2 $Id: Encode.xs,v 1.39 2002/04/26 03:02:04 dankogai Exp $
5 #define PERL_NO_GET_CONTEXT
11 # define PERLIO_FILENAME "PerlIO/encoding.pm"
13 /* set 1 or more to profile. t/encoding.t dumps core because of
14 Perl_warner and PerlIO don't work well */
15 #define ENCODE_XS_PROFILE 0
17 /* set 0 to disable floating point to calculate buffer size for
18 encode_method(). 1 is recommended. 2 restores NI-S original */
19 #define ENCODE_XS_USEFP 1
21 #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
22 Perl_croak(aTHX_ "panic_unimplemented"); \
23 return (y)0; /* fool picky compilers */ \
27 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
28 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
31 Encode_XSEncoding(pTHX_ encode_t * enc)
34 HV *stash = gv_stashpv("Encode::XS", TRUE);
35 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
39 while (enc->name[i]) {
40 const char *name = enc->name[i++];
41 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
44 call_pv("Encode::define_encoding", G_DISCARD);
49 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
51 /* Exists for breakpointing */
56 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
60 U8 *s = (U8 *) SvPV(src, slen);
65 /* We allocate slen+1.
66 PerlIO dumps core if this value is smaller than this. */
67 SV *dst = sv_2mortal(newSV(slen+1));
68 U8 *d = (U8 *)SvPVX(dst);
69 STRLEN dlen = SvLEN(dst)-1;
78 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
80 SvCUR_set(dst, dlen+ddone);
83 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
89 STRLEN more = 0; /* make sure you initialize! */
94 #if ENCODE_XS_PROFILE >= 2
96 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
97 more, sdone, sleft, SvLEN(dst));
99 if (sdone != 0) { /* has src ever been processed ? */
100 #if ENCODE_XS_USEFP == 2
101 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
103 #elif ENCODE_XS_USEFP
104 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
106 /* safe until SvLEN(dst) == MAX_INT/16 */
107 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
110 more += UTF8_MAXLEN; /* insurance policy */
111 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
112 /* dst need to grow need MORE bytes! */
113 if (ddone >= SvLEN(dst)) {
114 Perl_croak(aTHX_ "Destination couldn't be grown.");
116 dlen = SvLEN(dst)-ddone-1;
124 if (dir == enc->f_utf8) {
127 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
128 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
129 if (check & ENCODE_DIE_ON_ERR) {
131 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
132 ch, enc->name[0], __LINE__);
134 if (check & ENCODE_RETURN_ON_ERR){
135 if (check & ENCODE_WARN_ON_ERR){
137 aTHX_ packWARN(WARN_UTF8),
138 "\"\\N{U+%" UVxf "}\" does not map to %s",
142 }else if (check & ENCODE_PERLQQ){
144 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", ch));
145 sdone += slen + clen;
146 ddone += dlen + SvCUR(perlqq);
147 sv_catsv(dst, perlqq);
148 }else if (check & ENCODE_HTMLCREF){
150 sv_2mortal(newSVpvf("&#%" UVuf ";", ch));
151 sdone += slen + clen;
152 ddone += dlen + SvCUR(htmlcref);
153 sv_catsv(dst, htmlcref);
154 }else if (check & ENCODE_XMLCREF){
156 sv_2mortal(newSVpvf("&#x%" UVxf ";", ch));
157 sdone += slen + clen;
158 ddone += dlen + SvCUR(xmlcref);
159 sv_catsv(dst, xmlcref);
162 sdone += slen + clen;
163 ddone += dlen + enc->replen;
164 sv_catpvn(dst, (char*)enc->rep, enc->replen);
170 if (check & ENCODE_DIE_ON_ERR){
172 aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)",
173 enc->name[0], (U8) s[slen], code);
175 if (check & ENCODE_RETURN_ON_ERR){
176 if (check & ENCODE_WARN_ON_ERR){
178 aTHX_ packWARN(WARN_UTF8),
179 "%s \"\\x%02X\" does not map to Unicode (%d)",
180 enc->name[0], (U8) s[slen], code);
184 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
186 sv_2mortal(newSVpvf("\\x%02X", s[slen]));
