small corrections
[p5sagit/p5-mst-13.2.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
0ab8f81e 2 $Id: Encode.xs,v 1.33 2002/04/22 03:43:05 dankogai Exp $
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"
0ab8f81e 11# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 12
85982a32 13/* set 1 or more to profile. t/encoding.t dumps core because of
14 Perl_warner and PerlIO don't work well */
0ab8f81e 15#define ENCODE_XS_PROFILE 0
39cf9a5e 16
85982a32 17/* set 0 to disable floating point to calculate buffer size for
18 encode_method(). 1 is recommended. 2 restores NI-S original */
0ab8f81e 19#define ENCODE_XS_USEFP 1
39cf9a5e 20
fcf2db38 21#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
2f5768b8 22 Perl_croak(aTHX_ "panic_unimplemented"); \
4a83738a 23 return (y)0; /* fool picky compilers */ \
87714904 24 }
85982a32 25/**/
7e9a885a 26UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
27UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
33af2bc7 28
b0b300a3 29void
aa0053b7 30Encode_XSEncoding(pTHX_ encode_t * enc)
2f2b4ff2 31{
aa0053b7 32 dSP;
33 HV *stash = gv_stashpv("Encode::XS", TRUE);
34 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
35 int i = 0;
36 PUSHMARK(sp);
37 XPUSHs(sv);
38 while (enc->name[i]) {
39 const char *name = enc->name[i++];
40 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
41 }
42 PUTBACK;
43 call_pv("Encode::define_encoding", G_DISCARD);
44 SvREFCNT_dec(sv);
2f2b4ff2 45}
46
aa0053b7 47void
48call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
49{
85982a32 50 /* Exists for breakpointing */
aa0053b7 51}
67e989fb 52
85982a32 53
2f2b4ff2 54static SV *
aa0053b7 55encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
85982a32 56 int check)
2f2b4ff2 57{
aa0053b7 58 STRLEN slen;
59 U8 *s = (U8 *) SvPV(src, slen);
3aececda 60 STRLEN tlen = slen;
61 STRLEN ddone = 0;
62 STRLEN sdone = 0;
39cf9a5e 63
3c49ab08 64 /* We allocate slen+1.
85982a32 65 PerlIO dumps core if this value is smaller than this. */
3c49ab08 66 SV *dst = sv_2mortal(newSV(slen+1));
85982a32 67 U8 *d = (U8 *)SvPVX(dst);
68 STRLEN dlen = SvLEN(dst)-1;
69 int code;
70
71 if (!slen){
72 SvCUR_set(dst, 0);
73 SvPOK_only(dst);
74 goto ENCODE_END;
75 }
76
77 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
78 {
79 SvCUR_set(dst, dlen+ddone);
80 SvPOK_only(dst);
0b3236bb 81
85982a32 82 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
83 break;
84 }
85 switch (code) {
86 case ENCODE_NOSPACE:
87 {
88 STRLEN more = 0; /* make sure you initialize! */
89 STRLEN sleft;
90 sdone += slen;
91 ddone += dlen;
92 sleft = tlen - sdone;
fcb875d4 93#if ENCODE_XS_PROFILE >= 2
85982a32 94 Perl_warn(aTHX_
95 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
96 more, sdone, sleft, SvLEN(dst));
fcb875d4 97#endif
85982a32 98 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 99#if ENCODE_XS_USEFP == 2
85982a32 100 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
101 - SvLEN(dst);
39cf9a5e 102#elif ENCODE_XS_USEFP
85982a32 103 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
39cf9a5e 104#else
85982a32 105 /* safe until SvLEN(dst) == MAX_INT/16 */
106 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 107#endif
39cf9a5e 108 }
85982a32 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.");
114 }
115 dlen = SvLEN(dst)-ddone-1;
116 d += ddone;
117 s += slen;
118 slen = tlen-sdone;
119 continue;
120 }
121 case ENCODE_NOREP:
122 /* encoding */
0ab8f81e 123 if (dir == enc->f_utf8) {
85982a32 124 STRLEN clen;
125 UV ch =
0ab8f81e 126 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
b0b300a3 127 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
85982a32 128 if (check & ENCODE_DIE_ON_ERR) {
129 Perl_croak(
0ab8f81e 130 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
85982a32 131 ch, enc->name[0], __LINE__);
132 }else{
133 if (check & ENCODE_RETURN_ON_ERR){
134 if (check & ENCODE_WARN_ON_ERR){
135 Perl_warner(
136 aTHX_ packWARN(WARN_UTF8),
0ab8f81e 137 "\"\\N{U+%" UVxf "}\" does not map to %s",
85982a32 138 ch,enc->name[0]);
139 }
140 goto ENCODE_SET_SRC;
141 }else if (check & ENCODE_PERLQQ){
0ab8f81e 142 SV* perlqq =
85982a32 143 sv_2mortal(newSVpvf("\\x{%04x}", ch));
b2704119 144 sdone += slen + clen;
85982a32 145 ddone += dlen + SvCUR(perlqq);
