2.62 xs_json.h
  1/* copyright (c) 2022 - 2024 grunfink et al. / MIT license */
  2
  3#ifndef _XS_JSON_H
  4
  5#define _XS_JSON_H
  6
  7int xs_json_dump(const xs_val *data, int indent, FILE *f);
  8xs_str *xs_json_dumps(const xs_val *data, int indent);
  9
 10xs_val *xs_json_load(FILE *f);
 11xs_val *xs_json_loads(const xs_str *json);
 12
 13xstype xs_json_load_type(FILE *f);
 14int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c);
 15int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c);
 16xs_list *xs_json_load_array(FILE *f);
 17xs_dict *xs_json_load_object(FILE *f);
 18
 19
 20#ifdef XS_IMPLEMENTATION
 21
 22/** IMPLEMENTATION **/
 23
 24/** JSON dumps **/
 25
 26static void _xs_json_dump_str(const char *data, FILE *f)
 27/* dumps a string in JSON format */
 28{
 29    unsigned char c;
 30    fputs("\"", f);
 31
 32    while ((c = *data)) {
 33        if (c == '\n')
 34            fputs("\\n", f);
 35        else
 36        if (c == '\r')
 37            fputs("\\r", f);
 38        else
 39        if (c == '\t')
 40            fputs("\\t", f);
 41        else
 42        if (c == '\\')
 43            fputs("\\\\", f);
 44        else
 45        if (c == '"')
 46            fputs("\\\"", f);
 47        else
 48        if (c < 32)
 49            fprintf(f, "\\u%04x", (unsigned int) c);
 50        else
 51            fputc(c, f);
 52
 53        data++;
 54    }
 55
 56    fputs("\"", f);
 57}
 58
 59
 60static void _xs_json_indent(int level, int indent, FILE *f)
 61/* adds indentation */
 62{
 63    if (indent) {
 64        int n;
 65
 66        fputc('\n', f);
 67
 68        for (n = 0; n < level * indent; n++)
 69            fputc(' ', f);
 70    }
 71}
 72
 73
 74static void _xs_json_dump(const xs_val *data, int level, int indent, FILE *f)
 75/* dumps partial data as JSON */
 76{
 77    int c = 0;
 78    const xs_val *v;
 79
 80    switch (xs_type(data)) {
 81    case XSTYPE_NULL:
 82        fputs("null", f);
 83        break;
 84
 85    case XSTYPE_TRUE:
 86        fputs("true", f);
 87        break;
 88
 89    case XSTYPE_FALSE:
 90        fputs("false", f);
 91        break;
 92
 93    case XSTYPE_NUMBER:
 94        fputs(xs_number_str(data), f);
 95        break;
 96
 97    case XSTYPE_LIST:
 98        fputc('[', f);
 99
100        xs_list_foreach(data, v) {
101            if (c != 0)
102                fputc(',', f);
103
104            _xs_json_indent(level + 1, indent, f);
105            _xs_json_dump(v, level + 1, indent, f);
106
107            c++;
108        }
109
110        _xs_json_indent(level, indent, f);
111        fputc(']', f);
112
113        break;
114
115    case XSTYPE_DICT:
116        fputc('{', f);
117
118        const xs_str *k;
119
120        xs_dict_foreach(data, k, v) {
121            if (c != 0)
122                fputc(',', f);
123
124            _xs_json_indent(level + 1, indent, f);
125
126            _xs_json_dump_str(k, f);
127            fputc(':', f);
128
129            if (indent)
130                fputc(' ', f);
131
132            _xs_json_dump(v, level + 1, indent, f);
133
134            c++;
135        }
136
137        _xs_json_indent(level, indent, f);
138        fputc('}', f);
139        break;
140
141    case XSTYPE_STRING:
142        _xs_json_dump_str(data, f);
143        break;
144
145    default:
146        break;
147    }
148}
149
150
151xs_str *xs_json_dumps(const xs_val *data, int indent)
152/* dumps data as a JSON string */
153{
154    xs_str *s = NULL;
155    size_t sz;
156    FILE *f;
157
158    if ((f = open_memstream(&s, &sz)) != NULL) {
159        int r = xs_json_dump(data, indent, f);
160        fclose(f);
161
162        if (!r)
163            s = xs_free(s);
164    }
165
166    return s;
167}
168
169
170int xs_json_dump(const xs_val *data, int indent, FILE *f)
171/* dumps data into a file as JSON */
172{
173    xstype t = xs_type(data);
174
175    if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
176        _xs_json_dump(data, 0, indent, f);
177        return 1;
178    }
179
180    return 0;
181}
182
183
184/** JSON loads **/
185
186typedef enum {
187    JS_ERROR = -1,
188    JS_INCOMPLETE,
189    JS_OCURLY,
190    JS_OBRACK,
191    JS_CCURLY,
192    JS_CBRACK,
193    JS_COMMA,
194    JS_COLON,
