2 $Id: Encode.xs,v 1.43 2002/05/01 05:41:06 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)
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_ "\"\\x{%04" UVxf "}\" does not map to %s",
134 (UV)ch, enc->name[0]);
135 return &PL_sv_undef; /* never reaches but be safe */
137 if (check & ENCODE_WARN_ON_ERR){
138 Perl_warner(aTHX_ packWARN(WARN_UTF8),
139 "\"\\x{%" UVxf "}\" does not map to %s",
140 (UV)ch, enc->name[0]);
142 if (check & ENCODE_RETURN_ON_ERR){
145 if (check & ENCODE_PERLQQ){
147 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch));
148 sdone += slen + clen;
149 ddone += dlen + SvCUR(perlqq);
150 sv_catsv(dst, perlqq);
151 }else if (check & ENCODE_HTMLCREF){
153 sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch));
154 sdone += slen + clen;
155 ddone += dlen + SvCUR(htmlcref);
156 sv_catsv(dst, htmlcref);
157 }else if (check & ENCODE_XMLCREF){
159 sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch));
160 sdone += slen + clen;
161 ddone += dlen + SvCUR(xmlcref);
162 sv_catsv(dst, xmlcref);
165 sdone += slen + clen;
166 ddone += dlen + enc->replen;
167 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 (UV)enc->name[0], (U8)s[slen], code);
177 return &PL_sv_undef; /* never reaches but be safe */
179 if (check & ENCODE_WARN_ON_ERR){
181 aTHX_ packWARN(WARN_UTF8),
183 "\" does not map to Unicode (%d)",
184 (UV)enc->name[0], (U8)s[slen], code);
186 if (check & ENCODE_RETURN_ON_ERR){
190 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
192 sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen]));
194 ddone += dlen + SvCUR(perlqq);
195 sv_catsv(dst, perlqq);
198 ddone += dlen + strlen(FBCHAR_UTF8);
199 sv_catpv(dst, FBCHAR_UTF8);
202 /* settle variables when fallback */
203 d = (U8 *)SvEND(dst);
204 dlen = SvLEN(dst) - ddone - 1;
205 s = (U8*)SvPVX(src) + sdone;
210 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
211 code, (dir == enc->f_utf8) ? "to" : "from",
217 if (check && !(check & ENCODE_LEAVE_SRC)){
218 sdone = SvCUR(src) - (slen+sdone);
220 sv_setpvn(src, (char*)s+slen, sdone);
222 SvCUR_set(src, sdone);
224 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
226 SvCUR_set(dst, dlen+ddone);
229 #if ENCODE_XS_PROFILE
230 if (SvCUR(dst) > SvCUR(src)){
232 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
233 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
234 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
243 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
252 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
253 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
258 Method_decode(obj,src,check = 0)
264 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
265 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
271 Method_encode(obj,src,check = 0)
277 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
278 sv_utf8_upgrade(src);
279 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
284 Method_needs_lines(obj)
288 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
294 Method_perlio_ok(obj)
298 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
299 /* require_pv(PERLIO_FILENAME); */
301 eval_pv("require PerlIO::encoding", 0);
303 if (SvTRUE(get_sv("@", 0))) {
311 MODULE = Encode PACKAGE = Encode
316 _bytes_to_utf8(sv, ...)
320 SV * encoding = items == 2 ? ST(1) : Nullsv;
323 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
326 U8* s = (U8*)SvPV(sv, len);
329 converted = bytes_to_utf8(s, &len); /* This allocs */
330 sv_setpvn(sv, (char *)converted, len);
331 SvUTF8_on(sv); /* XXX Should we? */
332 Safefree(converted); /* ... so free it */
340 _utf8_to_bytes(sv, ...)
344 SV * to = items > 1 ? ST(1) : Nullsv;
345 SV * check = items > 2 ? ST(2) : Nullsv;
348 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
351 U8 *s = (U8*)SvPV(sv, len);
355 /* Must do things the slow way */
357 /* We need a copy to pass to check() */
358 U8 *src = (U8*)savepv((char *)s);
361 New(83, dest, len, U8); /* I think */
370 /* Have to do it all ourselves because of error routine,
372 if (!(uv & 0x40)){ goto failure; }
373 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
374 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
375 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
376 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
377 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
378 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
379 else { ulen = 13; uv = 0; }
381 /* Note change to utf8.c variable naming, for variety */
383 if ((*s & 0xc0) != 0x80){
386 uv = (uv << 6) | (*s++ & 0x3f);
391 call_failure(check, s, dest, src);
392 /* Now what happens? */
398 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
406 is_utf8(sv, check = 0)
411 if (SvGMAGICAL(sv)) /* it could be $1, for example */
412 sv = newSVsv(sv); /* GMAGIG will be done */
414 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
417 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
423 SvREFCNT_dec(sv); /* it was a temp copy */
434 SV *rsv = newSViv(SvUTF8(sv));
438 RETVAL = &PL_sv_undef;
450 SV *rsv = newSViv(SvUTF8(sv));
454 RETVAL = &PL_sv_undef;
463 RETVAL = ENCODE_DIE_ON_ERR;
470 RETVAL = ENCODE_WARN_ON_ERR;
477 RETVAL = ENCODE_LEAVE_SRC;
484 RETVAL = ENCODE_RETURN_ON_ERR;
491 RETVAL = ENCODE_PERLQQ;
498 RETVAL = ENCODE_HTMLCREF;
505 RETVAL = ENCODE_XMLCREF;
512 RETVAL = ENCODE_FB_DEFAULT;
519 RETVAL = ENCODE_FB_CROAK;
526 RETVAL = ENCODE_FB_QUIET;
533 RETVAL = ENCODE_FB_WARN;
540 RETVAL = ENCODE_FB_PERLQQ;
547 RETVAL = ENCODE_FB_HTMLCREF;
554 RETVAL = ENCODE_FB_XMLCREF;