Commit | Line | Data |
85982a32 |
1 | /* |
2 | $Id: Encode.xs,v 1.29 2002/04/19 05:36:43 dankogai Exp $ |
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" |
aae85ceb |
11 | |
85982a32 |
12 | /* set 1 or more to profile. t/encoding.t dumps core because of |
13 | Perl_warner and PerlIO don't work well */ |
14 | #define ENCODE_XS_PROFILE 0 |
39cf9a5e |
15 | |
85982a32 |
16 | /* set 0 to disable floating point to calculate buffer size for |
17 | encode_method(). 1 is recommended. 2 restores NI-S original */ |
18 | #define ENCODE_XS_USEFP 1 |
39cf9a5e |
19 | |
fcf2db38 |
20 | #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \ |
2f5768b8 |
21 | Perl_croak(aTHX_ "panic_unimplemented"); \ |
4a83738a |
22 | return (y)0; /* fool picky compilers */ \ |
87714904 |
23 | } |
85982a32 |
24 | /**/ |
67e989fb |
25 | UNIMPLEMENTED(_encoded_utf8_to_bytes, I32) |
85982a32 |
26 | UNIMPLEMENTED(_encoded_bytes_to_utf8, I32) |
33af2bc7 |
27 | |
85982a32 |
28 | void |
aa0053b7 |
29 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
30 | { |
aa0053b7 |
31 | dSP; |
32 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
33 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
34 | int i = 0; |
35 | PUSHMARK(sp); |
36 | XPUSHs(sv); |
37 | while (enc->name[i]) { |
38 | const char *name = enc->name[i++]; |
39 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
40 | } |
41 | PUTBACK; |
42 | call_pv("Encode::define_encoding", G_DISCARD); |
43 | SvREFCNT_dec(sv); |
2f2b4ff2 |
44 | } |
45 | |
aa0053b7 |
46 | void |
47 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
48 | { |
85982a32 |
49 | /* Exists for breakpointing */ |
aa0053b7 |
50 | } |
67e989fb |
51 | |
85982a32 |
52 | |
2f2b4ff2 |
53 | static SV * |
aa0053b7 |
54 | encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src, |
85982a32 |
55 | int check) |
2f2b4ff2 |
56 | { |
aa0053b7 |
57 | STRLEN slen; |
58 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
59 | STRLEN tlen = slen; |
60 | STRLEN ddone = 0; |
61 | STRLEN sdone = 0; |
39cf9a5e |
62 | |
3c49ab08 |
63 | /* We allocate slen+1. |
85982a32 |
64 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
65 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
66 | U8 *d = (U8 *)SvPVX(dst); |
67 | STRLEN dlen = SvLEN(dst)-1; |
68 | int code; |
69 | |
70 | if (!slen){ |
71 | SvCUR_set(dst, 0); |
72 | SvPOK_only(dst); |
73 | goto ENCODE_END; |
74 | } |
75 | |
76 | while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check)) |
77 | { |
78 | SvCUR_set(dst, dlen+ddone); |
79 | SvPOK_only(dst); |
0b3236bb |
80 | |
85982a32 |
81 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){ |
82 | break; |
83 | } |
84 | switch (code) { |
85 | case ENCODE_NOSPACE: |
86 | { |
87 | STRLEN more = 0; /* make sure you initialize! */ |
88 | STRLEN sleft; |
89 | sdone += slen; |
90 | ddone += dlen; |
91 | sleft = tlen - sdone; |
fcb875d4 |
92 | #if ENCODE_XS_PROFILE >= 2 |
85982a32 |
93 | Perl_warn(aTHX_ |
94 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
95 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
96 | #endif |
85982a32 |
97 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
98 | #if ENCODE_XS_USEFP == 2 |
85982a32 |
99 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
100 | - SvLEN(dst); |
39cf9a5e |
101 | #elif ENCODE_XS_USEFP |
85982a32 |
102 | more = (1.0*SvLEN(dst)+1)/sdone * sleft; |
39cf9a5e |
103 | #else |
85982a32 |
104 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
105 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
