c/libxml2
1
0
Fork 0
mirror of https://gitlab.gnome.org/GNOME/libxml2 synced 2025-03-28 21:33:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

encoding: Readd some UTF-8 validation to encoders

Browse Source 
This isn't strictly needed but avoids generating invalid UTF-16 and
unsigned integer overflows.
This commit is contained in:
Nick Wellnhofer 2024-07-10 22:26:19 +02:00
parent ae6e2ee7ec
commit d099795611
1 changed files with 121 additions and 66 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

187
encoding.c
View File

@ -2134,7 +2134,7 @@ UTF8ToLatin1(unsigned char* out, int *outlen,
if (c < 0x80) { if (c < 0x80) {
*out++ = c; *out++ = c;
} else if (c < 0xC4) { } else if ((c >= 0xC2) && (c <= 0xC3)) {
if (inend - in < 2) if (inend - in < 2)
break; break;
in++; in++;
@ -2272,48 +2272,75 @@ UTF8ToUTF16LE(unsigned char *out, int *outlen,
inend = in + *inlen; inend = in + *inlen;
outend = out + (*outlen & ~1); outend = out + (*outlen & ~1);
while (in < inend) { while (in < inend) {
if (out >= outend)
goto done;
c = in[0]; c = in[0];
if (c < 0x80) { if (c < 0x80) {
if (out >= outend)
goto done;
out[0] = c; out[0] = c;
out[1] = 0; out[1] = 0;
in += 1; in += 1;
out += 2; out += 2;
} else if (c < 0xE0) { } else {
if (inend - in < 2) int i, len;
unsigned min;
if (c < 0xE0) {
if (c < 0xC2) {
ret = XML_ENC_ERR_INPUT;
goto done;
}
c &= 0x1F;
len = 2;
min = 0x80;
} else if (c < 0xF0) {
c &= 0x0F;
len = 3;
min = 0x800;
} else {
c &= 0x0F;
len = 4;
min = 0x10000;
}
if (inend - in < len)
break; break;
c = ((c & 0x1F) << 6) | (in[1] & 0x3F);
out[0] = c & 0xFF; for (i = 1; i < len; i++) {
out[1] = c >> 8; if ((in[i] & 0xC0) != 0x80) {
in += 2; ret = XML_ENC_ERR_INPUT;
out += 2; goto done;
} else if (c < 0xF0) { }
if (inend - in < 3) c = (c << 6) | (in[i] & 0x3F);
break; }
c = ((c & 0x0F) << 12) | ((in[1] & 0x3F) << 6) | (in[2] & 0x3F);
out[0] = c & 0xFF; if ((c < min) ||
out[1] = c >> 8; ((c >= 0xD800) && (c <= 0xDFFF)) ||
in += 3; (c > 0x10FFFF)) {
out += 2; ret = XML_ENC_ERR_INPUT;
} else { /* c >= 0xF0 */
if (inend - in < 4)
break;
if (outend - out < 4)
goto done; goto done;
c = ((c & 0x0F) << 18) | ((in[1] & 0x3F) << 12) | }
((in[2] & 0x3F) << 6) | (in[3] & 0x3F);
c -= 0x10000; if (c < 0x10000) {
d = (c & 0x03FF) | 0xDC00; if (out >= outend)
c = (c >> 10) | 0xD800; goto done;
out[0] = c & 0xFF; out[0] = c & 0xFF;
out[1] = c >> 8; out[1] = c >> 8;
out[2] = d & 0xFF; out += 2;
out[3] = d >> 8; } else {
in += 4; if (outend - out < 4)
out += 4; goto done;
c -= 0x10000;
d = (c & 0x03FF) | 0xDC00;
c = (c >> 10) | 0xD800;
out[0] = c & 0xFF;
out[1] = c >> 8;
out[2] = d & 0xFF;
out[3] = d >> 8;
out += 4;
}
in += len;
} }
} }
@ -2438,47 +2465,75 @@ UTF8ToUTF16BE(unsigned char *out, int *outlen,
inend = in + *inlen; inend = in + *inlen;
outend = out + (*outlen & ~1); outend = out + (*outlen & ~1);
while (in < inend) { while (in < inend) {
if (out >= outend)
goto done;
c = in[0]; c = in[0];
if (c < 0x80) { if (c < 0x80) {
if (out >= outend)
goto done;
out[0] = 0; out[0] = 0;
out[1] = c; out[1] = c;
in += 1; in += 1;
out += 2; out += 2;
} else if (c < 0xE0) { } else {
if (inend - in < 2) int i, len;
unsigned min;
if (c < 0xE0) {
if (c < 0xC2) {
ret = XML_ENC_ERR_INPUT;
goto done;
}
c &= 0x1F;
len = 2;
min = 0x80;
} else if (c < 0xF0) {
c &= 0x0F;
len = 3;
min = 0x800;
} else {
c &= 0x0F;
len = 4;
min = 0x10000;
}
if (inend - in < len)
break; break;
c = ((c & 0x1F) << 6) | (in[1] & 0x3F);
out[0] = c >> 8; for (i = 1; i < len; i++) {
out[1] = c & 0xFF; if ((in[i] & 0xC0) != 0x80) {
in += 2; ret = XML_ENC_ERR_INPUT;
out += 2; goto done;
} else if (c < 0xF0) { }
if (inend - in < 3) c = (c << 6) | (in[i] & 0x3F);
break; }
c = ((c & 0x0F) << 12) | ((in[1] & 0x3F) << 6) | (in[2] & 0x3F);
out[0] = c >> 8; if ((c < min) ||
out[1] = c & 0xFF; ((c >= 0xD800) && (c <= 0xDFFF)) ||
in += 3; (c > 0x10FFFF)) {
out += 2; ret = XML_ENC_ERR_INPUT;
} else { /* c >= 0xF0 */
if (inend - in < 4)
break;
if (outend - out < 4)
goto done; goto done;
c = ((c & 0x0F) << 18) | ((in[1] & 0x3F) << 12) | }
((in[2] & 0x3F) << 6) | (in[3] & 0x3F);
c -= 0x10000; if (c < 0x10000) {
d = (c & 0x03FF) | 0xDC00; if (out >= outend)
c = (c >> 10) | 0xD800; goto done;
out[0] = c >> 8; out[0] = c >> 8;
out[1] = c & 0xFF; out[1] = c & 0xFF;
out[2] = d >> 8; out += 2;
out[3] = d & 0xFF; } else {
in += 4; if (outend - out < 4)
out += 4; goto done;
c -= 0x10000;
d = (c & 0x03FF) | 0xDC00;
c = (c >> 10) | 0xD800;
out[0] = c >> 8;
out[1] = c & 0xFF;
out[2] = d >> 8;
out[3] = d & 0xFF;
out += 4;
}
in += len;
} }
} }