summaryrefslogtreecommitdiff
path: root/Master/tlpkg/tlperl/lib/CORE/utf8.h
diff options
context:
space:
mode:
Diffstat (limited to 'Master/tlpkg/tlperl/lib/CORE/utf8.h')
-rw-r--r--Master/tlpkg/tlperl/lib/CORE/utf8.h468
1 files changed, 294 insertions, 174 deletions
diff --git a/Master/tlpkg/tlperl/lib/CORE/utf8.h b/Master/tlpkg/tlperl/lib/CORE/utf8.h
index 3e15707ff8a..c954b42ff5d 100644
--- a/Master/tlpkg/tlperl/lib/CORE/utf8.h
+++ b/Master/tlpkg/tlperl/lib/CORE/utf8.h
@@ -30,9 +30,9 @@
#include "unicode_constants.h"
/* For to_utf8_fold_flags, q.v. */
-#define FOLD_FLAGS_LOCALE 0x1
-#define FOLD_FLAGS_FULL 0x2
-#define FOLD_FLAGS_NOMIX_ASCII 0x4
+#define FOLD_FLAGS_LOCALE 0x1
+#define FOLD_FLAGS_FULL 0x2
+#define FOLD_FLAGS_NOMIX_ASCII 0x4
/* For _core_swash_init(), internal core use only */
#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
@@ -97,6 +97,13 @@ than just the ASCII characters, so C<is_invariant_string> is preferred.
#else /* ! EBCDIC */
START_EXTERN_C
+/* How wide can a single UTF-8 encoded character become in bytes. */
+/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
+ * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
+ * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
+ * non-negative integers in a binary format, even those above Unicode */
+#define UTF8_MAXBYTES 13
+
#ifdef DOINIT
EXTCONST unsigned char PL_utf8skip[] = {
/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
@@ -112,12 +119,14 @@ EXTCONST unsigned char PL_utf8skip[] = {
/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
/* 0xC0 */ 2,2, /* overlong */
-/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
+/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
-/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
- allowed (64-bit + reserved). */
+ /* Perl extended (never was official UTF-8). Up to 36 bit */
+/* 0xFE */ 7,
+ /* More extended, Up to 72 bits (64-bit + reserved) */
+/* 0xFF */ UTF8_MAXBYTES
};
#else
EXTCONST unsigned char PL_utf8skip[];
@@ -125,10 +134,20 @@ EXTCONST unsigned char PL_utf8skip[];
END_EXTERN_C
+#if defined(_MSC_VER) && _MSC_VER < 1400
+/* older MSVC versions have a smallish macro buffer */
+#define PERL_SMALL_MACRO_BUFFER
+#endif
+
/* Native character to/from iso-8859-1. Are the identity functions on ASCII
* platforms */
-#define NATIVE_TO_LATIN1(ch) (ch)
-#define LATIN1_TO_NATIVE(ch) (ch)
+#ifdef PERL_SMALL_MACRO_BUFFER
+#define NATIVE_TO_LATIN1(ch) ((U8)(ch))
+#define LATIN1_TO_NATIVE(ch) ((U8)(ch))
+#else
+#define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
+#define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
+#endif
/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
* consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
@@ -136,12 +155,17 @@ END_EXTERN_C
* because they are 8-bit encodings that serve the same purpose in Perl, and
* rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
* whichever one is applicable on the current platform */
+#ifdef PERL_SMALL_MACRO_BUFFER
#define NATIVE_UTF8_TO_I8(ch) (ch)
#define I8_TO_NATIVE_UTF8(ch) (ch)
+#else
+#define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
+#define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
+#endif
/* Transforms in wide UV chars */
-#define UNI_TO_NATIVE(ch) (ch)
-#define NATIVE_TO_UNI(ch) (ch)
+#define UNI_TO_NATIVE(ch) ((UV) (ch))
+#define NATIVE_TO_UNI(ch) ((UV) (ch))
/*
@@ -190,46 +214,50 @@ Perl's extended UTF-8 means we can have start bytes up to FF.
