2 $Id: Encode.xs,v 2.5 2005/08/05 10:58:25 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 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
68 int check, STRLEN * offset, SV * term, int * retcode)
71 U8 *s = (U8 *) SvPV(src, slen);
76 /* We allocate slen+1.
77 PerlIO dumps core if this value is smaller than this. */
78 SV *dst = sv_2mortal(newSV(slen+1));
79 U8 *d = (U8 *)SvPVX(dst);
80 STRLEN dlen = SvLEN(dst)-1;
83 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
87 if (slen > *offset){ /* safeguard against slen overflow */
101 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
104 SvCUR_set(dst, dlen+ddone);
107 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
108 code == ENCODE_FOUND_TERM) {
114 STRLEN more = 0; /* make sure you initialize! */
118 sleft = tlen - sdone;
119 #if ENCODE_XS_PROFILE >= 2
121 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
122 more, sdone, sleft, SvLEN(dst));
124 if (sdone != 0) { /* has src ever been processed ? */
125 #if ENCODE_XS_USEFP == 2
126 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
128 #elif ENCODE_XS_USEFP
129 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
131 /* safe until SvLEN(dst) == MAX_INT/16 */
132 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
135 more += UTF8_MAXLEN; /* insurance policy */
136 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
137 /* dst need to grow need MORE bytes! */
138 if (ddone >= SvLEN(dst)) {
139 Perl_croak(aTHX_ "Destination couldn't be grown.");
141 dlen = SvLEN(dst)-ddone-1;
149 if (dir == enc->f_utf8) {
152 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
153 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
154 /* if non-representable multibyte prefix at end of current buffer - break*/
155 if (clen > tlen - sdone) break;
156 if (check & ENCODE_DIE_ON_ERR) {
157 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
158 (UV)ch, enc->name[0]);
159 return &PL_sv_undef; /* never reaches but be safe */
161 if (check & ENCODE_WARN_ON_ERR){
162 Perl_warner(aTHX_ packWARN(WARN_UTF8),
163 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
165 if (check & ENCODE_RETURN_ON_ERR){
168 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
170 newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
171 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
172 "&#x%" UVxf ";", (UV)ch);
173 sdone += slen + clen;
174 ddone += dlen + SvCUR(subchar);
175 sv_catsv(dst, subchar);
176 SvREFCNT_dec(subchar);
179 sdone += slen + clen;
180 ddone += dlen + enc->replen;
181 sv_catpvn(dst, (char*)enc->rep, enc->replen);
186 if (check & ENCODE_DIE_ON_ERR){
187 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
188 enc->name[0], (UV)s[slen]);
189 return &PL_sv_undef; /* never reaches but be safe */
191 if (check & ENCODE_WARN_ON_ERR){
193 aTHX_ packWARN(WARN_UTF8),
195 enc->name[0], (UV)s[slen]);
197 if (check & ENCODE_RETURN_ON_ERR){
201 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
202 SV* subchar = newSVpvf("\\x%02" UVXf, (UV)s[slen]);
204 ddone += dlen + SvCUR(subchar);
205 sv_catsv(dst, subchar);
206 SvREFCNT_dec(subchar);
209 ddone += dlen + strlen(FBCHAR_UTF8);
210 sv_catpv(dst, FBCHAR_UTF8);
213 /* settle variables when fallback */
214 d = (U8 *)SvEND(dst);
215 dlen = SvLEN(dst) - ddone - 1;
216 s = (U8*)SvPVX(src) + sdone;
221 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
222 code, (dir == enc->f_utf8) ? "to" : "from",
228 if (check && !(check & ENCODE_LEAVE_SRC)){
229 sdone = SvCUR(src) - (slen+sdone);
231 sv_setpvn(src, (char*)s+slen, sdone);
233 SvCUR_set(src, sdone);
235 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
237 SvCUR_set(dst, dlen+ddone);
240 #if ENCODE_XS_PROFILE
241 if (SvCUR(dst) > SvCUR(src)){
243 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
244 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
245 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
250 *offset += sdone + slen;
254 if (retcode) *retcode = code;
259 strict_utf8(pTHX_ SV* sv)
264 if (!sv || SvTYPE(sv) != SVt_PVHV)
267 svp = hv_fetch(hv, "strict_utf8", 11, 0);
274 process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check,
275 bool encode, bool strict, bool stop_at_partial)
284 if (UTF8_IS_INVARIANT(*s)) {
285 sv_catpvn(dst, (char *)s, 1);
290 if (UTF8_IS_START(*s)) {
291 U8 skip = UTF8SKIP(s);
292 if ((s + skip) > e) {
293 /* Partial character */
294 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
295 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
301 uv = utf8n_to_uvuni(s, e - s, &ulen,
302 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
303 UTF8_ALLOW_NONSTRICT)
305 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
306 if (strict && uv > PERL_UNICODE_MAX)
311 uv = utf8n_to_uvuni(s, e - s, &ulen,
312 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
321 /* Whole char is good */
322 sv_catpvn(dst,(char *)s,skip);
327 /* If we get here there is something wrong with alleged UTF-8 */
333 if (check & ENCODE_DIE_ON_ERR){
335 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
337 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
339 if (check & ENCODE_WARN_ON_ERR){
341 Perl_warner(aTHX_ packWARN(WARN_UTF8),
342 ERR_ENCODE_NOMAP, uv, "utf8");
344 Perl_warner(aTHX_ packWARN(WARN_UTF8),
345 ERR_DECODE_NOMAP, "utf8", uv);
347 if (check & ENCODE_RETURN_ON_ERR) {
350 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
351 SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"):
