commit 7c8a6b7

default  ·  2023-08-08 11:04:55 +0000 UTC
parent 135e690
Backport from xs.
2 files changed,  +87, -128
+86, -127
  1@@ -6,8 +6,6 @@
  2 
  3 int xs_json_dump_pp(const xs_val *data, int indent, FILE *f);
  4 xs_str *xs_json_dumps_pp(const xs_val *data, int indent);
  5-#define xs_json_dumps(data) xs_json_dumps_pp(data, 0)
  6-#define xs_json_dump(data, f) xs_json_dumps_pp(data, 0, f)
  7 xs_val *xs_json_loads(const xs_str *json);
  8 xs_val *xs_json_load(FILE *f);
  9 
 10@@ -201,17 +199,15 @@ typedef enum {
 11 } js_type;
 12 
 13 
 14-static xs_val *_xs_json_loads_lexer(const char **json, js_type *t)
 15+static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
 16 {
 17-    char c;
 18-    const char *s = *json;
 19+    int c;
 20     xs_val *v = NULL;
 21 
 22-    /* skip blanks */
 23-    while (*s == L' ' || *s == L'\t' || *s == L'\n' || *s == L'\r')
 24-        s++;
 25+    *t = JS_ERROR;
 26 
 27-    c = *s++;
 28+    /* skip blanks */
 29+    while ((c = fgetc(f)) == L' ' || c == L'\t' || c == L'\n' || c == L'\r');
 30 
 31     if (c == '{')
 32         *t = JS_OCURLY;
 33@@ -236,141 +232,103 @@ static xs_val *_xs_json_loads_lexer(const char **json, js_type *t)
 34 
 35         v = xs_str_new(NULL);
 36 
 37-        while ((c = *s) != '"' && c != '\0') {
 38-            char tmp[5];
 39-            int cp, i;
 40-
 41+        while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
 42             if (c == '\\') {
 43-                s++;
 44-                c = *s;
 45-                switch (c) {
 46-                case 'n': c = '\n'; break;
 47-                case 'r': c = '\r'; break;
 48-                case 't': c = '\t'; break;
 49-                case 'u': /* Unicode codepoint as an hex char */
 50-                    s++;
 51-                    strncpy(tmp, s, 4);
 52-                    tmp[4] = '\0';
 53+                unsigned int cp = fgetc(f);
 54 
 55-                    if (strlen(tmp) != 4) {
 56+                switch (cp) {
 57+                case 'n': cp = '\n'; break;
 58+                case 'r': cp = '\r'; break;
 59+                case 't': cp = '\t'; break;
 60+                case 'u': /* Unicode codepoint as an hex char */
 61+                    if (fscanf(f, "%04x", &cp) != 1) {
 62                         *t = JS_ERROR;
 63                         break;
 64                     }
 65 
 66-                    s += 3; /* skip as it was one byte */
 67-
 68-                    sscanf(tmp, "%04x", &i);
 69-
 70-                    if (i >= 0xd800 && i <= 0xdfff) {
 71+                    if (cp >= 0xd800 && cp <= 0xdfff) {
 72                         /* it's a surrogate pair */
 73-                        cp = (i & 0x3ff) << 10;
 74+                        cp = (cp & 0x3ff) << 10;
 75 
 76-                        /* skip to the next value (last char + \ + u)  */
 77-                        s++;
 78-                        if (memcmp(s, "\\u", 2) != 0) {
 79+                        /* \u must follow */
 80+                        if (fgetc(f) != '\\' || fgetc(f) != 'u') {
 81                             *t = JS_ERROR;
 82                             break;
 83                         }
 84-                        s += 2;
 85 
 86-                        strncpy(tmp, s, 4);
 87-                        tmp[4] = '\0';
 88-
 89-                        if (strlen(tmp) != 4) {
 90+                        unsigned int i;
 91+                        if (fscanf(f, "%04x", &i) != 1) {
 92                             *t = JS_ERROR;
 93                             break;
 94                         }
 95 
 96-                        s += 3; /* skip as it was one byte */
 97-
 98-                        sscanf(tmp, "%04x", &i);
 99                         cp |= (i & 0x3ff);
100                         cp += 0x10000;
101                     }
102-                    else
103-                        cp = i;
104 
105                     /* replace dangerous control codes with their visual representations */
106-                    if (cp >= '\0' && cp < ' ' && !strchr("\r\n\t", cp))
107+                    if (cp < ' ' && !strchr("\r\n\t", cp))
108                         cp += 0x2400;
109 
110-                    v = xs_utf8_enc(v, cp);
111-                    c = '\0';
112-
113                     break;
114                 }
115-            }
116-
117-            if (c)
118-                v = xs_append_m(v, &c, 1);
119 
120-            s++;
121+                v = xs_utf8_enc(v, cp);
122+            }
123+            else {
124+                char cc = c;
125+                v = xs_append_m(v, &cc, 1);
126+            }
127         }
128-
129-        if (c != '\0')
130-            s++;
131     }
132     else
133     if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
134-        xs *vn = NULL;
135-
136-        *t = JS_INTEGER;
137-
138-        vn = xs_str_new(NULL);
139-        vn = xs_append_m(vn, &c, 1);
140-
141-        while (((c = *s) >= '0' && c <= '9') || c == '.') {
142-            if (c == '.')
