Commit | Line | Data |
85982a32 |
1 | /* |
370462a2 |
2 | $Id: Encode.xs,v 2.14 2007/05/29 18:15:32 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"); \ |
d1256cb1 |
25 | return (y)0; /* fool picky compilers */ \ |
87714904 |
26 | } |
85982a32 |
27 | /**/ |
011b2d2f |
28 | |
7e9a885a |
29 | UNIMPLEMENTED(_encoded_utf8_to_bytes, I32) |
30 | UNIMPLEMENTED(_encoded_bytes_to_utf8, I32) |
33af2bc7 |
31 | |
7f0d54d7 |
32 | #define UTF8_ALLOW_STRICT 0 |
33 | #define UTF8_ALLOW_NONSTRICT (UTF8_ALLOW_ANY & \ |
34 | ~(UTF8_ALLOW_CONTINUATION | \ |
35 | UTF8_ALLOW_NON_CONTINUATION | \ |
36 | UTF8_ALLOW_LONG)) |
37 | |
b0b300a3 |
38 | void |
aa0053b7 |
39 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
40 | { |
aa0053b7 |
41 | dSP; |
42 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
43 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
44 | int i = 0; |
45 | PUSHMARK(sp); |
46 | XPUSHs(sv); |
47 | while (enc->name[i]) { |
d1256cb1 |
48 | const char *name = enc->name[i++]; |
49 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
aa0053b7 |
50 | } |
51 | PUTBACK; |
52 | call_pv("Encode::define_encoding", G_DISCARD); |
53 | SvREFCNT_dec(sv); |
2f2b4ff2 |
54 | } |
55 | |
aa0053b7 |
56 | void |
57 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
58 | { |
85982a32 |
59 | /* Exists for breakpointing */ |
aa0053b7 |
60 | } |
67e989fb |
61 | |
85982a32 |
62 | |
2fc614e0 |
63 | #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s" |
64 | #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode" |
65 | |
2f2b4ff2 |
66 | static SV * |
370462a2 |
67 | do_fallback_cb(pTHX_ UV ch, SV *fallback_cb) |
8e180e82 |
68 | { |
69 | dSP; |
70 | int argc; |
7828f908 |
71 | SV *temp, *retval; |
8e180e82 |
72 | ENTER; |
73 | SAVETMPS; |
74 | PUSHMARK(sp); |
75 | XPUSHs(sv_2mortal(newSVnv((UV)ch))); |
76 | PUTBACK; |
77 | argc = call_sv(fallback_cb, G_SCALAR); |
78 | SPAGAIN; |
79 | if (argc != 1){ |
370462a2 |
80 | croak("fallback sub must return scalar!"); |
8e180e82 |
81 | } |
7828f908 |
82 | temp = newSVsv(POPs); |
8e180e82 |
83 | PUTBACK; |
84 | FREETMPS; |
85 | LEAVE; |
7828f908 |
86 | retval = newSVpv("",0); |
87 | sv_catsv(retval, temp); |
6ead8143 |
88 | SvREFCNT_dec(temp); |
8e180e82 |
89 | return retval; |
90 | } |
91 | |
92 | static SV * |
0629a5b3 |
93 | encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src, |
370462a2 |
94 | int check, STRLEN * offset, SV * term, int * retcode, |
95 | SV *fallback_cb) |
2f2b4ff2 |
96 | { |
aa0053b7 |
97 | STRLEN slen; |
98 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
99 | STRLEN tlen = slen; |
100 | STRLEN ddone = 0; |
101 | STRLEN sdone = 0; |
39cf9a5e |
102 | |
3c49ab08 |
103 | /* We allocate slen+1. |
85982a32 |
104 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
105 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
106 | U8 *d = (U8 *)SvPVX(dst); |
107 | STRLEN dlen = SvLEN(dst)-1; |
220e2d4e |
108 | int code = 0; |
109 | STRLEN trmlen = 0; |
cc7dbc11 |
110 | U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL; |
220e2d4e |
111 | |
112 | if (offset) { |
113 | s += *offset; |
6be7c101 |
114 | if (slen > *offset){ /* safeguard against slen overflow */ |
d1256cb1 |
115 | slen -= *offset; |
6be7c101 |
116 | }else{ |
d1256cb1 |
117 | slen = 0; |
6be7c101 |
118 | } |
220e2d4e |
119 | tlen = slen; |
120 | } |
85982a32 |
121 | |
6be7c101 |
122 | if (slen == 0){ |
d1256cb1 |
123 | SvCUR_set(dst, 0); |
