Backport from xs.

This commit is contained in:
default 2023-08-08 13:04:55 +02:00
parent 135e690a27
commit 7c8a6b71d9
2 changed files with 91 additions and 132 deletions

221
xs_json.h
View File

@ -6,8 +6,6 @@
int xs_json_dump_pp(const xs_val *data, int indent, FILE *f); int xs_json_dump_pp(const xs_val *data, int indent, FILE *f);
xs_str *xs_json_dumps_pp(const xs_val *data, int indent); xs_str *xs_json_dumps_pp(const xs_val *data, int indent);
#define xs_json_dumps(data) xs_json_dumps_pp(data, 0)
#define xs_json_dump(data, f) xs_json_dumps_pp(data, 0, f)
xs_val *xs_json_loads(const xs_str *json); xs_val *xs_json_loads(const xs_str *json);
xs_val *xs_json_load(FILE *f); xs_val *xs_json_load(FILE *f);
@ -201,17 +199,15 @@ typedef enum {
} js_type; } js_type;
static xs_val *_xs_json_loads_lexer(const char **json, js_type *t) static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
{ {
char c; int c;
const char *s = *json;
xs_val *v = NULL; xs_val *v = NULL;
/* skip blanks */ *t = JS_ERROR;
while (*s == L' ' || *s == L'\t' || *s == L'\n' || *s == L'\r')
s++;
c = *s++; /* skip blanks */
while ((c = fgetc(f)) == L' ' || c == L'\t' || c == L'\n' || c == L'\r');
if (c == '{') if (c == '{')
*t = JS_OCURLY; *t = JS_OCURLY;
@ -236,141 +232,103 @@ static xs_val *_xs_json_loads_lexer(const char **json, js_type *t)
v = xs_str_new(NULL); v = xs_str_new(NULL);
while ((c = *s) != '"' && c != '\0') { while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
char tmp[5];
int cp, i;
if (c == '\\') { if (c == '\\') {
s++; unsigned int cp = fgetc(f);
c = *s;
switch (c) {
case 'n': c = '\n'; break;
case 'r': c = '\r'; break;
case 't': c = '\t'; break;
case 'u': /* Unicode codepoint as an hex char */
s++;
strncpy(tmp, s, 4);
tmp[4] = '\0';
if (strlen(tmp) != 4) { switch (cp) {
case 'n': cp = '\n'; break;
case 'r': cp = '\r'; break;
case 't': cp = '\t'; break;
case 'u': /* Unicode codepoint as an hex char */
if (fscanf(f, "%04x", &cp) != 1) {
*t = JS_ERROR; *t = JS_ERROR;
break; break;
} }
s += 3; /* skip as it was one byte */ if (cp >= 0xd800 && cp <= 0xdfff) {
sscanf(tmp, "%04x", &i);
if (i >= 0xd800 && i <= 0xdfff) {
/* it's a surrogate pair */ /* it's a surrogate pair */
cp = (i & 0x3ff) << 10; cp = (cp & 0x3ff) << 10;
/* skip to the next value (last char + \ + u) */ /* \u must follow */
s++; if (fgetc(f) != '\\' || fgetc(f) != 'u') {
if (memcmp(s, "\\u", 2) != 0) {
*t = JS_ERROR;
break;
}
s += 2;
strncpy(tmp, s, 4);
tmp[4] = '\0';
if (strlen(tmp) != 4) {
*t = JS_ERROR; *t = JS_ERROR;
break; break;
} }
s += 3; /* skip as it was one byte */ unsigned int i;
if (fscanf(f, "%04x", &i) != 1) {
*t = JS_ERROR;
break;
}
sscanf(tmp, "%04x", &i);
cp |= (i & 0x3ff); cp |= (i & 0x3ff);
cp += 0x10000; cp += 0x10000;
} }
else
cp = i;
/* replace dangerous control codes with their visual representations */ /* replace dangerous control codes with their visual representations */
if (cp >= '\0' && cp < ' ' && !strchr("\r\n\t", cp)) if (cp < ' ' && !strchr("\r\n\t", cp))
cp += 0x2400; cp += 0x2400;
v = xs_utf8_enc(v, cp);
c = '\0';
break; break;
} }
v = xs_utf8_enc(v, cp);
}
else {
char cc = c;
v = xs_append_m(v, &cc, 1);
} }
if (c)
v = xs_append_m(v, &c, 1);
s++;
} }
if (c != '\0')
s++;
} }
else else
if (c == '-' || (c >= '0' && c <= '9') || c == '.') { if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