143-                *t = JS_REAL;
144+        double d;
145 
146-            vn = xs_append_m(vn, &c, 1);
147-            s++;
148+        ungetc(c, f);
149+        if (fscanf(f, "%lf", &d) == 1) {
150+            *t = JS_REAL;
151+            v = xs_number_new(d);
152         }
153-
154-        /* convert to XSTYPE_NUMBER */
155-        v = xs_number_new(atof(vn));
156     }
157     else
158-    if (c == 't' && strncmp(s, "rue", 3) == 0) {
159-        s += 3;
160-        *t = JS_TRUE;
161-
162-        v = xs_val_new(XSTYPE_TRUE);
163+    if (c == 't') {
164+        if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
165+            *t = JS_TRUE;
166+            v = xs_val_new(XSTYPE_TRUE);
167+        }
168     }
169     else
170-    if (c == 'f' && strncmp(s, "alse", 4) == 0) {
171-        s += 4;
172-        *t = JS_FALSE;
173-
174-        v = xs_val_new(XSTYPE_FALSE);
175+    if (c == 'f') {
176+        if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
177+            fgetc(f) == 's' && fgetc(f) == 'e') {
178+            *t = JS_FALSE;
179+            v = xs_val_new(XSTYPE_FALSE);
180+        }
181     }
182     else
183-    if (c == 'n' && strncmp(s, "ull", 3) == 0) {
184-        s += 3;
185-        *t = JS_NULL;
186-
187-        v = xs_val_new(XSTYPE_NULL);
188+    if (c == 'n') {
189+        if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
190+            *t = JS_NULL;
191+            v = xs_val_new(XSTYPE_NULL);
192+        }
193     }
194-    else
195-        *t = JS_ERROR;
196-
197-    *json = s;
198 
199     return v;
200 }
201 
202 
203-static xs_list *_xs_json_loads_array(const char **json, js_type *t);
204-static xs_dict *_xs_json_loads_object(const char **json, js_type *t);
205+static xs_list *_xs_json_load_array(FILE *f, js_type *t);
206+static xs_dict *_xs_json_load_object(FILE *f, js_type *t);
207 
208-static xs_val *_xs_json_loads_value(const char **json, js_type *t, xs_val *v)
209+static xs_val *_xs_json_load_value(FILE *f, js_type *t, xs_val *v)
210 /* parses a JSON value */
211 {
212     if (*t == JS_OBRACK)
213-        v = _xs_json_loads_array(json, t);
214+        v = _xs_json_load_array(f, t);
215     else
216     if (*t == JS_OCURLY)
217-        v = _xs_json_loads_object(json, t);
218+        v = _xs_json_load_object(f, t);
219 
220     if (*t >= JS_VALUE)
221         *t = JS_VALUE;
222@@ -381,10 +339,9 @@ static xs_val *_xs_json_loads_value(const char **json, js_type *t, xs_val *v)
223 }
224 
225 
226-static xs_list *_xs_json_loads_array(const char **json, js_type *t)
227+static xs_list *_xs_json_load_array(FILE *f, js_type *t)
228 /* parses a JSON array */
229 {
230-    const char *s = *json;
231     xs *v;
232     xs_list *l;
233     js_type tt;
234@@ -393,18 +350,18 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
235 
236     *t = JS_INCOMPLETE;
237 
238-    v = _xs_json_loads_lexer(&s, &tt);
239+    v = _xs_json_load_lexer(f, &tt);
240 
241     if (tt == JS_CBRACK)
242         *t = JS_ARRAY;
243     else {
244-        v = _xs_json_loads_value(&s, &tt, v);
245+        v = _xs_json_load_value(f, &tt, v);
246 
247         if (tt == JS_VALUE) {
248             l = xs_list_append(l, v);
249 
250             while (*t == JS_INCOMPLETE) {
251-                xs_free(_xs_json_loads_lexer(&s, &tt));
252+                xs_free(_xs_json_load_lexer(f, &tt));
253 
254                 if (tt == JS_CBRACK)
255                     *t = JS_ARRAY;
256@@ -412,8 +369,8 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
257                 if (tt == JS_COMMA) {
258                     xs *v2;
259 
260-                    v2 = _xs_json_loads_lexer(&s, &tt);
261-                    v2 = _xs_json_loads_value(&s, &tt, v2);
262+                    v2 = _xs_json_load_lexer(f, &tt);
263+                    v2 = _xs_json_load_value(f, &tt, v2);
264 
265                     if (tt == JS_VALUE)
266                         l = xs_list_append(l, v2);
