.line;
90
col = lex->stream.column;
91
pos = lex->stream.position;
93
if(saved_text && saved_text[0])
95
if(lex->saved_text.length <= 20) {
96
snprintf(msg_with_context, JSON_ERROR_TEXT_LENGTH,
97
"%s near '%s'", msg_text, saved_text);
98
result = msg_with_context;
103
if(lex->stream.state == STREAM_STATE_ERROR) {
104
/* No context for UTF-8 decoding errors */
108
snprintf(msg_with_context, JSON_ERROR_TEXT_LENGTH,
109
"%s near end of file", msg_text);
110
result = msg_with_context;
115
jsonp_error_set(error, line, col, pos, "%s", result);
119
/*** lexical analyzer ***/
122
stream_init(stream_t *stream, get_func get, void *data)
126
stream->buffer[0] = '\0';
127
stream->buffer_pos = 0;
129
stream->state = STREAM_STATE_OK;
132
stream->position = 0;
135
static int stream_get(stream_t *stream, json_error_t *error)
139
if(stream->state != STREAM_STATE_OK)
140
return stream->state;
142
if(!stream->buffer[stream->buffer_pos])
144
c = stream->get(stream->data);
146
stream->state = STREAM_STATE_EOF;
147
return STREAM_STATE_EOF;
150
stream->buffer[0] = c;
151
stream->buffer_pos = 0;
153
if(0x80 <= c && c <= 0xFF)
155
/* multi-byte UTF-8 sequence */
158
count = utf8_check_first(c);
164
for(i = 1; i < count; i++)
165
stream->buffer[i] = stream->get(stream->data);
167
if(!utf8_check_full(stream->buffer, count, NULL))
170
stream->buffer[count] = '\0';
173
stream->buffer[1] = '\0';
176
c = stream->buffer[stream->buffer_pos++];
181
stream->last_column = stream->column;
184
else if(utf8_check_first(c)) {
185
/* track the Unicode character column, so increment only if
186
this is the first character of a UTF-8 sequence */
193
stream->state = STREAM_STATE_ERROR;
194
error_set(error, stream_to_lex(stream), "unable to decode byte 0x%x", c);
195
return STREAM_STATE_ERROR;
198
static void stream_unget(stream_t *stream, int c)
200
if(c == STREAM_STATE_EOF || c == STREAM_STATE_ERROR)
206
stream->column = stream->last_column;
208
else if(utf8_check_first(c))
211
assert(stream->buffer_pos > 0);
212
stream->buffer_pos--;
213
assert(stream->buffer[stream->buffer_pos] == c);
217
static int lex_get(lex_t *lex, json_error_t *error)
219
return stream_get(&lex->stream, error);
222
static void lex_save(lex_t *lex, int c)
224
strbuffer_append_byte(&lex->saved_text, c);
227
static int lex_get_save(lex_t *lex, json_error_t *error)
229
int c = stream_get(&lex->stream, error);
230
if(c != STREAM_STATE_EOF && c != STREAM_STATE_ERROR)
235
static void lex_unget(lex_t *lex, int c)
237
stream_unget(&lex->stream, c);
240
static void lex_unget_unsave(lex_t *lex, int c)
242
if(c != STREAM_STATE_EOF && c != STREAM_STATE_ERROR) {
244
stream_unget(&lex->stream, c);
245
d = strbuffer_pop(&lex->saved_text);
250
static void lex_save_cached(lex_t *lex)
252
while(lex->stream.buffer[lex->stream.buffer_pos] != '\0')
254
lex_save(lex, lex->stream.buffer[lex->stream.buffer_pos]);
255
lex->stream.buffer_pos++;
256
lex->stream.position++;
260
/* assumes that str points to 'u' plus at least 4 valid hex digits */
261
static int32_t decode_unicode_escape(const char *str)
266
assert(str[0] == 'u');
268
for(i = 1; i <= 4; i++) {
274
value += c - 'a' + 10;
276
value += c - 'A' + 10;
284
static void lex_scan_string(lex_t *lex, json_error_t *error)
291
lex->value.string = NULL;
292
lex->token = TOKEN_INVALID;
294
c = lex_get_save(lex, error);
297
if(c == STREAM_STATE_ERROR)
300
else if(c == STREAM_STATE_EOF) {
301
error_set(error, lex, "premature end of input");
305
else if(0 <= c && c <= 0x1F) {
306
/* control character */
307
lex_unget_unsave(lex, c);
309
error_set(error, lex, "unexpected newline", c);
311
error_set(error, lex, "control character 0x%x", c);
316
c = lex_get_save(lex, error);
318
c = lex_get_save(lex, error);
319
for(i = 0; i < 4; i++) {
321
error_set(error, lex, "invalid escape");
324
c = lex_get_save(lex, error);
327
else if(c == '"' || c == '\\' || c == '/' || c == 'b' ||
328
c == 'f' || c == 'n' || c == 'r' || c == 't')
329
c = lex_get_save(lex, error);
331
error_set(error, lex, "invalid escape");
336
c = lex_get_save(lex, error);
339
/* the actual value is at most of the same length as the source
341
- shortcut escapes (e.g. "\t") (length 2) are converted to 1 byte