195    JS_VALUE,
196    JS_STRING,
197    JS_NUMBER,
198    JS_TRUE,
199    JS_FALSE,
200    JS_NULL,
201    JS_ARRAY,
202    JS_OBJECT
203} js_type;
204
205
206static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
207{
208    int c;
209    xs_val *v = NULL;
210    int offset;
211
212    *t = JS_ERROR;
213
214    /* skip blanks */
215    while ((c = fgetc(f)) == ' ' || c == '\t' || c == '\n' || c == '\r');
216
217    if (c == '{')
218        *t = JS_OCURLY;
219    else
220    if (c == '}')
221        *t = JS_CCURLY;
222    else
223    if (c == '[')
224        *t = JS_OBRACK;
225    else
226    if (c == ']')
227        *t = JS_CBRACK;
228    else
229    if (c == ',')
230        *t = JS_COMMA;
231    else
232    if (c == ':')
233        *t = JS_COLON;
234    else
235    if (c == '"') {
236        *t = JS_STRING;
237
238        v = xs_str_new(NULL);
239        offset = 0;
240
241        while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
242            if (c == '\\') {
243                unsigned int cp = fgetc(f);
244
245                switch (cp) {
246                case 'n': cp = '\n'; break;
247                case 'r': cp = '\r'; break;
248                case 't': cp = '\t'; break;
249                case 'u': /* Unicode codepoint as an hex char */
250                    if (fscanf(f, "%04x", &cp) != 1) {
251                        *t = JS_ERROR;
252                        break;
253                    }
254
255                    if (xs_is_surrogate(cp)) {
256                        /* \u must follow */
257                        if (fgetc(f) != '\\' || fgetc(f) != 'u') {
258                            *t = JS_ERROR;
259                            break;
260                        }
261
262                        unsigned int p2;
263                        if (fscanf(f, "%04x", &p2) != 1) {
264                            *t = JS_ERROR;
265                            break;
266                        }
267
268                        cp = xs_surrogate_dec(cp, p2);
269                    }
270
271                    /* replace dangerous control codes with their visual representations */
272                    if (cp < ' ' && !strchr("\r\n\t", cp))
273                        cp += 0x2400;
274
275                    break;
276                }
277
278                v = xs_utf8_insert(v, cp, &offset);
279            }
280            else {
281                char cc = c;
282                v = xs_insert_m(v, offset, &cc, 1);
283                offset++;
284            }
285        }
286
287        if (c == EOF)
288            *t = JS_ERROR;
289    }
290    else
291    if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
292        double d;
293
294        ungetc(c, f);
295        if (fscanf(f, "%lf", &d) == 1) {
296            *t = JS_NUMBER;
297            v = xs_number_new(d);
298        }
299    }
300    else
301    if (c == 't') {
302        if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
303            *t = JS_TRUE;
304            v = xs_val_new(XSTYPE_TRUE);
305        }
306    }
307    else
308    if (c == 'f') {
309        if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
310            fgetc(f) == 's' && fgetc(f) == 'e') {
311            *t = JS_FALSE;
312            v = xs_val_new(XSTYPE_FALSE);
313        }
314    }
315    else
316    if (c == 'n') {
317        if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
318            *t = JS_NULL;
319            v = xs_val_new(XSTYPE_NULL);
320        }
321    }
322
323    if (*t == JS_ERROR)
324        v = xs_free(v);
325
326    return v;
327}
328
329
330int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c)
331/* loads the next scalar value from the JSON stream */
332/* if the value ahead is compound, value is NULL and pt is set */
333{
334    js_type t;
335
336    *value = _xs_json_load_lexer(f, &t);
337
338    if (t == JS_ERROR)
339        return -1;
340
341    if (t == JS_CBRACK)
342        return 0;
343
344    if (*c > 0) {
345        if (t == JS_COMMA)
346            *value = _xs_json_load_lexer(f, &t);
347        else
348            return -1;
349    }
350
351    if (*value == NULL) {
352        /* possible compound type ahead */
353        if (t == JS_OBRACK)
354            *pt = XSTYPE_LIST;