*/
-/* Is the representation of the Unicode code point 'c' the same regardless of
+/* Is the representation of the Unicode code point 'cp' the same regardless of
* being encoded in UTF-8 or not? */
-#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
+#define OFFUNI_IS_INVARIANT(cp) isASCII(cp)
-/* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is
- * the inverse of UTF8_IS_INVARIANT */
-#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
+/* Is the representation of the code point 'cp' the same regardless of
+ * being encoded in UTF-8 or not? 'cp' is native if < 256; Unicode otherwise
+ * */
+#define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp)
+
+/* This defines the bits that are to be in the continuation bytes of a multi-byte
+ * UTF-8 encoded character that mark it is a continuation byte. */
+#define UTF_CONTINUATION_MARK 0x80
+
+/* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence
+ * in UTF-8? This is the inverse of UTF8_IS_INVARIANT. The |0 makes sure this
+ * isn't mistakenly called with a ptr argument */
+#define UTF8_IS_CONTINUED(c) (((U8)((c) | 0)) & UTF_CONTINUATION_MARK)
/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
* This doesn't catch invariants (they are single-byte). It also excludes the
- * illegal overlong sequences that begin with C0 and C1. */
-#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
+ * illegal overlong sequences that begin with C0 and C1. The |0 makes sure
+ * this isn't mistakenly called with a ptr argument */
+#define UTF8_IS_START(c) (((U8)((c) | 0)) >= 0xc2)
+
+/* For use in UTF8_IS_CONTINUATION() below */
+#define UTF_IS_CONTINUATION_MASK 0xC0
/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
- * first byte thereof? */
-#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
+ * first byte thereof? The |0 makes sure this isn't mistakenly called with a
+ * ptr argument */
+#define UTF8_IS_CONTINUATION(c) \
+ ((((U8)((c) | 0)) & UTF_IS_CONTINUATION_MASK) == UTF_CONTINUATION_MARK)
/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
* UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
* be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
- * this matches 0xc[23]. */
-#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
+ * this matches 0xc[23]. The |0 makes sure this isn't mistakenly called with a
+ * ptr argument */
+#define UTF8_IS_DOWNGRADEABLE_START(c) ((((U8)((c) | 0)) & 0xfe) == 0xc2)
/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
- * represent a code point > 255? */
-#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
-
-/* This defines the 1-bits that are to be in the first byte of a multi-byte
- * UTF-8 encoded character that give the number of bytes that comprise the
- * character. 'len' is the number of bytes in the multi-byte sequence. */
-#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
-
-/* Masks out the initial one bits in a start byte, leaving the real data ones.
- * Doesn't work on an invariant byte. 'len' is the number of bytes in the
- * multi-byte sequence that comprises the character. */
-#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
-
-/* This defines the bits that are to be in the continuation bytes of a multi-byte
- * UTF-8 encoded character that indicate it is a continuation byte. */
-#define UTF_CONTINUATION_MARK 0x80
+ * represent a code point > 255? The |0 makes sure this isn't mistakenly
+ * called with a ptr argument */
+#define UTF8_IS_ABOVE_LATIN1(c) (((U8)((c) | 0)) >= 0xc4)
/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
* sequence contributes to the specification of the code point. In the bit
@@ -237,63 +265,98 @@ Perl's extended UTF-8 means we can have start bytes up to FF.
* real information */
#define UTF_ACCUMULATION_SHIFT 6
-/* 2**UTF_ACCUMULATION_SHIFT - 1 */
-#define UTF_CONTINUATION_MASK ((U8)0x3f)
+/* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
+ * for more */
+#define QUESTION_MARK_CTRL DEL_NATIVE
-/* If a value is anded with this, and the result is non-zero, then using the
- * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
- * */
-#define UTF_ACCUMULATION_OVERFLOW_MASK \
- (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
- - UTF_ACCUMULATION_SHIFT))
+/* Surrogates, non-character code points and above-Unicode code points are
+ * problematic in some contexts. This allows code that needs to check for
+ * those to to quickly exclude the vast majority of code points it will
+ * encounter */
+#define isUTF8_POSSIBLY_PROBLEMATIC(c) ((U8) c >= 0xED)
-#if UVSIZE >= 8
-# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
+#endif /* EBCDIC vs ASCII */
-/* Input is a true Unicode (not-native) code point */
-#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
- (uv) < 0x800 ? 2 : \
- (uv) < 0x10000 ? 3 : \
- (uv) < 0x200000 ? 4 : \
- (uv) < 0x4000000 ? 5 : \
- (uv) < 0x80000000 ? 6 : \
- (uv) < UTF8_QUAD_MAX ? 7 : 13 )
+/* 2**UTF_ACCUMULATION_SHIFT - 1 */
+#define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1))
+
+/* Internal macro to be used only in this file to aid in constructing other
+ * publicly accessible macros.