106 | #endif |
39cf9a5e |
107 | } |
85982a32 |
108 | more += UTF8_MAXLEN; /* insurance policy */ |
109 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
110 | /* dst need to grow need MORE bytes! */ |
111 | if (ddone >= SvLEN(dst)) { |
112 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
113 | } |
114 | dlen = SvLEN(dst)-ddone-1; |
115 | d += ddone; |
116 | s += slen; |
117 | slen = tlen-sdone; |
118 | continue; |
119 | } |
120 | case ENCODE_NOREP: |
121 | /* encoding */ |
122 | if (dir == enc->f_utf8) { |
123 | STRLEN clen; |
124 | UV ch = |
125 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), &clen, 0); |
126 | if (check & ENCODE_DIE_ON_ERR) { |
127 | Perl_croak( |
128 | aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d", |
129 | ch, enc->name[0], __LINE__); |
130 | }else{ |
131 | if (check & ENCODE_RETURN_ON_ERR){ |
132 | if (check & ENCODE_WARN_ON_ERR){ |
133 | Perl_warner( |
134 | aTHX_ packWARN(WARN_UTF8), |
135 | "\"\\N{U+%" UVxf "}\" does not map to %s", |
136 | ch,enc->name[0]); |
137 | } |
138 | goto ENCODE_SET_SRC; |
139 | }else if (check & ENCODE_PERLQQ){ |
140 | SV* perlqq = |
141 | sv_2mortal(newSVpvf("\\x{%04x}", ch)); |
b2704119 |
142 | sdone += slen + clen; |
85982a32 |
143 | ddone += dlen + SvCUR(perlqq); |
144 | sv_catsv(dst, perlqq); |
145 | } else { |
146 | /* fallback char */ |
147 | sdone += slen + clen; |
148 | ddone += dlen + enc->replen; |
149 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
b2704119 |
150 | } |
85982a32 |
151 | } |
b2704119 |
152 | } |
85982a32 |
153 | /* decoding */ |
154 | else { |
155 | if (check & ENCODE_DIE_ON_ERR){ |
156 | Perl_croak( |
157 | aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)", |
158 | enc->name[0], (U8) s[slen], code); |
159 | }else{ |
160 | if (check & ENCODE_RETURN_ON_ERR){ |
161 | if (check & ENCODE_WARN_ON_ERR){ |
162 | Perl_warner( |
163 | aTHX_ packWARN(WARN_UTF8), |
164 | "%s \"\\x%02X\" does not map to Unicode (%d)", |
165 | enc->name[0], (U8) s[slen], code); |
166 | } |
167 | goto ENCODE_SET_SRC; |
168 | }else if (check & ENCODE_PERLQQ){ |
169 | SV* perlqq = |
b2704119 |
170 | sv_2mortal(newSVpvf("\\x%02X", s[slen])); |
85982a32 |
171 | sdone += slen + 1; |
172 | ddone += dlen + SvCUR(perlqq); |
173 | sv_catsv(dst, perlqq); |
174 | } else { |
175 | sdone += slen + 1; |
176 | ddone += dlen + strlen(FBCHAR_UTF8); |
177 | sv_catpv(dst, FBCHAR_UTF8); |
178 | } |
aa0053b7 |
179 | } |
b2704119 |
180 | } |
85982a32 |
181 | /* settle variables when fallback */ |
b2704119 |
182 | dlen = SvCUR(dst); |
35a51b20 |
183 | d = (U8*)SvPVX(dst) + dlen; |
184 | s = (U8*)SvPVX(src) + sdone; |
b2704119 |
185 | slen = tlen - sdone; |
186 | break; |
2f2b4ff2 |
187 | |
85982a32 |
188 | default: |
189 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
190 | code, (dir == enc->f_utf8) ? "to" : "from", |
191 | enc->name[0]); |
192 | return &PL_sv_undef; |
aa0053b7 |
193 | } |
85982a32 |
194 | } |
195 | ENCODE_SET_SRC: |
196 | if (check & ~ENCODE_LEAVE_SRC){ |
197 | sdone = SvCUR(src) - (slen+sdone); |
198 | if (sdone) { |
199 | sv_setpvn(src, (char*)s+slen, sdone); |
aa0053b7 |
200 | } |
85982a32 |
201 | SvCUR_set(src, sdone); |
2f2b4ff2 |
202 | } |
85982a32 |
203 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
204 | if (code && !(check & ENCODE_RETURN_ON_ERR)) { |
205 | return &PL_sv_undef; |
2f2b4ff2 |
206 | } |
85982a32 |
207 | |
208 | SvCUR_set(dst, dlen+ddone); |
209 | SvPOK_only(dst); |
210 | |
39cf9a5e |
211 | #if ENCODE_XS_PROFILE |
