2 $Id: Encode.xs,v 2.20 2010/12/31 22:48:48 dankogai Exp $
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 #ifdef UTF8_DISALLOW_ILLEGAL_INTERCHANGE
33 # define UTF8_ALLOW_STRICT UTF8_DISALLOW_ILLEGAL_INTERCHANGE
35 # define UTF8_ALLOW_STRICT 0
38 #define UTF8_ALLOW_NONSTRICT (UTF8_ALLOW_ANY & \
39 ~(UTF8_ALLOW_CONTINUATION | \
40 UTF8_ALLOW_NON_CONTINUATION | \
44 Encode_XSEncoding(pTHX_ encode_t * enc)
47 HV *stash = gv_stashpv("Encode::XS", TRUE);
48 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
52 while (enc->name[i]) {
53 const char *name = enc->name[i++];
54 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
57 call_pv("Encode::define_encoding", G_DISCARD);
62 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
64 /* Exists for breakpointing */
68 #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
69 #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
72 do_fallback_cb(pTHX_ UV ch, SV *fallback_cb)
76 SV *retval = newSVpv("",0);
80 XPUSHs(sv_2mortal(newSVnv((UV)ch)));
82 argc = call_sv(fallback_cb, G_SCALAR);
85 croak("fallback sub must return scalar!");
87 sv_catsv(retval, POPs);
95 encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src,
96 int check, STRLEN * offset, SV * term, int * retcode,
100 U8 *s = (U8 *) SvPV(src, slen);
105 /* We allocate slen+1.
106 PerlIO dumps core if this value is smaller than this. */
107 SV *dst = sv_2mortal(newSV(slen+1));
108 U8 *d = (U8 *)SvPVX(dst);
109 STRLEN dlen = SvLEN(dst)-1;
112 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
116 if (slen > *offset){ /* safeguard against slen overflow */
130 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
133 SvCUR_set(dst, dlen+ddone);
136 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
137 code == ENCODE_FOUND_TERM) {
143 STRLEN more = 0; /* make sure you initialize! */
147 sleft = tlen - sdone;
148 #if ENCODE_XS_PROFILE >= 2
150 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
151 more, sdone, sleft, SvLEN(dst));
153 if (sdone != 0) { /* has src ever been processed ? */
154 #if ENCODE_XS_USEFP == 2
155 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
157 #elif ENCODE_XS_USEFP
158 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
160 /* safe until SvLEN(dst) == MAX_INT/16 */
161 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
164 more += UTF8_MAXLEN; /* insurance policy */
165 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
166 /* dst need to grow need MORE bytes! */
167 if (ddone >= SvLEN(dst)) {
168 Perl_croak(aTHX_ "Destination couldn't be grown.");
170 dlen = SvLEN(dst)-ddone-1;
178 if (dir == enc->f_utf8) {
181 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
182 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
183 /* if non-representable multibyte prefix at end of current buffer - break*/
184 if (clen > tlen - sdone) break;
185 if (check & ENCODE_DIE_ON_ERR) {
186 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
187 (UV)ch, enc->name[0]);
188 return &PL_sv_undef; /* never reaches but be safe */
190 if (check & ENCODE_WARN_ON_ERR){
191 Perl_warner(aTHX_ packWARN(WARN_UTF8),
192 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
194 if (check & ENCODE_RETURN_ON_ERR){
197 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
199 (fallback_cb != &PL_sv_undef)
200 ? do_fallback_cb(aTHX_ ch, fallback_cb)
201 : newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
202 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
203 "&#x%" UVxf ";", (UV)ch);
204 SvUTF8_off(subchar); /* make sure no decoded string gets in */
205 sdone += slen + clen;
206 ddone += dlen + SvCUR(subchar);
207 sv_catsv(dst, subchar);
208 SvREFCNT_dec(subchar);
211 sdone += slen + clen;
212 ddone += dlen + enc->replen;
213 sv_catpvn(dst, (char*)enc->rep, enc->replen);
218 if (check & ENCODE_DIE_ON_ERR){
219 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
220 enc->name[0], (UV)s[slen]);
221 return &PL_sv_undef; /* never reaches but be safe */
223 if (check & ENCODE_WARN_ON_ERR){
225 aTHX_ packWARN(WARN_UTF8),
227 enc->name[0], (UV)s[slen]);
229 if (check & ENCODE_RETURN_ON_ERR){
233 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
235 (fallback_cb != &PL_sv_undef)
236 ? do_fallback_cb(aTHX_ (UV)s[slen], fallback_cb)
237 : newSVpvf("\\x%02" UVXf, (UV)s[slen]);
239 ddone += dlen + SvCUR(subchar);
240 sv_catsv(dst, subchar);
241 SvREFCNT_dec(subchar);
244 ddone += dlen + strlen(FBCHAR_UTF8);
245 sv_catpv(dst, FBCHAR_UTF8);
248 /* settle variables when fallback */
249 d = (U8 *)SvEND(dst);
250 dlen = SvLEN(dst) - ddone - 1;
251 s = (U8*)SvPVX(src) + sdone;
256 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