146 sv_catsv(dst, perlqq);
0ab8f81e 147 } else {
85982a32 148 /* fallback char */
149 sdone += slen + clen;
0ab8f81e 150 ddone += dlen + enc->replen;
151 sv_catpvn(dst, (char*)enc->rep, enc->replen);
b2704119 152 }
0ab8f81e 153 }
b2704119 154 }
85982a32 155 /* decoding */
0ab8f81e 156 else {
85982a32 157 if (check & ENCODE_DIE_ON_ERR){
158 Perl_croak(
159 aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)",
160 enc->name[0], (U8) s[slen], code);
161 }else{
162 if (check & ENCODE_RETURN_ON_ERR){
163 if (check & ENCODE_WARN_ON_ERR){
164 Perl_warner(
165 aTHX_ packWARN(WARN_UTF8),
166 "%s \"\\x%02X\" does not map to Unicode (%d)",
167 enc->name[0], (U8) s[slen], code);
168 }
169 goto ENCODE_SET_SRC;
170 }else if (check & ENCODE_PERLQQ){
0ab8f81e 171 SV* perlqq =
b2704119 172 sv_2mortal(newSVpvf("\\x%02X", s[slen]));
85982a32 173 sdone += slen + 1;
174 ddone += dlen + SvCUR(perlqq);
175 sv_catsv(dst, perlqq);
176 } else {
177 sdone += slen + 1;
0ab8f81e 178 ddone += dlen + strlen(FBCHAR_UTF8);
179 sv_catpv(dst, FBCHAR_UTF8);
85982a32 180 }
aa0053b7 181 }
b2704119 182 }
85982a32 183 /* settle variables when fallback */
b0b300a3 184 d = (U8 *)SvEND(dst);
0ab8f81e 185 dlen = SvLEN(dst) - ddone - 1;
186 s = (U8*)SvPVX(src) + sdone;
b2704119 187 slen = tlen - sdone;
188 break;
2f2b4ff2 189
85982a32 190 default:
191 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
192 code, (dir == enc->f_utf8) ? "to" : "from",
193 enc->name[0]);
194 return &PL_sv_undef;
aa0053b7 195 }
85982a32 196 }
197 ENCODE_SET_SRC:
6d1c0808 198 if (check && !(check & ENCODE_LEAVE_SRC)){
199 sdone = SvCUR(src) - (slen+sdone);
85982a32 200 if (sdone) {
201 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 202 }
85982a32 203 SvCUR_set(src, sdone);
2f2b4ff2 204 }
85982a32 205 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
206 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
207 return &PL_sv_undef;
2f2b4ff2 208 }
0ab8f81e 209
85982a32 210 SvCUR_set(dst, dlen+ddone);
211 SvPOK_only(dst);
0ab8f81e 212
39cf9a5e 213#if ENCODE_XS_PROFILE
214 if (SvCUR(dst) > SvCUR(src)){
85982a32 215 Perl_warn(aTHX_
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);
39cf9a5e 219 }
3c49ab08 220#endif
0ab8f81e 221
85982a32 222 ENCODE_END:
0b3236bb 223 *SvEND(dst) = '\0';
aa0053b7 224 return dst;
2f2b4ff2 225}
226
50d26985 227MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 228
229PROTOTYPES: ENABLE
230
231void
0a95303c 232Method_name(obj)
233SV * obj
234CODE:
85982a32 235{
236 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
237 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
238 XSRETURN(1);
239}
0a95303c 240
241void
b2704119 242Method_decode(obj,src,check = 0)
2f2b4ff2 243SV * obj
244SV * src
b2704119 245int check
2f2b4ff2 246CODE:
aae85ceb 247{
85982a32 248 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
249 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
250 SvUTF8_on(ST(0));
aae85ceb 251 XSRETURN(1);
252}
253
254void
85982a32 255Method_encode(obj,src,check = 0)
aae85ceb 256SV * obj
85982a32 257SV * src
258int check
aae85ceb 259CODE:
260{
85982a32 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);
aae85ceb 264 XSRETURN(1);
265}
266
0ab8f81e 267void
268Method_needs_lines(obj)
269SV * obj
270CODE:
271{
272 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
273 ST(0) = &PL_sv_no;
274 XSRETURN(1);
275}
276
277void
278Method_perlio_ok(obj)
279SV * obj
280CODE:
281{
282 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
283 if (hv_exists(get_hv("INC", 0),
284 PERLIO_FILENAME, strlen(PERLIO_FILENAME)))
285 {
286 ST(0) = &PL_sv_yes;
287 }else{
288 ST(0) = &PL_sv_no;
289 }
290 XSRETURN(1);
291}
292
67e989fb 293MODULE = Encode PACKAGE = Encode
2c674647 294
295PROTOTYPES: ENABLE
296
67e989fb 297I32
2c674647 298_bytes_to_utf8(sv, ...)
85982a32 299SV * sv
300CODE:
301{
302 SV * encoding = items == 2 ? ST(1) : Nullsv;
0ab8f81e 303
85982a32 304 if (encoding)
305 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
306 else {
307 STRLEN len;
308 U8* s = (U8*)SvPV(sv, len);
309 U8* converted;
310
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 */
315 RETVAL = len;
316 }
317}
318OUTPUT:
319 RETVAL
2c674647 320
67e989fb 321I32
2c674647 322_utf8_to_bytes(sv, ...)
