PCRE_EXPORT BOOL
_pcre_xclass(int c, const uschar *data)
{
int t;
BOOL negated = (*data & XCL_NOT) != 0;
/* Character values < 256 are matched against a bitmap, if one is present. If
not, we still carry on, because there may be ranges that start below 256 in the
additional data. */
if (c < 256)
{
if ((*data & XCL_MAP) != 0 && (data[1 + c/8] & (1 << (c&7))) != 0)
return !negated; /* char found */
}
/* First skip the bit map if present. Then match against the list of Unicode
properties or large chars or ranges that end with a large char. We won't ever
encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */
if ((*data++ & XCL_MAP) != 0) data += 32;
while ((t = *data++) != XCL_END)
{
int x, y;
if (t == XCL_SINGLE)
{
GETCHARINC(x, data);
if (c == x) return !negated;
}
else if (t == XCL_RANGE)
{
GETCHARINC(x, data);
GETCHARINC(y, data);
if (c >= x && c <= y) return !negated;
}
#ifdef SUPPORT_UCP
else /* XCL_PROP & XCL_NOTPROP */
{
int chartype, othercase;
int rqdtype = *data++;
int category = _pcre_ucp_findchar(c, &chartype, &othercase);
if (rqdtype >= 128)
{
if ((rqdtype - 128 == category) == (t == XCL_PROP)) return !negated;
}
else
{
if ((rqdtype == chartype) == (t == XCL_PROP)) return !negated;
}
}
#endif /* SUPPORT_UCP */
}
return negated; /* char did not match */
}
static const uschar *
set_table_bit(uschar *start_bits, const uschar *p, BOOL caseless,
compile_data *cd, BOOL utf8)
{
unsigned int c = *p;
SET_BIT(c);
#ifdef SUPPORT_UTF8
if (utf8 && c > 127)
{
GETCHARINC(c, p);
#ifdef SUPPORT_UCP
if (caseless)
{
uschar buff[8];
c = UCD_OTHERCASE(c);
(void)_pcre_ord2utf8(c, buff);
SET_BIT(buff[0]);
}
#endif
return p;
}
#endif
/* Not UTF-8 mode, or character is less than 127. */
if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]);
return p + 1;
}
BOOL
_pcre_xclass(int c, const uschar *data)
{
int t;
BOOL negated = (*data & XCL_NOT) != 0;
/* Character values < 256 are matched against a bitmap, if one is present. If
not, we still carry on, because there may be ranges that start below 256 in the
additional data. */
if (c < 256)
{
if ((*data & XCL_MAP) != 0 && (data[1 + c/8] & (1 << (c&7))) != 0)
return !negated; /* char found */
}
/* First skip the bit map if present. Then match against the list of Unicode
properties or large chars or ranges that end with a large char. We won't ever
encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */
if ((*data++ & XCL_MAP) != 0) data += 32;
while ((t = *data++) != XCL_END)
{
int x, y;
if (t == XCL_SINGLE)
{
GETCHARINC(x, data);
if (c == x) return !negated;
}
else if (t == XCL_RANGE)
{
GETCHARINC(x, data);
GETCHARINC(y, data);
if (c >= x && c <= y) return !negated;
}
#ifdef SUPPORT_UCP
else /* XCL_PROP & XCL_NOTPROP */
{
const ucd_record *prop = GET_UCD(c);
switch(*data)
{
case PT_ANY:
if (t == XCL_PROP) return !negated;
break;
case PT_LAMP:
if ((prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt) == (t == XCL_PROP)) return !negated;
break;
case PT_GC:
if ((data[1] == _pcre_ucp_gentype[prop->chartype]) == (t == XCL_PROP))
return !negated;
break;
case PT_PC:
if ((data[1] == prop->chartype) == (t == XCL_PROP)) return !negated;
break;
case PT_SC:
if ((data[1] == prop->script) == (t == XCL_PROP)) return !negated;
break;
case PT_ALNUM:
if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
_pcre_ucp_gentype[prop->chartype] == ucp_N) == (t == XCL_PROP))
return !negated;
break;
case PT_SPACE: /* Perl space */
if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
