This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Upgrade to Encode 1.67.
[perl5.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
2fc614e0 2 $Id: Encode.xs,v 1.44 2002/05/02 07:33:09 dankogai Exp $
85982a32
JH
3 */
4
fcf2db38 5#define PERL_NO_GET_CONTEXT
2c674647
JH
6#include "EXTERN.h"
7#include "perl.h"
8#include "XSUB.h"
2f2b4ff2
NIS
9#define U8 U8
10#include "encode.h"
10c5ecbb
JH
11
12# define PERLIO_MODNAME "PerlIO::encoding"
0ab8f81e 13# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 14
85982a32
JH
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
JH
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
JH
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
NIS
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
NIS
48}
49
aa0053b7
NIS
50void
51call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
52{
85982a32 53 /* Exists for breakpointing */
aa0053b7 54}
67e989fb 55
85982a32 56
2fc614e0
JH
57#define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
58#define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
59
2f2b4ff2 60static SV *
aa0053b7 61encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
85982a32 62 int check)
2f2b4ff2 63{
aa0053b7
NIS
64 STRLEN slen;
65 U8 *s = (U8 *) SvPV(src, slen);
3aececda
NIS
66 STRLEN tlen = slen;
67 STRLEN ddone = 0;
68 STRLEN sdone = 0;
39cf9a5e 69
3c49ab08 70 /* We allocate slen+1.
85982a32 71 PerlIO dumps core if this value is smaller than this. */
3c49ab08 72 SV *dst = sv_2mortal(newSV(slen+1));
85982a32
JH
73 U8 *d = (U8 *)SvPVX(dst);
74 STRLEN dlen = SvLEN(dst)-1;
75 int code;
76
77 if (!slen){
78 SvCUR_set(dst, 0);
79 SvPOK_only(dst);
80 goto ENCODE_END;
81 }
82
83 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
84 {
85 SvCUR_set(dst, dlen+ddone);
86 SvPOK_only(dst);
0b3236bb 87
85982a32
JH
88 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
89 break;
90 }
91 switch (code) {
92 case ENCODE_NOSPACE:
93 {
94 STRLEN more = 0; /* make sure you initialize! */
95 STRLEN sleft;
96 sdone += slen;
97 ddone += dlen;
98 sleft = tlen - sdone;
fcb875d4 99#if ENCODE_XS_PROFILE >= 2
85982a32
JH
100 Perl_warn(aTHX_
101 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
102 more, sdone, sleft, SvLEN(dst));
fcb875d4 103#endif
85982a32 104 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 105#if ENCODE_XS_USEFP == 2
85982a32
JH
106 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
107 - SvLEN(dst);
39cf9a5e 108#elif ENCODE_XS_USEFP
85982a32 109 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
39cf9a5e 110#else
85982a32
JH
111 /* safe until SvLEN(dst) == MAX_INT/16 */
112 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 113#endif
39cf9a5e 114 }
85982a32
JH
115 more += UTF8_MAXLEN; /* insurance policy */
116 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
117 /* dst need to grow need MORE bytes! */
118 if (ddone >= SvLEN(dst)) {
119 Perl_croak(aTHX_ "Destination couldn't be grown.");
120 }
121 dlen = SvLEN(dst)-ddone-1;
122 d += ddone;
123 s += slen;
124 slen = tlen-sdone;
125 continue;
126 }
127 case ENCODE_NOREP:
128 /* encoding */
c6a7db43 129 if (dir == enc->f_utf8) {
85982a32
JH
130 STRLEN clen;
131 UV ch =
3e952a88 132 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
