X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=utf8.c;h=fda99209335c094720d4450a63c8c9eb77f30759;hb=07766739ad671051f274806a83c05be36e7ca89a;hp=7302bb77e2256f72ca6d2d5b71c0bd06c009ce88;hpb=c7f1f0165ac822994a67426c2d8003c3974e49ef;p=p5sagit%2Fp5-mst-13.2.git diff --git a/utf8.c b/utf8.c index 7302bb7..fda9920 100644 --- a/utf8.c +++ b/utf8.c @@ -27,7 +27,7 @@ /* Unicode support */ /* -=for apidoc A|U8*|uvuni_to_utf8|U8 *d|UV uv +=for apidoc A|U8 *|uvuni_to_utf8|U8 *d|UV uv Adds the UTF8 representation of the Unicode codepoint C to the end of the string C; C should be have at least C free @@ -46,10 +46,22 @@ is the recommended Unicode-aware way of saying U8 * Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv) { - if (uv < 0x80) { - *d++ = uv; + if (UNI_IS_INVARIANT(uv)) { + *d++ = UTF_TO_NATIVE(uv); return d; } +#if defined(EBCDIC) || 1 /* always for testing */ + else { + STRLEN len = UNISKIP(uv); + U8 *p = d+len-1; + while (p > d) { + *p-- = UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK); + uv >>= UTF_ACCUMULATION_SHIFT; + } + *p = UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len)); + return d+len; + } +#else /* Non loop style */ if (uv < 0x800) { *d++ = (( uv >> 6) | 0xc0); *d++ = (( uv & 0x3f) | 0x80); @@ -116,43 +128,21 @@ Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv) return d; } #endif +#endif /* Loop style */ } -/* -=for apidoc A|U8*|uvchr_to_utf8|U8 *d|UV uv - -Adds the UTF8 representation of the Native codepoint C to the end -of the string C; C should be have at least C free -bytes available. The return value is the pointer to the byte after the -end of the new character. In other words, - - d = uvchr_to_utf8(d, uv); - -is the recommended wide native character-aware way of saying - - *(d++) = uv; - -=cut -*/ - -U8 * -Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv) -{ - if (uv < 0x100) - uv = NATIVE_TO_ASCII(uv); - return Perl_uvuni_to_utf8(aTHX_ d, uv); -} /* =for apidoc A|STRLEN|is_utf8_char|U8 *s Tests if some arbitrary number of bytes begins in a valid UTF-8 -character. Note that an ASCII character is a valid UTF-8 character. +character. Note that an INVARIANT (i.e. ASCII) character is a valid UTF-8 character. The actual number of bytes in the UTF-8 character will be returned if it is valid, otherwise 0. -=cut */ +=cut +*/ STRLEN Perl_is_utf8_char(pTHX_ U8 *s) { @@ -160,7 +150,7 @@ Perl_is_utf8_char(pTHX_ U8 *s) STRLEN slen, len; UV uv, ouv; - if (UTF8_IS_ASCII(u)) + if (UTF8_IS_INVARIANT(u)) return 1; if (!UTF8_IS_START(u)) @@ -173,13 +163,14 @@ Perl_is_utf8_char(pTHX_ U8 *s) slen = len - 1; s++; - uv = u; + u &= UTF_START_MASK(len); + uv = u; ouv = uv; while (slen--) { if (!UTF8_IS_CONTINUATION(*s)) return 0; uv = UTF8_ACCUMULATE(uv, *s); - if (uv < ouv) + if (uv < ouv) return 0; ouv = uv; s++; @@ -246,10 +237,11 @@ the strict UTF-8 encoding (see F). Most code should use utf8_to_uvchr() rather than call this directly. -=cut */ +=cut +*/ UV -Perl_utf8n_to_uvuni(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) +Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags) { UV uv = *s, ouv; STRLEN len = 1; @@ -276,10 +268,10 @@ Perl_utf8n_to_uvuni(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) goto malformed; } - if (UTF8_IS_ASCII(uv)) { + if (UTF8_IS_INVARIANT(uv)) { if (retlen) *retlen = 1; - return (UV) (*s); + return (UV) (NATIVE_TO_UTF(*s)); } if (UTF8_IS_CONTINUATION(uv) && @@ -294,20 +286,29 @@ Perl_utf8n_to_uvuni(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) goto malformed; } +#ifdef EBCDIC + uv = NATIVE_TO_UTF(uv); +#else if ((uv == 0xfe || uv == 0xff) && !(flags & UTF8_ALLOW_FE_FF)) { warning = UTF8_WARN_FE_FF; goto malformed; } - +#endif + if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } +#ifdef EBCDIC + else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } + else { len = 7; uv &= 0x01; } +#else else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } else if (!