main xs_json.h
  1/* copyright (c) 2022 - 2026 grunfink et al. / MIT license */
  2
  3#ifndef _XS_JSON_H
  4
  5#define _XS_JSON_H
  6
  7#include "xs.h"
  8
  9#ifndef MAX_JSON_DEPTH
 10#define MAX_JSON_DEPTH 32
 11#endif
 12
 13void xs_json_dump_value(const xs_val *data, int level, int indent, FILE *f);
 14int xs_json_dump(const xs_val *data, int indent, FILE *f);
 15xs_str *xs_json_dumps(const xs_val *data, int indent);
 16
 17xs_val *xs_json_load_full(FILE *f, int maxdepth);
 18xs_val *xs_json_loads_full(const xs_str *json, int maxdepth);
 19#define xs_json_load(f) xs_json_load_full(f, MAX_JSON_DEPTH)
 20#define xs_json_loads(s) xs_json_loads_full(s, MAX_JSON_DEPTH)
 21
 22xstype xs_json_load_type(FILE *f);
 23int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c);
 24int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c);
 25int xs_json_load_array(FILE *f, int maxdepth, xs_list **l);
 26int xs_json_load_object(FILE *f, int maxdepth, xs_dict **d);
 27
 28
 29#ifdef XS_IMPLEMENTATION
 30
 31/** IMPLEMENTATION **/
 32
 33#include "xs_unicode.h"
 34
 35/** JSON dumps **/
 36
 37static void _xs_json_dump_str(const char *data, FILE *f)
 38/* dumps a string in JSON format */
 39{
 40    unsigned char c;
 41    fputs("\"", f);
 42
 43    while ((c = *data)) {
 44        if (c == '\n')
 45            fputs("\\n", f);
 46        else
 47        if (c == '\r')
 48            fputs("\\r", f);
 49        else
 50        if (c == '\t')
 51            fputs("\\t", f);
 52        else
 53        if (c == '\\')
 54            fputs("\\\\", f);
 55        else
 56        if (c == '"')
 57            fputs("\\\"", f);
 58        else
 59        if (c < 32)
 60            fprintf(f, "\\u%04x", (unsigned int) c);
 61        else
 62            fputc(c, f);
 63
 64        data++;
 65    }
 66
 67    fputs("\"", f);
 68}
 69
 70
 71static void _xs_json_indent(int level, int indent, FILE *f)
 72/* adds indentation */
 73{
 74    if (indent) {
 75        int n;
 76
 77        fputc('\n', f);
 78
 79        for (n = 0; n < level * indent; n++)
 80            fputc(' ', f);
 81    }
 82}
 83
 84
 85void xs_json_dump_value(const xs_val *data, int level, int indent, FILE *f)
 86/* dumps partial data as JSON */
 87{
 88    int c = 0;
 89    const xs_val *v;
 90
 91    switch (xs_type(data)) {
 92    case XSTYPE_NULL:
 93        fputs("null", f);
 94        break;
 95
 96    case XSTYPE_TRUE:
 97        fputs("true", f);
 98        break;
 99
100    case XSTYPE_FALSE:
101        fputs("false", f);
102        break;
103
104    case XSTYPE_NUMBER:
105        fputs(xs_number_str(data), f);
106        break;
107
108    case XSTYPE_LIST:
109        fputc('[', f);
110
111        xs_list_foreach(data, v) {
112            if (c != 0)
113                fputc(',', f);
114
115            _xs_json_indent(level + 1, indent, f);
116            xs_json_dump_value(v, level + 1, indent, f);
117
118            c++;
119        }
120
121        _xs_json_indent(level, indent, f);
122        fputc(']', f);
123
124        break;
125
126    case XSTYPE_DICT:
127        fputc('{', f);
128
129        const xs_str *k;
130
131        xs_dict_foreach(data, k, v) {
132            if (c != 0)
133                fputc(',', f);
134
135            _xs_json_indent(level + 1, indent, f);
136
137            _xs_json_dump_str(k, f);
138            fputc(':', f);
139
140            if (indent)
141                fputc(' ', f);
142
143            xs_json_dump_value(v, level + 1, indent, f);
144
145            c++;
146        }
147
148        _xs_json_indent(level, indent, f);
149        fputc('}', f);
150        break;
151
152    case XSTYPE_STRING:
153        _xs_json_dump_str(data, f);
154        break;
155
156    default:
157        break;
158    }
159}
160
161
162int xs_json_dump(const xs_val *data, int indent, FILE *f)
163/* dumps data into a file as JSON */
164{
165    xstype t = xs_type(data);
166
167    if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
168        xs_json_dump_value(data, 0, indent, f);
169        return 1;
170    }
171
172    return 0;
173}
174
175
176xs_str *xs_json_dumps(const xs_val *data, int indent)
177/* dumps data as a JSON string */
178{
179    xs_str *s = NULL;
180    size_t sz;
181    FILE *f;
182
183    if ((f = open_memstream(&s, &sz)) != NULL) {
184        int r = xs_json_dump(data, indent, f);
185        fclose(f);
186
187        if (!r)
188            s = xs_free(s);
189    }
190
191    return s;
192}
193
194
195/** JSON loads **/
196
197typedef enum {
198    JS_ERROR = -1,
199    JS_INCOMPLETE,
200    JS_OCURLY,
201    JS_OBRACK,
202    JS_CCURLY,
203    JS_CBRACK,
204    JS_COMMA,
205    JS_COLON,
206    JS_VALUE,
207    JS_STRING,
208    JS_NUMBER,
209    JS_TRUE,
210    JS_FALSE,
211    JS_NULL,
212    JS_ARRAY,
213    JS_OBJECT
214} js_type;
215
216