257 code, (dir == enc->f_utf8) ? "to" : "from",
263 if (check && !(check & ENCODE_LEAVE_SRC)){
264 sdone = SvCUR(src) - (slen+sdone);
266 sv_setpvn(src, (char*)s+slen, sdone);
268 SvCUR_set(src, sdone);
270 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
272 SvCUR_set(dst, dlen+ddone);
275 #if ENCODE_XS_PROFILE
276 if (SvCUR(dst) > SvCUR(src)){
278 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
279 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
280 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
285 *offset += sdone + slen;
289 if (retcode) *retcode = code;
294 strict_utf8(pTHX_ SV* sv)
299 if (!sv || SvTYPE(sv) != SVt_PVHV)
302 svp = hv_fetch(hv, "strict_utf8", 11, 0);
309 process_utf8(pTHX_ SV* dst, U8* s, U8* e, SV *check_sv,
310 bool encode, bool strict, bool stop_at_partial)
317 if (SvROK(check_sv)) {
318 /* croak("UTF-8 decoder doesn't support callback CHECK"); */
319 fallback_cb = check_sv;
320 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as perlqq */
323 fallback_cb = &PL_sv_undef;
324 check = SvIV(check_sv);
331 if (UTF8_IS_INVARIANT(*s)) {
332 sv_catpvn(dst, (char *)s, 1);
337 if (UTF8_IS_START(*s)) {
338 U8 skip = UTF8SKIP(s);
339 if ((s + skip) > e) {
340 /* Partial character */
341 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
342 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
348 uv = utf8n_to_uvuni(s, e - s, &ulen,
349 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
350 UTF8_ALLOW_NONSTRICT)
352 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
353 if (strict && uv > PERL_UNICODE_MAX)
358 uv = utf8n_to_uvuni(s, e - s, &ulen,
359 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
368 /* Whole char is good */
369 sv_catpvn(dst,(char *)s,skip);
374 /* If we get here there is something wrong with alleged UTF-8 */
380 if (check & ENCODE_DIE_ON_ERR){
382 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
384 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
386 if (check & ENCODE_WARN_ON_ERR){
388 Perl_warner(aTHX_ packWARN(WARN_UTF8),
389 ERR_ENCODE_NOMAP, uv, "utf8");
391 Perl_warner(aTHX_ packWARN(WARN_UTF8),
392 ERR_DECODE_NOMAP, "utf8", uv);
394 if (check & ENCODE_RETURN_ON_ERR) {
397 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
399 (fallback_cb != &PL_sv_undef)
400 ? do_fallback_cb(aTHX_ uv, fallback_cb)
401 : newSVpvf(check & ENCODE_PERLQQ
402 ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}")
403 : check & ENCODE_HTMLCREF ? "&#%" UVuf ";"
404 : "&#x%" UVxf ";", uv);
406 SvUTF8_off(subchar); /* make sure no decoded string gets in */
408 sv_catsv(dst, subchar);
409 SvREFCNT_dec(subchar);
411 sv_catpv(dst, FBCHAR_UTF8);
421 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
426 Method_decode_xs(obj,src,check_sv = &PL_sv_no)
439 dSP; ENTER; SAVETMPS;
440 if (src == &PL_sv_undef || SvROK(src)) src = sv_2mortal(newSV(0));
441 s = (U8 *) SvPV(src, slen);
442 e = (U8 *) SvEND(src);
443 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
444 check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV(check_sv);
446 * PerlIO check -- we assume the object is of PerlIO if renewed
451 if (call_method("renewed",G_SCALAR) == 1) {
453 renewed = (bool)POPi;
456 fprintf(stderr, "renewed == %d\n", renewed);
460 /* end PerlIO check */
463 s = utf8_to_bytes(s,&slen);
470 croak("Cannot decode string with wide characters");
474 s = process_utf8(aTHX_ dst, s, e, check_sv, 0, strict_utf8(aTHX_ obj), renewed);
476 /* Clear out translated part of source unless asked not to */
477 if (check && !(check & ENCODE_LEAVE_SRC)){
480 sv_setpvn(src, (char*)s, slen);
482 SvCUR_set(src, slen);
485 ST(0) = sv_2mortal(dst);
490 Method_encode_xs(obj,src,check_sv = &PL_sv_no)
503 check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV(check_sv);
504 if (src == &PL_sv_undef || SvROK(src)) src = sv_2mortal(newSV(0));
505 s = (U8 *) SvPV(src, slen);
506 e = (U8 *) SvEND(src);
507 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
509 /* Already encoded */
510 if (strict_utf8(aTHX_ obj)) {
511 s = process_utf8(aTHX_ dst, s, e, check_sv, 1, 1, 0);
514 /* trust it and just copy the octets */
515 sv_setpvn(dst,(char *)s,(e-s));
520 /* Native bytes - can always encode */
521 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
523 UV uv = NATIVE_TO_UNI((UV) *s);
524 s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */
525 if (UNI_IS_INVARIANT(uv))
526 *d++ = (U8)UTF_TO_NATIVE(uv);
528 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
529 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
532 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
536 /* Clear out translated part of source unless asked not to */
