static value_t read_string(void)
{
char *buf, *temp;
char eseq[10];
size_t i=0, j, sz = 64, ndig;
int c;
value_t s;
u_int32_t wc=0;
buf = malloc(sz);
while (1) {
if (i >= sz-4) { // -4: leaves room for longest utf8 sequence
sz *= 2;
temp = realloc(buf, sz);
if (temp == NULL) {
free(buf);
lerror(ParseError, "read: out of memory reading string");
}
buf = temp;
}
c = ios_getc(F);
if (c == IOS_EOF) {
free(buf);
lerror(ParseError, "read: unexpected end of input in string");
}
if (c == '"')
break;
else if (c == '\\') {
c = ios_getc(F);
if (c == IOS_EOF) {
free(buf);
lerror(ParseError, "read: end of input in escape sequence");
}
j=0;
if (octal_digit(c)) {
do {
eseq[j++] = c;
c = ios_getc(F);
} while (octal_digit(c) && j<3 && (c!=IOS_EOF));
if (c!=IOS_EOF) ios_ungetc(c, F);
eseq[j] = '\0';
wc = strtol(eseq, NULL, 8);
// \DDD and \xXX read bytes, not characters
buf[i++] = ((char)wc);
}
else if ((c=='x' && (ndig=2)) ||
(c=='u' && (ndig=4)) ||
(c=='U' && (ndig=8))) {
c = ios_getc(F);
while (hex_digit(c) && j<ndig && (c!=IOS_EOF)) {
eseq[j++] = c;
c = ios_getc(F);
}
if (c!=IOS_EOF) ios_ungetc(c, F);
eseq[j] = '\0';
if (j) wc = strtol(eseq, NULL, 16);
if (!j || wc > 0x10ffff) {
free(buf);
lerror(ParseError, "read: invalid escape sequence");
}
if (ndig == 2)
buf[i++] = ((char)wc);
else
i += u8_wc_toutf8(&buf[i], wc);
}
else {
buf[i++] = read_escape_control_char((char)c);
}
}
else {
buf[i++] = c;
}
}
s = cvalue_string(i);
memcpy(cvalue_data(s), buf, i);
free(buf);
return s;
}
std::shared_ptr<Object> read_string(std::istream &in) {
std::stringstream buf;
// lambda to get the next char, or throw EOF exception
std::function<int ()> getc = [&in] () -> int {
int c = in.get();
if (in.eof()) {
throw "EOF while reading string";
}
return c;
};
while (1) {
int c = getc();
if (c == '"') {
return std::make_shared<String>(buf.str());
}
if (c == '\\') {
c = getc();
if (octal_digit(c)) {
int j = 0;
char eseq[4] = { 0, 0, 0, 0 } ;
do {
eseq[j++] = c;
c = getc();
} while (octal_digit(c) && j < 3);
in.unget();
if (!octal_digit(c) && j < 3) {
if (c == '\\') {
} else {
buf.str("Invalid digit: ");
buf.put(c);
throw buf.str();
}
}
u8_wc_toutf8(buf, std::strtol(eseq, 0, 8));
} else if (c == 'u') {
int j = 0;
char eseq[5] = { 0, 0, 0, 0, 0 };
c = getc();
while (hex_digit(c) && j < 4) {
eseq[j++] = c;
c = getc();
}
if (j != 4) {
buf.str("Invalid character length: ");
buf << j << ", should be: 4";
throw buf.str();
}
in.unget();
long uni = std::strtol(eseq, 0, 16);
u8_wc_toutf8(buf, uni);
} else {
if (c == 'b') {
buf.put('\b');
} else if (c == 'n') {
buf.put('\n');
} else if (c == 't') {
buf.put('\t');
} else if (c == 'f') {
buf.put('\f');
} else if (c == 'r') {
buf.put('\r');
} else if (c == '"') {
buf.put('\'');
} else if (c == '\\') {
buf.put('\\');
} else {
buf.str("Unsupported escape character: \\");
buf.put(c);
throw buf.str();
}
}
} else {
buf.put(c);
}
}
return Object::nil;
}