+ * The number of bytes required to express this uv in UTF-8, for just those
+ * uv's requiring 2 through 6 bytes, as these are common to all platforms and
+ * word sizes. The number of bytes needed is given by the number of leading 1
+ * bits in the start byte. There are 32 start bytes that have 2 initial 1 bits
+ * (C0-DF); there are 16 that have 3 initial 1 bits (E0-EF); 8 that have 4
+ * initial 1 bits (F0-F8); 4 that have 5 initial 1 bits (F9-FB), and 2 that
+ * have 6 initial 1 bits (FC-FD). The largest number a string of n bytes can
+ * represent is (the number of possible start bytes for 'n')
+ * * (the number of possiblities for each start byte
+ * The latter in turn is
+ * 2 ** ( (how many continuation bytes there are)
+ * * (the number of bits of information each
+ * continuation byte holds))
+ *
+ * If we were on a platform where we could use a fast find first set bit
+ * instruction (or count leading zeros instruction) this could be replaced by
+ * using that to find the log2 of the uv, and divide that by the number of bits
+ * of information in each continuation byte, adjusting for large cases and how
+ * much information is in a start byte for that length */
+#define __COMMON_UNI_SKIP(uv) \
+ (UV) (uv) < (32 * (1U << ( UTF_ACCUMULATION_SHIFT))) ? 2 : \
+ (UV) (uv) < (16 * (1U << (2 * UTF_ACCUMULATION_SHIFT))) ? 3 : \
+ (UV) (uv) < ( 8 * (1U << (3 * UTF_ACCUMULATION_SHIFT))) ? 4 : \
+ (UV) (uv) < ( 4 * (1U << (4 * UTF_ACCUMULATION_SHIFT))) ? 5 : \
+ (UV) (uv) < ( 2 * (1U << (5 * UTF_ACCUMULATION_SHIFT))) ? 6 :
+
+/* Internal macro to be used only in this file.
+ * This adds to __COMMON_UNI_SKIP the details at this platform's upper range.
+ * For any-sized EBCDIC platforms, or 64-bit ASCII ones, we need one more test
+ * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit
+ * ASCII platforms, everything is representable by 7 bytes */
+#if defined(UV_IS_QUAD) || defined(EBCDIC)
+# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \
+ (UV) (uv) < ((UV) 1U << (6 * UTF_ACCUMULATION_SHIFT)) ? 7 : UTF8_MAXBYTES)
#else
-/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
-#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
- (uv) < 0x800 ? 2 : \
- (uv) < 0x10000 ? 3 : \
- (uv) < 0x200000 ? 4 : \
- (uv) < 0x4000000 ? 5 : \
- (uv) < 0x80000000 ? 6 : 7 )
+# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7)
#endif
-/* How wide can a single UTF-8 encoded character become in bytes. */
-/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
- * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
- * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
- * non-negative integers in a binary format, even those above Unicode */
-#define UTF8_MAXBYTES 13
+/* The next two macros use the base macro defined above, and add in the tests
+ * at the low-end of the range, for just 1 byte, yielding complete macros,
+ * publicly accessible. */
-/* The maximum number of UTF-8 bytes a single Unicode character can
- * uppercase/lowercase/fold into. Unicode guarantees that the maximum
- * expansion is 3 characters. On ASCIIish platforms, the highest Unicode
- * character occupies 4 bytes, therefore this number would be 12, but this is
- * smaller than the maximum width a single above-Unicode character can occupy,
- * so use that instead */
-#if UTF8_MAXBYTES < 12
-#error UTF8_MAXBYTES must be at least 12
-#endif
+/* Input is a true Unicode (not-native) code point */
+#define OFFUNISKIP(uv) (OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
-/* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
- * for more */
-#define QUESTION_MARK_CTRL DEL_NATIVE
+/*
-#define MAX_UTF8_TWO_BYTE 0x7FF
+=for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
+returns the number of bytes required to represent the code point C<cp> when
+encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
+255; a Unicode code point otherwise.
