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);
42 strbuffer_t saved_text;
53 /*** error reporting ***/
55 static void error_set(json_error_t *error, const lex_t *lex,
59 char text[JSON_ERROR_TEXT_LENGTH];
65 vsnprintf(text, JSON_ERROR_TEXT_LENGTH, msg, ap);
70 const char *saved_text = strbuffer_value(&lex->saved_text);
71 error->line = lex->line;
72 if(saved_text && saved_text[0])
74 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
75 "%s near '%s'", text, saved_text);
79 snprintf(error->text, JSON_ERROR_TEXT_LENGTH,
80 "%s near end of file", text);
86 snprintf(error->text, JSON_ERROR_TEXT_LENGTH, "%s", text);
91 /*** lexical analyzer ***/
93 void stream_init(stream_t *stream, get_func get, eof_func eof, void *data)
98 stream->buffer[0] = '\0';
99 stream->buffer_pos = 0;
102 static char stream_get(stream_t *stream)
104 if(!stream->buffer[stream->buffer_pos])
108 stream->buffer[0] = stream->get(stream->data);
109 stream->buffer_pos = 0;
111 c = stream->buffer[0];
113 if(c == EOF && stream->eof(stream->data))
118 /* multi-byte UTF-8 sequence */
121 count = utf8_check_first(c);
127 for(i = 1; i < count; i++)
128 stream->buffer[i] = stream->get(stream->data);
130 if(!utf8_check_full(stream->buffer, count))
133 stream->buffer[count] = '\0';
136 stream->buffer[1] = '\0';
139 return (char)stream->buffer[stream->buffer_pos++];
142 static void stream_unget(stream_t *stream, char c)
144 assert(stream->buffer_pos > 0);
145 stream->buffer_pos--;
146 assert(stream->buffer[stream->buffer_pos] == (unsigned char)c);
150 static int lex_get(lex_t *lex)
152 return stream_get(&lex->stream);
155 static int lex_eof(lex_t *lex)
157 return lex->stream.eof(lex->stream.data);
160 static void lex_save(lex_t *lex, char c)
162 strbuffer_append_byte(&lex->saved_text, c);
165 static int lex_get_save(lex_t *lex)
167 char c = stream_get(&lex->stream);
172 static void lex_unget_unsave(lex_t *lex, char c)
175 stream_unget(&lex->stream, c);
176 d = strbuffer_pop(&lex->saved_text);
180 static void lex_scan_string(lex_t *lex)
186 lex->token = TOKEN_INVALID;
189 c = lex_get_save(lex);
192 if(c == EOF && lex_eof(lex))
195 else if(0 <= c && c <= 0x1F) {
196 /* control character */
197 lex_unget_unsave(lex, c);
202 c = lex_get_save(lex);
204 c = lex_get_save(lex);
205 for(int i = 0; i < 4; i++) {
207 lex_unget_unsave(lex, c);
210 c = lex_get_save(lex);
213 else if(c == '"' || c == '\\' || c == '/' || c == 'b' ||
214 c == 'f' || c == 'n' || c == 'r' || c == 't')
215 c = lex_get_save(lex);
217 lex_unget_unsave(lex, c);
222 c = lex_get_save(lex);
225 /* the actual value is at most of the same length as the source
227 - shortcut escapes (e.g. "\t") (length 2) are converted to 1 byte
228 - a single \uXXXX escape (length 6) is converted to at most 3 bytes
229 - two \uXXXX escapes (length 12) forming an UTF-16 surrogate pair
230 are converted to 4 bytes
232 lex->value.string = malloc(lex->saved_text.length + 1);
233 if(!lex->value.string) {
234 /* this is not very nice, since TOKEN_INVALID is returned */
239 t = lex->value.string;
241 /* + 1 to skip the " */
242 p = strbuffer_value(&lex->saved_text) + 1;
248 /* TODO: \uXXXX not supported yet */
249 free(lex->value.string);
250 lex->value.string = NULL;
254 case '"': case '\\': case '/':
256 case 'b': *t = '\b'; break;
257 case 'f': *t = '\f'; break;
258 case 'n': *t = '\n'; break;
259 case 'r': *t = '\r'; break;
260 case 't': *t = '\t'; break;
272 lex->token = TOKEN_STRING;
278 static void lex_scan_number(lex_t *lex, char c)
280 const char *saved_text;
283 lex->token = TOKEN_INVALID;
286 c = lex_get_save(lex);
289 c = lex_get_save(lex);
291 lex_unget_unsave(lex, c);
295 else /* c != '0' */ {
296 c = lex_get_save(lex);
298 c = lex_get_save(lex);
301 if(c != '.' && c != 'E' && c != 'e') {
302 lex_unget_unsave(lex, c);
303 lex->token = TOKEN_INTEGER;
305 saved_text = strbuffer_value(&lex->saved_text);
306 lex->value.integer = strtol(saved_text, &end, 10);
307 assert(end == saved_text + lex->saved_text.length);
318 c = lex_get_save(lex);
320 c = lex_get_save(lex);
323 if(c == 'E' || c == 'e') {
