2 $Id: Encode.xs,v 1.55 2003/02/28 01:40:27 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)
33 Encode_XSEncoding(pTHX_ encode_t * enc)
36 HV *stash = gv_stashpv("Encode::XS", TRUE);
37 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
41 while (enc->name[i]) {
42 const char *name = enc->name[i++];
43 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
46 call_pv("Encode::define_encoding", G_DISCARD);
51 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
53 /* Exists for breakpointing */
57 #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
58 #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
61 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
62 int check, STRLEN * offset, SV * term, int * retcode)
65 U8 *s = (U8 *) SvPV(src, slen);
70 /* We allocate slen+1.
71 PerlIO dumps core if this value is smaller than this. */
72 SV *dst = sv_2mortal(newSV(slen+1));
73 U8 *d = (U8 *)SvPVX(dst);
74 STRLEN dlen = SvLEN(dst)-1;
77 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
81 if (slen > *offset){ /* safeguard against slen overflow */
95 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
98 SvCUR_set(dst, dlen+ddone);
101 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
102 code == ENCODE_FOUND_TERM) {
108 STRLEN more = 0; /* make sure you initialize! */
112 sleft = tlen - sdone;
113 #if ENCODE_XS_PROFILE >= 2
115 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
116 more, sdone, sleft, SvLEN(dst));
118 if (sdone != 0) { /* has src ever been processed ? */
119 #if ENCODE_XS_USEFP == 2
120 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
122 #elif ENCODE_XS_USEFP
123 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
125 /* safe until SvLEN(dst) == MAX_INT/16 */
126 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
129 more += UTF8_MAXLEN; /* insurance policy */
130 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
131 /* dst need to grow need MORE bytes! */
132 if (ddone >= SvLEN(dst)) {
133 Perl_croak(aTHX_ "Destination couldn't be grown.");
135 dlen = SvLEN(dst)-ddone-1;
143 if (dir == enc->f_utf8) {
146 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
147 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
148 if (check & ENCODE_DIE_ON_ERR) {
149 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
150 (UV)ch, enc->name[0]);
151 return &PL_sv_undef; /* never reaches but be safe */
153 if (check & ENCODE_WARN_ON_ERR){
154 Perl_warner(aTHX_ packWARN(WARN_UTF8),
155 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
157 if (check & ENCODE_RETURN_ON_ERR){
160 if (check & ENCODE_PERLQQ){
162 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch));
163 sdone += slen + clen;
164 ddone += dlen + SvCUR(perlqq);
165 sv_catsv(dst, perlqq);
166 }else if (check & ENCODE_HTMLCREF){
168 sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch));
169 sdone += slen + clen;
170 ddone += dlen + SvCUR(htmlcref);
171 sv_catsv(dst, htmlcref);
172 }else if (check & ENCODE_XMLCREF){
174 sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch));
175 sdone += slen + clen;
176 ddone += dlen + SvCUR(xmlcref);
177 sv_catsv(dst, xmlcref);
180 sdone += slen + clen;
181 ddone += dlen + enc->replen;
182 sv_catpvn(dst, (char*)enc->rep, enc->replen);
187 if (check & ENCODE_DIE_ON_ERR){
188 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
189 enc->name[0], (UV)s[slen]);
190 return &PL_sv_undef; /* never reaches but be safe */
192 if (check & ENCODE_WARN_ON_ERR){
194 aTHX_ packWARN(WARN_UTF8),
196 enc->name[0], (UV)s[slen]);
198 if (check & ENCODE_RETURN_ON_ERR){
202 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
204 sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen]));
206 ddone += dlen + SvCUR(perlqq);
207 sv_catsv(dst, perlqq);
210 ddone += dlen + strlen(FBCHAR_UTF8);
211 sv_catpv(dst, FBCHAR_UTF8);
214 /* settle variables when fallback */
215 d = (U8 *)SvEND(dst);
216 dlen = SvLEN(dst) - ddone - 1;
217 s = (U8*)SvPVX(src) + sdone;
222 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
223 code, (dir == enc->f_utf8) ? "to" : "from",
229 if (check && !(check & ENCODE_LEAVE_SRC)){
230 sdone = SvCUR(src) - (slen+sdone);
232 sv_setpvn(src, (char*)s+slen, sdone);
234 SvCUR_set(src, sdone);
236 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
238 SvCUR_set(dst, dlen+ddone);
241 #if ENCODE_XS_PROFILE
242 if (SvCUR(dst) > SvCUR(src)){
244 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
245 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
246 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
251 *offset += sdone + slen;
255 if (retcode) *retcode = code;
259 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
262 Method_decode_xs(obj,src,check = 0)
269 U8 *s = (U8 *) SvPV(src, slen);
270 U8 *e = (U8 *) SvEND(src);
271 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
275 s = utf8_to_bytes(s,&slen);
282 croak("Cannot decode string with wide characters");
286 if (UTF8_IS_INVARIANT(*s) || UTF8_IS_START(*s)) {
287 U8 skip = UTF8SKIP(s);
288 if ((s + skip) > e) {
289 /* Partial character - done */
292 else if (is_utf8_char(s)) {
293 /* Whole char is good */
294 sv_catpvn(dst,(char *)s,skip);
