2 $Id: Encode.xs,v 2.18 2009/11/26 09:23:59 dankogai Exp dankogai $
5 #define PERL_NO_GET_CONTEXT
12 # define PERLIO_MODNAME "PerlIO::encoding"
13 # define PERLIO_FILENAME "PerlIO/encoding.pm"
15 /* set 1 or more to profile. t/encoding.t dumps core because of
16 Perl_warner and PerlIO don't work well */
17 #define ENCODE_XS_PROFILE 0
19 /* set 0 to disable floating point to calculate buffer size for
20 encode_method(). 1 is recommended. 2 restores NI-S original */
21 #define ENCODE_XS_USEFP 1
23 #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
24 Perl_croak(aTHX_ "panic_unimplemented"); \
25 return (y)0; /* fool picky compilers */ \
29 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
30 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
32 #define UTF8_ALLOW_STRICT 0
33 #define UTF8_ALLOW_NONSTRICT (UTF8_ALLOW_ANY & \
34 ~(UTF8_ALLOW_CONTINUATION | \
35 UTF8_ALLOW_NON_CONTINUATION | \
39 Encode_XSEncoding(pTHX_ encode_t * enc)
42 HV *stash = gv_stashpv("Encode::XS", TRUE);
43 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
47 while (enc->name[i]) {
48 const char *name = enc->name[i++];
49 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
52 call_pv("Encode::define_encoding", G_DISCARD);
57 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
59 /* Exists for breakpointing */
63 #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
64 #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
67 do_fallback_cb(pTHX_ UV ch, SV *fallback_cb)
71 SV *retval = newSVpv("",0);
75 XPUSHs(sv_2mortal(newSVnv((UV)ch)));
77 argc = call_sv(fallback_cb, G_SCALAR);
80 croak("fallback sub must return scalar!");
82 sv_catsv(retval, POPs);
90 encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src,
91 int check, STRLEN * offset, SV * term, int * retcode,
95 U8 *s = (U8 *) SvPV(src, slen);
100 /* We allocate slen+1.
101 PerlIO dumps core if this value is smaller than this. */
102 SV *dst = sv_2mortal(newSV(slen+1));
103 U8 *d = (U8 *)SvPVX(dst);
104 STRLEN dlen = SvLEN(dst)-1;
107 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
111 if (slen > *offset){ /* safeguard against slen overflow */
125 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
128 SvCUR_set(dst, dlen+ddone);
131 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
132 code == ENCODE_FOUND_TERM) {
138 STRLEN more = 0; /* make sure you initialize! */
142 sleft = tlen - sdone;
143 #if ENCODE_XS_PROFILE >= 2
145 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
146 more, sdone, sleft, SvLEN(dst));
148 if (sdone != 0) { /* has src ever been processed ? */
149 #if ENCODE_XS_USEFP == 2
150 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
152 #elif ENCODE_XS_USEFP
153 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
155 /* safe until SvLEN(dst) == MAX_INT/16 */
156 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
159 more += UTF8_MAXLEN; /* insurance policy */
160 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
161 /* dst need to grow need MORE bytes! */
162 if (ddone >= SvLEN(dst)) {
163 Perl_croak(aTHX_ "Destination couldn't be grown.");
165 dlen = SvLEN(dst)-ddone-1;
173 if (dir == enc->f_utf8) {
176 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
177 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
178 /* if non-representable multibyte prefix at end of current buffer - break*/
179 if (clen > tlen - sdone) break;
180 if (check & ENCODE_DIE_ON_ERR) {
181 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
182 (UV)ch, enc->name[0]);
183 return &PL_sv_undef; /* never reaches but be safe */
185 if (check & ENCODE_WARN_ON_ERR){
186 Perl_warner(aTHX_ packWARN(WARN_UTF8),
187 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
189 if (check & ENCODE_RETURN_ON_ERR){
192 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
194 (fallback_cb != &PL_sv_undef)
195 ? do_fallback_cb(aTHX_ ch, fallback_cb)
196 : newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
197 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
198 "&#x%" UVxf ";", (UV)ch);