xs *vn = NULL; double d;
*t = JS_INTEGER; ungetc(c, f);
if (fscanf(f, "%lf", &d) == 1) {
vn = xs_str_new(NULL); *t = JS_REAL;
vn = xs_append_m(vn, &c, 1); v = xs_number_new(d);
while (((c = *s) >= '0' && c <= '9') || c == '.') {
if (c == '.')
*t = JS_REAL;
vn = xs_append_m(vn, &c, 1);
s++;
} }
/* convert to XSTYPE_NUMBER */
v = xs_number_new(atof(vn));
} }
else else
if (c == 't' && strncmp(s, "rue", 3) == 0) { if (c == 't') {
s += 3; if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
*t = JS_TRUE; *t = JS_TRUE;
v = xs_val_new(XSTYPE_TRUE);
v = xs_val_new(XSTYPE_TRUE); }
} }
else else
if (c == 'f' && strncmp(s, "alse", 4) == 0) { if (c == 'f') {
s += 4; if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
*t = JS_FALSE; fgetc(f) == 's' && fgetc(f) == 'e') {
*t = JS_FALSE;
v = xs_val_new(XSTYPE_FALSE); v = xs_val_new(XSTYPE_FALSE);
}
} }
else else
if (c == 'n' && strncmp(s, "ull", 3) == 0) { if (c == 'n') {
s += 3; if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
*t = JS_NULL; *t = JS_NULL;
v = xs_val_new(XSTYPE_NULL);
v = xs_val_new(XSTYPE_NULL); }
} }
else
*t = JS_ERROR;
*json = s;
return v; return v;
} }
static xs_list *_xs_json_loads_array(const char **json, js_type *t); static xs_list *_xs_json_load_array(FILE *f, js_type *t);
static xs_dict *_xs_json_loads_object(const char **json, js_type *t); static xs_dict *_xs_json_load_object(FILE *f, js_type *t);
static xs_val *_xs_json_loads_value(const char **json, js_type *t, xs_val *v) static xs_val *_xs_json_load_value(FILE *f, js_type *t, xs_val *v)
/* parses a JSON value */ /* parses a JSON value */
{ {
if (*t == JS_OBRACK) if (*t == JS_OBRACK)
v = _xs_json_loads_array(json, t); v = _xs_json_load_array(f, t);
else else
if (*t == JS_OCURLY) if (*t == JS_OCURLY)
v = _xs_json_loads_object(json, t); v = _xs_json_load_object(f, t);
if (*t >= JS_VALUE) if (*t >= JS_VALUE)
*t = JS_VALUE; *t = JS_VALUE;
@ -381,10 +339,9 @@ static xs_val *_xs_json_loads_value(const char **json, js_type *t, xs_val *v)
} }
static xs_list *_xs_json_loads_array(const char **json, js_type *t) static xs_list *_xs_json_load_array(FILE *f, js_type *t)
/* parses a JSON array */ /* parses a JSON array */
{ {
const char *s = *json;
xs *v; xs *v;
xs_list *l; xs_list *l;
js_type tt; js_type tt;
@ -393,18 +350,18 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
*t = JS_INCOMPLETE; *t = JS_INCOMPLETE;
v = _xs_json_loads_lexer(&s, &tt); v = _xs_json_load_lexer(f, &tt);
if (tt == JS_CBRACK) if (tt == JS_CBRACK)
*t = JS_ARRAY; *t = JS_ARRAY;
else { else {
v = _xs_json_loads_value(&s, &tt, v); v = _xs_json_load_value(f, &tt, v);
if (tt == JS_VALUE) { if (tt == JS_VALUE) {
l = xs_list_append(l, v); l = xs_list_append(l, v);
while (*t == JS_INCOMPLETE) { while (*t == JS_INCOMPLETE) {
xs_free(_xs_json_loads_lexer(&s, &tt)); xs_free(_xs_json_load_lexer(f, &tt));
if (tt == JS_CBRACK) if (tt == JS_CBRACK)
*t = JS_ARRAY; *t = JS_ARRAY;
@ -412,8 +369,8 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
if (tt == JS_COMMA) { if (tt == JS_COMMA) {
xs *v2; xs *v2;
v2 = _xs_json_loads_lexer(&s, &tt); v2 = _xs_json_load_lexer(f, &tt);
v2 = _xs_json_loads_value(&s, &tt, v2); v2 = _xs_json_load_value(f, &tt, v2);
if (tt == JS_VALUE) if (tt == JS_VALUE)
l = xs_list_append(l, v2); l = xs_list_append(l, v2);