-#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
+=cut
+ */
+#define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
-#endif /* EBCDIC vs ASCII */
+/* As explained in the comments for __COMMON_UNI_SKIP, 32 start bytes with
+ * UTF_ACCUMULATION_SHIFT bits of information each */
+#define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1)
+
+/* constrained by EBCDIC which has 5 bits per continuation byte */
+#define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1)
+
+/* The maximum number of UTF-8 bytes a single Unicode character can
+ * uppercase/lowercase/fold into. Unicode guarantees that the maximum
+ * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode
+ * code point will fold to itself, so we only have to look at the expansion of
+ * the maximum Unicode code point. But this number may be less than the space
+ * occupied by a very large code point under Perl's extended UTF-8. We have to
+ * make it large enough to fit any single character. (It turns out that ASCII
+ * and EBCDIC differ in which is larger) */
+#define UTF8_MAXBYTES_CASE \
+ (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \
+ ? UTF8_MAXBYTES \
+ : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)))
/* Rest of these are attributes of Unicode and perl's internals rather than the
* encoding, or happen to be the same in both ASCII and EBCDIC (at least at
@@ -312,6 +375,17 @@ Perl's extended UTF-8 means we can have start bytes up to FF.
#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
+/* This defines the 1-bits that are to be in the first byte of a multi-byte
+ * UTF-8 encoded character that mark it as a start byte and give the number of
+ * bytes that comprise the character. 'len' is the number of bytes in the
+ * multi-byte sequence. */
+#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
+
+/* Masks out the initial one bits in a start byte, leaving the real data ones.
+ * Doesn't work on an invariant byte. 'len' is the number of bytes in the
+ * multi-byte sequence that comprises the character. */
+#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
+
/* Adds a UTF8 continuation byte 'new' of information to a running total code
* point 'old' of all the continuation bytes so far. This is designed to be
* used in a loop to convert from UTF-8 to the code point represented. Note
@@ -322,30 +396,44 @@ Perl's extended UTF-8 means we can have start bytes up to FF.
| ((NATIVE_UTF8_TO_I8((U8)new)) \
& UTF_CONTINUATION_MASK))
+/* If a value is anded with this, and the result is non-zero, then using the
+ * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
+ * */
+#define UTF_ACCUMULATION_OVERFLOW_MASK \
+ (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
+ - UTF_ACCUMULATION_SHIFT))
+
/* This works in the face of malformed UTF-8. */
#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
&& ( (e) - (s) > 1) \
&& UTF8_IS_CONTINUATION(*((s)+1)))
/* Number of bytes a code point occupies in UTF-8. */
-#define NATIVE_SKIP(uv) OFFUNISKIP(NATIVE_TO_UNI(uv))
-
-/*
-
-=for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
-returns the number of bytes required to represent the code point C<cp> when
-encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
-255; a Unicode code point otherwise.
-
-=cut
- */
+#define NATIVE_SKIP(uv) UVCHR_SKIP(uv)
/* Most code which says UNISKIP is really thinking in terms of native code
* points (0-255) plus all those beyond. This is an imprecise term, but having
* it means existing code continues to work. For precision, use UVCHR_SKIP,
- * NATIVE_SKIP, and OFFUNISKIP */
-#define UNISKIP(uv) NATIVE_SKIP(uv)
-#define UVCHR_SKIP(uv) NATIVE_SKIP(uv)
+ * NATIVE_SKIP, or OFFUNISKIP */
+#define UNISKIP(uv) UVCHR_SKIP(uv)
+
+/* Longer, but more accurate name */
+#define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c)
+
+/* Convert a UTF-8 variant Latin1 character to a native code point value.
+ * Needs just one iteration of accumulate. Should be used only if it is known
+ * that the code point is < 256, and is not UTF-8 invariant. Use the slower
+ * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point
+ * representable by two bytes (which turns out to be up through
+ * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are:
+ * HI: a downgradable start byte;
+ * LO: continuation.