199 SvUTF8_off(subchar); /* make sure no decoded string gets in */
200 sdone += slen + clen;
201 ddone += dlen + SvCUR(subchar);
202 sv_catsv(dst, subchar);
203 SvREFCNT_dec(subchar);
206 sdone += slen + clen;
207 ddone += dlen + enc->replen;
208 sv_catpvn(dst, (char*)enc->rep, enc->replen);
213 if (check & ENCODE_DIE_ON_ERR){
214 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
215 enc->name[0], (UV)s[slen]);
216 return &PL_sv_undef; /* never reaches but be safe */
218 if (check & ENCODE_WARN_ON_ERR){
220 aTHX_ packWARN(WARN_UTF8),
222 enc->name[0], (UV)s[slen]);
224 if (check & ENCODE_RETURN_ON_ERR){
228 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
230 (fallback_cb != &PL_sv_undef)
231 ? do_fallback_cb(aTHX_ (UV)s[slen], fallback_cb)
232 : newSVpvf("\\x%02" UVXf, (UV)s[slen]);
234 ddone += dlen + SvCUR(subchar);
235 sv_catsv(dst, subchar);
236 SvREFCNT_dec(subchar);
239 ddone += dlen + strlen(FBCHAR_UTF8);
240 sv_catpv(dst, FBCHAR_UTF8);
243 /* settle variables when fallback */
244 d = (U8 *)SvEND(dst);
245 dlen = SvLEN(dst) - ddone - 1;
246 s = (U8*)SvPVX(src) + sdone;
251 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
252 code, (dir == enc->f_utf8) ? "to" : "from",
258 if (check && !(check & ENCODE_LEAVE_SRC)){
259 sdone = SvCUR(src) - (slen+sdone);
261 sv_setpvn(src, (char*)s+slen, sdone);
263 SvCUR_set(src, sdone);
265 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
267 SvCUR_set(dst, dlen+ddone);
270 #if ENCODE_XS_PROFILE
271 if (SvCUR(dst) > SvCUR(src)){
273 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
274 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
275 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
280 *offset += sdone + slen;
284 if (retcode) *retcode = code;
289 strict_utf8(pTHX_ SV* sv)
294 if (!sv || SvTYPE(sv) != SVt_PVHV)
297 svp = hv_fetch(hv, "strict_utf8", 11, 0);
304 process_utf8(pTHX_ SV* dst, U8* s, U8* e, SV *check_sv,
305 bool encode, bool strict, bool stop_at_partial)
312 if (SvROK(check_sv)) {
313 /* croak("UTF-8 decoder doesn't support callback CHECK"); */
314 fallback_cb = check_sv;
315 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as perlqq */
318 fallback_cb = &PL_sv_undef;
319 check = SvIV(check_sv);
326 if (UTF8_IS_INVARIANT(*s)) {
327 sv_catpvn(dst, (char *)s, 1);
332 if (UTF8_IS_START(*s)) {
333 U8 skip = UTF8SKIP(s);
334 if ((s + skip) > e) {
335 /* Partial character */
336 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
337 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
343 uv = utf8n_to_uvuni(s, e - s, &ulen,
344 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
345 UTF8_ALLOW_NONSTRICT)
347 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
348 if (strict && uv > PERL_UNICODE_MAX)
353 uv = utf8n_to_uvuni(s, e - s, &ulen,
354 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
363 /* Whole char is good */
364 sv_catpvn(dst,(char *)s,skip);
369 /* If we get here there is something wrong with alleged UTF-8 */
375 if (check & ENCODE_DIE_ON_ERR){
377 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
379 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
381 if (check & ENCODE_WARN_ON_ERR){
383 Perl_warner(aTHX_ packWARN(WARN_UTF8),
384 ERR_ENCODE_NOMAP, uv, "utf8");
386 Perl_warner(aTHX_ packWARN(WARN_UTF8),
387 ERR_DECODE_NOMAP, "utf8", uv);
389 if (check & ENCODE_RETURN_ON_ERR) {
392 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
394 (fallback_cb != &PL_sv_undef)
395 ? do_fallback_cb(aTHX_ uv, fallback_cb)
396 : newSVpvf(check & ENCODE_PERLQQ
397 ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}")
398 : check & ENCODE_HTMLCREF ? "&#%" UVuf ";"
399 : "&#x%" UVxf ";", uv);
401 SvUTF8_off(subchar); /* make sure no decoded string gets in */
403 sv_catsv(dst, subchar);
404 SvREFCNT_dec(subchar);
406 sv_catpv(dst, FBCHAR_UTF8);
416 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
421 Method_decode_xs(obj,src,check_sv = &PL_sv_no)
434 dSP; ENTER; SAVETMPS;
435 if (src == &PL_sv_undef) src = newSV(0);
436 s = (U8 *) SvPV(src, slen);
437 e = (U8 *) SvEND(src);
