2 $Id: Encode.xs,v 2.13 2007/05/29 07:35:27 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 | \
38 static SV* fallback_cb = (SV*)NULL ;
41 Encode_XSEncoding(pTHX_ encode_t * enc)
44 HV *stash = gv_stashpv("Encode::XS", TRUE);
45 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
49 while (enc->name[i]) {
50 const char *name = enc->name[i++];
51 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
54 call_pv("Encode::define_encoding", G_DISCARD);
59 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
61 /* Exists for breakpointing */
65 #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
66 #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
69 do_fallback_cb(pTHX_ UV ch)
77 XPUSHs(sv_2mortal(newSVnv((UV)ch)));
79 argc = call_sv(fallback_cb, G_SCALAR);
82 croak("fallback sub must return scalar!");
88 retval = newSVpv("",0);
89 sv_catsv(retval, temp);
95 encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src,
96 int check, STRLEN * offset, SV * term, int * retcode)
99 U8 *s = (U8 *) SvPV(src, slen);
104 /* We allocate slen+1.
105 PerlIO dumps core if this value is smaller than this. */
106 SV *dst = sv_2mortal(newSV(slen+1));
107 U8 *d = (U8 *)SvPVX(dst);
108 STRLEN dlen = SvLEN(dst)-1;
111 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
115 if (slen > *offset){ /* safeguard against slen overflow */
129 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
132 SvCUR_set(dst, dlen+ddone);
135 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
136 code == ENCODE_FOUND_TERM) {
142 STRLEN more = 0; /* make sure you initialize! */
146 sleft = tlen - sdone;
147 #if ENCODE_XS_PROFILE >= 2
149 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
150 more, sdone, sleft, SvLEN(dst));
152 if (sdone != 0) { /* has src ever been processed ? */
153 #if ENCODE_XS_USEFP == 2
154 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
156 #elif ENCODE_XS_USEFP
157 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
159 /* safe until SvLEN(dst) == MAX_INT/16 */
160 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
163 more += UTF8_MAXLEN; /* insurance policy */
164 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
165 /* dst need to grow need MORE bytes! */
166 if (ddone >= SvLEN(dst)) {
167 Perl_croak(aTHX_ "Destination couldn't be grown.");
169 dlen = SvLEN(dst)-ddone-1;
177 if (dir == enc->f_utf8) {
180 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
181 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
182 /* if non-representable multibyte prefix at end of current buffer - break*/
183 if (clen > tlen - sdone) break;
184 if (check & ENCODE_DIE_ON_ERR) {
185 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
186 (UV)ch, enc->name[0]);
187 return &PL_sv_undef; /* never reaches but be safe */
189 if (check & ENCODE_WARN_ON_ERR){
190 Perl_warner(aTHX_ packWARN(WARN_UTF8),
191 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
193 if (check & ENCODE_RETURN_ON_ERR){
196 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
198 (fallback_cb != (SV*)NULL) ? do_fallback_cb(aTHX_ ch) :
199 newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
200 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
201 "&#x%" UVxf ";", (UV)ch);
202 sdone += slen + clen;
203 ddone += dlen + SvCUR(subchar);
204 sv_catsv(dst, subchar);
205 SvREFCNT_dec(subchar);
208 sdone += slen + clen;
209 ddone += dlen + enc->replen;
210 sv_catpvn(dst, (char*)enc->rep, enc->replen);
215 if (check & ENCODE_DIE_ON_ERR){
216 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
217 enc->name[0], (UV)s[slen]);
218 return &PL_sv_undef; /* never reaches but be safe */
220 if (check & ENCODE_WARN_ON_ERR){
222 aTHX_ packWARN(WARN_UTF8),
224 enc->name[0], (UV)s[slen]);
226 if (check & ENCODE_RETURN_ON_ERR){
230 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
232 (fallback_cb != (SV*)NULL) ?