324 c = lex_get_save(lex);
325 if(c == '+' || c == '-')
326 c = lex_get_save(lex);
329 lex_unget_unsave(lex, c);
333 c = lex_get_save(lex);
335 c = lex_get_save(lex);
338 lex_unget_unsave(lex, c);
339 lex->token = TOKEN_REAL;
341 saved_text = strbuffer_value(&lex->saved_text);
342 lex->value.real = strtod(saved_text, &end);
343 assert(end == saved_text + lex->saved_text.length);
349 static int lex_scan(lex_t *lex)
353 strbuffer_clear(&lex->saved_text);
355 if(lex->token == TOKEN_STRING) {
356 free(lex->value.string);
357 lex->value.string = NULL;
361 while(c == ' ' || c == '\t' || c == '\n' || c == '\r')
369 if(c == EOF && lex_eof(lex)) {
370 lex->token = TOKEN_EOF;
376 if(c == '{' || c == '}' || c == '[' || c == ']' || c == ':' || c == ',')
380 lex_scan_string(lex);
382 else if(isdigit(c) || c == '-')
383 lex_scan_number(lex, c);
385 else if(isupper(c) || islower(c)) {
386 /* eat up the whole identifier for clearer error messages */
387 const char *saved_text;
389 c = lex_get_save(lex);
390 while(isupper(c) || islower(c))
391 c = lex_get_save(lex);
392 lex_unget_unsave(lex, c);
394 saved_text = strbuffer_value(&lex->saved_text);
396 if(strcmp(saved_text, "true") == 0)
397 lex->token = TOKEN_TRUE;
398 else if(strcmp(saved_text, "false") == 0)
399 lex->token = TOKEN_FALSE;
400 else if(strcmp(saved_text, "null") == 0)
401 lex->token = TOKEN_NULL;
403 lex->token = TOKEN_INVALID;
407 lex->token = TOKEN_INVALID;
413 static int lex_init(lex_t *lex, get_func get, eof_func eof, void *data)
415 stream_init(&lex->stream, get, eof, data);
416 if(strbuffer_init(&lex->saved_text))
419 lex->token = TOKEN_INVALID;
425 static void lex_close(lex_t *lex)
427 if(lex->token == TOKEN_STRING)
428 free(lex->value.string);
434 static json_t *parse_value(lex_t *lex, json_error_t *error);
436 static json_t *parse_object(lex_t *lex, json_error_t *error)
438 json_t *object = json_object();
443 if(lex->token == '}')
450 if(lex->token != TOKEN_STRING) {
451 error_set(error, lex, "string or '}' expected");
455 key = strdup(lex->value.string);
460 if(lex->token != ':') {
462 error_set(error, lex, "':' expected");
467 value = parse_value(lex, error);
473 if(json_object_set_nocheck(object, key, value)) {
483 if(lex->token != ',')
489 if(lex->token != '}') {
490 error_set(error, lex, "'}' expected");
501 static json_t *parse_array(lex_t *lex, json_error_t *error)
503 json_t *array = json_array();
508 if(lex->token == ']')
512 json_t *elem = parse_value(lex, error);
516 if(json_array_append(array, elem)) {
523 if(lex->token != ',')
529 if(lex->token != ']') {
530 error_set(error, lex, "']' expected");
541 static json_t *parse_value(lex_t *lex, json_error_t *error)
547 json = json_string_nocheck(lex->value.string);
551 case TOKEN_INTEGER: {
552 json = json_integer(lex->value.integer);
557 json = json_real(lex->value.real);
574 json = parse_object(lex, error);
578 json = parse_array(lex, error);
582 error_set(error, lex, "invalid token");
586 error_set(error, lex, "unexpected token");
596 json_t *parse_json(lex_t *lex, json_error_t *error)
600 if(lex->token != '[' && lex->token != '{') {
601 error_set(error, lex, "'[' or '{' expected");
605 return parse_value(lex, error);
608 json_t *json_load(const char *path, json_error_t *error)
613 fp = fopen(path, "r");
616 error_set(error, NULL, "unable to open %s: %s",
617 path, strerror(errno));
621 result = json_loadf(fp, error);
633 static int string_get(void *data)
636 string_data_t *stream = (string_data_t *)data;
637 c = stream->data[stream->pos++];
644 static int string_eof(void *data)
646 string_data_t *stream = (string_data_t *)data;
647 return (stream->data[stream->pos] == '\0');
650 json_t *json_loads(const char *string, json_error_t *error)
655 string_data_t stream_data = {
660 if(lex_init(&lex, string_get, string_eof, (void *)&stream_data))
663 result = parse_json(&lex, error);
668 if(lex.token != TOKEN_EOF) {
669 error_set(error, &lex, "end of file expected");
679 json_t *json_loadf(FILE *input, json_error_t *error)
684 if(lex_init(&lex, (get_func)fgetc, (eof_func)feof, input))
687 result = parse_json(&lex, error);