438 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
439 check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV(check_sv);
441 * PerlIO check -- we assume the object is of PerlIO if renewed
446 if (call_method("renewed",G_SCALAR) == 1) {
448 renewed = (bool)POPi;
451 fprintf(stderr, "renewed == %d\n", renewed);
455 /* end PerlIO check */
458 s = utf8_to_bytes(s,&slen);
465 croak("Cannot decode string with wide characters");
469 s = process_utf8(aTHX_ dst, s, e, check_sv, 0, strict_utf8(aTHX_ obj), renewed);
471 /* Clear out translated part of source unless asked not to */
472 if (check && !(check & ENCODE_LEAVE_SRC)){
475 sv_setpvn(src, (char*)s, slen);
477 SvCUR_set(src, slen);
480 ST(0) = sv_2mortal(dst);
485 Method_encode_xs(obj,src,check_sv = &PL_sv_no)
498 check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV(check_sv);
499 if (src == &PL_sv_undef) src = newSV(0);
500 s = (U8 *) SvPV(src, slen);
501 e = (U8 *) SvEND(src);
502 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
504 /* Already encoded */
505 if (strict_utf8(aTHX_ obj)) {
506 s = process_utf8(aTHX_ dst, s, e, check_sv, 1, 1, 0);
509 /* trust it and just copy the octets */
510 sv_setpvn(dst,(char *)s,(e-s));
515 /* Native bytes - can always encode */
516 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
518 UV uv = NATIVE_TO_UNI((UV) *s);
519 s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */
520 if (UNI_IS_INVARIANT(uv))
521 *d++ = (U8)UTF_TO_NATIVE(uv);
523 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
524 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
527 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
531 /* Clear out translated part of source unless asked not to */
532 if (check && !(check & ENCODE_LEAVE_SRC)){
535 sv_setpvn(src, (char*)s, slen);
537 SvCUR_set(src, slen);
541 ST(0) = sv_2mortal(dst);
545 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
570 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
571 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
576 Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no)
586 SV *fallback_cb = &PL_sv_undef;
587 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
588 STRLEN offset = (STRLEN)SvIV(off);
591 sv_utf8_downgrade(src, FALSE);
593 if (SvROK(check_sv)){
594 fallback_cb = check_sv;
595 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
597 check = SvIV(check_sv);
599 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
600 &offset, term, &code, fallback_cb));
601 SvIV_set(off, (IV)offset);
602 if (code == ENCODE_FOUND_TERM) {
611 Method_decode(obj,src,check_sv = &PL_sv_no)
618 SV *fallback_cb = &PL_sv_undef;
619 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
621 sv_utf8_downgrade(src, FALSE);
623 if (SvROK(check_sv)){
624 fallback_cb = check_sv;
625 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
627 check = SvIV(check_sv);
629 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
630 NULL, Nullsv, NULL, fallback_cb);
636 Method_encode(obj,src,check_sv = &PL_sv_no)
643 SV *fallback_cb = &PL_sv_undef;
644 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
645 sv_utf8_upgrade(src);
646 if (SvROK(check_sv)){
647 fallback_cb = check_sv;
648 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
650 check = SvIV(check_sv);
652 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
653 NULL, Nullsv, NULL, fallback_cb);
658 Method_needs_lines(obj)
662 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
668 Method_perlio_ok(obj)
672 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
673 /* require_pv(PERLIO_FILENAME); */
675 eval_pv("require PerlIO::encoding", 0);
677 if (SvTRUE(get_sv("@", 0))) {
686 Method_mime_name(obj)
690 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
692 eval_pv("require Encode::MIME::Name", 0);
694 if (SvTRUE(get_sv("@", 0))) {
695 ST(0) = &PL_sv_undef;
700 XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0]))));
702 call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR);
704 retval = newSVsv(POPs);
714 MODULE = Encode PACKAGE = Encode
719 _bytes_to_utf8(sv, ...)