233 do_fallback_cb(aTHX_ (UV)s[slen]) :
234 newSVpvf("\\x%02" UVXf, (UV)s[slen]);
236 ddone += dlen + SvCUR(subchar);
237 sv_catsv(dst, subchar);
238 SvREFCNT_dec(subchar);
241 ddone += dlen + strlen(FBCHAR_UTF8);
242 sv_catpv(dst, FBCHAR_UTF8);
245 /* settle variables when fallback */
246 d = (U8 *)SvEND(dst);
247 dlen = SvLEN(dst) - ddone - 1;
248 s = (U8*)SvPVX(src) + sdone;
253 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
254 code, (dir == enc->f_utf8) ? "to" : "from",
260 if (check && !(check & ENCODE_LEAVE_SRC)){
261 sdone = SvCUR(src) - (slen+sdone);
263 sv_setpvn(src, (char*)s+slen, sdone);
265 SvCUR_set(src, sdone);
267 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
269 SvCUR_set(dst, dlen+ddone);
272 #if ENCODE_XS_PROFILE
273 if (SvCUR(dst) > SvCUR(src)){
275 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
276 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
277 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
282 *offset += sdone + slen;
286 if (retcode) *retcode = code;
291 strict_utf8(pTHX_ SV* sv)
296 if (!sv || SvTYPE(sv) != SVt_PVHV)
299 svp = hv_fetch(hv, "strict_utf8", 11, 0);
306 process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check,
307 bool encode, bool strict, bool stop_at_partial)
316 if (UTF8_IS_INVARIANT(*s)) {
317 sv_catpvn(dst, (char *)s, 1);
322 if (UTF8_IS_START(*s)) {
323 U8 skip = UTF8SKIP(s);
324 if ((s + skip) > e) {
325 /* Partial character */
326 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
327 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
333 uv = utf8n_to_uvuni(s, e - s, &ulen,
334 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
335 UTF8_ALLOW_NONSTRICT)
337 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
338 if (strict && uv > PERL_UNICODE_MAX)
343 uv = utf8n_to_uvuni(s, e - s, &ulen,
344 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
353 /* Whole char is good */
354 sv_catpvn(dst,(char *)s,skip);
359 /* If we get here there is something wrong with alleged UTF-8 */
365 if (check & ENCODE_DIE_ON_ERR){
367 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
369 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
371 if (check & ENCODE_WARN_ON_ERR){
373 Perl_warner(aTHX_ packWARN(WARN_UTF8),
374 ERR_ENCODE_NOMAP, uv, "utf8");
376 Perl_warner(aTHX_ packWARN(WARN_UTF8),
377 ERR_DECODE_NOMAP, "utf8", uv);
379 if (check & ENCODE_RETURN_ON_ERR) {
382 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
383 SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"):
384 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
385 "&#x%" UVxf ";", uv);
386 sv_catsv(dst, subchar);
387 SvREFCNT_dec(subchar);
389 sv_catpv(dst, FBCHAR_UTF8);
399 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
404 Method_decode_xs(obj,src,check = 0)
411 U8 *s = (U8 *) SvPV(src, slen);
412 U8 *e = (U8 *) SvEND(src);
413 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
416 * PerlIO check -- we assume the object is of PerlIO if renewed
419 dSP; ENTER; SAVETMPS;
423 if (call_method("renewed",G_SCALAR) == 1) {
425 renewed = (bool)POPi;
428 fprintf(stderr, "renewed == %d\n", renewed);
432 /* end PerlIO check */
435 s = utf8_to_bytes(s,&slen);
442 croak("Cannot decode string with wide characters");
446 s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed);
448 /* Clear out translated part of source unless asked not to */
449 if (check && !(check & ENCODE_LEAVE_SRC)){
452 sv_setpvn(src, (char*)s, slen);
454 SvCUR_set(src, slen);
457 ST(0) = sv_2mortal(dst);
462 Method_encode_xs(obj,src,check = 0)
469 U8 *s = (U8 *) SvPV(src, slen);
470 U8 *e = (U8 *) SvEND(src);
471 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
473 /* Already encoded */
474 if (strict_utf8(aTHX_ obj)) {
475 s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0);
478 /* trust it and just copy the octets */
479 sv_setpvn(dst,(char *)s,(e-s));
484 /* Native bytes - can always encode */
485 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
487 UV uv = NATIVE_TO_UNI((UV) *s);
488 s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */
489 if (UNI_IS_INVARIANT(uv))
490 *d++ = (U8)UTF_TO_NATIVE(uv);
492 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
493 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
496 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
500 /* Clear out translated part of source unless asked not to */
501 if (check && !(check & ENCODE_LEAVE_SRC)){
504 sv_setpvn(src, (char*)s, slen);
506 SvCUR_set(src, slen);
510 ST(0) = sv_2mortal(dst);