212 | if (SvCUR(dst) > SvCUR(src)){ |
85982a32 |
213 | Perl_warn(aTHX_ |
214 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
215 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
216 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
217 | } |
3c49ab08 |
218 | #endif |
85982a32 |
219 | |
220 | ENCODE_END: |
0b3236bb |
221 | *SvEND(dst) = '\0'; |
aa0053b7 |
222 | return dst; |
2f2b4ff2 |
223 | } |
224 | |
50d26985 |
225 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
226 | |
227 | PROTOTYPES: ENABLE |
228 | |
229 | void |
0a95303c |
230 | Method_name(obj) |
231 | SV * obj |
232 | CODE: |
85982a32 |
233 | { |
234 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
235 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
236 | XSRETURN(1); |
237 | } |
0a95303c |
238 | |
239 | void |
b2704119 |
240 | Method_decode(obj,src,check = 0) |
2f2b4ff2 |
241 | SV * obj |
242 | SV * src |
b2704119 |
243 | int check |
2f2b4ff2 |
244 | CODE: |
aae85ceb |
245 | { |
85982a32 |
246 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
247 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check); |
248 | SvUTF8_on(ST(0)); |
aae85ceb |
249 | XSRETURN(1); |
250 | } |
251 | |
252 | void |
85982a32 |
253 | Method_encode(obj,src,check = 0) |
aae85ceb |
254 | SV * obj |
85982a32 |
255 | SV * src |
256 | int check |
aae85ceb |
257 | CODE: |
258 | { |
85982a32 |
259 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
260 | sv_utf8_upgrade(src); |
261 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check); |
aae85ceb |
262 | XSRETURN(1); |
263 | } |
264 | |
67e989fb |
265 | MODULE = Encode PACKAGE = Encode |
2c674647 |
266 | |
267 | PROTOTYPES: ENABLE |
268 | |
67e989fb |
269 | I32 |
2c674647 |
270 | _bytes_to_utf8(sv, ...) |
85982a32 |
271 | SV * sv |
272 | CODE: |
273 | { |
274 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
275 | |
276 | if (encoding) |
277 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
278 | else { |
279 | STRLEN len; |
280 | U8* s = (U8*)SvPV(sv, len); |
281 | U8* converted; |
282 | |
283 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
284 | sv_setpvn(sv, (char *)converted, len); |
285 | SvUTF8_on(sv); /* XXX Should we? */ |
286 | Safefree(converted); /* ... so free it */ |
287 | RETVAL = len; |
288 | } |
289 | } |
290 | OUTPUT: |
291 | RETVAL |
2c674647 |
292 | |
67e989fb |
293 | I32 |
2c674647 |
294 | _utf8_to_bytes(sv, ...) |
85982a32 |
295 | SV * sv |
296 | CODE: |
297 | { |
298 | SV * to = items > 1 ? ST(1) : Nullsv; |
299 | SV * check = items > 2 ? ST(2) : Nullsv; |
300 | |
301 | if (to) { |
302 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
303 | } else { |
304 | STRLEN len; |
305 | U8 *s = (U8*)SvPV(sv, len); |
306 | |
307 | RETVAL = 0; |
308 | if (SvTRUE(check)) { |
309 | /* Must do things the slow way */ |
310 | U8 *dest; |
311 | /* We need a copy to pass to check() */ |
312 | U8 *src = (U8*)savepv((char *)s); |
313 | U8 *send = s + len; |
314 | |
315 | New(83, dest, len, U8); /* I think */ |
316 | |
317 | while (s < send) { |
318 | if (*s < 0x80){ |
319 | *dest++ = *s++; |
320 | } else { |
321 | STRLEN ulen; |
322 | UV uv = *s++; |
323 | |
324 | /* Have to do it all ourselves because of error routine, |
325 | aargh. */ |
326 | if (!(uv & 0x40)){ goto failure; } |
327 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
328 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
329 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
330 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
