Integrate perlio:
[p5sagit/p5-mst-13.2.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
4089adc4 2 $Id: Encode.xs,v 1.43 2002/05/01 05:41:06 dankogai Exp dankogai $
85982a32 3 */
4
fcf2db38 5#define PERL_NO_GET_CONTEXT
2c674647 6#include "EXTERN.h"
7#include "perl.h"
8#include "XSUB.h"
2f2b4ff2 9#define U8 U8
10#include "encode.h"
10c5ecbb 11
12# define PERLIO_MODNAME "PerlIO::encoding"
0ab8f81e 13# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 14
85982a32 15/* set 1 or more to profile. t/encoding.t dumps core because of
16 Perl_warner and PerlIO don't work well */
c6a7db43 17#define ENCODE_XS_PROFILE 0
39cf9a5e 18
85982a32 19/* set 0 to disable floating point to calculate buffer size for
20 encode_method(). 1 is recommended. 2 restores NI-S original */
c6a7db43 21#define ENCODE_XS_USEFP 1
39cf9a5e 22
fcf2db38 23#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
2f5768b8 24 Perl_croak(aTHX_ "panic_unimplemented"); \
4a83738a 25 return (y)0; /* fool picky compilers */ \
87714904 26 }
85982a32 27/**/
011b2d2f 28
7e9a885a 29UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
30UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
33af2bc7 31
b0b300a3 32void
aa0053b7 33Encode_XSEncoding(pTHX_ encode_t * enc)
2f2b4ff2 34{
aa0053b7 35 dSP;
36 HV *stash = gv_stashpv("Encode::XS", TRUE);
37 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
38 int i = 0;
39 PUSHMARK(sp);
40 XPUSHs(sv);
41 while (enc->name[i]) {
42 const char *name = enc->name[i++];
43 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
44 }
45 PUTBACK;
46 call_pv("Encode::define_encoding", G_DISCARD);
47 SvREFCNT_dec(sv);
2f2b4ff2 48}
49
aa0053b7 50void
51call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
52{
85982a32 53 /* Exists for breakpointing */
aa0053b7 54}
67e989fb 55
85982a32 56
2f2b4ff2 57static SV *
aa0053b7 58encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
85982a32 59 int check)
2f2b4ff2 60{
aa0053b7 61 STRLEN slen;
62 U8 *s = (U8 *) SvPV(src, slen);
3aececda 63 STRLEN tlen = slen;
64 STRLEN ddone = 0;
65 STRLEN sdone = 0;
39cf9a5e 66
3c49ab08 67 /* We allocate slen+1.
85982a32 68 PerlIO dumps core if this value is smaller than this. */
3c49ab08 69 SV *dst = sv_2mortal(newSV(slen+1));
85982a32 70 U8 *d = (U8 *)SvPVX(dst);
71 STRLEN dlen = SvLEN(dst)-1;
72 int code;
73
74 if (!slen){
75 SvCUR_set(dst, 0);
76 SvPOK_only(dst);
77 goto ENCODE_END;
78 }
79
80 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
81 {
82 SvCUR_set(dst, dlen+ddone);
83 SvPOK_only(dst);
0b3236bb 84
85982a32 85 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
86 break;
87 }
88 switch (code) {
89 case ENCODE_NOSPACE:
90 {
91 STRLEN more = 0; /* make sure you initialize! */
92 STRLEN sleft;
93 sdone += slen;
94 ddone += dlen;
95 sleft = tlen - sdone;
fcb875d4 96#if ENCODE_XS_PROFILE >= 2
85982a32 97 Perl_warn(aTHX_
98 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
99 more, sdone, sleft, SvLEN(dst));
fcb875d4 100#endif
85982a32 101 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 102#if ENCODE_XS_USEFP == 2
85982a32 103 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
104 - SvLEN(dst);
39cf9a5e 105#elif ENCODE_XS_USEFP