217static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
218{
219    int c;
220    xs_val *v = NULL;
221    int offset;
222
223    *t = JS_ERROR;
224
225    /* skip blanks */
226    while ((c = fgetc(f)) == ' ' || c == '\t' || c == '\n' || c == '\r');
227
228    if (c == '{')
229        *t = JS_OCURLY;
230    else
231    if (c == '}')
232        *t = JS_CCURLY;
233    else
234    if (c == '[')
235        *t = JS_OBRACK;
236    else
237    if (c == ']')
238        *t = JS_CBRACK;
239    else
240    if (c == ',')
241        *t = JS_COMMA;
242    else
243    if (c == ':')
244        *t = JS_COLON;
245    else
246    if (c == '"') {
247        *t = JS_STRING;
248
249        v = xs_str_new(NULL);
250        offset = 0;
251
252        while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
253            if (c == '\\') {
254                unsigned int cp = fgetc(f);
255
256                switch (cp) {
257                case 'n': cp = '\n'; break;
258                case 'r': cp = '\r'; break;
259                case 't': cp = '\t'; break;
260                case '"': cp = '"'; break;
261                case '\\': cp = '\\'; break;
262                case '/': cp = '/'; break;
263                case 'u': /* Unicode codepoint as an hex char */
264                    if (fscanf(f, "%04x", &cp) != 1) {
265                        *t = JS_ERROR;
266                        break;
267                    }
268
269                    if (xs_is_surrogate(cp)) {
270                        /* \u must follow */
271                        if (fgetc(f) != '\\' || fgetc(f) != 'u') {
272                            *t = JS_ERROR;
273                            break;
274                        }
275
276                        unsigned int p2;
277                        if (fscanf(f, "%04x", &p2) != 1) {
278                            *t = JS_ERROR;
279                            break;
280                        }
281
282                        cp = xs_surrogate_dec(cp, p2);
283                    }
284
285                    /* replace dangerous control codes with their visual representations */
286                    if (cp == 0 || (cp < ' ' && !strchr("\r\n\t", cp)))
287                        cp += 0x2400;
288
289                    break;
290
291                default:
292                    *t = JS_ERROR;
293                    break;
294                }
295
296                v = xs_utf8_insert(v, cp, &offset);
297            }
298            else {
299                if (c >= 0 && c < ' ') {
300                    v = xs_utf8_insert(v, c + 0x2400, &offset);
301                }
302                else {
303                    char cc = c;
304                    v = xs_insert_m(v, offset, &cc, 1);
305                    offset++;
306                }
307
308                if (!xs_is_string(v)) {
309                    *t = JS_ERROR;
310                    break;
311                }
312            }
313        }
314
315        if (c == EOF)
316            *t = JS_ERROR;
317    }
318    else
319    if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
320        double d;
321
322        ungetc(c, f);
323        if (fscanf(f, "%lf", &d) == 1) {
324            *t = JS_NUMBER;
325            v = xs_number_new(d);
326        }
327    }
328    else
329    if (c == 't') {
330        if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
331            *t = JS_TRUE;
332            v = xs_val_new(XSTYPE_TRUE);
333        }
334    }
335    else
336    if (c == 'f') {
337        if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
338            fgetc(f) == 's' && fgetc(f) == 'e') {
339            *t = JS_FALSE;
340            v = xs_val_new(XSTYPE_FALSE);
341        }
342    }
343    else
344    if (c == 'n') {
345        if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
346            *t = JS_NULL;
347            v = xs_val_new(XSTYPE_NULL);
348        }
349    }
350
351    if (*t == JS_ERROR)
352        v = xs_free(v);
353
354    return v;
355}
356
357
358int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c)
359/* loads the next scalar value from the JSON stream */
360/* if the value ahead is compound, value is NULL and pt is set */
361{
362    js_type t;
363
364    *value = _xs_json_load_lexer(f, &t);
365
366    if (t == JS_ERROR)
367        return -1;
368
369    if (t == JS_CBRACK)
370        return 0;
371
372    if (*c > 0) {
373        if (t == JS_COMMA)
374            *value = _xs_json_load_lexer(f, &t);
375        else
376            return -1;
377    }
378
379    if (*value == NULL) {
380        /* possible compound type ahead */
381        if (t == JS_OBRACK)
382            *pt = XSTYPE_LIST;
383        else
384        if (t == JS_OCURLY)
385            *pt = XSTYPE_DICT;
386        else
387            return -1;
388    }
389    else
390        *pt = xs_type(*value);