85982a32 323SV * sv
324CODE:
325{
326 SV * to = items > 1 ? ST(1) : Nullsv;
327 SV * check = items > 2 ? ST(2) : Nullsv;
328
329 if (to) {
330 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
331 } else {
332 STRLEN len;
333 U8 *s = (U8*)SvPV(sv, len);
334
335 RETVAL = 0;
336 if (SvTRUE(check)) {
337 /* Must do things the slow way */
338 U8 *dest;
339 /* We need a copy to pass to check() */
0ab8f81e 340 U8 *src = (U8*)savepv((char *)s);
85982a32 341 U8 *send = s + len;
342
343 New(83, dest, len, U8); /* I think */
344
345 while (s < send) {
346 if (*s < 0x80){
347 *dest++ = *s++;
348 } else {
349 STRLEN ulen;
350 UV uv = *s++;
351
352 /* Have to do it all ourselves because of error routine,
353 aargh. */
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; }
87714904 362
85982a32 363 /* Note change to utf8.c variable naming, for variety */
364 while (ulen--) {
0ab8f81e 365 if ((*s & 0xc0) != 0x80){
366 goto failure;
85982a32 367 } else {
368 uv = (uv << 6) | (*s++ & 0x3f);
369 }
87714904 370 }
67e989fb 371 if (uv > 256) {
372 failure:
85982a32 373 call_failure(check, s, dest, src);
374 /* Now what happens? */
67e989fb 375 }
376 *dest++ = (U8)uv;
85982a32 377 }
378 }
379 } else {
380 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 381 }
85982a32 382 }
383}
384OUTPUT:
385 RETVAL
2c674647 386
2c674647 387bool
b2704119 388is_utf8(sv, check = 0)
4411f3b6 389SV * sv
b2704119 390int check
85982a32 391CODE:
392{
393 if (SvGMAGICAL(sv)) /* it could be $1, for example */
394 sv = newSVsv(sv); /* GMAGIG will be done */
395 if (SvPOK(sv)) {
396 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
397 if (RETVAL &&
398 check &&
399 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 400 RETVAL = FALSE;
85982a32 401 } else {
402 RETVAL = FALSE;
403 }
404 if (sv != ST(0))
405 SvREFCNT_dec(sv); /* it was a temp copy */
406}
407OUTPUT:
408 RETVAL
2c674647 409
410SV *
4411f3b6 411_utf8_on(sv)
85982a32 412SV * sv
413CODE:
414{
415 if (SvPOK(sv)) {
416 SV *rsv = newSViv(SvUTF8(sv));
417 RETVAL = rsv;
418 SvUTF8_on(sv);
419 } else {
420 RETVAL = &PL_sv_undef;
421 }
422}
423OUTPUT:
424 RETVAL
2c674647 425
426SV *
4411f3b6 427_utf8_off(sv)
85982a32 428SV * sv
429CODE:
430{
431 if (SvPOK(sv)) {
432 SV *rsv = newSViv(SvUTF8(sv));
433 RETVAL = rsv;
434 SvUTF8_off(sv);
435 } else {
436 RETVAL = &PL_sv_undef;
437 }
438}
439OUTPUT:
440 RETVAL
441
442PROTOTYPES: DISABLE
443
444
445int
446DIE_ON_ERR()
447CODE:
448 RETVAL = ENCODE_DIE_ON_ERR;
449OUTPUT:
450 RETVAL
451
0ab8f81e 452int
85982a32 453WARN_ON_ERR()
454CODE:
455 RETVAL = ENCODE_WARN_ON_ERR;
456OUTPUT:
457 RETVAL
458
459int
460LEAVE_SRC()
461CODE:
462 RETVAL = ENCODE_LEAVE_SRC;
463OUTPUT:
464 RETVAL
465
466int
467RETURN_ON_ERR()
468CODE:
469 RETVAL = ENCODE_RETURN_ON_ERR;
470OUTPUT:
471 RETVAL
472
473int
474PERLQQ()
475CODE:
476 RETVAL = ENCODE_PERLQQ;
477OUTPUT:
478 RETVAL
479
480int
481FB_DEFAULT()
482CODE:
483 RETVAL = ENCODE_FB_DEFAULT;
484OUTPUT:
485 RETVAL
486
487int
488FB_CROAK()
489CODE:
490 RETVAL = ENCODE_FB_CROAK;
491OUTPUT:
492 RETVAL
493
494int
495FB_QUIET()
496CODE:
497 RETVAL = ENCODE_FB_QUIET;
498OUTPUT:
499 RETVAL
500
501int
502FB_WARN()
503CODE:
504 RETVAL = ENCODE_FB_WARN;
505OUTPUT:
506 RETVAL
507
508int
509FB_PERLQQ()
510CODE:
511 RETVAL = ENCODE_FB_PERLQQ;
512OUTPUT:
513 RETVAL
2c674647 514
33af2bc7 515BOOT:
516{
85982a32 517#include "def_t.h"
e7cbefb8 518#include "def_t.exh"
33af2bc7 519}