(uv & 0x01)) { len = 7; uv = 0; } - else { len = 13; uv = 0; } /* whoa! */ - + else { len = 13; uv = 0; } /* whoa! */ +#endif + if (retlen) *retlen = len; @@ -440,28 +441,7 @@ malformed: } /* -=for apidoc A|U8* s|utf8n_to_uvchr|STRLEN curlen, STRLEN *retlen, U32 flags - -Returns the native character value of the first character in the string C -which is assumed to be in UTF8 encoding; C will be set to the -length, in bytes, of that character. - -Allows length and flags to be passed to low level routine. - -=cut -*/ - -UV -Perl_utf8n_to_uvchr(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) -{ - UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags); - if (uv < 0x100) - return (UV) ASCII_TO_NATIVE(uv); - return uv; -} - -/* -=for apidoc A|U8* s|utf8_to_uvchr|STRLEN *retlen +=for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen Returns the native character value of the first character in the string C which is assumed to be in UTF8 encoding; C will be set to the @@ -474,13 +454,13 @@ returned and retlen is set, if possible, to -1. */ UV -Perl_utf8_to_uvchr(pTHX_ U8* s, STRLEN* retlen) +Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen) { return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0); } /* -=for apidoc A|U8* s|utf8_to_uvuni|STRLEN *retlen +=for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen Returns the Unicode code point of the first character in the string C which is assumed to be in UTF8 encoding; C will be set to the @@ -496,14 +476,14 @@ returned and retlen is set, if possible, to -1. */ UV -Perl_utf8_to_uvuni(pTHX_ U8* s, STRLEN* retlen) +Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen) { /* Call the low level routine asking for checks */ return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0); } /* -=for apidoc A|STRLEN|utf8_length|U8* s|U8 *e +=for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e Return the length of the UTF-8 char encoded string C in characters. Stops at C (inclusive). If C s> or if the scan would end @@ -513,7 +493,7 @@ up past C, croaks. */ STRLEN -Perl_utf8_length(pTHX_ U8* s, U8* e) +Perl_utf8_length(pTHX_ U8 *s, U8 *e) { STRLEN len = 0; @@ -527,7 +507,7 @@ Perl_utf8_length(pTHX_ U8* s, U8* e) U8 t = UTF8SKIP(s); if (e - s < t) - Perl_croak(aTHX_ "panic: utf8_length: unaligned end"); + Perl_croak(aTHX_ "panic: utf8_length: s=%p (%02X) e=%p l=%d - unaligned end",s,*s,e,t); s += t; len++; } @@ -544,7 +524,8 @@ and C. WARNING: use only if you *know* that the pointers point inside the same UTF-8 buffer. -=cut */ +=cut +*/ IV Perl_utf8_distance(pTHX_ U8 *a, U8 *b) @@ -580,7 +561,7 @@ Perl_utf8_distance(pTHX_ U8 *a, U8 *b) } /* -=for apidoc A|U8*|utf8_hop|U8 *s|I32 off +=for apidoc A|U8 *|utf8_hop|U8 *s|I32 off Return the UTF-8 pointer C displaced by C characters, either forward or backward. @@ -589,7 +570,8 @@ WARNING: do not use the following unless you *know* C is within the UTF-8 data pointed to by C *and* that on entry C is aligned on the first byte of character or just after the last byte of a character. -=cut */ +=cut +*/ U8 * Perl_utf8_hop(pTHX_ U8 *s, I32 off) @@ -624,7 +606,7 @@ Returns zero on failure, setting C to -1. */ U8 * -Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len) +Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len) { U8 *send; U8 *d; @@ -634,9 +616,9 @@ Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len) for (send = s + *len; s < send; ) { U8 c = *s++; - if (c >= 0x80 && - ((s >= send) || - ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) { + if (!UTF8_IS_INVARIANT(c) && + (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send) + || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) { *len = -1; return 0; } @@ -663,14 +645,15 @@ length. Returns the original string if no conversion occurs, C is unchanged. Do nothing if C points to 0. Sets C to 0 if C is converted or contains all 7bit characters. -=cut */ +=cut +*/ U8 * -Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8) +Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8) { - U8 *send; U8 *d; U8 *start = s; + U8 *send; I32 count = 0; if (!