static char *parse_key(const char **key, char *p, nx_json_unicode_encoder encoder) {
// on '}' return with *p=='}'
char c;
while ((c = *p++)) {
if (c == '"') {
*key = unescape_string(p, &p, encoder);
if (!*key)
return 0; // propagate error
while (*p && IS_WHITESPACE(*p))
p++;
if (*p == ':')
return p + 1;
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0;
}
else if (IS_WHITESPACE(c) || c == ',') {
// continue
}
else if (c == '}') {
return p - 1;
}
else if (c == '/') {
if (*p == '/') { // line comment
char *ps = p - 1;
p = strchr(p + 1, '\n');
if (!p) {
NX_JSON_REPORT_ERROR("endless comment", ps);
return 0; // error
}
p++;
}
else if (*p == '*') { // block comment
p = skip_block_comment(p + 1);
if (!p)
return 0;
}
else {
NX_JSON_REPORT_ERROR("unexpected chars", p - 1);
return 0; // error
}
}
else {
NX_JSON_REPORT_ERROR("unexpected chars", p - 1);
return 0; // error
}
}
NX_JSON_REPORT_ERROR("unexpected chars", p - 1);
return 0; // error
}
static char* skip_block_comment(char* p) {
// assume p[-2]=='/' && p[-1]=='*'
char* ps=p-2;
if (!*p) {
NX_JSON_REPORT_ERROR("endless comment", ps);
return 0;
}
REPEAT:
p=strchr(p+1, '/');
if (!p) {
NX_JSON_REPORT_ERROR("endless comment", ps);
return 0;
}
if (p[-1]!='*') goto REPEAT;
return p+1;
}
static char* parse_value(nx_json* parent, const char* key, char* p, nx_json_unicode_encoder encoder) {
nx_json* js;
while (1) {
switch (*p) {
case '\0':
NX_JSON_REPORT_ERROR("unexpected end of text", p);
return 0; // error
case ' ': case '\t': case '\n': case '\r':
case ',':
// skip
p++;
break;
case '{':
js=create_json(NX_JSON_OBJECT, key, parent);
if (!js)
return 0;
p++;
while (1) {
const char* new_key = NULL;
p=parse_key(&new_key, p, encoder);
if (!p) return 0; // error
if (*p=='}') return p+1; // end of object
p=parse_value(js, new_key, p, encoder);
if (!p) return 0; // error
}
case '[':
js=create_json(NX_JSON_ARRAY, key, parent);
if (!js)
return 0;
p++;
while (1) {
p=parse_value(js, 0, p, encoder);
if (!p) return 0; // error
if (*p==']') return p+1; // end of array
}
case ']':
return p;
case '"':
p++;
js=create_json(NX_JSON_STRING, key, parent);
if (!js)
return 0;
js->text_value=unescape_string(p, &p, encoder);
if (!js->text_value) return 0; // propagate error
return p;
case '-': case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9':
{
char* pe;
js=create_json(NX_JSON_INTEGER, key, parent);
if (!js)
return 0;
js->int_value=strtol(p, &pe, 0);
if (pe==p) {
NX_JSON_REPORT_ERROR("invalid number", p);
return 0; // error
}
#ifndef __KERNEL__
if (*pe=='.' || *pe=='e' || *pe=='E') { // double value
js->type=NX_JSON_DOUBLE;
js->dbl_value=strtod(p, &pe);
if (pe==p) {
NX_JSON_REPORT_ERROR("invalid number", p);
return 0; // error
}
}
else {
js->dbl_value=js->int_value;
}
#endif
return pe;
}
case 't':
if (!strncmp(p, "true", 4)) {
js=create_json(NX_JSON_BOOL, key, parent);
if (!js)
return 0;
js->int_value=1;
return p+4;
}
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0; // error
case 'f':
if (!strncmp(p, "false", 5)) {
js=create_json(NX_JSON_BOOL, key, parent);
if (!js)
return 0;
js->int_value=0;
return p+5;
}
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0; // error
case 'n':
if (!strncmp(p, "null", 4)) {
js = create_json(NX_JSON_NULL, key, parent);
if (!js)
return 0;
return p+4;
}
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0; // error
case '/': // comment
if (p[1]=='/') { // line comment
char* ps=p;
p=strchr(p+2, '\n');
if (!p) {
NX_JSON_REPORT_ERROR("endless comment", ps);
return 0; // error
}
p++;
}
else if (p[1]=='*') { // block comment
p=skip_block_comment(p+2);
if (!p) return 0;
}
else {
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0; // error
}
break;
default:
NX_JSON_REPORT_ERROR("unexpected chars", p);
return 0; // error
}
}
}
static char* unescape_string(char* s, char** end, nx_json_unicode_encoder encoder) {
char* p=s;
char* d=s;
char c;
while ((c=*p++)) {
if (c=='"') {
*d='\0';
*end=p;
return s;
}
else if (c=='\\') {
switch (*p) {
case '\\':
case '/':
case '"':
*d++=*p++;
break;
case 'b':
*d++='\b'; p++;
break;
case 'f':
*d++='\f'; p++;
break;
case 'n':
*d++='\n'; p++;
break;
case 'r':
*d++='\r'; p++;
break;
case 't':
*d++='\t'; p++;
break;
case 'u': // unicode
{
char *ps;
int h1, h2, h3, h4;
unsigned int codepoint;
if (!encoder) {
// leave untouched
*d++=c;
break;
}
ps=p-1;
if ((h1=hex_val(p[1]))<0 || (h2=hex_val(p[2]))<0 || (h3=hex_val(p[3]))<0 || (h4=hex_val(p[4]))<0) {
NX_JSON_REPORT_ERROR("invalid unicode escape", p-1);
return 0;
}
codepoint=h1<<12|h2<<8|h3<<4|h4;
if ((codepoint & 0xfc00)==0xd800) { // high surrogate; need one more unicode to succeed
unsigned int codepoint2;
p+=6;
if (p[-1]!='\\' || *p!='u' || (h1=hex_val(p[1]))<0 || (h2=hex_val(p[2]))<0 || (h3=hex_val(p[3]))<0 || (h4=hex_val(p[4]))<0) {
NX_JSON_REPORT_ERROR("invalid unicode surrogate", ps);
return 0;
}
codepoint2=h1<<12|h2<<8|h3<<4|h4;
if ((codepoint2 & 0xfc00)!=0xdc00) {
NX_JSON_REPORT_ERROR("invalid unicode surrogate", ps);
return 0;
}
codepoint=0x10000+((codepoint-0xd800)<<10)+(codepoint2-0xdc00);
}
if (!encoder(codepoint, d, &d)) {
NX_JSON_REPORT_ERROR("invalid codepoint", ps);
return 0;
}
p+=5;
break;
}
default:
// leave untouched
*d++=c;
break;
}
}
else {
*d++=c;
}
}
NX_JSON_REPORT_ERROR("no closing quote for string", s);
return 0;
}