b0b300a3 133 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
85982a32 134 if (check & ENCODE_DIE_ON_ERR) {
2fc614e0
JH
135 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
136 (UV)ch, enc->name[0]);
4089adc4
JH
137 return &PL_sv_undef; /* never reaches but be safe */
138 }
139 if (check & ENCODE_WARN_ON_ERR){
140 Perl_warner(aTHX_ packWARN(WARN_UTF8),
2fc614e0 141 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
4089adc4
JH
142 }
143 if (check & ENCODE_RETURN_ON_ERR){
144 goto ENCODE_SET_SRC;
145 }
146 if (check & ENCODE_PERLQQ){
147 SV* perlqq =
148 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch));
149 sdone += slen + clen;
150 ddone += dlen + SvCUR(perlqq);
151 sv_catsv(dst, perlqq);
152 }else if (check & ENCODE_HTMLCREF){
153 SV* htmlcref =
154 sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch));
155 sdone += slen + clen;
156 ddone += dlen + SvCUR(htmlcref);
157 sv_catsv(dst, htmlcref);
158 }else if (check & ENCODE_XMLCREF){
159 SV* xmlcref =
160 sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch));
161 sdone += slen + clen;
162 ddone += dlen + SvCUR(xmlcref);
163 sv_catsv(dst, xmlcref);
164 } else {
165 /* fallback char */
166 sdone += slen + clen;
167 ddone += dlen + enc->replen;
168 sv_catpvn(dst, (char*)enc->rep, enc->replen);
c6a7db43 169 }
b2704119 170 }
85982a32 171 /* decoding */
c6a7db43 172 else {
85982a32 173 if (check & ENCODE_DIE_ON_ERR){
2fc614e0
JH
174 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
175 (UV)enc->name[0], (U8)s[slen]);
4089adc4
JH
176 return &PL_sv_undef; /* never reaches but be safe */
177 }
178 if (check & ENCODE_WARN_ON_ERR){
179 Perl_warner(
180 aTHX_ packWARN(WARN_UTF8),
2fc614e0
JH
181 ERR_DECODE_NOMAP,
182 (UV)enc->name[0], (U8)s[slen]);
4089adc4
JH
183 }
184 if (check & ENCODE_RETURN_ON_ERR){
185 goto ENCODE_SET_SRC;
186 }
187 if (check &
188 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
189 SV* perlqq =
190 sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen]));
191 sdone += slen + 1;
192 ddone += dlen + SvCUR(perlqq);
193 sv_catsv(dst, perlqq);
194 } else {
195 sdone += slen + 1;
196 ddone += dlen + strlen(FBCHAR_UTF8);
197 sv_catpv(dst, FBCHAR_UTF8);
aa0053b7 198 }
b2704119 199 }
85982a32 200 /* settle variables when fallback */
b0b300a3
JH
201 d = (U8 *)SvEND(dst);
202 dlen = SvLEN(dst) - ddone - 1;
3e952a88 203 s = (U8*)SvPVX(src) + sdone;
b2704119
JH
204 slen = tlen - sdone;
205 break;
2f2b4ff2 206
85982a32
JH
207 default:
208 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
209 code, (dir == enc->f_utf8) ? "to" : "from",
210 enc->name[0]);
211 return &PL_sv_undef;
aa0053b7 212 }
85982a32
JH
213 }
214 ENCODE_SET_SRC:
ca777f1c
NIS
215 if (check && !(check & ENCODE_LEAVE_SRC)){
216 sdone = SvCUR(src) - (slen+sdone);
85982a32
JH
217 if (sdone) {
218 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 219 }
85982a32 220 SvCUR_set(src, sdone);
2f2b4ff2 221 }
85982a32 222 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
c6a7db43 223
85982a32
JH
224 SvCUR_set(dst, dlen+ddone);
225 SvPOK_only(dst);
c6a7db43 226
39cf9a5e
DK
227#if ENCODE_XS_PROFILE
228 if (SvCUR(dst) > SvCUR(src)){
85982a32
JH
229 Perl_warn(aTHX_
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);
39cf9a5e 233 }
3c49ab08 234#endif