537 if (check && !(check & ENCODE_LEAVE_SRC)){
540 sv_setpvn(src, (char*)s, slen);
542 SvCUR_set(src, slen);
546 ST(0) = sv_2mortal(dst);
550 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
575 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
576 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
581 Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no)
591 SV *fallback_cb = &PL_sv_undef;
592 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
593 STRLEN offset = (STRLEN)SvIV(off);
596 sv_utf8_downgrade(src, FALSE);
598 if (SvROK(check_sv)){
599 fallback_cb = check_sv;
600 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
602 check = SvIV(check_sv);
604 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
605 &offset, term, &code, fallback_cb));
606 SvIV_set(off, (IV)offset);
607 if (code == ENCODE_FOUND_TERM) {
616 Method_decode(obj,src,check_sv = &PL_sv_no)
623 SV *fallback_cb = &PL_sv_undef;
624 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
626 sv_utf8_downgrade(src, FALSE);
628 if (SvROK(check_sv)){
629 fallback_cb = check_sv;
630 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
632 check = SvIV(check_sv);
634 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
635 NULL, Nullsv, NULL, fallback_cb);
641 Method_encode(obj,src,check_sv = &PL_sv_no)
648 SV *fallback_cb = &PL_sv_undef;
649 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
650 sv_utf8_upgrade(src);
651 if (SvROK(check_sv)){
652 fallback_cb = check_sv;
653 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
655 check = SvIV(check_sv);
657 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
658 NULL, Nullsv, NULL, fallback_cb);
663 Method_needs_lines(obj)
667 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
673 Method_perlio_ok(obj)
677 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
678 /* require_pv(PERLIO_FILENAME); */
680 eval_pv("require PerlIO::encoding", 0);
682 if (SvTRUE(get_sv("@", 0))) {
691 Method_mime_name(obj)
695 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
697 eval_pv("require Encode::MIME::Name", 0);
699 if (SvTRUE(get_sv("@", 0))) {
700 ST(0) = &PL_sv_undef;
705 XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0]))));
707 call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR);
709 retval = newSVsv(POPs);
719 MODULE = Encode PACKAGE = Encode
724 _bytes_to_utf8(sv, ...)
728 SV * encoding = items == 2 ? ST(1) : Nullsv;
731 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
734 U8* s = (U8*)SvPV(sv, len);
737 converted = bytes_to_utf8(s, &len); /* This allocs */
738 sv_setpvn(sv, (char *)converted, len);
739 SvUTF8_on(sv); /* XXX Should we? */
740 Safefree(converted); /* ... so free it */
748 _utf8_to_bytes(sv, ...)
752 SV * to = items > 1 ? ST(1) : Nullsv;
753 SV * check = items > 2 ? ST(2) : Nullsv;
756 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
759 U8 *s = (U8*)SvPV(sv, len);
763 /* Must do things the slow way */
765 /* We need a copy to pass to check() */
770 New(83, dest, len, U8); /* I think */
780 /* Have to do it all ourselves because of error routine,
782 if (!(uv & 0x40)){ goto failure; }
783 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
784 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
785 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
786 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
787 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
788 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
789 else { ulen = 13; uv = 0; }
791 /* Note change to utf8.c variable naming, for variety */
793 if ((*s & 0xc0) != 0x80){
796 uv = (uv << 6) | (*s++ & 0x3f);
801 call_failure(check, s, dest, src);
802 /* Now what happens? */
808 sv_usepvn(sv, (char *)dest, RETVAL);
811 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
819 is_utf8(sv, check = 0)
824 if (SvGMAGICAL(sv)) /* it could be $1, for example */
825 sv = newSVsv(sv); /* GMAGIG will be done */
826 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
829 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
832 SvREFCNT_dec(sv); /* it was a temp copy */
843 SV *rsv = newSViv(SvUTF8(sv));
847 RETVAL = &PL_sv_undef;
859 SV *rsv = newSViv(SvUTF8(sv));
863 RETVAL = &PL_sv_undef;
872 RETVAL = ENCODE_DIE_ON_ERR;
879 RETVAL = ENCODE_WARN_ON_ERR;
886 RETVAL = ENCODE_LEAVE_SRC;
893 RETVAL = ENCODE_RETURN_ON_ERR;
900 RETVAL = ENCODE_PERLQQ;
907 RETVAL = ENCODE_HTMLCREF;
914 RETVAL = ENCODE_XMLCREF;
921 RETVAL = ENCODE_STOP_AT_PARTIAL;
928 RETVAL = ENCODE_FB_DEFAULT;
935 RETVAL = ENCODE_FB_CROAK;
942 RETVAL = ENCODE_FB_QUIET;
949 RETVAL = ENCODE_FB_WARN;
956 RETVAL = ENCODE_FB_PERLQQ;
963 RETVAL = ENCODE_FB_HTMLCREF;
970 RETVAL = ENCODE_FB_XMLCREF;