Commit | Line | Data |
85982a32 |
1 | /* |
b0b300a3 |
2 | $Id: Encode.xs,v 1.30 2002/04/20 09:58:23 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" |
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 */ |
b0b300a3 |
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 */ |
b0b300a3 |
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 | /**/ |
b0b300a3 |
25 | UNIMPLEMENTED(_encoded_utf8_to_bytes, I32); |
26 | UNIMPLEMENTED(_encoded_bytes_to_utf8, I32); |
33af2bc7 |
27 | |
b0b300a3 |
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 */ |
b0b300a3 |
122 | if (dir == enc->f_utf8) { |
85982a32 |
123 | STRLEN clen; |
124 | UV ch = |
b0b300a3 |
125 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
126 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
85982a32 |
127 | if (check & ENCODE_DIE_ON_ERR) { |
128 | Perl_croak( |
b0b300a3 |
129 | aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d", |
85982a32 |
130 | ch, enc->name[0], __LINE__); |
131 | }else{ |
132 | if (check & ENCODE_RETURN_ON_ERR){ |
133 | if (check & ENCODE_WARN_ON_ERR){ |
134 | Perl_warner( |
135 | aTHX_ packWARN(WARN_UTF8), |
b0b300a3 |
136 | "\"\\N{U+%" UVxf "}\" does not map to %s", |
85982a32 |
137 | ch,enc->name[0]); |
138 | } |
139 | goto ENCODE_SET_SRC; |
140 | }else if (check & ENCODE_PERLQQ){ |
b0b300a3 |
141 | SV* perlqq = |
85982a32 |
142 | sv_2mortal(newSVpvf("\\x{%04x}", ch)); |
b2704119 |
143 | sdone += slen + clen; |
85982a32 |
144 | ddone += dlen + SvCUR(perlqq); |
145 | sv_catsv(dst, perlqq); |
b0b300a3 |
146 | } else { |
85982a32 |
147 | /* fallback char */ |
148 | sdone += slen + clen; |
b0b300a3 |
149 | ddone += dlen + enc->replen; |
150 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
b2704119 |
151 | } |
b0b300a3 |
152 | } |
b2704119 |
153 | } |
85982a32 |
154 | /* decoding */ |
b0b300a3 |
155 | else { |
85982a32 |
156 | if (check & ENCODE_DIE_ON_ERR){ |
157 | Perl_croak( |
158 | aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)", |
159 | enc->name[0], (U8) s[slen], code); |
160 | }else{ |
161 | if (check & ENCODE_RETURN_ON_ERR){ |
162 | if (check & ENCODE_WARN_ON_ERR){ |
163 | Perl_warner( |
164 | aTHX_ packWARN(WARN_UTF8), |
165 | "%s \"\\x%02X\" does not map to Unicode (%d)", |
166 | enc->name[0], (U8) s[slen], code); |
167 | } |
168 | goto ENCODE_SET_SRC; |
169 | }else if (check & ENCODE_PERLQQ){ |
b0b300a3 |
170 | SV* perlqq = |
b2704119 |
171 | sv_2mortal(newSVpvf("\\x%02X", s[slen])); |
85982a32 |
172 | sdone += slen + 1; |
173 | ddone += dlen + SvCUR(perlqq); |
174 | sv_catsv(dst, perlqq); |
175 | } else { |
176 | sdone += slen + 1; |
b0b300a3 |
177 | ddone += dlen + strlen(FBCHAR_UTF8); |
178 | sv_catpv(dst, FBCHAR_UTF8); |
85982a32 |
179 | } |
aa0053b7 |
180 | } |
b2704119 |
181 | } |
85982a32 |
182 | /* settle variables when fallback */ |
b0b300a3 |
183 | d = (U8 *)SvEND(dst); |
184 | dlen = SvLEN(dst) - ddone - 1; |
185 | s = (U8*)SvPVX(src) + sdone; |
b2704119 |
186 | slen = tlen - sdone; |
187 | break; |
2f2b4ff2 |
188 | |
85982a32 |
189 | default: |
190 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
191 | code, (dir == enc->f_utf8) ? "to" : "from", |
192 | enc->name[0]); |
193 | return &PL_sv_undef; |
aa0053b7 |
194 | } |
85982a32 |
195 | } |
196 | ENCODE_SET_SRC: |
b0b300a3 |
197 | if (check & ~ENCODE_LEAVE_SRC){ |
85982a32 |
198 | sdone = SvCUR(src) - (slen+sdone); |
199 | if (sdone) { |
200 | sv_setpvn(src, (char*)s+slen, sdone); |
aa0053b7 |
201 | } |
85982a32 |
202 | SvCUR_set(src, sdone); |
2f2b4ff2 |
203 | } |
85982a32 |
204 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
205 | if (code && !(check & ENCODE_RETURN_ON_ERR)) { |