85982a32 106 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
39cf9a5e 107#else
85982a32 108 /* safe until SvLEN(dst) == MAX_INT/16 */
109 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 110#endif
39cf9a5e 111 }
85982a32 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.");
117 }
118 dlen = SvLEN(dst)-ddone-1;
119 d += ddone;
120 s += slen;
121 slen = tlen-sdone;
122 continue;
123 }
124 case ENCODE_NOREP:
125 /* encoding */
c6a7db43 126 if (dir == enc->f_utf8) {
85982a32 127 STRLEN clen;
128 UV ch =
3e952a88 129 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
b0b300a3 130 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
85982a32 131 if (check & ENCODE_DIE_ON_ERR) {
132 Perl_croak(
83049857 133 aTHX_ "\"\\x{%04" UVxf "}\" does not map to %s",
4089adc4 134 (UV)ch, enc->name[0]);
135 return &PL_sv_undef; /* never reaches but be safe */
136 }
137 if (check & ENCODE_WARN_ON_ERR){
138 Perl_warner(aTHX_ packWARN(WARN_UTF8),
83049857 139 "\"\\x{%" UVxf "}\" does not map to %s",
4089adc4 140 (UV)ch, enc->name[0]);
141 }
142 if (check & ENCODE_RETURN_ON_ERR){
143 goto ENCODE_SET_SRC;
144 }
145 if (check & ENCODE_PERLQQ){
146 SV* 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){
152 SV* 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){
158 SV* xmlcref =
159 sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch));
160 sdone += slen + clen;
161 ddone += dlen + SvCUR(xmlcref);
162 sv_catsv(dst, xmlcref);
163 } else {
164 /* fallback char */
165 sdone += slen + clen;
166 ddone += dlen + enc->replen;
167 sv_catpvn(dst, (char*)enc->rep, enc->replen);
c6a7db43 168 }
b2704119 169 }
85982a32 170 /* decoding */
c6a7db43 171 else {
85982a32 172 if (check & ENCODE_DIE_ON_ERR){
173 Perl_croak(
4089adc4 174 aTHX_ "%s \"\\x%02" UVXf
fc17bd48 175 "\" does not map to Unicode (%d)",
4089adc4 176 (UV)enc->name[0], (U8)s[slen], code);
177 return &PL_sv_undef; /* never reaches but be safe */
178 }
179 if (check & ENCODE_WARN_ON_ERR){
180 Perl_warner(
181 aTHX_ packWARN(WARN_UTF8),
182 "%s \"\\x%02" UVXf
183 "\" does not map to Unicode (%d)",
184 (UV)enc->name[0], (U8)s[slen], code);
185 }
186 if (check & ENCODE_RETURN_ON_ERR){
187 goto ENCODE_SET_SRC;
188 }
189 if (check &
190 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
191 SV* perlqq =
192 sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen]));
193 sdone += slen + 1;
194 ddone += dlen + SvCUR(perlqq);
195 sv_catsv(dst, perlqq);
196 } else {
197 sdone += slen + 1;
198 ddone += dlen + strlen(FBCHAR_UTF8);
199 sv_catpv(dst, FBCHAR_UTF8);
aa0053b7 200 }
b2704119 201 }
85982a32 202 /* settle variables when fallback */
b0b300a3 203 d = (U8 *)SvEND(dst);
204 dlen = SvLEN(dst) - ddone - 1;
3e952a88 205 s = (U8*)SvPVX(src) + sdone;
b2704119 206 slen = tlen - sdone;
207 break;
2f2b4ff2 208
85982a32 209 default:
210 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
211 code, (dir == enc->f_utf8) ? "to" : "from",
212 enc->name[0]);
213 return &PL_sv_undef;
aa0053b7 214 }
85982a32 215 }
216 ENCODE_SET_SRC:
ca777f1c 217 if (check && !(check & ENCODE_LEAVE_SRC)){
218 sdone = SvCUR(src) - (slen+sdone);
85982a32 219 if (sdone) {
