2 $Id: Encode.xs,v 2.7 2006/01/15 15:43:36 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!");
84 retval = newSVsv(POPs);
92 encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src,
93 int check, STRLEN * offset, SV * term, int * retcode)
96 U8 *s = (U8 *) SvPV(src, slen);
101 /* We allocate slen+1.
102 PerlIO dumps core if this value is smaller than this. */
103 SV *dst = sv_2mortal(newSV(slen+1));
104 U8 *d = (U8 *)SvPVX(dst);
105 STRLEN dlen = SvLEN(dst)-1;
108 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
112 if (slen > *offset){ /* safeguard against slen overflow */
126 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
129 SvCUR_set(dst, dlen+ddone);
132 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
133 code == ENCODE_FOUND_TERM) {
139 STRLEN more = 0; /* make sure you initialize! */
143 sleft = tlen - sdone;
144 #if ENCODE_XS_PROFILE >= 2
146 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
147 more, sdone, sleft, SvLEN(dst));
149 if (sdone != 0) { /* has src ever been processed ? */
150 #if ENCODE_XS_USEFP == 2
151 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
153 #elif ENCODE_XS_USEFP
154 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
156 /* safe until SvLEN(dst) == MAX_INT/16 */
157 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
160 more += UTF8_MAXLEN; /* insurance policy */
161 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
162 /* dst need to grow need MORE bytes! */
163 if (ddone >= SvLEN(dst)) {
164 Perl_croak(aTHX_ "Destination couldn't be grown.");
166 dlen = SvLEN(dst)-ddone-1;
174 if (dir == enc->f_utf8) {
177 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
178 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
179 /* if non-representable multibyte prefix at end of current buffer - break*/
180 if (clen > tlen - sdone) break;
181 if (check & ENCODE_DIE_ON_ERR) {
182 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
183 (UV)ch, enc->name[0]);
184 return &PL_sv_undef; /* never reaches but be safe */
186 if (check & ENCODE_WARN_ON_ERR){
187 Perl_warner(aTHX_ packWARN(WARN_UTF8),
188 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
190 if (check & ENCODE_RETURN_ON_ERR){
193 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
195 (fallback_cb != (SV*)NULL) ? do_fallback_cb(aTHX_ ch) :
196 newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
197 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
198 "&#x%" UVxf ";", (UV)ch);
199 sdone += slen + clen;
200 ddone += dlen + SvCUR(subchar);
201 sv_catsv(dst, subchar);
202 SvREFCNT_dec(subchar);
205 sdone += slen + clen;
206 ddone += dlen + enc->replen;
207 sv_catpvn(dst, (char*)enc->rep, enc->replen);
212 if (check & ENCODE_DIE_ON_ERR){
213 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
214 enc->name[0], (UV)s[slen]);
215 return &PL_sv_undef; /* never reaches but be safe */
217 if (check & ENCODE_WARN_ON_ERR){
219 aTHX_ packWARN(WARN_UTF8),
221 enc->name[0], (UV)s[slen]);
223 if (check & ENCODE_RETURN_ON_ERR){
227 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
229 (fallback_cb != (SV*)NULL) ?