206 | return &PL_sv_undef; |
2f2b4ff2 |
207 | } |
b0b300a3 |
208 | |
85982a32 |
209 | SvCUR_set(dst, dlen+ddone); |
210 | SvPOK_only(dst); |
b0b300a3 |
211 | |
39cf9a5e |
212 | #if ENCODE_XS_PROFILE |
213 | if (SvCUR(dst) > SvCUR(src)){ |
85982a32 |
214 | Perl_warn(aTHX_ |
215 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
216 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
217 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
218 | } |
3c49ab08 |
219 | #endif |
b0b300a3 |
220 | |
85982a32 |
221 | ENCODE_END: |
0b3236bb |
222 | *SvEND(dst) = '\0'; |
aa0053b7 |
223 | return dst; |
2f2b4ff2 |
224 | } |
225 | |
50d26985 |
226 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
227 | |
228 | PROTOTYPES: ENABLE |
229 | |
230 | void |
0a95303c |
231 | Method_name(obj) |
232 | SV * obj |
233 | CODE: |
85982a32 |
234 | { |
235 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
236 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
237 | XSRETURN(1); |
238 | } |
0a95303c |
239 | |
240 | void |
b2704119 |
241 | Method_decode(obj,src,check = 0) |
2f2b4ff2 |
242 | SV * obj |
243 | SV * src |
b2704119 |
244 | int check |
2f2b4ff2 |
245 | CODE: |
aae85ceb |
246 | { |
85982a32 |
247 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
248 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check); |
249 | SvUTF8_on(ST(0)); |
aae85ceb |
250 | XSRETURN(1); |
251 | } |
252 | |
253 | void |
85982a32 |
254 | Method_encode(obj,src,check = 0) |
aae85ceb |
255 | SV * obj |
85982a32 |
256 | SV * src |
257 | int check |
aae85ceb |
258 | CODE: |
259 | { |
85982a32 |
260 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
261 | sv_utf8_upgrade(src); |
262 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check); |
aae85ceb |
263 | XSRETURN(1); |
264 | } |
265 | |
67e989fb |
266 | MODULE = Encode PACKAGE = Encode |
2c674647 |
267 | |
268 | PROTOTYPES: ENABLE |
269 | |
67e989fb |
270 | I32 |
2c674647 |
271 | _bytes_to_utf8(sv, ...) |
85982a32 |
272 | SV * sv |
273 | CODE: |
274 | { |
275 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
b0b300a3 |
276 | |
85982a32 |
277 | if (encoding) |
278 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
279 | else { |
280 | STRLEN len; |
281 | U8* s = (U8*)SvPV(sv, len); |
282 | U8* converted; |
283 | |
284 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
285 | sv_setpvn(sv, (char *)converted, len); |
286 | SvUTF8_on(sv); /* XXX Should we? */ |
287 | Safefree(converted); /* ... so free it */ |
288 | RETVAL = len; |
289 | } |
290 | } |
291 | OUTPUT: |
292 | RETVAL |
2c674647 |
293 | |
67e989fb |
294 | I32 |
2c674647 |
295 | _utf8_to_bytes(sv, ...) |
85982a32 |
296 | SV * sv |
297 | CODE: |
298 | { |
299 | SV * to = items > 1 ? ST(1) : Nullsv; |
300 | SV * check = items > 2 ? ST(2) : Nullsv; |
301 | |
302 | if (to) { |
303 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
304 | } else { |
305 | STRLEN len; |
306 | U8 *s = (U8*)SvPV(sv, len); |
307 | |
308 | RETVAL = 0; |
309 | if (SvTRUE(check)) { |
310 | /* Must do things the slow way */ |
311 | U8 *dest; |
312 | /* We need a copy to pass to check() */ |
b0b300a3 |
313 | U8 *src = (U8*)savepv((char *)s); |
85982a32 |
314 | U8 *send = s + len; |
315 | |
316 | New(83, dest, len, U8); /* I think */ |
317 | |
318 | while (s < send) { |
319 | if (*s < 0x80){ |
320 | *dest++ = *s++; |
321 | } else { |
322 | STRLEN ulen; |
323 | UV uv = *s++; |
324 | |
325 | /* Have to do it all ourselves because of error routine, |
326 | aargh. */ |
327 | if (!(uv & 0x40)){ goto failure; } |
328 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
329 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