220 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 221 }
85982a32 222 SvCUR_set(src, sdone);
2f2b4ff2 223 }
85982a32 224 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
c6a7db43 225
85982a32 226 SvCUR_set(dst, dlen+ddone);
227 SvPOK_only(dst);
c6a7db43 228
39cf9a5e 229#if ENCODE_XS_PROFILE
230 if (SvCUR(dst) > SvCUR(src)){
85982a32 231 Perl_warn(aTHX_
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);
39cf9a5e 235 }
3c49ab08 236#endif
c6a7db43 237
85982a32 238 ENCODE_END:
0b3236bb 239 *SvEND(dst) = '\0';
aa0053b7 240 return dst;
2f2b4ff2 241}
242
50d26985 243MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 244
245PROTOTYPES: ENABLE
246
247void
0a95303c 248Method_name(obj)
249SV * obj
250CODE:
85982a32 251{
252 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
253 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
254 XSRETURN(1);
255}
0a95303c 256
257void
b2704119 258Method_decode(obj,src,check = 0)
2f2b4ff2 259SV * obj
260SV * src
b2704119 261int check
2f2b4ff2 262CODE:
aae85ceb 263{
85982a32 264 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
265 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
266 SvUTF8_on(ST(0));
aae85ceb 267 XSRETURN(1);
268}
269
270void
85982a32 271Method_encode(obj,src,check = 0)
aae85ceb 272SV * obj
85982a32 273SV * src
274int check
aae85ceb 275CODE:
276{
85982a32 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);
aae85ceb 280 XSRETURN(1);
281}
282
0ab8f81e 283void
284Method_needs_lines(obj)
285SV * obj
286CODE:
287{
288 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
289 ST(0) = &PL_sv_no;
290 XSRETURN(1);
291}
292
293void
294Method_perlio_ok(obj)
295SV * obj
296CODE:
297{
298 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
10c5ecbb 299 /* require_pv(PERLIO_FILENAME); */
300
301 eval_pv("require PerlIO::encoding", 0);
302
303 if (SvTRUE(get_sv("@", 0))) {
0ab8f81e 304 ST(0) = &PL_sv_no;
10c5ecbb 305 }else{
306 ST(0) = &PL_sv_yes;
0ab8f81e 307 }
308 XSRETURN(1);
309}
310
67e989fb 311MODULE = Encode PACKAGE = Encode
2c674647 312
313PROTOTYPES: ENABLE
314
67e989fb 315I32
2c674647 316_bytes_to_utf8(sv, ...)
85982a32 317SV * sv
318CODE:
319{
320 SV * encoding = items == 2 ? ST(1) : Nullsv;
c6a7db43 321
85982a32 322 if (encoding)
323 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
324 else {
325 STRLEN len;
326 U8* s = (U8*)SvPV(sv, len);
327 U8* converted;
328
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 */
333 RETVAL = len;
334 }
335}
336OUTPUT:
337 RETVAL
2c674647 338
67e989fb 339I32
2c674647 340_utf8_to_bytes(sv, ...)
85982a32 341SV * sv
342CODE:
343{
344 SV * to = items > 1 ? ST(1) : Nullsv;
345 SV * check = items > 2 ? ST(2) : Nullsv;
346
347 if (to) {
348 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
349 } else {
350 STRLEN len;
351 U8 *s = (U8*)SvPV(sv, len);
352
353 RETVAL = 0;
354 if (SvTRUE(check)) {
355 /* Must do things the slow way */
356 U8 *dest;
357 /* We need a copy to pass to check() */
c6a7db43 358 U8 *src = (U8*)savepv((char *)s);
85982a32 359 U8 *send = s + len;
360
361 New(83, dest, len, U8); /* I think */
362
363 while (s < send) {
364 if (*s < 0x80){
365 *dest++ = *s++;
366 } else {