391
392    *c = *c + 1;
393
394    return 1;
395}
396
397
398int xs_json_load_array(FILE *f, int maxdepth, xs_list **l)
399/* loads a full JSON array (after the initial OBRACK) */
400/* l can be NULL for the content to be dropped */
401{
402    xstype t;
403    int r = 0;
404    int c = 0;
405
406    for (;;) {
407        xs *v = NULL;
408        r = xs_json_load_array_iter(f, &v, &t, &c);
409
410        if (r == 1) {
411            /* partial load? */
412            if (v == NULL && maxdepth != 0) {
413                if (t == XSTYPE_LIST) {
414                    if (l)
415                        v = xs_list_new();
416
417                    r = xs_json_load_array(f, maxdepth - 1, &v);
418                }
419                else
420                if (t == XSTYPE_DICT) {
421                    if (l)
422                        v = xs_dict_new();
423
424                    r = xs_json_load_object(f, maxdepth - 1, &v);
425                }
426            }
427
428            /* error? */
429            if (r < 0)
430                break;
431
432            if (l)
433                *l = xs_list_append(*l, v);
434        }
435        else
436            break;
437    }
438
439    if (r < 0 && l)
440        *l = xs_free(*l);
441
442    return r;
443}
444
445
446int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c)
447/* loads the next key and scalar value from the JSON stream */
448/* if the value ahead is compound, value is NULL and pt is set */
449{
450    js_type t;
451
452    *key = _xs_json_load_lexer(f, &t);
453
454    if (t == JS_ERROR)
455        return -1;
456
457    if (t == JS_CCURLY)
458        return 0;
459
460    if (*c > 0) {
461        if (t == JS_COMMA)
462            *key = _xs_json_load_lexer(f, &t);
463        else
464            return -1;
465    }
466
467    if (t != JS_STRING)
468        return -1;
469
470    xs_free(_xs_json_load_lexer(f, &t));
471
472    if (t != JS_COLON)
473        return -1;
474
475    *value = _xs_json_load_lexer(f, &t);
476
477    if (*value == NULL) {
478        /* possible complex type ahead */
479        if (t == JS_OBRACK)
480            *pt = XSTYPE_LIST;
481        else
482        if (t == JS_OCURLY)
483            *pt = XSTYPE_DICT;
484        else
485            return -1;
486    }
487    else
488        *pt = xs_type(*value);
489
490    *c = *c + 1;
491
492    return 1;
493}
494
495
496int xs_json_load_object(FILE *f, int maxdepth, xs_dict **d)
497/* loads a full JSON object (after the initial OCURLY) */
498/* d can be NULL for the content to be dropped */
499{
500    xstype t;
501    int r = 0;
502    int c = 0;
503
504    for (;;) {
505        xs *k = NULL;
506        xs *v = NULL;
507        r = xs_json_load_object_iter(f, &k, &v, &t, &c);
508
509        if (r == 1) {
510            /* partial load? */
511            if (v == NULL && maxdepth != 0) {
512                if (t == XSTYPE_LIST) {
513                    if (d)
514                        v = xs_list_new();
515
516                    r = xs_json_load_array(f, maxdepth - 1, &v);
517                }
518                else
519                if (t == XSTYPE_DICT) {
520                    if (d)
521                        v = xs_dict_new();
522
523                    r = xs_json_load_object(f, maxdepth - 1, &v);
524                }
525            }
526
527            /* error? */
528            if (r < 0)
529                break;
530
531            if (d)
532                *d = xs_dict_append(*d, k, v);
533        }
534        else
535            break;
536    }
537
538    if (r < 0 && d)
539        *d = xs_free(*d);
540
541    return r;
542}
543
544
545xstype xs_json_load_type(FILE *f)
546/* identifies the type of a JSON stream */
547{
548    xstype t = XSTYPE_NULL;
549    js_type jt;
550
551    xs_free(_xs_json_load_lexer(f, &jt));
552
553    if (jt == JS_OBRACK)
554        t = XSTYPE_LIST;
555    else
556    if (jt == JS_OCURLY)
557        t = XSTYPE_DICT;
558
559    return t;
560}
561
562
563xs_val *xs_json_load_full(FILE *f, int maxdepth)
564/* loads a JSON file */
565{
566    xs_val *v = NULL;
567    xstype t = xs_json_load_type(f);
568
569    if (t == XSTYPE_LIST) {
570        v = xs_list_new();
571        xs_json_load_array(f, maxdepth, &v);
572    }
573    else
574    if (t == XSTYPE_DICT) {
575        v = xs_dict_new();
576        xs_json_load_object(f, maxdepth, &v);
577    }
578
579    return v;
580}
581
582
583xs_val *xs_json_loads_full(const xs_str *json, int maxdepth)
584/* loads a string in JSON format and converts to a multiple data */
585{
586    FILE *f;
587    xs_val *v = NULL;
588
589    if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
590        v = xs_json_load_full(f, maxdepth);
591        fclose(f);
592    }
593
594    return v;
595}
596
597
598#endif /* XS_IMPLEMENTATION */
599
600#endif /* _XS_JSON_H */