352 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
353 "&#x%" UVxf ";", uv);
354 sv_catsv(dst, subchar);
355 SvREFCNT_dec(subchar);
357 sv_catpv(dst, FBCHAR_UTF8);
367 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
372 Method_decode_xs(obj,src,check = 0)
379 U8 *s = (U8 *) SvPV(src, slen);
380 U8 *e = (U8 *) SvEND(src);
381 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
384 * PerlIO check -- we assume the object is of PerlIO if renewed
387 dSP; ENTER; SAVETMPS;
391 if (call_method("renewed",G_SCALAR) == 1) {
396 fprintf(stderr, "renewed == %d\n", renewed);
400 /* end PerlIO check */
403 s = utf8_to_bytes(s,&slen);
410 croak("Cannot decode string with wide characters");
414 s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed);
416 /* Clear out translated part of source unless asked not to */
417 if (check && !(check & ENCODE_LEAVE_SRC)){
420 sv_setpvn(src, (char*)s, slen);
422 SvCUR_set(src, slen);
425 ST(0) = sv_2mortal(dst);
430 Method_encode_xs(obj,src,check = 0)
437 U8 *s = (U8 *) SvPV(src, slen);
438 U8 *e = (U8 *) SvEND(src);
439 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
441 /* Already encoded */
442 if (strict_utf8(aTHX_ obj)) {
443 s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0);
446 /* trust it and just copy the octets */
447 sv_setpvn(dst,(char *)s,(e-s));
452 /* Native bytes - can always encode */
453 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
455 UV uv = NATIVE_TO_UNI((UV) *s++);
456 if (UNI_IS_INVARIANT(uv))
457 *d++ = (U8)UTF_TO_NATIVE(uv);
459 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
460 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
463 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
467 /* Clear out translated part of source unless asked not to */
468 if (check && !(check & ENCODE_LEAVE_SRC)){
471 sv_setpvn(src, (char*)s, slen);
473 SvCUR_set(src, slen);
477 ST(0) = sv_2mortal(dst);
481 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
506 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
507 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
512 Method_cat_decode(obj, dst, src, off, term, check = 0)
521 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
522 STRLEN offset = (STRLEN)SvIV(off);
525 sv_utf8_downgrade(src, FALSE);
527 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
528 &offset, term, &code));
529 SvIV_set(off, (IV)offset);
530 if (code == ENCODE_FOUND_TERM) {
539 Method_decode(obj,src,check = 0)
545 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
547 sv_utf8_downgrade(src, FALSE);
549 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
556 Method_encode(obj,src,check = 0)
562 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
563 sv_utf8_upgrade(src);
564 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
570 Method_needs_lines(obj)
574 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
580 Method_perlio_ok(obj)
584 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
585 /* require_pv(PERLIO_FILENAME); */
587 eval_pv("require PerlIO::encoding", 0);
589 if (SvTRUE(get_sv("@", 0))) {
597 MODULE = Encode PACKAGE = Encode
602 _bytes_to_utf8(sv, ...)
606 SV * encoding = items == 2 ? ST(1) : Nullsv;
609 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
612 U8* s = (U8*)SvPV(sv, len);
615 converted = bytes_to_utf8(s, &len); /* This allocs */
616 sv_setpvn(sv, (char *)converted, len);
617 SvUTF8_on(sv); /* XXX Should we? */
618 Safefree(converted); /* ... so free it */
626 _utf8_to_bytes(sv, ...)
630 SV * to = items > 1 ? ST(1) : Nullsv;
631 SV * check = items > 2 ? ST(2) : Nullsv;
634 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
637 U8 *s = (U8*)SvPV(sv, len);
641 /* Must do things the slow way */
643 /* We need a copy to pass to check() */
644 U8 *src = (U8*)savepv((char *)s);
647 New(83, dest, len, U8); /* I think */
656 /* Have to do it all ourselves because of error routine,
658 if (!(uv & 0x40)){ goto failure; }
659 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
660 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
661 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
662 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
663 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
664 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
665 else { ulen = 13; uv = 0; }
667 /* Note change to utf8.c variable naming, for variety */
669 if ((*s & 0xc0) != 0x80){
672 uv = (uv << 6) | (*s++ & 0x3f);
677 call_failure(check, s, dest, src);
678 /* Now what happens? */
684 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
692 is_utf8(sv, check = 0)
697 if (SvGMAGICAL(sv)) /* it could be $1, for example */
698 sv = newSVsv(sv); /* GMAGIG will be done */
700 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
703 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
709 SvREFCNT_dec(sv); /* it was a temp copy */
720 SV *rsv = newSViv(SvUTF8(sv));
724 RETVAL = &PL_sv_undef;
736 SV *rsv = newSViv(SvUTF8(sv));
740 RETVAL = &PL_sv_undef;
749 RETVAL = ENCODE_DIE_ON_ERR;
756 RETVAL = ENCODE_WARN_ON_ERR;
763 RETVAL = ENCODE_LEAVE_SRC;
770 RETVAL = ENCODE_RETURN_ON_ERR;
777 RETVAL = ENCODE_PERLQQ;
784 RETVAL = ENCODE_HTMLCREF;
791 RETVAL = ENCODE_XMLCREF;
798 RETVAL = ENCODE_STOP_AT_PARTIAL;
805 RETVAL = ENCODE_FB_DEFAULT;
812 RETVAL = ENCODE_FB_CROAK;
819 RETVAL = ENCODE_FB_QUIET;
826 RETVAL = ENCODE_FB_WARN;
833 RETVAL = ENCODE_FB_PERLQQ;
840 RETVAL = ENCODE_FB_HTMLCREF;
847 RETVAL = ENCODE_FB_XMLCREF;