2.78 xs_json.h
  1/* copyright (c) 2022 - 2025 grunfink et al. / MIT license */
  2
  3#ifndef _XS_JSON_H
  4
  5#define _XS_JSON_H
  6
  7#ifndef MAX_JSON_DEPTH
  8#define MAX_JSON_DEPTH 32
  9#endif
 10
 11int xs_json_dump(const xs_val *data, int indent, FILE *f);
 12xs_str *xs_json_dumps(const xs_val *data, int indent);
 13
 14xs_val *xs_json_load_full(FILE *f, int maxdepth);
 15xs_val *xs_json_loads_full(const xs_str *json, int maxdepth);
 16#define xs_json_load(f) xs_json_load_full(f, MAX_JSON_DEPTH)
 17#define xs_json_loads(s) xs_json_loads_full(s, MAX_JSON_DEPTH)
 18
 19xstype xs_json_load_type(FILE *f);
 20int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c);
 21int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c);
 22xs_list *xs_json_load_array(FILE *f, int maxdepth);
 23xs_dict *xs_json_load_object(FILE *f, int maxdepth);
 24
 25
 26#ifdef XS_IMPLEMENTATION
 27
 28/** IMPLEMENTATION **/
 29
 30/** JSON dumps **/
 31
 32static void _xs_json_dump_str(const char *data, FILE *f)
 33/* dumps a string in JSON format */
 34{
 35    unsigned char c;
 36    fputs("\"", f);
 37
 38    while ((c = *data)) {
 39        if (c == '\n')
 40            fputs("\\n", f);
 41        else
 42        if (c == '\r')
 43            fputs("\\r", f);
 44        else
 45        if (c == '\t')
 46            fputs("\\t", f);
 47        else
 48        if (c == '\\')
 49            fputs("\\\\", f);
 50        else
 51        if (c == '"')
 52            fputs("\\\"", f);
 53        else
 54        if (c < 32)
 55            fprintf(f, "\\u%04x", (unsigned int) c);
 56        else
 57            fputc(c, f);
 58
 59        data++;
 60    }
 61
 62    fputs("\"", f);
 63}
 64
 65
 66static void _xs_json_indent(int level, int indent, FILE *f)
 67/* adds indentation */
 68{
 69    if (indent) {
 70        int n;
 71
 72        fputc('\n', f);
 73
 74        for (n = 0; n < level * indent; n++)
 75            fputc(' ', f);
 76    }
 77}
 78
 79
 80static void _xs_json_dump(const xs_val *data, int level, int indent, FILE *f)
 81/* dumps partial data as JSON */
 82{
 83    int c = 0;
 84    const xs_val *v;
 85
 86    switch (xs_type(data)) {
 87    case XSTYPE_NULL:
 88        fputs("null", f);
 89        break;
 90
 91    case XSTYPE_TRUE:
 92        fputs("true", f);
 93        break;
 94
 95    case XSTYPE_FALSE:
 96        fputs("false", f);
 97        break;
 98
 99    case XSTYPE_NUMBER:
100        fputs(xs_number_str(data), f);
101        break;
102
103    case XSTYPE_LIST:
104        fputc('[', f);
105
106        xs_list_foreach(data, v) {
107            if (c != 0)
108                fputc(',', f);
109
110            _xs_json_indent(level + 1, indent, f);
111            _xs_json_dump(v, level + 1, indent, f);
112
113            c++;
114        }
115
116        _xs_json_indent(level, indent, f);
117        fputc(']', f);
118
119        break;
120
121    case XSTYPE_DICT:
122        fputc('{', f);
123
124        const xs_str *k;
125
126        xs_dict_foreach(data, k, v) {
127            if (c != 0)
128                fputc(',', f);
129
130            _xs_json_indent(level + 1, indent, f);
131
132            _xs_json_dump_str(k, f);
133            fputc(':', f);
134
135            if (indent)
136                fputc(' ', f);
137
138            _xs_json_dump(v, level + 1, indent, f);
139
140            c++;
141        }
142
143        _xs_json_indent(level, indent, f);
144        fputc('}', f);
145        break;
146
147    case XSTYPE_STRING:
148        _xs_json_dump_str(data, f);
149        break;
150
151    default:
152        break;
153    }
154}
155
156
157xs_str *xs_json_dumps(const xs_val *data, int indent)
158/* dumps data as a JSON string */
159{
160    xs_str *s = NULL;
161    size_t sz;
162    FILE *f;
163
164    if ((f = open_memstream(&s, &sz)) != NULL) {
165        int r = xs_json_dump(data, indent, f);
166        fclose(f);
167
168        if (!r)
169            s = xs_free(s);
170    }
171
172    return s;
173}
174
175
176int xs_json_dump(const xs_val *data, int indent, FILE *f)
177/* dumps data into a file as JSON */
178{
179    xstype t = xs_type(data);
180
181    if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
182        _xs_json_dump(data, 0, indent, f);
183        return 1;
184    }
185
186    return 0;
187}
188
189
190/** JSON loads **/
191
192typedef enum {
193    JS_ERROR = -1,
194    JS_INCOMPLETE,
195    JS_OCURLY,
196    JS_OBRACK,
