2 $Id: Encode.xs,v 1.33 2002/04/22 03:43:05 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 */ \
26 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
27 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
30 Encode_XSEncoding(pTHX_ encode_t * enc)
33 HV *stash = gv_stashpv("Encode::XS", TRUE);
34 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
38 while (enc->name[i]) {
39 const char *name = enc->name[i++];
40 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
43 call_pv("Encode::define_encoding", G_DISCARD);
48 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
50 /* Exists for breakpointing */
55 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
59 U8 *s = (U8 *) SvPV(src, slen);
64 /* We allocate slen+1.
65 PerlIO dumps core if this value is smaller than this. */
66 SV *dst = sv_2mortal(newSV(slen+1));
67 U8 *d = (U8 *)SvPVX(dst);
68 STRLEN dlen = SvLEN(dst)-1;
77 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
79 SvCUR_set(dst, dlen+ddone);
82 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
88 STRLEN more = 0; /* make sure you initialize! */
93 #if ENCODE_XS_PROFILE >= 2
95 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
96 more, sdone, sleft, SvLEN(dst));
98 if (sdone != 0) { /* has src ever been processed ? */
99 #if ENCODE_XS_USEFP == 2
100 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
102 #elif ENCODE_XS_USEFP
103 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
105 /* safe until SvLEN(dst) == MAX_INT/16 */
106 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
109 more += UTF8_MAXLEN; /* insurance policy */
110 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
111 /* dst need to grow need MORE bytes! */
112 if (ddone >= SvLEN(dst)) {
113 Perl_croak(aTHX_ "Destination couldn't be grown.");
115 dlen = SvLEN(dst)-ddone-1;
123 if (dir == enc->f_utf8) {
126 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
127 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
128 if (check & ENCODE_DIE_ON_ERR) {
130 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
131 ch, enc->name[0], __LINE__);
133 if (check & ENCODE_RETURN_ON_ERR){
134 if (check & ENCODE_WARN_ON_ERR){
136 aTHX_ packWARN(WARN_UTF8),
137 "\"\\N{U+%" UVxf "}\" does not map to %s",
141 }else if (check & ENCODE_PERLQQ){
143 sv_2mortal(newSVpvf("\\x{%04x}", ch));
144 sdone += slen + clen;
145 ddone += dlen + SvCUR(perlqq);
146 sv_catsv(dst, perlqq);
149 sdone += slen + clen;
150 ddone += dlen + enc->replen;
151 sv_catpvn(dst, (char*)enc->rep, enc->replen);
157 if (check & ENCODE_DIE_ON_ERR){
159 aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)",
160 enc->name[0], (U8) s[slen], code);
162 if (check & ENCODE_RETURN_ON_ERR){
163 if (check & ENCODE_WARN_ON_ERR){
165 aTHX_ packWARN(WARN_UTF8),
166 "%s \"\\x%02X\" does not map to Unicode (%d)",
167 enc->name[0], (U8) s[slen], code);
170 }else if (check & ENCODE_PERLQQ){
172 sv_2mortal(newSVpvf("\\x%02X", s[slen]));
174 ddone += dlen + SvCUR(perlqq);
175 sv_catsv(dst, perlqq);
178 ddone += dlen + strlen(FBCHAR_UTF8);
179 sv_catpv(dst, FBCHAR_UTF8);
183 /* settle variables when fallback */
184 d = (U8 *)SvEND(dst);
185 dlen = SvLEN(dst) - ddone - 1;
186 s = (U8*)SvPVX(src) + sdone;
191 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
192 code, (dir == enc->f_utf8) ? "to" : "from",
198 if (check && !(check & ENCODE_LEAVE_SRC)){
199 sdone = SvCUR(src) - (slen+sdone);
201 sv_setpvn(src, (char*)s+slen, sdone);
203 SvCUR_set(src, sdone);
205 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
206 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
210 SvCUR_set(dst, dlen+ddone);
213 #if ENCODE_XS_PROFILE
214 if (SvCUR(dst) > SvCUR(src)){
216 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
217 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
218 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
227 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
236 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
237 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
242 Method_decode(obj,src,check = 0)
248 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
249 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
255 Method_encode(obj,src,check = 0)
261 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
262 sv_utf8_upgrade(src);
263 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
268 Method_needs_lines(obj)
272 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
278 Method_perlio_ok(obj)
282 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
283 if (hv_exists(get_hv("INC", 0),
284 PERLIO_FILENAME, strlen(PERLIO_FILENAME)))
293 MODULE = Encode PACKAGE = Encode
298 _bytes_to_utf8(sv, ...)
302 SV * encoding = items == 2 ? ST(1) : Nullsv;
305 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
308 U8* s = (U8*)SvPV(sv, len);
311 converted = bytes_to_utf8(s, &len); /* This allocs */
312 sv_setpvn(sv, (char *)converted, len);
313 SvUTF8_on(sv); /* XXX Should we? */
314 Safefree(converted); /* ... so free it */
322 _utf8_to_bytes(sv, ...)
326 SV * to = items > 1 ? ST(1) : Nullsv;
327 SV * check = items > 2 ? ST(2) : Nullsv;
330 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
333 U8 *s = (U8*)SvPV(sv, len);
337 /* Must do things the slow way */
339 /* We need a copy to pass to check() */
340 U8 *src = (U8*)savepv((char *)s);
343 New(83, dest, len, U8); /* I think */
352 /* Have to do it all ourselves because of error routine,
354 if (!(uv & 0x40)){ goto failure; }
355 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
356 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
357 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
358 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
359 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
360 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
361 else { ulen = 13; uv = 0; }
363 /* Note change to utf8.c variable naming, for variety */
365 if ((*s & 0xc0) != 0x80){
368 uv = (uv << 6) | (*s++ & 0x3f);
373 call_failure(check, s, dest, src);
374 /* Now what happens? */
380 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
388 is_utf8(sv, check = 0)
393 if (SvGMAGICAL(sv)) /* it could be $1, for example */
394 sv = newSVsv(sv); /* GMAGIG will be done */
396 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
399 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
405 SvREFCNT_dec(sv); /* it was a temp copy */
416 SV *rsv = newSViv(SvUTF8(sv));
420 RETVAL = &PL_sv_undef;
432 SV *rsv = newSViv(SvUTF8(sv));
436 RETVAL = &PL_sv_undef;
448 RETVAL = ENCODE_DIE_ON_ERR;
455 RETVAL = ENCODE_WARN_ON_ERR;
462 RETVAL = ENCODE_LEAVE_SRC;
469 RETVAL = ENCODE_RETURN_ON_ERR;
476 RETVAL = ENCODE_PERLQQ;
483 RETVAL = ENCODE_FB_DEFAULT;
490 RETVAL = ENCODE_FB_CROAK;
497 RETVAL = ENCODE_FB_QUIET;
504 RETVAL = ENCODE_FB_WARN;
511 RETVAL = ENCODE_FB_PERLQQ;