c6a7db43 235
85982a32 236 ENCODE_END:
0b3236bb 237 *SvEND(dst) = '\0';
aa0053b7 238 return dst;
2f2b4ff2
NIS
239}
240
50d26985 241MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2
NIS
242
243PROTOTYPES: ENABLE
244
245void
0a95303c
NIS
246Method_name(obj)
247SV * obj
248CODE:
85982a32
JH
249{
250 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
251 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
252 XSRETURN(1);
253}
0a95303c
NIS
254
255void
b2704119 256Method_decode(obj,src,check = 0)
2f2b4ff2
NIS
257SV * obj
258SV * src
b2704119 259int check
2f2b4ff2 260CODE:
aae85ceb 261{
85982a32
JH
262 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
263 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
264 SvUTF8_on(ST(0));
aae85ceb
DK
265 XSRETURN(1);
266}
267
268void
85982a32 269Method_encode(obj,src,check = 0)
aae85ceb 270SV * obj
85982a32
JH
271SV * src
272int check
aae85ceb
DK
273CODE:
274{
85982a32
JH
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);
aae85ceb
DK
278 XSRETURN(1);
279}
280
0ab8f81e
JH
281void
282Method_needs_lines(obj)
283SV * obj
284CODE:
285{
286 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
287 ST(0) = &PL_sv_no;
288 XSRETURN(1);
289}
290
291void
292Method_perlio_ok(obj)
293SV * obj
294CODE:
295{
296 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
10c5ecbb
JH
297 /* require_pv(PERLIO_FILENAME); */
298
299 eval_pv("require PerlIO::encoding", 0);
300
301 if (SvTRUE(get_sv("@", 0))) {
0ab8f81e 302 ST(0) = &PL_sv_no;
10c5ecbb
JH
303 }else{
304 ST(0) = &PL_sv_yes;
0ab8f81e
JH
305 }
306 XSRETURN(1);
307}
308
67e989fb 309MODULE = Encode PACKAGE = Encode
2c674647
JH
310
311PROTOTYPES: ENABLE
312
67e989fb 313I32
2c674647 314_bytes_to_utf8(sv, ...)
85982a32
JH
315SV * sv
316CODE:
317{
318 SV * encoding = items == 2 ? ST(1) : Nullsv;
c6a7db43 319
85982a32
JH
320 if (encoding)
321 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
322 else {
323 STRLEN len;
324 U8* s = (U8*)SvPV(sv, len);
325 U8* converted;
326
327 converted = bytes_to_utf8(s, &len); /* This allocs */
328 sv_setpvn(sv, (char *)converted, len);
329 SvUTF8_on(sv); /* XXX Should we? */
330 Safefree(converted); /* ... so free it */
331 RETVAL = len;
332 }
333}
334OUTPUT:
335 RETVAL
2c674647 336
67e989fb 337I32
2c674647 338_utf8_to_bytes(sv, ...)
85982a32
JH
339SV * sv
340CODE:
341{
342 SV * to = items > 1 ? ST(1) : Nullsv;
343 SV * check = items > 2 ? ST(2) : Nullsv;
344
345 if (to) {
346 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
347 } else {
348 STRLEN len;
349 U8 *s = (U8*)SvPV(sv, len);
350
351 RETVAL = 0;
352 if (SvTRUE(check)) {
353 /* Must do things the slow way */
354 U8 *dest;
355 /* We need a copy to pass to check() */
c6a7db43 356 U8 *src = (U8*)savepv((char *)s);
85982a32
JH
357 U8 *send = s + len;
358
359 New(83, dest, len, U8); /* I think */
360
361 while (s < send) {
362 if (*s < 0x80){
363 *dest++ = *s++;
364 } else {
365 STRLEN ulen;
366 UV uv = *s++;
367
368 /* Have to do it all ourselves because of error routine,
369 aargh. */
370 if (!(uv & 0x40)){ goto failure; }
371 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
372 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
373 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
