// returns # of characters in utf-8 string (excluding nul terminator)
size_t String::utf8_len(const char *s) {
if (s == nullptr) {
throw ReferenceError();
}
if (!*s) {
return 0;
}
size_t n = 0;
while(*s) {
if ((*s & 0x80) == 0x80) { // negative value
if ((*s & 0xe0) == 0xc0) { // 1 more byte
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
n++;
} else {
if ((*s & 0xf0) == 0xe0) { // 2 more bytes
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
n++;
} else {
if ((*s & 0xf8) == 0xf0) { // 3 more bytes
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
s++;
n++;
} else {
throw StringEncodingError();
}
}
}
} else {
s++;
n++;
}
}
return n;
}
void VbpValidator::processReference(Reference reference) {
if (!reference.existsInRegistry()) {
ReferenceError referenceError = ReferenceError(reference);
this->referenceErrors.push_back(referenceError);
}
//TODO: include filepath checks
}
void String::utf8_encode(const rune *r, char *s, size_t n) {
if (r == nullptr || s == nullptr) {
throw ReferenceError();
}
if (!n) {
throw ValueError();
}
rune code;
while(*r && n > 0) {
code = *r;
if (code < 0x80) {
if (n <= 1) {
throw StringEncodingError("String::utf8_encode(): buffer too small");
}
*s++ = code;
n--;
} else {
if (code < 0x800) {
if (n <= 2) {
throw StringEncodingError("String::utf8_encode(): buffer too small");
}
*s++ = ((code & 0x7c0) >> 6) | 0xc0;
*s++ = (code & 0x3f) | 0x80;
n -= 2;
} else {
if ((code >= 0x800 && code < 0xd800) || (code >= 0xe000 && code < 0x10000)) {
if (n <= 3) {
throw StringEncodingError("String::utf8_encode(): buffer too small");
}
*s++ = ((code & 0xf000) >> 12) | 0xe0;
*s++ = ((code & 0xfc0) >> 6) | 0x80;
*s++ = (code & 0x3f) | 0x80;
n -= 3;
} else {
if (code > 0x10000 && code <= 0x10ffff) {
if (n <= 4) {
throw StringEncodingError("String::utf8_encode(): buffer too small");
}
*s++ = ((code & 0x1c0000) >> 18) | 0xf0;
*s++ = ((code & 0x3f000) >> 12) | 0x80;
*s++ = ((code & 0xfc0) >> 6) | 0x80;
*s++ = (code & 0x3f) | 0x80;
n -= 4;
} else {
throw StringEncodingError();
}
}
}
void drop_privs(const char *username) {
if (username == nullptr) {
throw ReferenceError();
}
struct passwd *pw = ::getpwnam(username);
if (pw == nullptr) {
char buf[256];
std::snprintf(buf, sizeof(buf),
"failed to drop privileges: no such user '%s'", username);
panic(buf);
}
drop_privs(pw->pw_uid, pw->pw_gid);
}
// default argument sep = ' '
String String::join(const Array<String>& a, rune sep) const {
if (!a.len()) {
return String();
}
// first calculate total size
size_t rune_size = utf8_encoded_size(sep);
size_t total = 0;
const char *p;
for(size_t i = 0; i < a.len(); i++) {
p = a[i].c_str();
if (p == nullptr) {
throw ReferenceError();
}
total += std::strlen(p) + rune_size;
}
total -= rune_size;
// make the joined string
String s;
s.grow(total + 1);
s.s_data[0] = 0;
if (a[0].s_data != nullptr) {
std::strcpy(s.s_data, a[0].s_data);
}
rune r[2] = { sep, 0 };
char rs[5];
utf8_encode(r, rs, sizeof(rs));
for(size_t i = 1; i < a.len(); i++) {
std::strcat(s.s_data, rs);
if (a[i].s_data) {
std::strcat(s.s_data, a[i].s_data);
}
}
s.s_len = utf8_len(s.s_data);
return s;
}
void String::utf8_decode(const char *s, rune *r, size_t n) {
if (s == nullptr || r == nullptr) {
throw ReferenceError();
}
if (!n) {
throw ValueError();
}
rune code;
while(*s && n > 0) {
if ((*s & 0x80) == 0x80) { // negative value
if ((*s & 0xe0) == 0xc0) { // 1 more byte
code = (*s & 0x1f) << 6;
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= (*s & 0x3f);
s++;
*r = code;
r++;
n--;
} else {
if ((*s & 0xf0) == 0xe0) { // 2 more bytes
code = (*s & 0xf) << 12;
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= ((*s & 0x3f) << 6);
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= (*s & 0x3f);
s++;
*r = code;
r++;
n--;
} else {
if ((*s & 0xf8) == 0xf0) { // 3 more bytes
code = (*s & 7) << 18;
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= ((*s & 0x3f) << 12);
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= ((*s & 0x3f) << 6);
s++;
if ((*s & 0xc0) != 0x80) {
throw StringEncodingError();
}
code |= (*s & 0x3f);
s++;
*r = code;
r++;
n--;
} else {
throw StringEncodingError();
}
}
}
} else {
*r = *s & 0x7f;
r++;
s++;
n--;
}
}
*r = 0;
}