197    JS_CCURLY,
198    JS_CBRACK,
199    JS_COMMA,
200    JS_COLON,
201    JS_VALUE,
202    JS_STRING,
203    JS_NUMBER,
204    JS_TRUE,
205    JS_FALSE,
206    JS_NULL,
207    JS_ARRAY,
208    JS_OBJECT
209} js_type;
210
211
212static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
213{
214    int c;
215    xs_val *v = NULL;
216    int offset;
217
218    *t = JS_ERROR;
219
220    /* skip blanks */
221    while ((c = fgetc(f)) == ' ' || c == '\t' || c == '\n' || c == '\r');
222
223    if (c == '{')
224        *t = JS_OCURLY;
225    else
226    if (c == '}')
227        *t = JS_CCURLY;
228    else
229    if (c == '[')
230        *t = JS_OBRACK;
231    else
232    if (c == ']')
233        *t = JS_CBRACK;
234    else
235    if (c == ',')
236        *t = JS_COMMA;
237    else
238    if (c == ':')
239        *t = JS_COLON;
240    else
241    if (c == '"') {
242        *t = JS_STRING;
243
244        v = xs_str_new(NULL);
245        offset = 0;
246
247        while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
248            if (c == '\\') {
249                unsigned int cp = fgetc(f);
250
251                switch (cp) {
252                case 'n': cp = '\n'; break;
253                case 'r': cp = '\r'; break;
254                case 't': cp = '\t'; break;
255                case 'u': /* Unicode codepoint as an hex char */
256                    if (fscanf(f, "%04x", &cp) != 1) {
257                        *t = JS_ERROR;
258                        break;
259                    }
260
261                    if (xs_is_surrogate(cp)) {
262                        /* \u must follow */
263                        if (fgetc(f) != '\\' || fgetc(f) != 'u') {
264                            *t = JS_ERROR;
265                            break;
266                        }
267
268                        unsigned int p2;
269                        if (fscanf(f, "%04x", &p2) != 1) {
270                            *t = JS_ERROR;
271                            break;
272                        }
273
274                        cp = xs_surrogate_dec(cp, p2);
275                    }
276
277                    /* replace dangerous control codes with their visual representations */
278                    if (cp < ' ' && !strchr("\r\n\t", cp))
279                        cp += 0x2400;
280
281                    break;
282                }
283
284                v = xs_utf8_insert(v, cp, &offset);
285            }
286            else {
287                char cc = c;
288                v = xs_insert_m(v, offset, &cc, 1);
289
290                if (!xs_is_string(v)) {
291                    *t = JS_ERROR;
292                    break;
293                }
294
295                offset++;
296            }
297        }
298
299        if (c == EOF)
300            *t = JS_ERROR;
301    }
302    else
303    if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
304        double d;
305
306        ungetc(c, f);
307        if (fscanf(f, "%lf", &d) == 1) {
308            *t = JS_NUMBER;
309            v = xs_number_new(d);
310        }
311    }
312    else
313    if (c == 't') {
314        if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
315            *t = JS_TRUE;
316            v = xs_val_new(XSTYPE_TRUE);
317        }
318    }
319    else
320    if (c == 'f') {
321        if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
322            fgetc(f) == 's' && fgetc(f) == 'e') {
323            *t = JS_FALSE;
324            v = xs_val_new(XSTYPE_FALSE);
325        }
326    }
327    else
328    if (c == 'n') {
329        if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
330            *t = JS_NULL;
331            v = xs_val_new(XSTYPE_NULL);
332        }
333    }
334
335    if (*t == JS_ERROR)
336        v = xs_free(v);
337
338    return v;
339}
340
341
342int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c)
343/* loads the next scalar value from the JSON stream */
344/* if the value ahead is compound, value is NULL and pt is set */
345{
346    js_type t;
347
348    *value = _xs_json_load_lexer(f, &t);
349
350    if (t == JS_ERROR)
351        return -1;
352
353    if (t == JS_CBRACK)
354        return 0;
355
356    if (*c > 0) {
357        if (t == JS_COMMA)
358            *value = _xs_json_load_lexer(f, &t);
359        else
360            return -1;
361    }
362
363    if (*value == NULL) {
364        /* possible compound type ahead */
365        if (t == JS_OBRACK)
