Commit 86bd92cc0 for imagemagick.org
commit 86bd92cc0fc7bba8224369dce7bb99e04d7bc5cd
Author: Cristy <urban-warrior@imagemagick.org>
Date: Mon Nov 16 20:03:59 2026 -0500
ensure compliance to RFC 3629
diff --git a/MagickCore/token-private.h b/MagickCore/token-private.h
index 9989d8102..07a32df13 100644
--- a/MagickCore/token-private.h
+++ b/MagickCore/token-private.h
@@ -40,7 +40,7 @@ typedef struct
utf_value;
} UTFInfo;
-static UTFInfo
+static const UTFInfo
utf_info[MaxMultibyteCodes] =
{
{ 0x80, 0x00, 0x000007f, 0x0000000 }, /* 1 byte sequence */
@@ -67,6 +67,8 @@ static inline unsigned char *ConvertLatin1ToUTF8(
*magick_restrict q,
*utf8;
+ if (content == (const unsigned char *) NULL)
+ return((unsigned char *) NULL);
length=0;
for (p=content; *p != '\0'; p++)
length+=(*p & 0x80) != 0 ? 2 : 1;
@@ -126,9 +128,11 @@ static inline unsigned char *ConvertMacRomanToUTF8(
*magick_restrict q,
*utf8;
+ if (content == (const unsigned char *) NULL)
+ return((unsigned char *) NULL);
length=0;
for (p=content; *p != '\0'; p++)
- length+=(*p & 0x80) != 0 ? 4 : 1;
+ length+=(*p & 0x80) != 0 ? 3 : 1; /* table maps to at most 3 bytes */
utf8=(unsigned char *) NULL;
if (~length >= 1)
utf8=(unsigned char *) AcquireQuantumMemory(length+1UL,sizeof(*utf8));
@@ -138,7 +142,7 @@ static inline unsigned char *ConvertMacRomanToUTF8(
for (p=content; *p != '\0'; p++)
{
c=(*p);
- if (c > 128)
+ if (c >= 128)
c=(int) macroman_unicode[c-128];
if ((c & 0x80) == 0)
*q++=(unsigned char) c;
@@ -170,41 +174,61 @@ static inline unsigned char *ConvertMacRomanToUTF8(
static inline int GetNextUTFCode(const char *magick_restrict text,
unsigned int *magick_restrict octets)
{
+ const unsigned char
+ *p;
+
int
code;
- ssize_t
+ unsigned int
+ count,
i;
- int
- c,
- unicode;
-
*octets=1;
if (text == (const char *) NULL)
{
errno=EINVAL;
return(-1);
}
- code=(int) (*text++) & 0xff;
- unicode=code;
- for (i=0; i < MaxMultibyteCodes; i++)
- {
- if ((code & utf_info[i].code_mask) == utf_info[i].code_value)
+ p=(const unsigned char *) text;
+ code=(int) p[0];
+ if (code < 0x80)
+ return(code);
+ if (code < 0xc2) /* stray continuation byte or overlong 2-byte lead */
+ goto invalid;
+ if (code < 0xe0)
+ {
+ count=2;
+ code&=0x1f;
+ }
+ else
+ if (code < 0xf0)
{
- unicode&=utf_info[i].utf_mask;
- if (unicode < utf_info[i].utf_value)
- break;
- *octets=(unsigned int) (i+1);
- return(unicode);
+ count=3;
+ code&=0x0f;
}
- c=(int) (*text++ ^ 0x80) & 0xff;
- if ((c & 0xc0) != 0)
- break;
- if (unicode > 0x10FFFF)
- break;
- unicode=(unicode << 6) | c;
+ else
+ if (code < 0xf5)
+ {
+ count=4;
+ code&=0x07;
+ }
+ else
+ goto invalid;
+ for (i=1; i < count; i++)
+ {
+ if ((p[i] & 0xc0) != 0x80)
+ goto invalid;
+ code=(code << 6) | (int) (p[i] & 0x3f);
}
+ if (((count == 3) && (code < 0x800)) || /* overlong */
+ ((count == 4) && (code < 0x10000)) || /* overlong */
+ (code > 0x10ffff) || /* beyond Unicode */
+ ((code >= 0xd800) && (code <= 0xdfff))) /* surrogates */
+ goto invalid;
+ *octets=count;
+ return(code);
+invalid:
errno=EILSEQ;
return(-1);
}
@@ -246,12 +270,11 @@ static inline MagickBooleanType IsUTFSpace(const int code)
static inline MagickBooleanType IsUTFValid(int code)
{
- int
- mask;
-
- mask=(int) 0x7fffffff;
- if (((code & ~mask) != 0) && ((code < 0xd800) || (code > 0xdfff)) &&
- (code != 0xfffe) && (code != 0xffff))
+ if ((code < 0) || (code > 0x10ffff))
+ return(MagickFalse);
+ if ((code >= 0xd800) && (code <= 0xdfff))
+ return(MagickFalse);
+ if ((code & 0xfffe) == 0xfffe) /* U+FFFE/U+FFFF on every plane */
return(MagickFalse);
return(MagickTrue);
}