2 $Id: Encode.xs,v 1.41 2002/04/27 18:59:50 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 */
58 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
62 U8 *s = (U8 *) SvPV(src, slen);
67 /* We allocate slen+1.
68 PerlIO dumps core if this value is smaller than this. */
69 SV *dst = sv_2mortal(newSV(slen+1));
70 U8 *d = (U8 *)SvPVX(dst);
71 STRLEN dlen = SvLEN(dst)-1;
80 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
82 SvCUR_set(dst, dlen+ddone);
85 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
91 STRLEN more = 0; /* make sure you initialize! */
96 #if ENCODE_XS_PROFILE >= 2
98 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
99 more, sdone, sleft, SvLEN(dst));
101 if (sdone != 0) { /* has src ever been processed ? */
102 #if ENCODE_XS_USEFP == 2
103 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
105 #elif ENCODE_XS_USEFP
106 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
108 /* safe until SvLEN(dst) == MAX_INT/16 */
109 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
112 more += UTF8_MAXLEN; /* insurance policy */
113 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
114 /* dst need to grow need MORE bytes! */
115 if (ddone >= SvLEN(dst)) {
116 Perl_croak(aTHX_ "Destination couldn't be grown.");
118 dlen = SvLEN(dst)-ddone-1;
126 if (dir == enc->f_utf8) {
129 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
130 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
131 if (check & ENCODE_DIE_ON_ERR) {
133 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
134 ch, enc->name[0], __LINE__);
136 if (check & ENCODE_RETURN_ON_ERR){
137 if (check & ENCODE_WARN_ON_ERR){
139 aTHX_ packWARN(WARN_UTF8),
140 "\"\\N{U+%" UVxf "}\" does not map to %s",
144 }else if (check & ENCODE_PERLQQ){
146 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", ch));
147 sdone += slen + clen;
148 ddone += dlen + SvCUR(perlqq);
149 sv_catsv(dst, perlqq);
150 }else if (check & ENCODE_HTMLCREF){
152 sv_2mortal(newSVpvf("&#%" UVuf ";", ch));
153 sdone += slen + clen;
154 ddone += dlen + SvCUR(htmlcref);
155 sv_catsv(dst, htmlcref);
156 }else if (check & ENCODE_XMLCREF){
158 sv_2mortal(newSVpvf("&#x%" UVxf ";", ch));
159 sdone += slen + clen;
160 ddone += dlen + SvCUR(xmlcref);
161 sv_catsv(dst, xmlcref);
164 sdone += slen + clen;
165 ddone += dlen + enc->replen;
166 sv_catpvn(dst, (char*)enc->rep, enc->replen);
172 if (check & ENCODE_DIE_ON_ERR){
174 aTHX_ "%s \"\\x%02" UVXf
175 "\" does not map to Unicode (%d)",
176 enc->name[0], (U8) s[slen], code);
178 if (check & ENCODE_RETURN_ON_ERR){
179 if (check & ENCODE_WARN_ON_ERR){
181 aTHX_ packWARN(WARN_UTF8),
183 "\" does not map to Unicode (%d)",
184 enc->name[0], (U8) s[slen], code);
188 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
190 sv_2mortal(newSVpvf("\\x%02" UVXf, s[slen]));
192 ddone += dlen + SvCUR(perlqq);
193 sv_catsv(dst, perlqq);
196 ddone += dlen + strlen(FBCHAR_UTF8);
197 sv_catpv(dst, FBCHAR_UTF8);
201 /* settle variables when fallback */
202 d = (U8 *)SvEND(dst);
203 dlen = SvLEN(dst) - ddone - 1;
204 s = (U8*)SvPVX(src) + sdone;
209 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
210 code, (dir == enc->f_utf8) ? "to" : "from",
216 if (check && !(check & ENCODE_LEAVE_SRC)){
217 sdone = SvCUR(src) - (slen+sdone);
219 sv_setpvn(src, (char*)s+slen, sdone);
221 SvCUR_set(src, sdone);
223 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
225 SvCUR_set(dst, dlen+ddone);
228 #if ENCODE_XS_PROFILE
229 if (SvCUR(dst) > SvCUR(src)){
231 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
232 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
233 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
242 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
251 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
252 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
257 Method_decode(obj,src,check = 0)
263 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
264 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
270 Method_encode(obj,src,check = 0)
276 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
277 sv_utf8_upgrade(src);
278 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
283 Method_needs_lines(obj)
287 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
293 Method_perlio_ok(obj)
297 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
298 /* require_pv(PERLIO_FILENAME); */
300 eval_pv("require PerlIO::encoding", 0);
302 if (SvTRUE(get_sv("@", 0))) {
310 MODULE = Encode PACKAGE = Encode
315 _bytes_to_utf8(sv, ...)
319 SV * encoding = items == 2 ? ST(1) : Nullsv;
322 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
325 U8* s = (U8*)SvPV(sv, len);
328 converted = bytes_to_utf8(s, &len); /* This allocs */
329 sv_setpvn(sv, (char *)converted, len);
330 SvUTF8_on(sv); /* XXX Should we? */
331 Safefree(converted); /* ... so free it */
339 _utf8_to_bytes(sv, ...)
343 SV * to = items > 1 ? ST(1) : Nullsv;
344 SV * check = items > 2 ? ST(2) : Nullsv;
347 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
350 U8 *s = (U8*)SvPV(sv, len);
354 /* Must do things the slow way */
356 /* We need a copy to pass to check() */
357 U8 *src = (U8*)savepv((char *)s);
360 New(83, dest, len, U8); /* I think */
369 /* Have to do it all ourselves because of error routine,
371 if (!(uv & 0x40)){ goto failure; }
372 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
373 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
374 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
375 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
376 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
377 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
378 else { ulen = 13; uv = 0; }
380 /* Note change to utf8.c variable naming, for variety */
382 if ((*s & 0xc0) != 0x80){
385 uv = (uv << 6) | (*s++ & 0x3f);
390 call_failure(check, s, dest, src);
391 /* Now what happens? */
397 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
405 is_utf8(sv, check = 0)
410 if (SvGMAGICAL(sv)) /* it could be $1, for example */
411 sv = newSVsv(sv); /* GMAGIG will be done */
413 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
416 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
422 SvREFCNT_dec(sv); /* it was a temp copy */
433 SV *rsv = newSViv(SvUTF8(sv));
437 RETVAL = &PL_sv_undef;
449 SV *rsv = newSViv(SvUTF8(sv));
453 RETVAL = &PL_sv_undef;
462 RETVAL = ENCODE_DIE_ON_ERR;
469 RETVAL = ENCODE_WARN_ON_ERR;
476 RETVAL = ENCODE_LEAVE_SRC;
483 RETVAL = ENCODE_RETURN_ON_ERR;
490 RETVAL = ENCODE_PERLQQ;
497 RETVAL = ENCODE_HTMLCREF;
504 RETVAL = ENCODE_XMLCREF;
511 RETVAL = ENCODE_FB_DEFAULT;
518 RETVAL = ENCODE_FB_CROAK;
525 RETVAL = ENCODE_FB_QUIET;
532 RETVAL = ENCODE_FB_WARN;
539 RETVAL = ENCODE_FB_PERLQQ;
546 RETVAL = ENCODE_FB_HTMLCREF;
553 RETVAL = ENCODE_FB_XMLCREF;