== (t == XCL_PROP))
return !negated;
break;
case PT_PXSPACE: /* POSIX space */
if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
c == CHAR_FF || c == CHAR_CR) == (t == XCL_PROP))
return !negated;
break;
case PT_WORD:
if ((_pcre_ucp_gentype[prop->chartype] == ucp_L ||
_pcre_ucp_gentype[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE)
== (t == XCL_PROP))
return !negated;
break;
/* This should never occur, but compilers may mutter if there is no
default. */
default:
return FALSE;
}
data += 2;
}
#endif /* SUPPORT_UCP */
}
return negated; /* char did not match */
}
static const pcre_uchar *
set_table_bit(pcre_uint8 *start_bits, const pcre_uchar *p, BOOL caseless,
compile_data *cd, BOOL utf)
{
pcre_uint32 c = *p;
#ifdef COMPILE_PCRE8
SET_BIT(c);
#ifdef SUPPORT_UTF
if (utf && c > 127)
{
GETCHARINC(c, p);
#ifdef SUPPORT_UCP
if (caseless)
{
pcre_uchar buff[6];
c = UCD_OTHERCASE(c);
(void)PRIV(ord2utf)(c, buff);
SET_BIT(buff[0]);
}
#endif /* Not SUPPORT_UCP */
return p;
}
#else /* Not SUPPORT_UTF */
(void)(utf); /* Stops warning for unused parameter */
#endif /* SUPPORT_UTF */
/* Not UTF-8 mode, or character is less than 127. */
if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]);
return p + 1;
#endif /* COMPILE_PCRE8 */
#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
if (c > 0xff)
{
c = 0xff;
caseless = FALSE;
}
SET_BIT(c);
#ifdef SUPPORT_UTF
if (utf && c > 127)
{
GETCHARINC(c, p);
#ifdef SUPPORT_UCP
if (caseless)
{
c = UCD_OTHERCASE(c);
if (c > 0xff)
c = 0xff;
SET_BIT(c);
}
#endif /* SUPPORT_UCP */
return p;
}
#else /* Not SUPPORT_UTF */
(void)(utf); /* Stops warning for unused parameter */
#endif /* SUPPORT_UTF */
if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]);
return p + 1;
#endif
}
BOOL
PRIV(xclass)(pcre_uint32 c, const pcre_uchar *data, BOOL utf)
{
pcre_uchar t;
BOOL negated = (*data & XCL_NOT) != 0;
(void)utf;
#ifdef COMPILE_PCRE8
/* In 8 bit mode, this must always be TRUE. Help the compiler to know that. */
utf = TRUE;
#endif
/* Character values < 256 are matched against a bitmap, if one is present. If
not, we still carry on, because there may be ranges that start below 256 in the
additional data. */
if (c < 256)
{
if ((*data & XCL_MAP) != 0 &&
(((pcre_uint8 *)(data + 1))[c/8] & (1 << (c&7))) != 0)
return !negated; /* char found */
}
/* First skip the bit map if present. Then match against the list of Unicode
properties or large chars or ranges that end with a large char. We won't ever
encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */
if ((*data++ & XCL_MAP) != 0) data += 32 / sizeof(pcre_uchar);
while ((t = *data++) != XCL_END)
{
pcre_uint32 x, y;
if (t == XCL_SINGLE)
{
#ifdef SUPPORT_UTF
if (utf)
{
GETCHARINC(x, data); /* macro generates multiple statements */
}
else
#endif
x = *data++;
if (c == x) return !negated;
}
else if (t == XCL_RANGE)
{
#ifdef SUPPORT_UTF
if (utf)
{
GETCHARINC(x, data); /* macro generates multiple statements */
GETCHARINC(y, data); /* macro generates multiple statements */
}
else
#endif
{
x = *data++;
y = *data++;
}
if (c >= x && c <= y) return !negated;
}
#ifdef SUPPORT_UCP
else /* XCL_PROP & XCL_NOTPROP */
{
const ucd_record *prop = GET_UCD(c);
switch(*data)
{
case PT_ANY:
if (t == XCL_PROP) return !negated;
break;
case PT_LAMP:
if ((prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt) == (t == XCL_PROP)) return !negated;
break;
case PT_GC:
if ((data[1] == PRIV(ucp_gentype)[prop->chartype]) == (t == XCL_PROP))