367 STRLEN ulen;
368 UV uv = *s++;
369
370 /* Have to do it all ourselves because of error routine,
371 aargh. */
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; }
87714904 380
85982a32 381 /* Note change to utf8.c variable naming, for variety */
382 while (ulen--) {
c6a7db43 383 if ((*s & 0xc0) != 0x80){
384 goto failure;
85982a32 385 } else {
386 uv = (uv << 6) | (*s++ & 0x3f);
387 }
87714904 388 }
67e989fb 389 if (uv > 256) {
390 failure:
85982a32 391 call_failure(check, s, dest, src);
392 /* Now what happens? */
67e989fb 393 }
394 *dest++ = (U8)uv;
85982a32 395 }
396 }
397 } else {
398 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 399 }
85982a32 400 }
401}
402OUTPUT:
403 RETVAL
2c674647 404
2c674647 405bool
b2704119 406is_utf8(sv, check = 0)
4411f3b6 407SV * sv
b2704119 408int check
85982a32 409CODE:
410{
411 if (SvGMAGICAL(sv)) /* it could be $1, for example */
412 sv = newSVsv(sv); /* GMAGIG will be done */
413 if (SvPOK(sv)) {
414 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
415 if (RETVAL &&
416 check &&
417 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 418 RETVAL = FALSE;
85982a32 419 } else {
420 RETVAL = FALSE;
421 }
422 if (sv != ST(0))
423 SvREFCNT_dec(sv); /* it was a temp copy */
424}
425OUTPUT:
426 RETVAL
2c674647 427
428SV *
4411f3b6 429_utf8_on(sv)
85982a32 430SV * sv
431CODE:
432{
433 if (SvPOK(sv)) {
434 SV *rsv = newSViv(SvUTF8(sv));
435 RETVAL = rsv;
436 SvUTF8_on(sv);
437 } else {
438 RETVAL = &PL_sv_undef;
439 }
440}
441OUTPUT:
442 RETVAL
2c674647 443
444SV *
4411f3b6 445_utf8_off(sv)
85982a32 446SV * sv
447CODE:
448{
449 if (SvPOK(sv)) {
450 SV *rsv = newSViv(SvUTF8(sv));
451 RETVAL = rsv;
452 SvUTF8_off(sv);
453 } else {
454 RETVAL = &PL_sv_undef;
455 }
456}
457OUTPUT:
458 RETVAL
459
85982a32 460int
461DIE_ON_ERR()
462CODE:
463 RETVAL = ENCODE_DIE_ON_ERR;
464OUTPUT:
465 RETVAL
466
c6a7db43 467int
85982a32 468WARN_ON_ERR()
469CODE:
470 RETVAL = ENCODE_WARN_ON_ERR;
471OUTPUT:
472 RETVAL
473
474int
475LEAVE_SRC()
476CODE:
477 RETVAL = ENCODE_LEAVE_SRC;
478OUTPUT:
479 RETVAL
480
481int
482RETURN_ON_ERR()
483CODE:
484 RETVAL = ENCODE_RETURN_ON_ERR;
485OUTPUT:
486 RETVAL
487
488int
489PERLQQ()
490CODE:
491 RETVAL = ENCODE_PERLQQ;
492OUTPUT:
493 RETVAL
494
495int
af1f55d9 496HTMLCREF()
497CODE:
498 RETVAL = ENCODE_HTMLCREF;
499OUTPUT:
500 RETVAL
501
502int
503XMLCREF()
504CODE:
505 RETVAL = ENCODE_XMLCREF;
506OUTPUT:
507 RETVAL
508
509int
85982a32 510FB_DEFAULT()
511CODE:
512 RETVAL = ENCODE_FB_DEFAULT;
513OUTPUT:
514 RETVAL
515
516int
517FB_CROAK()
518CODE:
519 RETVAL = ENCODE_FB_CROAK;
520OUTPUT:
521 RETVAL
522
523int
524FB_QUIET()
525CODE:
526 RETVAL = ENCODE_FB_QUIET;
527OUTPUT:
528 RETVAL
529
530int
531FB_WARN()
532CODE:
533 RETVAL = ENCODE_FB_WARN;
534OUTPUT:
535 RETVAL
536
537int
538FB_PERLQQ()
539CODE:
540 RETVAL = ENCODE_FB_PERLQQ;
541OUTPUT:
542 RETVAL
2c674647 543
af1f55d9 544int
545FB_HTMLCREF()
546CODE:
547 RETVAL = ENCODE_FB_HTMLCREF;
548OUTPUT:
549 RETVAL
550
551int
552FB_XMLCREF()
553CODE:
554 RETVAL = ENCODE_FB_XMLCREF;
555OUTPUT:
556 RETVAL
557
33af2bc7 558BOOT:
559{
85982a32 560#include "def_t.h"
e7cbefb8 561#include "def_t.exh"
33af2bc7 562}