+ * */
+#define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \
+ ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \
+ __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
+ LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \
+ NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO))))
/* Convert a two (not one) byte utf8 character to a native code point value.
* Needs just one iteration of accumulate. Should not be used unless it is
@@ -353,8 +441,10 @@ encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
* Note that the result can be larger than 255 if the input character is not
* downgradable */
#define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
+ ( __ASSERT_(PL_utf8skip[HI] == 2) \
+ __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
- (LO)))
+ (LO))))
/* Should never be used, and be deprecated */
#define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
@@ -367,22 +457,25 @@ only) byte is pointed to by C<s>.
=cut
*/
-#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
+#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
+#define UTF8_SKIP(s) UTF8SKIP(s)
+
+/* Most code that says 'UNI_' really means the native value for code points up
+ * through 255 */
+#define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp)
/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
* works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
- * each for the exact same set of bit patterns. (And it works on any byte in a
- * UTF-8 encoded string) */
-#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c))
-
-/* Like the above, but its name implies a non-UTF8 input, and is implemented
- * differently (for no particular reason) */
-#define NATIVE_BYTE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c))
-
-/* Like the above, but accepts any UV as input */
-#define UVCHR_IS_INVARIANT(uv) UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))
+ * each for the exact same set of bit patterns. It is valid on a subset of
+ * what UVCHR_IS_INVARIANT is valid on, so can just use that; and the compiler
+ * should optimize out anything extraneous given the implementation of the
+ * latter. The |0 makes sure this isn't mistakenly called with a ptr argument.
+ * */
+#define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT((c) | 0)
-#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
+/* Like the above, but its name implies a non-UTF8 input, which as the comments
+ * above show, doesn't matter as to its implementation */
+#define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
/* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
* bytes from an ordinal that is known to fit into exactly two (not one) bytes;
@@ -390,14 +483,16 @@ only) byte is pointed to by C<s>.
/* These two are helper macros for the other three sets, and should not be used
* directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
- * (which works for code points up to 0xFF) or NATIVE_TO_UNI which works for any
- * code point */
+ * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works
+ * for any code point */
#define __BASE_TWO_BYTE_HI(c, translate_function) \
+ (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
- | UTF_START_MARK(2))
+ | UTF_START_MARK(2)))
#define __BASE_TWO_BYTE_LO(c, translate_function) \
+ (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
- | UTF_CONTINUATION_MARK)
+ | UTF_CONTINUATION_MARK))
/* The next two macros should not be used. They were designed to be usable as
* the case label of a switch statement, but this doesn't work for EBCDIC. Use
@@ -408,9 +503,9 @@ only) byte is pointed to by C<s>.
/* The next two macros are used when the source should be a single byte
* character; checked for under DEBUGGING */
#define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
- ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
+ ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
#define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
- ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
+ (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
/* These final two macros in the series are used when the source can be any
* code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
@@ -421,11 +516,11 @@ only) byte is pointed to by C<s>.
#define UTF8_TWO_BYTE_HI(c) \
(__ASSERT_((sizeof(c) == 1) \
|| !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
- ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
+ (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
#define UTF8_TWO_BYTE_LO(c) \
(__ASSERT_((sizeof(c) == 1) \
|| !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
- ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
+ (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
/* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
* as it is only in overlongs. */
@@ -438,11 +533,11 @@ only) byte is pointed to by C<s>.
* beginning of a utf8 character. Now that foo_utf8() determines that itself,
* no need to do it again here
*/
-#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
- ? isIDFIRST(*(p)) \
+#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF) \
+ ? isIDFIRST(*(p)) \
: isIDFIRST_utf8((const U8*)p))
-#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
- ? isWORDCHAR(*(p)) \
+#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF)) \
+ ? isWORDCHAR(*(p)) \
: isWORDCHAR_utf8((const U8*)p))
#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
@@ -465,7 +560,12 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
=cut
*/
#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
-#define IN_UNI_8_BIT \
+
+/* Should all strings be treated as Unicode, and not just UTF-8 encoded ones?