return !negated;
break;
case PT_PC:
if ((data[1] == prop->chartype) == (t == XCL_PROP)) return !negated;
break;
case PT_SC:
if ((data[1] == prop->script) == (t == XCL_PROP)) return !negated;
break;
case PT_ALNUM:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (t == XCL_PROP))
return !negated;
break;
case PT_SPACE: /* Perl space */
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
== (t == XCL_PROP))
return !negated;
break;
case PT_PXSPACE: /* POSIX space */
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
c == CHAR_FF || c == CHAR_CR) == (t == XCL_PROP))
return !negated;
break;
case PT_WORD:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE)
== (t == XCL_PROP))
return !negated;
break;
case PT_UCNC:
if (c < 0xa0)
{
if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT) == (t == XCL_PROP))
return !negated;
}
else
{
if ((c < 0xd800 || c > 0xdfff) == (t == XCL_PROP))
return !negated;
}
break;
/* This should never occur, but compilers may mutter if there is no
default. */
default:
return FALSE;
}
data += 2;
}
#endif /* SUPPORT_UCP */
}
return negated; /* char did not match */
}
BOOL
_pcre_xclass(int c, const uschar *data)
{
int t;
BOOL negated = (*data & XCL_NOT) != 0;
/* Character values < 256 are matched against a bitmap, if one is present. If
not, we still carry on, because there may be ranges that start below 256 in the
additional data. */
if (c < 256)
{
if ((*data & XCL_MAP) != 0 && (data[1 + c/8] & (1 << (c&7))) != 0)
return !negated; /* char found */
}
/* First skip the bit map if present. Then match against the list of Unicode
properties or large chars or ranges that end with a large char. We won't ever
encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */
if ((*data++ & XCL_MAP) != 0) data += 32;
while ((t = *data++) != XCL_END)
{
int x, y;
if (t == XCL_SINGLE)
{
GETCHARINC(x, data);
if (c == x) return !negated;
}
else if (t == XCL_RANGE)
{
GETCHARINC(x, data);
GETCHARINC(y, data);
if (c >= x && c <= y) return !negated;
}
#ifdef SUPPORT_UCP
else /* XCL_PROP & XCL_NOTPROP */
{
int chartype, script;
int category = _pcre_ucp_findprop(c, &chartype, &script);
switch(*data)
{
case PT_ANY:
if (t == XCL_PROP) return !negated;
break;
case PT_LAMP:
if ((chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt) ==
(t == XCL_PROP)) return !negated;
break;
case PT_GC:
if ((data[1] == category) == (t == XCL_PROP)) return !negated;
break;
case PT_PC:
if ((data[1] == chartype) == (t == XCL_PROP)) return !negated;
break;
case PT_SC:
if ((data[1] == script) == (t == XCL_PROP)) return !negated;
break;
/* This should never occur, but compilers may mutter if there is no
default. */
default:
return FALSE;
}
data += 2;
}
#endif /* SUPPORT_UCP */
}
return negated; /* char did not match */
}
BOOL
PRIV(xclass)(pcre_uint32 c, const pcre_uchar *data, BOOL utf)
{
pcre_uchar t;
BOOL negated = (*data & XCL_NOT) != 0;
(void)utf;
#ifdef COMPILE_PCRE8
/* In 8 bit mode, this must always be TRUE. Help the compiler to know that. */
utf = TRUE;
#endif
/* Character values < 256 are matched against a bitmap, if one is present. If
not, we still carry on, because there may be ranges that start below 256 in the
additional data. */
if (c < 256)
{
if ((*data & XCL_HASPROP) == 0)
{
if ((*data & XCL_MAP) == 0) return negated;
return (((pcre_uint8 *)(data + 1))[c/8] & (1 << (c&7))) != 0;
}
if ((*data & XCL_MAP) != 0 &&
(((pcre_uint8 *)(data + 1))[c/8] & (1 << (c&7))) != 0)