@ -431,16 +388,13 @@ static xs_list *_xs_json_loads_array(const char **json, js_type *t)
if (*t == JS_ERROR) if (*t == JS_ERROR)
l = xs_free(l); l = xs_free(l);
*json = s;
return l; return l;
} }
static xs_dict *_xs_json_loads_object(const char **json, js_type *t) static xs_dict *_xs_json_load_object(FILE *f, js_type *t)
/* parses a JSON object */ /* parses a JSON object */
{ {
const char *s = *json;
xs *k1; xs *k1;
xs_dict *d; xs_dict *d;
js_type tt; js_type tt;
@ -449,40 +403,40 @@ static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
*t = JS_INCOMPLETE; *t = JS_INCOMPLETE;
k1 = _xs_json_loads_lexer(&s, &tt); k1 = _xs_json_load_lexer(f, &tt);
if (tt == JS_CCURLY) if (tt == JS_CCURLY)
*t = JS_OBJECT; *t = JS_OBJECT;
else else
if (tt == JS_STRING) { if (tt == JS_STRING) {
xs_free(_xs_json_loads_lexer(&s, &tt)); xs_free(_xs_json_load_lexer(f, &tt));
if (tt == JS_COLON) { if (tt == JS_COLON) {
xs *v1; xs *v1;
v1 = _xs_json_loads_lexer(&s, &tt); v1 = _xs_json_load_lexer(f, &tt);
v1 = _xs_json_loads_value(&s, &tt, v1); v1 = _xs_json_load_value(f, &tt, v1);
if (tt == JS_VALUE) { if (tt == JS_VALUE) {
d = xs_dict_append(d, k1, v1); d = xs_dict_append(d, k1, v1);
while (*t == JS_INCOMPLETE) { while (*t == JS_INCOMPLETE) {
xs_free(_xs_json_loads_lexer(&s, &tt)); xs_free(_xs_json_load_lexer(f, &tt));
if (tt == JS_CCURLY) if (tt == JS_CCURLY)
*t = JS_OBJECT; *t = JS_OBJECT;
else else
if (tt == JS_COMMA) { if (tt == JS_COMMA) {
xs *k = _xs_json_loads_lexer(&s, &tt); xs *k = _xs_json_load_lexer(f, &tt);
if (tt == JS_STRING) { if (tt == JS_STRING) {
xs_free(_xs_json_loads_lexer(&s, &tt)); xs_free(_xs_json_load_lexer(f, &tt));
if (tt == JS_COLON) { if (tt == JS_COLON) {
xs *v; xs *v;
v = _xs_json_loads_lexer(&s, &tt); v = _xs_json_load_lexer(f, &tt);
v = _xs_json_loads_value(&s, &tt, v); v = _xs_json_load_value(f, &tt, v);
if (tt == JS_VALUE) if (tt == JS_VALUE)
d = xs_dict_append(d, k, v); d = xs_dict_append(d, k, v);
@ -511,8 +465,6 @@ static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
if (*t == JS_ERROR) if (*t == JS_ERROR)
d = xs_free(d); d = xs_free(d);
*json = s;
return d; return d;
} }
@ -520,18 +472,13 @@ static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
xs_val *xs_json_loads(const xs_str *json) xs_val *xs_json_loads(const xs_str *json)
/* loads a string in JSON format and converts to a multiple data */ /* loads a string in JSON format and converts to a multiple data */
{ {
FILE *f;
xs_val *v = NULL; xs_val *v = NULL;
js_type t;
xs_free(_xs_json_loads_lexer(&json, &t)); if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
v = xs_json_load(f);
if (t == JS_OBRACK) fclose(f);
v = _xs_json_loads_array(&json, &t); }
else
if (t == JS_OCURLY)
v = _xs_json_loads_object(&json, &t);
else
t = JS_ERROR;
return v; return v;
} }
@ -540,8 +487,20 @@ xs_val *xs_json_loads(const xs_str *json)
xs_val *xs_json_load(FILE *f) xs_val *xs_json_load(FILE *f)
/* loads a JSON file */ /* loads a JSON file */
{ {
xs *o = xs_readall(f); xs_val *v = NULL;
return o ? xs_json_loads(o) : NULL; js_type t;
xs_free(_xs_json_load_lexer(f, &t));
if (t == JS_OBRACK)
v = _xs_json_load_array(f, &t);
else
if (t == JS_OCURLY)
v = _xs_json_load_object(f, &t);
else
t = JS_ERROR;
return v;
} }

View File

@ -1 +1 @@
/* 7ee8d1e20f4adc0a1cddd64b290efee94e055f32 */ /* 568c95296bb7d5b40e35e49220403358bc7dadfc */