2 $Id: Encode.xs,v 2.3 2004/12/03 19:16:53 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 (check & ENCODE_DIE_ON_ERR) {
155 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
156 (UV)ch, enc->name[0]);
157 return &PL_sv_undef; /* never reaches but be safe */
159 if (check & ENCODE_WARN_ON_ERR){
160 Perl_warner(aTHX_ packWARN(WARN_UTF8),
161 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
163 if (check & ENCODE_RETURN_ON_ERR){
166 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
168 newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
169 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
170 "&#x%" UVxf ";", (UV)ch);
171 sdone += slen + clen;
172 ddone += dlen + SvCUR(subchar);
173 sv_catsv(dst, subchar);
174 SvREFCNT_dec(subchar);
177 sdone += slen + clen;
178 ddone += dlen + enc->replen;
179 sv_catpvn(dst, (char*)enc->rep, enc->replen);
184 if (check & ENCODE_DIE_ON_ERR){
185 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
186 enc->name[0], (UV)s[slen]);
187 return &PL_sv_undef; /* never reaches but be safe */
189 if (check & ENCODE_WARN_ON_ERR){
191 aTHX_ packWARN(WARN_UTF8),
193 enc->name[0], (UV)s[slen]);
195 if (check & ENCODE_RETURN_ON_ERR){
199 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
200 SV* subchar = newSVpvf("\\x%02" UVXf, (UV)s[slen]);
202 ddone += dlen + SvCUR(subchar);
203 sv_catsv(dst, subchar);
204 SvREFCNT_dec(subchar);
207 ddone += dlen + strlen(FBCHAR_UTF8);
208 sv_catpv(dst, FBCHAR_UTF8);
211 /* settle variables when fallback */
212 d = (U8 *)SvEND(dst);
213 dlen = SvLEN(dst) - ddone - 1;
214 s = (U8*)SvPVX(src) + sdone;
219 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
220 code, (dir == enc->f_utf8) ? "to" : "from",
226 if (check && !(check & ENCODE_LEAVE_SRC)){
227 sdone = SvCUR(src) - (slen+sdone);
229 sv_setpvn(src, (char*)s+slen, sdone);
231 SvCUR_set(src, sdone);
233 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
235 SvCUR_set(dst, dlen+ddone);
238 #if ENCODE_XS_PROFILE
239 if (SvCUR(dst) > SvCUR(src)){
241 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
242 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
243 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
248 *offset += sdone + slen;
252 if (retcode) *retcode = code;
257 strict_utf8(pTHX_ SV* sv)
262 if (!sv || SvTYPE(sv) != SVt_PVHV)
265 svp = hv_fetch(hv, "strict_utf8", 11, 0);
272 process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check,
273 bool encode, bool strict, bool stop_at_partial)
282 if (UTF8_IS_INVARIANT(*s)) {
283 sv_catpvn(dst, (char *)s, 1);
288 if (UTF8_IS_START(*s)) {
289 U8 skip = UTF8SKIP(s);
290 if ((s + skip) > e) {
291 /* Partial character */
292 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
299 uv = utf8n_to_uvuni(s, e - s, &ulen,
300 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
301 UTF8_ALLOW_NONSTRICT)
303 #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
304 if (strict && uv > PERL_UNICODE_MAX)
309 uv = utf8n_to_uvuni(s, e - s, &ulen,
310 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
319 /* Whole char is good */
320 sv_catpvn(dst,(char *)s,skip);
325 /* If we get here there is something wrong with alleged UTF-8 */
331 if (check & ENCODE_DIE_ON_ERR){
333 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
335 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
337 if (check & ENCODE_WARN_ON_ERR){
339 Perl_warner(aTHX_ packWARN(WARN_UTF8),
340 ERR_ENCODE_NOMAP, uv, "utf8");
342 Perl_warner(aTHX_ packWARN(WARN_UTF8),
343 ERR_DECODE_NOMAP, "utf8", uv);
345 if (check & ENCODE_RETURN_ON_ERR) {
348 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
349 SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"):
350 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
351 "&#x%" UVxf ";", uv);
352 sv_catsv(dst, subchar);
353 SvREFCNT_dec(subchar);
355 sv_catpv(dst, FBCHAR_UTF8);
365 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
370 Method_decode_xs(obj,src,check = 0)
377 U8 *s = (U8 *) SvPV(src, slen);
378 U8 *e = (U8 *) SvEND(src);
379 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
382 * PerlIO check -- we assume the object is of PerlIO if renewed
385 dSP; ENTER; SAVETMPS;
389 if (call_method("renewed",G_SCALAR) == 1) {
394 fprintf(stderr, "renewed == %d\n", renewed);
398 /* end PerlIO check */
401 s = utf8_to_bytes(s,&slen);
408 croak("Cannot decode string with wide characters");
412 s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed);
414 /* Clear out translated part of source unless asked not to */
415 if (check && !(check & ENCODE_LEAVE_SRC)){