374 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
375 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
376 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
377 else { ulen = 13; uv = 0; }
87714904 378
85982a32
JH
379 /* Note change to utf8.c variable naming, for variety */
380 while (ulen--) {
c6a7db43
NIS
381 if ((*s & 0xc0) != 0x80){
382 goto failure;
85982a32
JH
383 } else {
384 uv = (uv << 6) | (*s++ & 0x3f);
385 }
87714904 386 }
67e989fb
JH
387 if (uv > 256) {
388 failure:
85982a32
JH
389 call_failure(check, s, dest, src);
390 /* Now what happens? */
67e989fb
JH
391 }
392 *dest++ = (U8)uv;
85982a32
JH
393 }
394 }
395 } else {
396 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 397 }
85982a32
JH
398 }
399}
400OUTPUT:
401 RETVAL
2c674647 402
2c674647 403bool
b2704119 404is_utf8(sv, check = 0)
4411f3b6 405SV * sv
b2704119 406int check
85982a32
JH
407CODE:
408{
409 if (SvGMAGICAL(sv)) /* it could be $1, for example */
410 sv = newSVsv(sv); /* GMAGIG will be done */
411 if (SvPOK(sv)) {
412 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
413 if (RETVAL &&
414 check &&
415 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 416 RETVAL = FALSE;
85982a32
JH
417 } else {
418 RETVAL = FALSE;
419 }
420 if (sv != ST(0))
421 SvREFCNT_dec(sv); /* it was a temp copy */
422}
423OUTPUT:
424 RETVAL
2c674647
JH
425
426SV *
4411f3b6 427_utf8_on(sv)
85982a32
JH
428SV * sv
429CODE:
430{
431 if (SvPOK(sv)) {
432 SV *rsv = newSViv(SvUTF8(sv));
433 RETVAL = rsv;
434 SvUTF8_on(sv);
435 } else {
436 RETVAL = &PL_sv_undef;
437 }
438}
439OUTPUT:
440 RETVAL
2c674647
JH
441
442SV *
4411f3b6 443_utf8_off(sv)
85982a32
JH
444SV * sv
445CODE:
446{
447 if (SvPOK(sv)) {
448 SV *rsv = newSViv(SvUTF8(sv));
449 RETVAL = rsv;
450 SvUTF8_off(sv);
451 } else {
452 RETVAL = &PL_sv_undef;
453 }
454}
455OUTPUT:
456 RETVAL
457
85982a32
JH
458int
459DIE_ON_ERR()
460CODE:
461 RETVAL = ENCODE_DIE_ON_ERR;
462OUTPUT:
463 RETVAL
464
c6a7db43 465int
85982a32
JH
466WARN_ON_ERR()
467CODE:
468 RETVAL = ENCODE_WARN_ON_ERR;
469OUTPUT:
470 RETVAL
471
472int
473LEAVE_SRC()
474CODE:
475 RETVAL = ENCODE_LEAVE_SRC;
476OUTPUT:
477 RETVAL
478
479int
480RETURN_ON_ERR()
481CODE:
482 RETVAL = ENCODE_RETURN_ON_ERR;
483OUTPUT:
484 RETVAL
485
486int
487PERLQQ()
488CODE:
489 RETVAL = ENCODE_PERLQQ;
490OUTPUT:
491 RETVAL
492
493int
af1f55d9
JH
494HTMLCREF()
495CODE:
496 RETVAL = ENCODE_HTMLCREF;
497OUTPUT:
498 RETVAL
499
500int
501XMLCREF()
502CODE:
503 RETVAL = ENCODE_XMLCREF;
504OUTPUT:
505 RETVAL
506
507int
85982a32
JH
508FB_DEFAULT()
509CODE:
510 RETVAL = ENCODE_FB_DEFAULT;
511OUTPUT:
512 RETVAL
513
514int
515FB_CROAK()
516CODE:
517 RETVAL = ENCODE_FB_CROAK;
518OUTPUT:
519 RETVAL
520
521int
522FB_QUIET()
523CODE:
524 RETVAL = ENCODE_FB_QUIET;
525OUTPUT:
526 RETVAL
527
528int
529FB_WARN()
530CODE:
531 RETVAL = ENCODE_FB_WARN;
532OUTPUT:
533 RETVAL
534
535int
536FB_PERLQQ()
537CODE:
538 RETVAL = ENCODE_FB_PERLQQ;
539OUTPUT:
540 RETVAL
2c674647 541
af1f55d9
JH
542int
543FB_HTMLCREF()
544CODE:
545 RETVAL = ENCODE_FB_HTMLCREF;
546OUTPUT:
547 RETVAL
548
549int
550FB_XMLCREF()
551CODE:
552 RETVAL = ENCODE_FB_XMLCREF;
553OUTPUT:
554 RETVAL
555
33af2bc7
NIS
556BOOT:
557{
85982a32 558#include "def_t.h"
e7cbefb8 559#include "def_t.exh"
33af2bc7 560}