330 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
331 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
332 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
333 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
334 | else { ulen = 13; uv = 0; } |
87714904 |
335 | |
85982a32 |
336 | /* Note change to utf8.c variable naming, for variety */ |
337 | while (ulen--) { |
b0b300a3 |
338 | if ((*s & 0xc0) != 0x80){ |
339 | goto failure; |
85982a32 |
340 | } else { |
341 | uv = (uv << 6) | (*s++ & 0x3f); |
342 | } |
87714904 |
343 | } |
67e989fb |
344 | if (uv > 256) { |
345 | failure: |
85982a32 |
346 | call_failure(check, s, dest, src); |
347 | /* Now what happens? */ |
67e989fb |
348 | } |
349 | *dest++ = (U8)uv; |
85982a32 |
350 | } |
351 | } |
352 | } else { |
353 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
2c674647 |
354 | } |
85982a32 |
355 | } |
356 | } |
357 | OUTPUT: |
358 | RETVAL |
2c674647 |
359 | |
2c674647 |
360 | bool |
b2704119 |
361 | is_utf8(sv, check = 0) |
4411f3b6 |
362 | SV * sv |
b2704119 |
363 | int check |
85982a32 |
364 | CODE: |
365 | { |
366 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
367 | sv = newSVsv(sv); /* GMAGIG will be done */ |
368 | if (SvPOK(sv)) { |
369 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
370 | if (RETVAL && |
371 | check && |
372 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
2c674647 |
373 | RETVAL = FALSE; |
85982a32 |
374 | } else { |
375 | RETVAL = FALSE; |
376 | } |
377 | if (sv != ST(0)) |
378 | SvREFCNT_dec(sv); /* it was a temp copy */ |
379 | } |
380 | OUTPUT: |
381 | RETVAL |
2c674647 |
382 | |
383 | SV * |
4411f3b6 |
384 | _utf8_on(sv) |
85982a32 |
385 | SV * sv |
386 | CODE: |
387 | { |
388 | if (SvPOK(sv)) { |
389 | SV *rsv = newSViv(SvUTF8(sv)); |
390 | RETVAL = rsv; |
391 | SvUTF8_on(sv); |
392 | } else { |
393 | RETVAL = &PL_sv_undef; |
394 | } |
395 | } |
396 | OUTPUT: |
397 | RETVAL |
2c674647 |
398 | |
399 | SV * |
4411f3b6 |
400 | _utf8_off(sv) |
85982a32 |
401 | SV * sv |
402 | CODE: |
403 | { |
404 | if (SvPOK(sv)) { |
405 | SV *rsv = newSViv(SvUTF8(sv)); |
406 | RETVAL = rsv; |
407 | SvUTF8_off(sv); |
408 | } else { |
409 | RETVAL = &PL_sv_undef; |
410 | } |
411 | } |
412 | OUTPUT: |
413 | RETVAL |
414 | |
415 | PROTOTYPES: DISABLE |
416 | |
417 | |
418 | int |
419 | DIE_ON_ERR() |
420 | CODE: |
421 | RETVAL = ENCODE_DIE_ON_ERR; |
422 | OUTPUT: |
423 | RETVAL |
424 | |
b0b300a3 |
425 | int |
85982a32 |
426 | WARN_ON_ERR() |
427 | CODE: |
428 | RETVAL = ENCODE_WARN_ON_ERR; |
429 | OUTPUT: |
430 | RETVAL |
431 | |
432 | int |
433 | LEAVE_SRC() |
434 | CODE: |
435 | RETVAL = ENCODE_LEAVE_SRC; |
436 | OUTPUT: |
437 | RETVAL |
438 | |
439 | int |
440 | RETURN_ON_ERR() |
441 | CODE: |
442 | RETVAL = ENCODE_RETURN_ON_ERR; |
443 | OUTPUT: |
444 | RETVAL |
445 | |
446 | int |
447 | PERLQQ() |
448 | CODE: |
449 | RETVAL = ENCODE_PERLQQ; |
450 | OUTPUT: |
451 | RETVAL |
452 | |
453 | int |
454 | FB_DEFAULT() |
455 | CODE: |
456 | RETVAL = ENCODE_FB_DEFAULT; |
457 | OUTPUT: |
458 | RETVAL |
459 | |
460 | int |
461 | FB_CROAK() |
462 | CODE: |
463 | RETVAL = ENCODE_FB_CROAK; |
464 | OUTPUT: |
465 | RETVAL |
466 | |
467 | int |
468 | FB_QUIET() |
469 | CODE: |
470 | RETVAL = ENCODE_FB_QUIET; |
471 | OUTPUT: |
472 | RETVAL |
473 | |
474 | int |
475 | FB_WARN() |
476 | CODE: |
477 | RETVAL = ENCODE_FB_WARN; |
478 | OUTPUT: |
479 | RETVAL |
480 | |
481 | int |
482 | FB_PERLQQ() |
483 | CODE: |
484 | RETVAL = ENCODE_FB_PERLQQ; |
485 | OUTPUT: |
486 | RETVAL |
2c674647 |
487 | |
33af2bc7 |
488 | BOOT: |
489 | { |
85982a32 |
490 | #include "def_t.h" |
e7cbefb8 |
491 | #include "def_t.exh" |
33af2bc7 |
492 | } |