342
- a single \uXXXX escape (length 6) is converted to at most 3 bytes
343
- two \uXXXX escapes (length 12) forming an UTF-16 surrogate pair
344
are converted to 4 bytes
346
lex->value.string = (char *) jsonp_malloc(lex->saved_text.length + 1);
347
if(!lex->value.string) {
348
/* this is not very nice, since TOKEN_INVALID is returned */
353
t = lex->value.string;
355
/* + 1 to skip the " */
356
p = strbuffer_value(&lex->saved_text) + 1;
366
value = decode_unicode_escape(p);
369
if(0xD800 <= value && value <= 0xDBFF) {
371
if(*p == '\\' && *(p + 1) == 'u') {
372
int32_t value2 = decode_unicode_escape(++p);
375
if(0xDC00 <= value2 && value2 <= 0xDFFF) {
376
/* valid second surrogate */
378
((value - 0xD800) << 10) +
383
/* invalid second surrogate */
384
error_set(error, lex,
385
"invalid Unicode '\\u%04X\\u%04X'",
391
/* no second surrogate */
392
error_set(error, lex, "invalid Unicode '\\u%04X'",
397
else if(0xDC00 <= value && value <= 0xDFFF) {
398
error_set(error, lex, "invalid Unicode '\\u%04X'", value);
403
error_set(error, lex, "\\u0000 is not allowed");
407
if(utf8_encode(value, buffer, &length))
410
memcpy(t, buffer, length);
415
case '"': case '\\': case '/':
417
case 'b': *t = '\b'; break;
418
case 'f': *t = '\f'; break;
419
case 'n': *t = '\n'; break;
420
case 'r': *t = '\r'; break;
421
case 't': *t = '\t'; break;
432
lex->token = TOKEN_STRING;
436
jsonp_free(lex->value.string);
439
#if JSON_INTEGER_IS_LONG_LONG
441
#define json_strtoint _strtoi64
443
#define json_strtoint strtoll
446
#define json_strtoint strtol
449
static int lex_scan_number(lex_t *lex, int c, json_error_t *error)
451
const char *saved_text;
455
lex->token = TOKEN_INVALID;
458
c = lex_get_save(lex, error);
461
c = lex_get_save(lex, error);
463
lex_unget_unsave(lex, c);
467
else if(isdigit(c)) {
468
c = lex_get_save(lex, error);
470
c = lex_get_save(lex, error);
473
lex_unget_unsave(lex, c);
477
if(c != '.' && c != 'E' && c != 'e') {
480
lex_unget_unsave(lex, c);
482
saved_text = strbuffer_value(&lex->saved_text);
485
value = json_strtoint(saved_text, &end, 10);
486
if(errno == ERANGE) {
488
error_set(error, lex, "too big negative integer");
490
error_set(error, lex, "too big integer");
494
assert(end == saved_text + lex->saved_text.length);
496
lex->token = TOKEN_INTEGER;
497
lex->value.integer = value;
502
c = lex_get(lex, error);
509
c = lex_get_save(lex, error);
511
c = lex_get_save(lex, error);
514
if(c == 'E' || c == 'e') {
515
c = lex_get_save(lex, error);
516
if(c == '+' || c == '-')
517
c = lex_get_save(lex, error);
520
lex_unget_unsave(lex, c);
524
c = lex_get_save(lex, error);
526
c = lex_get_save(lex, error);
529
lex_unget_unsave(lex, c);
531
saved_text = strbuffer_value(&lex->saved_text);
533
value = strtod(saved_text, &end);
534
assert(end == saved_text + lex->saved_text.length);
536
if(errno == ERANGE && value != 0) {
537
error_set(error, lex, "real number overflow");
541
lex->token = TOKEN_REAL;
542
lex->value.real = value;
549
static int lex_scan(lex_t *lex, json_error_t *error)
553
strbuffer_clear(&lex->saved_text);
555
if(lex->token == TOKEN_STRING) {
556
jsonp_free(lex->value.string);
557
lex->value.string = NULL;
560
c = lex_get(lex, error);
561
while(c == ' ' || c == '\t' || c == '\n' || c == '\r')
562
c = lex_get(lex, error);
564
if(c == STREAM_STATE_EOF) {
565
lex->token = TOKEN_EOF;
569
if(c == STREAM_STATE_ERROR) {
570
lex->token = TOKEN_INVALID;
576
if(c == '{' || c == '}' || c == '[' || c == ']' || c == ':' || c == ',')
580
lex_scan_string(lex, error);
582
else if(isdigit(c) || c == '-') {
583
if(lex_scan_number(lex, c, error))
587
else if(isupper(c) || islower(c)) {
588
/* eat up the whole identifier for clearer error messages */
589
const char *saved_text;
591
c = lex_get_save(lex, error);
592
while(isupper(c) || islower(c))
593
c = lex_get_save(lex, error);
594
lex_unget_unsave(lex, c);
596
saved_text = strbuffer_value(&lex->saved_text);
598
if(strcmp(saved_text, "true") == 0)
599
lex->token = TOKEN_TRUE;
600
else if(strcmp(saved_text, "false") == 0)
601
lex->token = TOKEN_FALSE;
602
else if(strcmp(saved_text, "null") == 0)
603
lex->token = TOKEN_NULL;
