2 $Id: Encode.xs,v 1.52 2002/11/18 17:28:49 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,
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;
83 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check)) )
85 SvCUR_set(dst, dlen+ddone);
88 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
94 STRLEN more = 0; /* make sure you initialize! */
99 #if ENCODE_XS_PROFILE >= 2
101 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
102 more, sdone, sleft, SvLEN(dst));
104 if (sdone != 0) { /* has src ever been processed ? */
105 #if ENCODE_XS_USEFP == 2
106 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
108 #elif ENCODE_XS_USEFP
109 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
111 /* safe until SvLEN(dst) == MAX_INT/16 */
112 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
115 more += UTF8_MAXLEN; /* insurance policy */
116 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
117 /* dst need to grow need MORE bytes! */
118 if (ddone >= SvLEN(dst)) {
119 Perl_croak(aTHX_ "Destination couldn't be grown.");
121 dlen = SvLEN(dst)-ddone-1;
129 if (dir == enc->f_utf8) {
132 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
133 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
134 if (check & ENCODE_DIE_ON_ERR) {
135 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
136 (UV)ch, enc->name[0]);
137 return &PL_sv_undef; /* never reaches but be safe */
139 if (check & ENCODE_WARN_ON_ERR){
140 Perl_warner(aTHX_ packWARN(WARN_UTF8),
141 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
143 if (check & ENCODE_RETURN_ON_ERR){
146 if (check & ENCODE_PERLQQ){
148 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch));
149 sdone += slen + clen;
150 ddone += dlen + SvCUR(perlqq);
151 sv_catsv(dst, perlqq);
152 }else if (check & ENCODE_HTMLCREF){
154 sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch));
155 sdone += slen + clen;
156 ddone += dlen + SvCUR(htmlcref);
157 sv_catsv(dst, htmlcref);
158 }else if (check & ENCODE_XMLCREF){
160 sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch));
161 sdone += slen + clen;
162 ddone += dlen + SvCUR(xmlcref);
163 sv_catsv(dst, xmlcref);
166 sdone += slen + clen;
167 ddone += dlen + enc->replen;
168 sv_catpvn(dst, (char*)enc->rep, enc->replen);
173 if (check & ENCODE_DIE_ON_ERR){
174 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
175 enc->name[0], (UV)s[slen]);
176 return &PL_sv_undef; /* never reaches but be safe */
178 if (check & ENCODE_WARN_ON_ERR){
180 aTHX_ packWARN(WARN_UTF8),
182 enc->name[0], (UV)s[slen]);
184 if (check & ENCODE_RETURN_ON_ERR){
188 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
190 sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen]));
192 ddone += dlen + SvCUR(perlqq);
193 sv_catsv(dst, perlqq);
196 ddone += dlen + strlen(FBCHAR_UTF8);
197 sv_catpv(dst, FBCHAR_UTF8);
200 /* settle variables when fallback */
201 d = (U8 *)SvEND(dst);
202 dlen = SvLEN(dst) - ddone - 1;
203 s = (U8*)SvPVX(src) + sdone;
208 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
209 code, (dir == enc->f_utf8) ? "to" : "from",
215 if (check && !(check & ENCODE_LEAVE_SRC)){
216 sdone = SvCUR(src) - (slen+sdone);
218 sv_setpvn(src, (char*)s+slen, sdone);
220 SvCUR_set(src, sdone);
222 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
224 SvCUR_set(dst, dlen+ddone);
227 #if ENCODE_XS_PROFILE
228 if (SvCUR(dst) > SvCUR(src)){
230 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
231 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
232 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
241 MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
244 Method_decode_xs(obj,src,check = 0)
251 U8 *s = (U8 *) SvPV(src, slen);
252 U8 *e = (U8 *) SvEND(src);
253 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
257 s = utf8_to_bytes(s,&slen);
264 croak("Cannot decode string with wide characters");
268 if (UTF8_IS_INVARIANT(*s) || UTF8_IS_START(*s)) {
269 U8 skip = UTF8SKIP(s);
270 if ((s + skip) > e) {
271 /* Partial character - done */
274 else if (is_utf8_char(s)) {
275 /* Whole char is good */
276 sv_catpvn(dst,(char *)s,skip);
281 /* starts ok but isn't "good" */
285 /* Invalid start byte */
287 /* If we get here there is something wrong with alleged UTF-8 */
288 if (check & ENCODE_DIE_ON_ERR){
289 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", (UV)*s);
292 if (check & ENCODE_WARN_ON_ERR){
293 Perl_warner(aTHX_ packWARN(WARN_UTF8),
294 ERR_DECODE_NOMAP, "utf8", (UV)*s);