267@@ -431,16 +388,13 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
268     if (*t == JS_ERROR)
269         l = xs_free(l);
270 
271-    *json = s;
272-
273     return l;
274 }
275 
276 
277-static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
278+static xs_dict *_xs_json_load_object(FILE *f, js_type *t)
279 /* parses a JSON object */
280 {
281-    const char *s = *json;
282     xs *k1;
283     xs_dict *d;
284     js_type tt;
285@@ -449,40 +403,40 @@ static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
286 
287     *t = JS_INCOMPLETE;
288 
289-    k1 = _xs_json_loads_lexer(&s, &tt);
290+    k1 = _xs_json_load_lexer(f, &tt);
291 
292     if (tt == JS_CCURLY)
293         *t = JS_OBJECT;
294     else
295     if (tt == JS_STRING) {
296-        xs_free(_xs_json_loads_lexer(&s, &tt));
297+        xs_free(_xs_json_load_lexer(f, &tt));
298 
299         if (tt == JS_COLON) {
300             xs *v1;
301 
302-            v1 = _xs_json_loads_lexer(&s, &tt);
303-            v1 = _xs_json_loads_value(&s, &tt, v1);
304+            v1 = _xs_json_load_lexer(f, &tt);
305+            v1 = _xs_json_load_value(f, &tt, v1);
306 
307             if (tt == JS_VALUE) {
308                 d = xs_dict_append(d, k1, v1);
309 
310                 while (*t == JS_INCOMPLETE) {
311-                    xs_free(_xs_json_loads_lexer(&s, &tt));
312+                    xs_free(_xs_json_load_lexer(f, &tt));
313 
314                     if (tt == JS_CCURLY)
315                         *t = JS_OBJECT;
316                     else
317                     if (tt == JS_COMMA) {
318-                        xs *k = _xs_json_loads_lexer(&s, &tt);
319+                        xs *k = _xs_json_load_lexer(f, &tt);
320 
321                         if (tt == JS_STRING) {
322-                            xs_free(_xs_json_loads_lexer(&s, &tt));
323+                            xs_free(_xs_json_load_lexer(f, &tt));
324 
325                             if (tt == JS_COLON) {
326                                 xs *v;
327 
328-                                v = _xs_json_loads_lexer(&s, &tt);
329-                                v = _xs_json_loads_value(&s, &tt, v);
330+                                v = _xs_json_load_lexer(f, &tt);
331+                                v = _xs_json_load_value(f, &tt, v);
332 
333                                 if (tt == JS_VALUE)
334                                     d = xs_dict_append(d, k, v);
335@@ -511,25 +465,38 @@ static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
336     if (*t == JS_ERROR)
337         d = xs_free(d);
338 
339-    *json = s;
340-
341     return d;
342 }
343 
344 
345 xs_val *xs_json_loads(const xs_str *json)
346 /* loads a string in JSON format and converts to a multiple data */
347+{
348+    FILE *f;
349+    xs_val *v = NULL;
350+
351+    if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
352+        v = xs_json_load(f);
353+        fclose(f);
354+    }
355+
356+    return v;
357+}
358+
359+
360+xs_val *xs_json_load(FILE *f)
361+/* loads a JSON file */
362 {
363     xs_val *v = NULL;
364     js_type t;
365 
366-    xs_free(_xs_json_loads_lexer(&json, &t));
367+    xs_free(_xs_json_load_lexer(f, &t));
368 
369     if (t == JS_OBRACK)
370-        v = _xs_json_loads_array(&json, &t);
371+        v = _xs_json_load_array(f, &t);
372     else
373     if (t == JS_OCURLY)
374-        v = _xs_json_loads_object(&json, &t);
375+        v = _xs_json_load_object(f, &t);
376     else
377         t = JS_ERROR;
378 
379@@ -537,14 +504,6 @@ xs_val *xs_json_loads(const xs_str *json)
380 }
381 
382 
383-xs_val *xs_json_load(FILE *f)
384-/* loads a JSON file */
385-{
386-    xs *o = xs_readall(f);
387-    return o ? xs_json_loads(o) : NULL;
388-}
389-
390-
391 #endif /* XS_IMPLEMENTATION */
392 
393 #endif /* _XS_JSON_H */
+1, -1
1@@ -1 +1 @@
2-/* 7ee8d1e20f4adc0a1cddd64b290efee94e055f32 */
3+/* 568c95296bb7d5b40e35e49220403358bc7dadfc */