605
lex->token = TOKEN_INVALID;
609
/* save the rest of the input UTF-8 sequence to get an error
610
message of valid UTF-8 */
611
lex_save_cached(lex);
612
lex->token = TOKEN_INVALID;
619
static char *lex_steal_string(lex_t *lex)
622
if(lex->token == TOKEN_STRING)
624
result = lex->value.string;
625
lex->value.string = NULL;
630
static int lex_init(lex_t *lex, get_func get, void *data)
632
stream_init(&lex->stream, get, data);
633
if(strbuffer_init(&lex->saved_text))
636
lex->token = TOKEN_INVALID;
640
static void lex_close(lex_t *lex)
642
if(lex->token == TOKEN_STRING)
643
jsonp_free(lex->value.string);
644
strbuffer_close(&lex->saved_text);
650
static json_t *parse_value(lex_t *lex, size_t flags, json_error_t *error);
652
static json_t *parse_object(lex_t *lex, size_t flags, json_error_t *error)
654
json_t *object = json_object();
658
lex_scan(lex, error);
659
if(lex->token == '}')
666
if(lex->token != TOKEN_STRING) {
667
error_set(error, lex, "string or '}' expected");
671
key = lex_steal_string(lex);
675
if(flags & JSON_REJECT_DUPLICATES) {
676
if(json_object_get(object, key)) {
678
error_set(error, lex, "duplicate object key");
683
lex_scan(lex, error);
684
if(lex->token != ':') {
686
error_set(error, lex, "':' expected");
690
lex_scan(lex, error);
691
value = parse_value(lex, flags, error);
697
if(json_object_set_nocheck(object, key, value)) {
706
lex_scan(lex, error);
707
if(lex->token != ',')
710
lex_scan(lex, error);
713
if(lex->token != '}') {
714
error_set(error, lex, "'}' expected");
725
static json_t *parse_array(lex_t *lex, size_t flags, json_error_t *error)
727
json_t *array = json_array();
731
lex_scan(lex, error);
732
if(lex->token == ']')
736
json_t *elem = parse_value(lex, flags, error);
740
if(json_array_append(array, elem)) {
746
lex_scan(lex, error);
747
if(lex->token != ',')
750
lex_scan(lex, error);
753
if(lex->token != ']') {
754
error_set(error, lex, "']' expected");
765
static json_t *parse_value(lex_t *lex, size_t flags, json_error_t *error)
771
json = json_string_nocheck(lex->value.string);
775
case TOKEN_INTEGER: {
776
json = json_integer(lex->value.integer);
781
json = json_real(lex->value.real);
798
json = parse_object(lex, flags, error);
802
json = parse_array(lex, flags, error);
806
error_set(error, lex, "invalid token");
810
error_set(error, lex, "unexpected token");
820
static json_t *parse_json(lex_t *lex, size_t flags, json_error_t *error)
824
lex_scan(lex, error);
825
if(lex->token != '[' && lex->token != '{') {
826
error_set(error, lex, "'[' or '{' expected");
830
result = parse_value(lex, flags, error);
834
if(!(flags & JSON_DISABLE_EOF_CHECK)) {
835
lex_scan(lex, error);
836
if(lex->token != TOKEN_EOF) {
837
error_set(error, lex, "end of file expected");
852
static int string_get(void *data)
855
string_data_t *stream = (string_data_t *)data;
856
c = stream->data[stream->pos];
862
return (unsigned char)c;
866
json_t *json_loads(const char *string, size_t flags, json_error_t *error)
870
string_data_t stream_data;
872
stream_data.data = string;
875
if(lex_init(&lex, string_get, (void *)&stream_data))
878
jsonp_error_init(error, "<string>");
879
result = parse_json(&lex, flags, error);
892
static int buffer_get(void *data)
895
buffer_data_t *stream = (buffer_data_t *) data;
896
if(stream->pos >= stream->len)
899
c = stream->data[stream->pos];
901
return (unsigned char)c;
904
json_t *json_loadb(const char *buffer, size_t buflen, size_t flags, json_error_t *error)
908
buffer_data_t stream_data;
910
stream_data.data = buffer;
912
stream_data.len = buflen;
914
if(lex_init(&lex, buffer_get, (void *)&stream_data))
917
jsonp_error_init(error, "<buffer>");
918
result = parse_json(&lex, flags, error);
924
json_t *json_loadf(FILE *input, size_t flags, json_error_t *error)
930
if(lex_init(&lex, (get_func)fgetc, input))
938
jsonp_error_init(error, source);
939
result = parse_json(&lex, flags, error);
945
json_t *json_load_file(const char *path, size_t flags, json_error_t *error)
950
jsonp_error_init(error, path);
952
fp = fopen(path, "r");
955
error_set(error, NULL, "unable to open %s: %s",
956
path, strerror(errno));
960
result = json_loadf(fp, flags, error);