*is_utf8) @@ -679,28 +662,27 @@ Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8) /* ensure valid UTF8 and chars < 256 before converting string */ for (send = s + *len; s < send;) { U8 c = *s++; - if (!UTF8_IS_ASCII(c)) { - if (UTF8_IS_CONTINUATION(c) || s >= send || - !UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c)) + if (!UTF8_IS_INVARIANT(c)) { + if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send && + (c = *s++) && UTF8_IS_CONTINUATION(c)) + count++; + else return start; - s++, count++; - } + } } *is_utf8 = 0; - if (!count) - return start; - Newz(801, d, (*len) - count + 1, U8); s = start; start = d; while (s < send) { U8 c = *s++; - - if (UTF8_IS_ASCII(c)) - *d++ = c; - else - *d++ = UTF8_ACCUMULATE(c, *s++); + if (!UTF8_IS_INVARIANT(c)) { + /* Then it is two-byte encoded */ + c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++); + c = ASCII_TO_NATIVE(c); + } + *d++ = c; } *d = '\0'; *len = d - start; @@ -718,7 +700,7 @@ reflect the new length. */ U8* -Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len) +Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len) { U8 *send; U8 *d; @@ -729,11 +711,10 @@ Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len) dst = d; while (s < send) { - if (UTF8_IS_ASCII(*s)) - *d++ = *s++; + UV uv = NATIVE_TO_ASCII(*s++); + if (UNI_IS_INVARIANT(uv)) + *d++ = UTF_TO_NATIVE(uv); else { - UV uv = *s++; - *d++ = UTF8_EIGHT_BIT_HI(uv); *d++ = UTF8_EIGHT_BIT_LO(uv); } @@ -818,7 +799,7 @@ bool Perl_is_uni_alnum(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_alnum(tmpbuf); } @@ -826,7 +807,7 @@ bool Perl_is_uni_alnumc(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_alnumc(tmpbuf); } @@ -834,7 +815,7 @@ bool Perl_is_uni_idfirst(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_idfirst(tmpbuf); } @@ -842,7 +823,7 @@ bool Perl_is_uni_alpha(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_alpha(tmpbuf); } @@ -850,7 +831,7 @@ bool Perl_is_uni_ascii(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_ascii(tmpbuf); } @@ -858,7 +839,7 @@ bool Perl_is_uni_space(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_space(tmpbuf); } @@ -866,7 +847,7 @@ bool Perl_is_uni_digit(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_digit(tmpbuf); } @@ -874,7 +855,7 @@ bool Perl_is_uni_upper(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_upper(tmpbuf); } @@ -882,7 +863,7 @@ bool Perl_is_uni_lower(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_lower(tmpbuf); } @@ -890,7 +871,7 @@ bool Perl_is_uni_cntrl(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_cntrl(tmpbuf); } @@ -898,7 +879,7 @@ bool Perl_is_uni_graph(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_graph(tmpbuf); } @@ -906,7 +887,7 @@ bool Perl_is_uni_print(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_print(tmpbuf); } @@ -914,7 +895,7 @@ bool Perl_is_uni_punct(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_punct(tmpbuf); } @@ -922,7 +903,7 @@ bool Perl_is_uni_xdigit(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return is_utf8_xdigit(tmpbuf); } @@ -930,7 +911,7 @@ U32 Perl_to_uni_upper(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return to_utf8_upper(tmpbuf); } @@ -938,7 +919,7 @@ U32 Perl_to_uni_title(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return to_utf8_title(tmpbuf); } @@ -946,7 +927,7 @@ U32 Perl_to_uni_lower(pTHX_ U32 c) { U8 tmpbuf[UTF8_MAXLEN+1]; - uvuni_to_utf8(tmpbuf, (UV)c); + uvchr_to_utf8(tmpbuf, (UV)c); return to_utf8_lower(tmpbuf); } @@ -1258,8 +1239,9 @@ Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) SV* retval; SV* tokenbufsv = sv_2mortal(NEWSV(0,0)); dSP; + HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE); - if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */ + if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */ ENTER; Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv); LEAVE; @@ -1303,14 +1285,34 @@ UV Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) { HV* hv = (HV*)SvRV(sv); + /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ + then the "swatch" is a vec() for al the chars which start + with 0xAA..0xYY + So the key in the hash is length of encoded char -1 + */ U32 klen = UTF8SKIP(ptr) - 1; - U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */ + U32 off = ptr[klen]; STRLEN slen; - STRLEN needents = (klen ? 64 : 128); + STRLEN needents; U8 *tmps; U32 bit; SV *retval; + if (klen == 0) + { + /* If char in invariant then swatch is for all the invariant chars + * In both UTF-8 and UTF8-MOD that happens to be UTF_CONTINUATION_MARK + */ + needents = UTF_CONTINUATION_MARK; + off = NATIVE_TO_UTF(ptr[klen]); + } + else + { + /* If char is encoded then swatch is for the prefix */ + needents = (1 << UTF_ACCUMULATION_SHIFT); + off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK; + } + /* * This single-entry cache saves about 1/3 of the utf8 overhead in test * suite. (That is, only 7-8% overall over just a hash cache. Still, @@ -1334,6 +1336,10 @@ Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) /* If not cached, generate it via utf8::SWASHGET */ if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { dSP; + /* We use utf8n_to_uvuni() as we want an index into + Unicode tables, not a native character number. + */ + UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, NULL, 0); ENTER; SAVETMPS; save_re_context(); @@ -1341,10 +1347,8 @@ Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) PUSHMARK(SP); EXTEND(SP,3); PUSHs((SV*)sv); - /* We call utf8_to_uni as we want and index into Unicode tables, - not a native character number. - */ - PUSHs(sv_2mortal(newSViv(utf8_to_uvuni(ptr, 0) & ~(needents - 1)))); + /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */ + PUSHs(sv_2mortal(newSViv((klen) ? (code_point & ~(needents - 1)) : 0))); PUSHs(sv_2mortal(newSViv(needents))); PUTBACK; if (call_method("SWASHGET", G_SCALAR)) @@ -1359,7 +1363,7 @@ Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) svp = hv_store(hv, (char*)ptr, klen, retval, 0); - if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8) + if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents) Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); } @@ -1388,3 +1392,56 @@ Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) Perl_croak(aTHX_ "panic: swash_fetch"); return 0; } + + +/* +=for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv + +Adds the UTF8 representation of the Native codepoint C to the end +of the string C; C should be have at least C free +bytes available. The return value is the pointer to the byte after the +end of the new character. In other words, + + d = uvchr_to_utf8(d, uv); + +is the recommended wide native character-aware way of saying + + *(d++) = uv; + +=cut +*/ + +/* On ASCII machines this is normally a macro but we want a + real function in case XS code wants it +*/ +#undef Perl_uvchr_to_utf8 +U8 * +Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv) +{ + return Perl_uvuni_to_utf8(aTHX_ d, NATIVE_TO_UNI(uv)); +} + + +/* +=for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags + +Returns the native character value of the first character in the string C +which is assumed to be in UTF8 encoding; C will be set to the +length, in bytes, of that character. + +Allows length and flags to be passed to low level routine. + +=cut +*/ +/* On ASCII machines this is normally a macro but we want a + real function in case XS code wants it +*/ +#undef Perl_utf8n_to_uvchr +UV +Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags) +{ + UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags); + return UNI_TO_NATIVE(uv); +} + +