230 do_fallback_cb(aTHX_ (UV)s[slen]) :
231 newSVpvf("\\x%02" UVXf, (UV)s[slen]);
233 ddone += dlen + SvCUR(subchar);
234 sv_catsv(dst, subchar);
235 SvREFCNT_dec(subchar);
238 ddone += dlen + strlen(FBCHAR_UTF8);
239 sv_catpv(dst, FBCHAR_UTF8);
242 /* settle variables when fallback */
243 d = (U8 *)SvEND(dst);
244 dlen = SvLEN(dst) - ddone - 1;
245 s = (U8*)SvPVX(src) + sdone;
250 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
251 code, (dir == enc->f_utf8) ? "to" : "from",
257 if (check && !(check & ENCODE_LEAVE_SRC)){
258 sdone = SvCUR(src) - (slen+sdone);
260 sv_setpvn(src, (char*)s+slen, sdone);
262 SvCUR_set(src, sdone);
264 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
266 SvCUR_set(dst, dlen+ddone);
269 #if ENCODE_XS_PROFILE
270 if (SvCUR(dst) > SvCUR(src)){
272 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
273 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
274 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
279 *offset += sdone + slen;
283 if (retcode) *retcode = code;
288 strict_utf8(pTHX_ SV* sv)
293 if (!sv || SvTYPE(sv) != SVt_PVHV)
296 svp = hv_fetch(hv, "strict_utf8", 11, 0);
303 process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check,
304 bool encode, bool strict, bool stop_at_partial)
313 if (UTF8_IS_INVARIANT(*s)) {
314 sv_catpvn(dst, (char *)s, 1);
319 if (UTF8_IS_START(*s)) {
320 U8 skip = UTF8SKIP(s);
321 if ((s + skip) > e) {
322 /* Partial character */
323 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
324 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
330 uv = utf8n_to_uvuni(s, e - s, &ulen,
331 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
332 UTF8_ALLOW_NONSTRICT)
334 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
335 if (strict && uv > PERL_UNICODE_MAX)
340 uv = utf8n_to_uvuni(s, e - s, &ulen,
341 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
350 /* Whole char is good */
351 sv_catpvn(dst,(char *)s,skip);
356 /* If we get here there is something wrong with alleged UTF-8 */
362 if (check & ENCODE_DIE_ON_ERR){
364 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
366 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
368 if (check & ENCODE_WARN_ON_ERR){
370 Perl_warner(aTHX_ packWARN(WARN_UTF8),
371 ERR_ENCODE_NOMAP, uv, "utf8");
373 Perl_warner(aTHX_ packWARN(WARN_UTF8),
374 ERR_DECODE_NOMAP, "utf8", uv);
376 if (check & ENCODE_RETURN_ON_ERR) {
379 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
380 SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"):
381 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
382 "&#x%" UVxf ";", uv);
383 sv_catsv(dst, subchar);
384 SvREFCNT_dec(subchar);
386 sv_catpv(dst, FBCHAR_UTF8);
396 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
401 Method_decode_xs(obj,src,check = 0)
408 U8 *s = (U8 *) SvPV(src, slen);
409 U8 *e = (U8 *) SvEND(src);
410 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
413 * PerlIO check -- we assume the object is of PerlIO if renewed
416 dSP; ENTER; SAVETMPS;
420 if (call_method("renewed",G_SCALAR) == 1) {
422 renewed = (bool)POPi;
425 fprintf(stderr, "renewed == %d\n", renewed);
429 /* end PerlIO check */
432 s = utf8_to_bytes(s,&slen);
439 croak("Cannot decode string with wide characters");
443 s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed);
445 /* Clear out translated part of source unless asked not to */
446 if (check && !(check & ENCODE_LEAVE_SRC)){
449 sv_setpvn(src, (char*)s, slen);
451 SvCUR_set(src, slen);
454 ST(0) = sv_2mortal(dst);
459 Method_encode_xs(obj,src,check = 0)
466 U8 *s = (U8 *) SvPV(src, slen);
467 U8 *e = (U8 *) SvEND(src);
468 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
470 /* Already encoded */
471 if (strict_utf8(aTHX_ obj)) {
472 s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0);
475 /* trust it and just copy the octets */
476 sv_setpvn(dst,(char *)s,(e-s));
481 /* Native bytes - can always encode */
482 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
484 UV uv = NATIVE_TO_UNI((UV) *s++);
485 if (UNI_IS_INVARIANT(uv))
486 *d++ = (U8)UTF_TO_NATIVE(uv);
488 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
489 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
492 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
496 /* Clear out translated part of source unless asked not to */