366            *pt = XSTYPE_LIST;
367        else
368        if (t == JS_OCURLY)
369            *pt = XSTYPE_DICT;
370        else
371            return -1;
372    }
373
374    *c = *c + 1;
375
376    return 1;
377}
378
379
380xs_list *xs_json_load_array(FILE *f, int maxdepth)
381/* loads a full JSON array (after the initial OBRACK) */
382{
383    xstype t;
384    xs_list *l = xs_list_new();
385    int c = 0;
386
387    for (;;) {
388        xs *v = NULL;
389        int r = xs_json_load_array_iter(f, &v, &t, &c);
390
391        if (r == -1)
392            l = xs_free(l);
393
394        if (r == 1) {
395            /* partial load? */
396            if (v == NULL && maxdepth != 0) {
397                if (t == XSTYPE_LIST)
398                    v = xs_json_load_array(f, maxdepth - 1);
399                else
400                if (t == XSTYPE_DICT)
401                    v = xs_json_load_object(f, maxdepth - 1);
402            }
403
404            /* still null? fail */
405            if (v == NULL) {
406                l = xs_free(l);
407                break;
408            }
409
410            l = xs_list_append(l, v);
411        }
412        else
413            break;
414    }
415
416    return l;
417}
418
419
420int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c)
421/* loads the next key and scalar value from the JSON stream */
422/* if the value ahead is compound, value is NULL and pt is set */
423{
424    js_type t;
425
426    *key = _xs_json_load_lexer(f, &t);
427
428    if (t == JS_ERROR)
429        return -1;
430
431    if (t == JS_CCURLY)
432        return 0;
433
434    if (*c > 0) {
435        if (t == JS_COMMA)
436            *key = _xs_json_load_lexer(f, &t);
437        else
438            return -1;
439    }
440
441    if (t != JS_STRING)
442        return -1;
443
444    xs_free(_xs_json_load_lexer(f, &t));
445
446    if (t != JS_COLON)
447        return -1;
448
449    *value = _xs_json_load_lexer(f, &t);
450
451    if (*value == NULL) {
452        /* possible complex type ahead */
453        if (t == JS_OBRACK)
454            *pt = XSTYPE_LIST;
455        else
456        if (t == JS_OCURLY)
457            *pt = XSTYPE_DICT;
458        else
459            return -1;
460    }
461
462    *c = *c + 1;
463
464    return 1;
465}
466
467
468xs_dict *xs_json_load_object(FILE *f, int maxdepth)
469/* loads a full JSON object (after the initial OCURLY) */
470{
471    xstype t;
472    xs_dict *d = xs_dict_new();
473    int c = 0;
474
475    for (;;) {
476        xs *k = NULL;
477        xs *v = NULL;
478        int r = xs_json_load_object_iter(f, &k, &v, &t, &c);
479
480        if (r == -1)
481            d = xs_free(d);
482
483        if (r == 1) {
484            /* partial load? */
485            if (v == NULL && maxdepth != 0) {
486                if (t == XSTYPE_LIST)
487                    v = xs_json_load_array(f, maxdepth - 1);
488                else
489                if (t == XSTYPE_DICT)
490                    v = xs_json_load_object(f, maxdepth - 1);
491            }
492
493            /* still null? fail */
494            if (v == NULL) {
495                d = xs_free(d);
496                break;
497            }
498
499            d = xs_dict_append(d, k, v);
500        }
501        else
502            break;
503    }
504
505    return d;
506}
507
508
509xs_val *xs_json_loads_full(const xs_str *json, int maxdepth)
510/* loads a string in JSON format and converts to a multiple data */
511{
512    FILE *f;
513    xs_val *v = NULL;
514
515    if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
516        v = xs_json_load_full(f, maxdepth);
517        fclose(f);
518    }
519
520    return v;
521}
522
523
524xstype xs_json_load_type(FILE *f)
525/* identifies the type of a JSON stream */
526{
527    xstype t = XSTYPE_NULL;
528    js_type jt;
529
530    xs_free(_xs_json_load_lexer(f, &jt));
531
532    if (jt == JS_OBRACK)
533        t = XSTYPE_LIST;
534    else
535    if (jt == JS_OCURLY)
536        t = XSTYPE_DICT;
537
538    return t;
539}
540
541
542xs_val *xs_json_load_full(FILE *f, int maxdepth)
543/* loads a JSON file */
544{
545    xs_val *v = NULL;
546    xstype t = xs_json_load_type(f);
547
548    if (t == XSTYPE_LIST)
549        v = xs_json_load_array(f, maxdepth);
550    else
551    if (t == XSTYPE_DICT)
552        v = xs_json_load_object(f, maxdepth);
553
554    return v;
555}
556
557
558#endif /* XS_IMPLEMENTATION */
559
560#endif /* _XS_JSON_H */