296 if (check & ENCODE_RETURN_ON_ERR) {
299 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
300 SV* perlqq = newSVpvf("\\x%02" UVXf, (UV)*s);
301 sv_catsv(dst, perlqq);
302 SvREFCNT_dec(perlqq);
304 sv_catpv(dst, FBCHAR_UTF8);
310 /* Clear out translated part of source unless asked not to */
311 if (check && !(check & ENCODE_LEAVE_SRC)){
314 sv_setpvn(src, (char*)s, slen);
316 SvCUR_set(src, slen);
319 ST(0) = sv_2mortal(dst);
324 Method_encode_xs(obj,src,check = 0)
331 U8 *s = (U8 *) SvPV(src, slen);
332 U8 *e = (U8 *) SvEND(src);
333 SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
335 /* Already encoded - trust it and just copy the octets */
336 sv_setpvn(dst,(char *)s,(e-s));
340 /* Native bytes - can always encode */
341 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
343 UV uv = NATIVE_TO_UNI((UV) *s++);
344 if (UNI_IS_INVARIANT(uv))
345 *d++ = (U8)UTF_TO_NATIVE(uv);
347 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
348 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
351 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
355 /* Clear out translated part of source unless asked not to */
356 if (check && !(check & ENCODE_LEAVE_SRC)){
359 sv_setpvn(src, (char*)s, slen);
361 SvCUR_set(src, slen);
365 ST(0) = sv_2mortal(dst);
369 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
378 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
379 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
384 Method_decode(obj,src,check = 0)
390 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
392 sv_utf8_downgrade(src, FALSE);
394 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
400 Method_encode(obj,src,check = 0)
406 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
407 sv_utf8_upgrade(src);
408 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
413 Method_needs_lines(obj)
417 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
423 Method_perlio_ok(obj)
427 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
428 /* require_pv(PERLIO_FILENAME); */
430 eval_pv("require PerlIO::encoding", 0);
432 if (SvTRUE(get_sv("@", 0))) {
440 MODULE = Encode PACKAGE = Encode
445 _bytes_to_utf8(sv, ...)
449 SV * encoding = items == 2 ? ST(1) : Nullsv;
452 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
455 U8* s = (U8*)SvPV(sv, len);
458 converted = bytes_to_utf8(s, &len); /* This allocs */
459 sv_setpvn(sv, (char *)converted, len);
460 SvUTF8_on(sv); /* XXX Should we? */
461 Safefree(converted); /* ... so free it */
469 _utf8_to_bytes(sv, ...)
473 SV * to = items > 1 ? ST(1) : Nullsv;
474 SV * check = items > 2 ? ST(2) : Nullsv;
477 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
480 U8 *s = (U8*)SvPV(sv, len);
484 /* Must do things the slow way */
486 /* We need a copy to pass to check() */
487 U8 *src = (U8*)savepv((char *)s);
490 New(83, dest, len, U8); /* I think */
499 /* Have to do it all ourselves because of error routine,
501 if (!(uv & 0x40)){ goto failure; }
502 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
503 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
504 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
505 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
506 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
507 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
508 else { ulen = 13; uv = 0; }
510 /* Note change to utf8.c variable naming, for variety */
512 if ((*s & 0xc0) != 0x80){
515 uv = (uv << 6) | (*s++ & 0x3f);
520 call_failure(check, s, dest, src);
521 /* Now what happens? */
527 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
535 is_utf8(sv, check = 0)
540 if (SvGMAGICAL(sv)) /* it could be $1, for example */
541 sv = newSVsv(sv); /* GMAGIG will be done */
543 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
546 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
552 SvREFCNT_dec(sv); /* it was a temp copy */
563 SV *rsv = newSViv(SvUTF8(sv));
567 RETVAL = &PL_sv_undef;
579 SV *rsv = newSViv(SvUTF8(sv));
583 RETVAL = &PL_sv_undef;
592 RETVAL = ENCODE_DIE_ON_ERR;
599 RETVAL = ENCODE_WARN_ON_ERR;
606 RETVAL = ENCODE_LEAVE_SRC;
613 RETVAL = ENCODE_RETURN_ON_ERR;
620 RETVAL = ENCODE_PERLQQ;
627 RETVAL = ENCODE_HTMLCREF;
634 RETVAL = ENCODE_XMLCREF;
641 RETVAL = ENCODE_FB_DEFAULT;
648 RETVAL = ENCODE_FB_CROAK;
655 RETVAL = ENCODE_FB_QUIET;
662 RETVAL = ENCODE_FB_WARN;
669 RETVAL = ENCODE_FB_PERLQQ;
676 RETVAL = ENCODE_FB_HTMLCREF;
683 RETVAL = ENCODE_FB_XMLCREF;