188 ddone += dlen + SvCUR(perlqq);
189 sv_catsv(dst, perlqq);
192 ddone += dlen + strlen(FBCHAR_UTF8);
193 sv_catpv(dst, FBCHAR_UTF8);
197 /* settle variables when fallback */
198 d = (U8 *)SvEND(dst);
199 dlen = SvLEN(dst) - ddone - 1;
200 s = (U8*)SvPVX(src) + sdone;
205 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
206 code, (dir == enc->f_utf8) ? "to" : "from",
212 if (check && !(check & ENCODE_LEAVE_SRC)){
213 sdone = SvCUR(src) - (slen+sdone);
215 sv_setpvn(src, (char*)s+slen, sdone);
217 SvCUR_set(src, sdone);
219 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
220 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
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::XS PREFIX = Method_
250 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
251 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
256 Method_decode(obj,src,check = 0)
262 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
263 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
269 Method_encode(obj,src,check = 0)
275 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
276 sv_utf8_upgrade(src);
277 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
282 Method_needs_lines(obj)
286 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
292 Method_perlio_ok(obj)
296 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
297 require_pv(PERLIO_FILENAME);
298 if (hv_exists(get_hv("INC", 0),
299 PERLIO_FILENAME, strlen(PERLIO_FILENAME)))
308 MODULE = Encode PACKAGE = Encode
313 _bytes_to_utf8(sv, ...)
317 SV * encoding = items == 2 ? ST(1) : Nullsv;
320 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
323 U8* s = (U8*)SvPV(sv, len);
326 converted = bytes_to_utf8(s, &len); /* This allocs */
327 sv_setpvn(sv, (char *)converted, len);
328 SvUTF8_on(sv); /* XXX Should we? */
329 Safefree(converted); /* ... so free it */
337 _utf8_to_bytes(sv, ...)
341 SV * to = items > 1 ? ST(1) : Nullsv;
342 SV * check = items > 2 ? ST(2) : Nullsv;
345 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
348 U8 *s = (U8*)SvPV(sv, len);
352 /* Must do things the slow way */
354 /* We need a copy to pass to check() */
355 U8 *src = (U8*)savepv((char *)s);
358 New(83, dest, len, U8); /* I think */
367 /* Have to do it all ourselves because of error routine,
369 if (!(uv & 0x40)){ goto failure; }
370 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
371 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
372 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
373 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
374 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
375 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
376 else { ulen = 13; uv = 0; }
378 /* Note change to utf8.c variable naming, for variety */
380 if ((*s & 0xc0) != 0x80){
383 uv = (uv << 6) | (*s++ & 0x3f);
388 call_failure(check, s, dest, src);
389 /* Now what happens? */
395 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
403 is_utf8(sv, check = 0)
408 if (SvGMAGICAL(sv)) /* it could be $1, for example */
409 sv = newSVsv(sv); /* GMAGIG will be done */
411 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
414 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
420 SvREFCNT_dec(sv); /* it was a temp copy */
431 SV *rsv = newSViv(SvUTF8(sv));
435 RETVAL = &PL_sv_undef;
447 SV *rsv = newSViv(SvUTF8(sv));
451 RETVAL = &PL_sv_undef;
460 RETVAL = ENCODE_DIE_ON_ERR;
467 RETVAL = ENCODE_WARN_ON_ERR;
474 RETVAL = ENCODE_LEAVE_SRC;
481 RETVAL = ENCODE_RETURN_ON_ERR;
488 RETVAL = ENCODE_PERLQQ;
495 RETVAL = ENCODE_HTMLCREF;
502 RETVAL = ENCODE_XMLCREF;
509 RETVAL = ENCODE_FB_DEFAULT;
516 RETVAL = ENCODE_FB_CROAK;
523 RETVAL = ENCODE_FB_QUIET;
530 RETVAL = ENCODE_FB_WARN;
537 RETVAL = ENCODE_FB_PERLQQ;
544 RETVAL = ENCODE_FB_HTMLCREF;
551 RETVAL = ENCODE_FB_XMLCREF;