+ * Is so within 'feature unicode_strings' or 'locale :not_characters', and not
+ * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps
+ * could be */
+#define IN_UNI_8_BIT \
(((CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT)) \
|| (CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
/* -1 below is for :not_characters */ \
@@ -496,12 +596,16 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
-/* Code points which never were part of the original UTF-8 standard, the first
- * byte of which is a FE or FF on ASCII platforms. If the first byte is FF, it
- * will overflow a 32-bit word. If the first byte is FE, it will overflow a
- * signed 32-bit word. */
-#define UTF8_DISALLOW_FE_FF 0x0800
-#define UTF8_WARN_FE_FF 0x1000
+/* Code points which never were part of the original UTF-8 standard, which only
+ * went up to 2 ** 31 - 1. Note that these all overflow a signed 32-bit word,
+ * The first byte of these code points is FE or FF on ASCII platforms. If the
+ * first byte is FF, it will overflow a 32-bit word. */
+#define UTF8_DISALLOW_ABOVE_31_BIT 0x0800
+#define UTF8_WARN_ABOVE_31_BIT 0x1000
+
+/* For back compat, these old names are misleading for UTF_EBCDIC */
+#define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_ABOVE_31_BIT
+#define UTF8_WARN_FE_FF UTF8_WARN_ABOVE_31_BIT
#define UTF8_CHECK_ONLY 0x2000
@@ -511,25 +615,25 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
#define UTF8_ALLOW_FFFF 0
#define UTF8_ALLOW_SURROGATE 0
-#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
+#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \
+ ( UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR \
+ |UTF8_DISALLOW_SURROGATE)
#define UTF8_WARN_ILLEGAL_INTERCHANGE \
- (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
-#define UTF8_ALLOW_ANY \
- (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
+ (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE)
+#define UTF8_ALLOW_ANY \
+ (~( UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_DISALLOW_ABOVE_31_BIT \
+ |UTF8_WARN_ILLEGAL_INTERCHANGE|UTF8_WARN_ABOVE_31_BIT))
#define UTF8_ALLOW_ANYUV \
(UTF8_ALLOW_EMPTY \
& ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
UTF8_ALLOW_ANYUV)
-/* Surrogates, non-character code points and above-Unicode code points are
- * problematic in some contexts. This allows code that needs to check for
- * those to to quickly exclude the vast majority of code points it will
- * encounter */
-#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
- FIRST_SURROGATE_UTF8_FIRST_BYTE
+/* Several of the macros below have a second parameter that is currently
+ * unused; but could be used in the future to make sure that the input is
+ * well-formed. */
-#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
+#define UTF8_IS_SURROGATE(s, e) cBOOL(is_SURROGATE_utf8(s))
#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
/* ASCII EBCDIC I8
@@ -540,20 +644,20 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
* BE AWARE that this test doesn't rule out malformed code points, in
* particular overlongs */
#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
-# define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
+# define UTF8_IS_SUPER(s, e) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
&& (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \
|| (NATIVE_UTF8_TO_I8(* ((U8*) (s) + 1)) >= 0xA2)))
#else
-# define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \
- && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
+# define UTF8_IS_SUPER(s, e) (*(U8*) (s) >= 0xF4 \
+ && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
#endif
/* These are now machine generated, and the 'given' clause is no longer
* applicable */
-#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
+#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \
cBOOL(is_NONCHAR_utf8(s))
-#define UTF8_IS_NONCHAR_(s) \
- UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
+#define UTF8_IS_NONCHAR(s, e) \
+ UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e)
#define UNICODE_SURROGATE_FIRST 0xD800
#define UNICODE_SURROGATE_LAST 0xDFFF
@@ -564,14 +668,14 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
* let's be conservative and do as Unicode says. */
#define PERL_UNICODE_MAX 0x10FFFF
-#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
-#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
-#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
-#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
-#define UNICODE_DISALLOW_SURROGATE 0x0010
-#define UNICODE_DISALLOW_NONCHAR 0x0020