418 sv_setpvn(src, (char*)s, slen);
420 SvCUR_set(src, slen);
423 ST(0) = sv_2mortal(dst);
428 Method_encode_xs(obj,src,check = 0)
435 U8 *s = (U8 *) SvPV(src, slen);
436 U8 *e = (U8 *) SvEND(src);
437 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
439 /* Already encoded */
440 if (strict_utf8(aTHX_ obj)) {
441 s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0);
444 /* trust it and just copy the octets */
445 sv_setpvn(dst,(char *)s,(e-s));
450 /* Native bytes - can always encode */
451 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
453 UV uv = NATIVE_TO_UNI((UV) *s++);
454 if (UNI_IS_INVARIANT(uv))
455 *d++ = (U8)UTF_TO_NATIVE(uv);
457 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
458 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
461 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
465 /* Clear out translated part of source unless asked not to */
466 if (check && !(check & ENCODE_LEAVE_SRC)){
469 sv_setpvn(src, (char*)s, slen);
471 SvCUR_set(src, slen);
475 ST(0) = sv_2mortal(dst);
479 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
504 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
505 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
510 Method_cat_decode(obj, dst, src, off, term, check = 0)
519 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
520 STRLEN offset = (STRLEN)SvIV(off);
523 sv_utf8_downgrade(src, FALSE);
525 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
526 &offset, term, &code));
527 SvIVX(off) = (IV)offset;
528 if (code == ENCODE_FOUND_TERM) {
537 Method_decode(obj,src,check = 0)
543 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
545 sv_utf8_downgrade(src, FALSE);
547 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
554 Method_encode(obj,src,check = 0)
560 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
561 sv_utf8_upgrade(src);
562 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
568 Method_needs_lines(obj)
572 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
578 Method_perlio_ok(obj)
582 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
583 /* require_pv(PERLIO_FILENAME); */
585 eval_pv("require PerlIO::encoding", 0);
587 if (SvTRUE(get_sv("@", 0))) {
595 MODULE = Encode PACKAGE = Encode
600 _bytes_to_utf8(sv, ...)
604 SV * encoding = items == 2 ? ST(1) : Nullsv;
607 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
610 U8* s = (U8*)SvPV(sv, len);
613 converted = bytes_to_utf8(s, &len); /* This allocs */
614 sv_setpvn(sv, (char *)converted, len);
615 SvUTF8_on(sv); /* XXX Should we? */
616 Safefree(converted); /* ... so free it */
624 _utf8_to_bytes(sv, ...)
628 SV * to = items > 1 ? ST(1) : Nullsv;
629 SV * check = items > 2 ? ST(2) : Nullsv;
632 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
635 U8 *s = (U8*)SvPV(sv, len);
639 /* Must do things the slow way */
641 /* We need a copy to pass to check() */
642 U8 *src = (U8*)savepv((char *)s);
645 New(83, dest, len, U8); /* I think */
654 /* Have to do it all ourselves because of error routine,
656 if (!(uv & 0x40)){ goto failure; }
657 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
658 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
659 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
660 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
661 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
662 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
663 else { ulen = 13; uv = 0; }
665 /* Note change to utf8.c variable naming, for variety */
667 if ((*s & 0xc0) != 0x80){
670 uv = (uv << 6) | (*s++ & 0x3f);
675 call_failure(check, s, dest, src);
676 /* Now what happens? */
682 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
690 is_utf8(sv, check = 0)
695 if (SvGMAGICAL(sv)) /* it could be $1, for example */
696 sv = newSVsv(sv); /* GMAGIG will be done */
698 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
701 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
707 SvREFCNT_dec(sv); /* it was a temp copy */
718 SV *rsv = newSViv(SvUTF8(sv));
722 RETVAL = &PL_sv_undef;
734 SV *rsv = newSViv(SvUTF8(sv));
738 RETVAL = &PL_sv_undef;
747 RETVAL = ENCODE_DIE_ON_ERR;
754 RETVAL = ENCODE_WARN_ON_ERR;
761 RETVAL = ENCODE_LEAVE_SRC;
768 RETVAL = ENCODE_RETURN_ON_ERR;
775 RETVAL = ENCODE_PERLQQ;
782 RETVAL = ENCODE_HTMLCREF;
789 RETVAL = ENCODE_XMLCREF;
796 RETVAL = ENCODE_FB_DEFAULT;
803 RETVAL = ENCODE_FB_CROAK;
810 RETVAL = ENCODE_FB_QUIET;
817 RETVAL = ENCODE_FB_WARN;
824 RETVAL = ENCODE_FB_PERLQQ;
831 RETVAL = ENCODE_FB_HTMLCREF;
838 RETVAL = ENCODE_FB_XMLCREF;