497 if (check && !(check & ENCODE_LEAVE_SRC)){
500 sv_setpvn(src, (char*)s, slen);
502 SvCUR_set(src, slen);
506 ST(0) = sv_2mortal(dst);
510 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
535 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
536 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
541 Method_cat_decode(obj, dst, src, off, term, check = 0)
550 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
551 STRLEN offset = (STRLEN)SvIV(off);
554 sv_utf8_downgrade(src, FALSE);
556 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
557 &offset, term, &code));
558 SvIV_set(off, (IV)offset);
559 if (code == ENCODE_FOUND_TERM) {
568 Method_decode(obj,src,check_sv = &PL_sv_no)
575 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
577 sv_utf8_downgrade(src, FALSE);
579 if (SvROK(check_sv)){
580 if (fallback_cb == (SV*)NULL){
581 fallback_cb = newSVsv(check_sv); /* First time */
583 SvSetSV(fallback_cb, check_sv); /* Been here before */
585 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
587 fallback_cb = (SV*)NULL;
588 check = SvIV(check_sv);
590 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
599 Method_encode(obj,src,check_sv = &PL_sv_no)
606 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
607 sv_utf8_upgrade(src);
608 if (SvROK(check_sv)){
609 if (fallback_cb == (SV*)NULL){
610 fallback_cb = newSVsv(check_sv); /* First time */
612 SvSetSV(fallback_cb, check_sv); /* Been here before */
614 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
616 fallback_cb = (SV*)NULL;
617 check = SvIV(check_sv);
619 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
625 Method_needs_lines(obj)
629 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
635 Method_perlio_ok(obj)
639 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
640 /* require_pv(PERLIO_FILENAME); */
642 eval_pv("require PerlIO::encoding", 0);
644 if (SvTRUE(get_sv("@", 0))) {
652 MODULE = Encode PACKAGE = Encode
657 _bytes_to_utf8(sv, ...)
661 SV * encoding = items == 2 ? ST(1) : Nullsv;
664 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
667 U8* s = (U8*)SvPV(sv, len);
670 converted = bytes_to_utf8(s, &len); /* This allocs */
671 sv_setpvn(sv, (char *)converted, len);
672 SvUTF8_on(sv); /* XXX Should we? */
673 Safefree(converted); /* ... so free it */
681 _utf8_to_bytes(sv, ...)
685 SV * to = items > 1 ? ST(1) : Nullsv;
686 SV * check = items > 2 ? ST(2) : Nullsv;
689 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
692 U8 *s = (U8*)SvPV(sv, len);
696 /* Must do things the slow way */
698 /* We need a copy to pass to check() */
699 U8 *src = (U8*)savepv((char *)s);
702 New(83, dest, len, U8); /* I think */
711 /* Have to do it all ourselves because of error routine,
713 if (!(uv & 0x40)){ goto failure; }
714 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
715 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
716 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
717 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
718 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
719 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
720 else { ulen = 13; uv = 0; }
722 /* Note change to utf8.c variable naming, for variety */
724 if ((*s & 0xc0) != 0x80){
727 uv = (uv << 6) | (*s++ & 0x3f);
732 call_failure(check, s, dest, src);
733 /* Now what happens? */
739 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
747 is_utf8(sv, check = 0)
752 if (SvGMAGICAL(sv)) /* it could be $1, for example */
753 sv = newSVsv(sv); /* GMAGIG will be done */
755 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
758 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
764 SvREFCNT_dec(sv); /* it was a temp copy */
775 SV *rsv = newSViv(SvUTF8(sv));
779 RETVAL = &PL_sv_undef;
791 SV *rsv = newSViv(SvUTF8(sv));
795 RETVAL = &PL_sv_undef;
804 RETVAL = ENCODE_DIE_ON_ERR;
811 RETVAL = ENCODE_WARN_ON_ERR;
818 RETVAL = ENCODE_LEAVE_SRC;
825 RETVAL = ENCODE_RETURN_ON_ERR;
832 RETVAL = ENCODE_PERLQQ;
839 RETVAL = ENCODE_HTMLCREF;
846 RETVAL = ENCODE_XMLCREF;
853 RETVAL = ENCODE_STOP_AT_PARTIAL;
860 RETVAL = ENCODE_FB_DEFAULT;
867 RETVAL = ENCODE_FB_CROAK;
874 RETVAL = ENCODE_FB_QUIET;
881 RETVAL = ENCODE_FB_WARN;
888 RETVAL = ENCODE_FB_PERLQQ;
895 RETVAL = ENCODE_FB_HTMLCREF;
902 RETVAL = ENCODE_FB_XMLCREF;