-#define UNICODE_DISALLOW_SUPER 0x0040
-#define UNICODE_DISALLOW_FE_FF 0x0080
+#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
+#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
+#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
+#define UNICODE_WARN_ABOVE_31_BIT 0x0008 /* Above 0x7FFF_FFFF */
+#define UNICODE_DISALLOW_SURROGATE 0x0010
+#define UNICODE_DISALLOW_NONCHAR 0x0020
+#define UNICODE_DISALLOW_SUPER 0x0040
+#define UNICODE_DISALLOW_ABOVE_31_BIT 0x0080
#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
(UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
@@ -582,19 +686,31 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
#define UNICODE_ALLOW_SUPER 0
#define UNICODE_ALLOW_ANY 0
-#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
- (c) <= UNICODE_SURROGATE_LAST)
-#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
-#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
-#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
- /* The other noncharacters end in FFFE or FFFF, which \
- * the mask below catches both of, but beyond the last \
- * official unicode code point, they aren't \
- * noncharacters, since those aren't Unicode \
- * characters at all */ \
- || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
-#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
-#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
+/* This matches the 2048 code points between UNICODE_SURROGATE_FIRST (0xD800) and
+ * UNICODE_SURROGATE_LAST (0xDFFF) */
+#define UNICODE_IS_SURROGATE(uv) (((UV) (uv) & (~0xFFFF | 0xF800)) \
+ == 0xD800)
+
+#define UNICODE_IS_REPLACEMENT(uv) ((UV) (uv) == UNICODE_REPLACEMENT)
+#define UNICODE_IS_BYTE_ORDER_MARK(uv) ((UV) (uv) == UNICODE_BYTE_ORDER_MARK)
+
+/* Is 'uv' one of the 32 contiguous-range noncharacters? */
+#define UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) ((UV) (uv) >= 0xFDD0 \
+ && (UV) (uv) <= 0xFDEF)
+
+/* Is 'uv' one of the 34 plane-ending noncharacters 0xFFFE, 0xFFFF, 0x1FFFE,
+ * 0x1FFFF, ... 0x10FFFE, 0x10FFFF, given that we know that 'uv' is not above
+ * the Unicode legal max */
+#define UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv) \
+ (((UV) (uv) & 0xFFFE) == 0xFFFE)
+
+#define UNICODE_IS_NONCHAR(uv) \
+ ( UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) \
+ || ( LIKELY( ! UNICODE_IS_SUPER(uv)) \
+ && UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv)))
+
+#define UNICODE_IS_SUPER(uv) ((UV) (uv) > PERL_UNICODE_MAX)
+#define UNICODE_IS_ABOVE_31_BIT(uv) ((UV) (uv) > 0x7FFFFFFF)
#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
@@ -611,7 +727,11 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
of MICRON */
#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
-#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
+#ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8
+# define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
+#endif
+#define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130
+#define LATIN_SMALL_LETTER_DOTLESS_I 0x131
#define LATIN_SMALL_LETTER_LONG_S 0x017F
#define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
#define LATIN_SMALL_LIGATURE_ST 0xFB06
@@ -654,9 +774,9 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
* on the order of 10 minutes to generate, and is never going to change, unless
* the generated code is improved.
*
- * The EBCDIC versions have been cut to not cover all of legal Unicode, so
- * don't take too long to generate, and there is a separate one for each code
- * page, so they are in regcharclass.h instead of here */
+ * The EBCDIC versions have been cut to not cover all of legal Unicode,
+ * otherwise they take too long to generate; besides there is a separate one
+ * for each code page, so they are in regcharclass.h instead of here */
/*
UTF8_CHAR: Matches legal UTF-8 encoded characters from 2 through 4 bytes
@@ -680,9 +800,9 @@ case any call to string overloading updates the internal UTF-8 encoding flag.
=for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
Returns the number of bytes beginning at C<s> which form a legal UTF-8 (or
-UTF-EBCDIC) encoded character, looking no further than C<e - s> bytes into
-C<s>. Returns 0 if the sequence starting at C<s> through C<e - 1> is not
-well-formed UTF-8
+UTF-EBCDIC) encoded character, looking no further than S<C<e - s>> bytes into
+C<s>. Returns 0 if the sequence starting at C<s> through S<C<e - 1>> is not
+well-formed UTF-8.
Note that an INVARIANT character (i.e. ASCII on non-EBCDIC
machines) is a valid UTF-8 character.