514 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
539 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
540 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
545 Method_cat_decode(obj, dst, src, off, term, check = 0)
554 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
555 STRLEN offset = (STRLEN)SvIV(off);
558 sv_utf8_downgrade(src, FALSE);
560 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
561 &offset, term, &code));
562 SvIV_set(off, (IV)offset);
563 if (code == ENCODE_FOUND_TERM) {
572 Method_decode(obj,src,check_sv = &PL_sv_no)
579 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
581 sv_utf8_downgrade(src, FALSE);
583 if (SvROK(check_sv)){
584 if (fallback_cb == (SV*)NULL){
585 fallback_cb = newSVsv(check_sv); /* First time */
587 SvSetSV(fallback_cb, check_sv); /* Been here before */
589 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
591 fallback_cb = (SV*)NULL;
592 check = SvIV(check_sv);
594 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
603 Method_encode(obj,src,check_sv = &PL_sv_no)
610 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
611 sv_utf8_upgrade(src);
612 if (SvROK(check_sv)){
613 if (fallback_cb == (SV*)NULL){
614 fallback_cb = newSVsv(check_sv); /* First time */
616 SvSetSV(fallback_cb, check_sv); /* Been here before */
618 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
620 fallback_cb = (SV*)NULL;
621 check = SvIV(check_sv);
623 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
629 Method_needs_lines(obj)
633 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
639 Method_perlio_ok(obj)
643 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
644 /* require_pv(PERLIO_FILENAME); */
646 eval_pv("require PerlIO::encoding", 0);
648 if (SvTRUE(get_sv("@", 0))) {
657 Method_mime_name(obj)
661 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
663 eval_pv("require Encode::MIME::Name", 0);
665 if (SvTRUE(get_sv("@", 0))) {
666 ST(0) = &PL_sv_undef;
671 XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0]))));
673 call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR);
675 retval = newSVsv(POPs);
685 MODULE = Encode PACKAGE = Encode
690 _bytes_to_utf8(sv, ...)
694 SV * encoding = items == 2 ? ST(1) : Nullsv;
697 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
700 U8* s = (U8*)SvPV(sv, len);
703 converted = bytes_to_utf8(s, &len); /* This allocs */
704 sv_setpvn(sv, (char *)converted, len);
705 SvUTF8_on(sv); /* XXX Should we? */
706 Safefree(converted); /* ... so free it */
714 _utf8_to_bytes(sv, ...)
718 SV * to = items > 1 ? ST(1) : Nullsv;
719 SV * check = items > 2 ? ST(2) : Nullsv;
722 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
725 U8 *s = (U8*)SvPV(sv, len);
729 /* Must do things the slow way */
731 /* We need a copy to pass to check() */
736 New(83, dest, len, U8); /* I think */
746 /* Have to do it all ourselves because of error routine,
748 if (!(uv & 0x40)){ goto failure; }
749 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
750 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
751 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
752 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
753 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
754 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
755 else { ulen = 13; uv = 0; }
757 /* Note change to utf8.c variable naming, for variety */
759 if ((*s & 0xc0) != 0x80){
762 uv = (uv << 6) | (*s++ & 0x3f);
767 call_failure(check, s, dest, src);
768 /* Now what happens? */
774 sv_usepvn(sv, (char *)dest, RETVAL);
777 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
785 is_utf8(sv, check = 0)
790 if (SvGMAGICAL(sv)) /* it could be $1, for example */
791 sv = newSVsv(sv); /* GMAGIG will be done */
792 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
795 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
798 SvREFCNT_dec(sv); /* it was a temp copy */
809 SV *rsv = newSViv(SvUTF8(sv));
813 RETVAL = &PL_sv_undef;
825 SV *rsv = newSViv(SvUTF8(sv));
829 RETVAL = &PL_sv_undef;
838 RETVAL = ENCODE_DIE_ON_ERR;
845 RETVAL = ENCODE_WARN_ON_ERR;
852 RETVAL = ENCODE_LEAVE_SRC;
859 RETVAL = ENCODE_RETURN_ON_ERR;
866 RETVAL = ENCODE_PERLQQ;
873 RETVAL = ENCODE_HTMLCREF;
880 RETVAL = ENCODE_XMLCREF;
887 RETVAL = ENCODE_STOP_AT_PARTIAL;
894 RETVAL = ENCODE_FB_DEFAULT;
901 RETVAL = ENCODE_FB_CROAK;
908 RETVAL = ENCODE_FB_QUIET;
915 RETVAL = ENCODE_FB_WARN;
922 RETVAL = ENCODE_FB_PERLQQ;
929 RETVAL = ENCODE_FB_HTMLCREF;
936 RETVAL = ENCODE_FB_XMLCREF;