723 SV * encoding = items == 2 ? ST(1) : Nullsv;
726 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
729 U8* s = (U8*)SvPV(sv, len);
732 converted = bytes_to_utf8(s, &len); /* This allocs */
733 sv_setpvn(sv, (char *)converted, len);
734 SvUTF8_on(sv); /* XXX Should we? */
735 Safefree(converted); /* ... so free it */
743 _utf8_to_bytes(sv, ...)
747 SV * to = items > 1 ? ST(1) : Nullsv;
748 SV * check = items > 2 ? ST(2) : Nullsv;
751 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
754 U8 *s = (U8*)SvPV(sv, len);
758 /* Must do things the slow way */
760 /* We need a copy to pass to check() */
765 New(83, dest, len, U8); /* I think */
775 /* Have to do it all ourselves because of error routine,
777 if (!(uv & 0x40)){ goto failure; }
778 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
779 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
780 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
781 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
782 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
783 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
784 else { ulen = 13; uv = 0; }
786 /* Note change to utf8.c variable naming, for variety */
788 if ((*s & 0xc0) != 0x80){
791 uv = (uv << 6) | (*s++ & 0x3f);
796 call_failure(check, s, dest, src);
797 /* Now what happens? */
803 sv_usepvn(sv, (char *)dest, RETVAL);
806 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
814 is_utf8(sv, check = 0)
819 if (SvGMAGICAL(sv)) /* it could be $1, for example */
820 sv = newSVsv(sv); /* GMAGIG will be done */
821 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
824 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
827 SvREFCNT_dec(sv); /* it was a temp copy */
838 SV *rsv = newSViv(SvUTF8(sv));
842 RETVAL = &PL_sv_undef;
854 SV *rsv = newSViv(SvUTF8(sv));
858 RETVAL = &PL_sv_undef;
867 RETVAL = ENCODE_DIE_ON_ERR;
874 RETVAL = ENCODE_WARN_ON_ERR;
881 RETVAL = ENCODE_LEAVE_SRC;
888 RETVAL = ENCODE_RETURN_ON_ERR;
895 RETVAL = ENCODE_PERLQQ;
902 RETVAL = ENCODE_HTMLCREF;
909 RETVAL = ENCODE_XMLCREF;
916 RETVAL = ENCODE_STOP_AT_PARTIAL;
923 RETVAL = ENCODE_FB_DEFAULT;
930 RETVAL = ENCODE_FB_CROAK;
937 RETVAL = ENCODE_FB_QUIET;
944 RETVAL = ENCODE_FB_WARN;
951 RETVAL = ENCODE_FB_PERLQQ;
958 RETVAL = ENCODE_FB_HTMLCREF;
965 RETVAL = ENCODE_FB_XMLCREF;