331 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
332 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
333 | else { ulen = 13; uv = 0; } |
87714904 |
334 | |
85982a32 |
335 | /* Note change to utf8.c variable naming, for variety */ |
336 | while (ulen--) { |
337 | if ((*s & 0xc0) != 0x80){ |
338 | goto failure; |
339 | } else { |
340 | uv = (uv << 6) | (*s++ & 0x3f); |
341 | } |
87714904 |
342 | } |
67e989fb |
343 | if (uv > 256) { |
344 | failure: |
85982a32 |
345 | call_failure(check, s, dest, src); |
346 | /* Now what happens? */ |
67e989fb |
347 | } |
348 | *dest++ = (U8)uv; |
85982a32 |
349 | } |
350 | } |
351 | } else { |
352 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
2c674647 |
353 | } |
85982a32 |
354 | } |
355 | } |
356 | OUTPUT: |
357 | RETVAL |
2c674647 |
358 | |
2c674647 |
359 | bool |
b2704119 |
360 | is_utf8(sv, check = 0) |
4411f3b6 |
361 | SV * sv |
b2704119 |
362 | int check |
85982a32 |
363 | CODE: |
364 | { |
365 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
366 | sv = newSVsv(sv); /* GMAGIG will be done */ |
367 | if (SvPOK(sv)) { |
368 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
369 | if (RETVAL && |
370 | check && |
371 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
2c674647 |
372 | RETVAL = FALSE; |
85982a32 |
373 | } else { |
374 | RETVAL = FALSE; |
375 | } |
376 | if (sv != ST(0)) |
377 | SvREFCNT_dec(sv); /* it was a temp copy */ |
378 | } |
379 | OUTPUT: |
380 | RETVAL |
2c674647 |
381 | |
382 | SV * |
4411f3b6 |
383 | _utf8_on(sv) |
85982a32 |
384 | SV * sv |
385 | CODE: |
386 | { |
387 | if (SvPOK(sv)) { |
388 | SV *rsv = newSViv(SvUTF8(sv)); |
389 | RETVAL = rsv; |
390 | SvUTF8_on(sv); |
391 | } else { |
392 | RETVAL = &PL_sv_undef; |
393 | } |
394 | } |
395 | OUTPUT: |
396 | RETVAL |
2c674647 |
397 | |
398 | SV * |
4411f3b6 |
399 | _utf8_off(sv) |
85982a32 |
400 | SV * sv |
401 | CODE: |
402 | { |
403 | if (SvPOK(sv)) { |
404 | SV *rsv = newSViv(SvUTF8(sv)); |
405 | RETVAL = rsv; |
406 | SvUTF8_off(sv); |
407 | } else { |
408 | RETVAL = &PL_sv_undef; |
409 | } |
410 | } |
411 | OUTPUT: |
412 | RETVAL |
413 | |
414 | PROTOTYPES: DISABLE |
415 | |
416 | |
417 | int |
418 | DIE_ON_ERR() |
419 | CODE: |
420 | RETVAL = ENCODE_DIE_ON_ERR; |
421 | OUTPUT: |
422 | RETVAL |
423 | |
424 | int |
425 | WARN_ON_ERR() |
426 | CODE: |
427 | RETVAL = ENCODE_WARN_ON_ERR; |
428 | OUTPUT: |
429 | RETVAL |
430 | |
431 | int |
432 | LEAVE_SRC() |
433 | CODE: |
434 | RETVAL = ENCODE_LEAVE_SRC; |
435 | OUTPUT: |
436 | RETVAL |
437 | |
438 | int |
439 | RETURN_ON_ERR() |
440 | CODE: |
441 | RETVAL = ENCODE_RETURN_ON_ERR; |
442 | OUTPUT: |
443 | RETVAL |
444 | |
445 | int |
446 | PERLQQ() |
447 | CODE: |
448 | RETVAL = ENCODE_PERLQQ; |
449 | OUTPUT: |
450 | RETVAL |
451 | |
452 | int |
453 | FB_DEFAULT() |
454 | CODE: |
455 | RETVAL = ENCODE_FB_DEFAULT; |
456 | OUTPUT: |
457 | RETVAL |
458 | |
459 | int |
460 | FB_CROAK() |
461 | CODE: |
462 | RETVAL = ENCODE_FB_CROAK; |
463 | OUTPUT: |
464 | RETVAL |
465 | |
466 | int |
467 | FB_QUIET() |
468 | CODE: |
469 | RETVAL = ENCODE_FB_QUIET; |
470 | OUTPUT: |
471 | RETVAL |
472 | |
473 | int |
474 | FB_WARN() |
475 | CODE: |
476 | RETVAL = ENCODE_FB_WARN; |
477 | OUTPUT: |
478 | RETVAL |
479 | |
480 | int |
481 | FB_PERLQQ() |
482 | CODE: |
483 | RETVAL = ENCODE_FB_PERLQQ; |
484 | OUTPUT: |
485 | RETVAL |
2c674647 |
486 | |
33af2bc7 |
487 | BOOT: |
488 | { |
85982a32 |
489 | #include "def_t.h" |
e7cbefb8 |
490 | #include "def_t.exh" |
33af2bc7 |
491 | } |