299 /* starts ok but isn't "good" */
303 /* Invalid start byte */
305 /* If we get here there is something wrong with alleged UTF-8 */
306 if (check & ENCODE_DIE_ON_ERR){
307 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", (UV)*s);
310 if (check & ENCODE_WARN_ON_ERR){
311 Perl_warner(aTHX_ packWARN(WARN_UTF8),
312 ERR_DECODE_NOMAP, "utf8", (UV)*s);
314 if (check & ENCODE_RETURN_ON_ERR) {
317 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
318 SV* perlqq = newSVpvf("\\x%02" UVXf, (UV)*s);
319 sv_catsv(dst, perlqq);
320 SvREFCNT_dec(perlqq);
322 sv_catpv(dst, FBCHAR_UTF8);
328 /* Clear out translated part of source unless asked not to */
329 if (check && !(check & ENCODE_LEAVE_SRC)){
332 sv_setpvn(src, (char*)s, slen);
334 SvCUR_set(src, slen);
337 ST(0) = sv_2mortal(dst);
342 Method_encode_xs(obj,src,check = 0)
349 U8 *s = (U8 *) SvPV(src, slen);
350 U8 *e = (U8 *) SvEND(src);
351 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
353 /* Already encoded - trust it and just copy the octets */
354 sv_setpvn(dst,(char *)s,(e-s));
358 /* Native bytes - can always encode */
359 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
361 UV uv = NATIVE_TO_UNI((UV) *s++);
362 if (UNI_IS_INVARIANT(uv))
363 *d++ = (U8)UTF_TO_NATIVE(uv);
365 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
366 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
369 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
373 /* Clear out translated part of source unless asked not to */
374 if (check && !(check & ENCODE_LEAVE_SRC)){
377 sv_setpvn(src, (char*)s, slen);
379 SvCUR_set(src, slen);
383 ST(0) = sv_2mortal(dst);
387 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
396 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
397 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
402 Method_cat_decode(obj, dst, src, off, term, check = 0)
411 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
412 STRLEN offset = (STRLEN)SvIV(off);
415 sv_utf8_downgrade(src, FALSE);
417 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
418 &offset, term, &code));
419 SvIVX(off) = (IV)offset;
420 if (code == ENCODE_FOUND_TERM) {
429 Method_decode(obj,src,check = 0)
435 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
437 sv_utf8_downgrade(src, FALSE);
439 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
446 Method_encode(obj,src,check = 0)
452 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
453 sv_utf8_upgrade(src);
454 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
460 Method_needs_lines(obj)
464 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
470 Method_perlio_ok(obj)
474 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
475 /* require_pv(PERLIO_FILENAME); */
477 eval_pv("require PerlIO::encoding", 0);
479 if (SvTRUE(get_sv("@", 0))) {
487 MODULE = Encode PACKAGE = Encode
492 _bytes_to_utf8(sv, ...)
496 SV * encoding = items == 2 ? ST(1) : Nullsv;
499 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
502 U8* s = (U8*)SvPV(sv, len);
505 converted = bytes_to_utf8(s, &len); /* This allocs */
506 sv_setpvn(sv, (char *)converted, len);
507 SvUTF8_on(sv); /* XXX Should we? */
508 Safefree(converted); /* ... so free it */
516 _utf8_to_bytes(sv, ...)
520 SV * to = items > 1 ? ST(1) : Nullsv;
521 SV * check = items > 2 ? ST(2) : Nullsv;
524 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
527 U8 *s = (U8*)SvPV(sv, len);
531 /* Must do things the slow way */
533 /* We need a copy to pass to check() */
534 U8 *src = (U8*)savepv((char *)s);
537 New(83, dest, len, U8); /* I think */
546 /* Have to do it all ourselves because of error routine,
548 if (!(uv & 0x40)){ goto failure; }
549 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
550 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
551 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
552 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
553 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
554 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
555 else { ulen = 13; uv = 0; }
557 /* Note change to utf8.c variable naming, for variety */
559 if ((*s & 0xc0) != 0x80){
562 uv = (uv << 6) | (*s++ & 0x3f);
567 call_failure(check, s, dest, src);
568 /* Now what happens? */
574 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
582 is_utf8(sv, check = 0)
587 if (SvGMAGICAL(sv)) /* it could be $1, for example */
588 sv = newSVsv(sv); /* GMAGIG will be done */
590 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
593 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
599 SvREFCNT_dec(sv); /* it was a temp copy */
610 SV *rsv = newSViv(SvUTF8(sv));
614 RETVAL = &PL_sv_undef;
626 SV *rsv = newSViv(SvUTF8(sv));
630 RETVAL = &PL_sv_undef;
639 RETVAL = ENCODE_DIE_ON_ERR;
646 RETVAL = ENCODE_WARN_ON_ERR;
653 RETVAL = ENCODE_LEAVE_SRC;
660 RETVAL = ENCODE_RETURN_ON_ERR;
667 RETVAL = ENCODE_PERLQQ;
674 RETVAL = ENCODE_HTMLCREF;
681 RETVAL = ENCODE_XMLCREF;
688 RETVAL = ENCODE_FB_DEFAULT;
695 RETVAL = ENCODE_FB_CROAK;
702 RETVAL = ENCODE_FB_QUIET;
709 RETVAL = ENCODE_FB_WARN;
716 RETVAL = ENCODE_FB_PERLQQ;
723 RETVAL = ENCODE_FB_HTMLCREF;
730 RETVAL = ENCODE_FB_XMLCREF;