355        else
356        if (t == JS_OCURLY)
357            *pt = XSTYPE_DICT;
358        else
359            return -1;
360    }
361
362    *c = *c + 1;
363
364    return 1;
365}
366
367
368xs_list *xs_json_load_array(FILE *f)
369/* loads a full JSON array (after the initial OBRACK) */
370{
371    xstype t;
372    xs_list *l = xs_list_new();
373    int c = 0;
374
375    for (;;) {
376        xs *v = NULL;
377        int r = xs_json_load_array_iter(f, &v, &t, &c);
378
379        if (r == -1)
380            l = xs_free(l);
381
382        if (r == 1) {
383            /* partial load? */
384            if (v == NULL) {
385                if (t == XSTYPE_LIST)
386                    v = xs_json_load_array(f);
387                else
388                if (t == XSTYPE_DICT)
389                    v = xs_json_load_object(f);
390            }
391
392            /* still null? fail */
393            if (v == NULL) {
394                l = xs_free(l);
395                break;
396            }
397
398            l = xs_list_append(l, v);
399        }
400        else
401            break;
402    }
403
404    return l;
405}
406
407
408int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c)
409/* loads the next key and scalar value from the JSON stream */
410/* if the value ahead is compound, value is NULL and pt is set */
411{
412    js_type t;
413
414    *key = _xs_json_load_lexer(f, &t);
415
416    if (t == JS_ERROR)
417        return -1;
418
419    if (t == JS_CCURLY)
420        return 0;
421
422    if (*c > 0) {
423        if (t == JS_COMMA)
424            *key = _xs_json_load_lexer(f, &t);
425        else
426            return -1;
427    }
428
429    if (t != JS_STRING)
430        return -1;
431
432    xs_free(_xs_json_load_lexer(f, &t));
433
434    if (t != JS_COLON)
435        return -1;
436
437    *value = _xs_json_load_lexer(f, &t);
438
439    if (*value == NULL) {
440        /* possible complex type ahead */
441        if (t == JS_OBRACK)
442            *pt = XSTYPE_LIST;
443        else
444        if (t == JS_OCURLY)
445            *pt = XSTYPE_DICT;
446        else
447            return -1;
448    }
449
450    *c = *c + 1;
451
452    return 1;
453}
454
455
456xs_dict *xs_json_load_object(FILE *f)
457/* loads a full JSON object (after the initial OCURLY) */
458{
459    xstype t;
460    xs_dict *d = xs_dict_new();
461    int c = 0;
462
463    for (;;) {
464        xs *k = NULL;
465        xs *v = NULL;
466        int r = xs_json_load_object_iter(f, &k, &v, &t, &c);
467
468        if (r == -1)
469            d = xs_free(d);
470
471        if (r == 1) {
472            /* partial load? */
473            if (v == NULL) {
474                if (t == XSTYPE_LIST)
475                    v = xs_json_load_array(f);
476                else
477                if (t == XSTYPE_DICT)
478                    v = xs_json_load_object(f);
479            }
480
481            /* still null? fail */
482            if (v == NULL) {
483                d = xs_free(d);
484                break;
485            }
486
487            d = xs_dict_append(d, k, v);
488        }
489        else
490            break;
491    }
492
493    return d;
494}
495
496
497xs_val *xs_json_loads(const xs_str *json)
498/* loads a string in JSON format and converts to a multiple data */
499{
500    FILE *f;
501    xs_val *v = NULL;
502
503    if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
504        v = xs_json_load(f);
505        fclose(f);
506    }
507
508    return v;
509}
510
511
512xstype xs_json_load_type(FILE *f)
513/* identifies the type of a JSON stream */
514{
515    xstype t = XSTYPE_NULL;
516    js_type jt;
517
518    xs_free(_xs_json_load_lexer(f, &jt));
519
520    if (jt == JS_OBRACK)
521        t = XSTYPE_LIST;
522    else
523    if (jt == JS_OCURLY)
524        t = XSTYPE_DICT;
525
526    return t;
527}
528
529
530xs_val *xs_json_load(FILE *f)
531/* loads a JSON file */
532{
533    xs_val *v = NULL;
534    xstype t = xs_json_load_type(f);
535
536    if (t == XSTYPE_LIST)
537        v = xs_json_load_array(f);
538    else
539    if (t == XSTYPE_DICT)
540        v = xs_json_load_object(f);
541
542    return v;
543}
544
545
546#endif /* XS_IMPLEMENTATION */
547
548#endif /* _XS_JSON_H */