return !negated; /* char found */
}
/* First skip the bit map if present. Then match against the list of Unicode
properties or large chars or ranges that end with a large char. We won't ever
encounter XCL_PROP or XCL_NOTPROP when UCP support is not compiled. */
if ((*data++ & XCL_MAP) != 0) data += 32 / sizeof(pcre_uchar);
while ((t = *data++) != XCL_END)
{
pcre_uint32 x, y;
if (t == XCL_SINGLE)
{
#ifdef SUPPORT_UTF
if (utf)
{
GETCHARINC(x, data); /* macro generates multiple statements */
}
else
#endif
x = *data++;
if (c == x) return !negated;
}
else if (t == XCL_RANGE)
{
#ifdef SUPPORT_UTF
if (utf)
{
GETCHARINC(x, data); /* macro generates multiple statements */
GETCHARINC(y, data); /* macro generates multiple statements */
}
else
#endif
{
x = *data++;
y = *data++;
}
if (c >= x && c <= y) return !negated;
}
#ifdef SUPPORT_UCP
else /* XCL_PROP & XCL_NOTPROP */
{
const ucd_record *prop = GET_UCD(c);
BOOL isprop = t == XCL_PROP;
switch(*data)
{
case PT_ANY:
if (isprop) return !negated;
break;
case PT_LAMP:
if ((prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt) == isprop) return !negated;
break;
case PT_GC:
if ((data[1] == PRIV(ucp_gentype)[prop->chartype]) == isprop)
return !negated;
break;
case PT_PC:
if ((data[1] == prop->chartype) == isprop) return !negated;
break;
case PT_SC:
if ((data[1] == prop->script) == isprop) return !negated;
break;
case PT_ALNUM:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N) == isprop)
return !negated;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
if (isprop) return !negated;
break;
default:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == isprop)
return !negated;
break;
}
break;
case PT_WORD:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE)
== isprop)
return !negated;
break;
case PT_UCNC:
if (c < 0xa0)
{
if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT) == isprop)
return !negated;
}
else
{
if ((c < 0xd800 || c > 0xdfff) == isprop)
return !negated;
}
break;
/* The following three properties can occur only in an XCLASS, as there
is no \p or \P coding for them. */
/* Graphic character. Implement this as not Z (space or separator) and
not C (other), except for Cf (format) with a few exceptions. This seems
to be what Perl does. The exceptional characters are:
U+061C Arabic Letter Mark
U+180E Mongolian Vowel Separator
U+2066 - U+2069 Various "isolate"s
*/
case PT_PXGRAPH:
if ((PRIV(ucp_gentype)[prop->chartype] != ucp_Z &&
(PRIV(ucp_gentype)[prop->chartype] != ucp_C ||
(prop->chartype == ucp_Cf &&
c != 0x061c && c != 0x180e && (c < 0x2066 || c > 0x2069))
)) == isprop)
return !negated;
break;
/* Printable character: same as graphic, with the addition of Zs, i.e.
not Zl and not Zp, and U+180E. */
case PT_PXPRINT:
if ((prop->chartype != ucp_Zl &&
prop->chartype != ucp_Zp &&
(PRIV(ucp_gentype)[prop->chartype] != ucp_C ||
(prop->chartype == ucp_Cf &&
c != 0x061c && (c < 0x2066 || c > 0x2069))
)) == isprop)
return !negated;
break;
/* Punctuation: all Unicode punctuation, plus ASCII characters that
Unicode treats as symbols rather than punctuation, for Perl
compatibility (these are $+<=>^`|~). */
case PT_PXPUNCT:
if ((PRIV(ucp_gentype)[prop->chartype] == ucp_P ||
(c < 256 && PRIV(ucp_gentype)[prop->chartype] == ucp_S)) == isprop)
return !negated;
break;
/* This should never occur, but compilers may mutter if there is no
default. */
default:
return FALSE;
}
data += 2;
}
#endif /* SUPPORT_UCP */
}
return negated; /* char did not match */
}