124 | SvPOK_only(dst); |
125 | goto ENCODE_END; |
85982a32 |
126 | } |
127 | |
220e2d4e |
128 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
d1256cb1 |
129 | trm, trmlen)) ) |
85982a32 |
130 | { |
d1256cb1 |
131 | SvCUR_set(dst, dlen+ddone); |
132 | SvPOK_only(dst); |
133 | |
134 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
135 | code == ENCODE_FOUND_TERM) { |
136 | break; |
137 | } |
138 | switch (code) { |
139 | case ENCODE_NOSPACE: |
140 | { |
141 | STRLEN more = 0; /* make sure you initialize! */ |
142 | STRLEN sleft; |
143 | sdone += slen; |
144 | ddone += dlen; |
145 | sleft = tlen - sdone; |
fcb875d4 |
146 | #if ENCODE_XS_PROFILE >= 2 |
d1256cb1 |
147 | Perl_warn(aTHX_ |
148 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
149 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
150 | #endif |
d1256cb1 |
151 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
152 | #if ENCODE_XS_USEFP == 2 |
d1256cb1 |
153 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
154 | - SvLEN(dst); |
39cf9a5e |
155 | #elif ENCODE_XS_USEFP |
d1256cb1 |
156 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
157 | #else |
d1256cb1 |
158 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
159 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
160 | #endif |
d1256cb1 |
161 | } |
162 | more += UTF8_MAXLEN; /* insurance policy */ |
163 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
164 | /* dst need to grow need MORE bytes! */ |
165 | if (ddone >= SvLEN(dst)) { |
166 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
167 | } |
168 | dlen = SvLEN(dst)-ddone-1; |
169 | d += ddone; |
170 | s += slen; |
171 | slen = tlen-sdone; |
172 | continue; |
173 | } |
174 | case ENCODE_NOREP: |
175 | /* encoding */ |
176 | if (dir == enc->f_utf8) { |
177 | STRLEN clen; |
178 | UV ch = |
179 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
180 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
181 | /* if non-representable multibyte prefix at end of current buffer - break*/ |
182 | if (clen > tlen - sdone) break; |
183 | if (check & ENCODE_DIE_ON_ERR) { |
184 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
185 | (UV)ch, enc->name[0]); |
186 | return &PL_sv_undef; /* never reaches but be safe */ |
187 | } |
188 | if (check & ENCODE_WARN_ON_ERR){ |
189 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
190 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
191 | } |
192 | if (check & ENCODE_RETURN_ON_ERR){ |
193 | goto ENCODE_SET_SRC; |
194 | } |
195 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
196 | SV* subchar = |
370462a2 |
197 | (fallback_cb != &PL_sv_undef) |
198 | ? do_fallback_cb(aTHX_ ch, fallback_cb) |
199 | : newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" : |
d1256cb1 |
200 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
201 | "&#x%" UVxf ";", (UV)ch); |
202 | sdone += slen + clen; |
203 | ddone += dlen + SvCUR(subchar); |
204 | sv_catsv(dst, subchar); |
205 | SvREFCNT_dec(subchar); |
206 | } else { |
207 | /* fallback char */ |
208 | sdone += slen + clen; |
209 | ddone += dlen + enc->replen; |
210 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
211 | } |
212 | } |
213 | /* decoding */ |
214 | else { |
215 | if (check & ENCODE_DIE_ON_ERR){ |
216 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
217 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
218 | return &PL_sv_undef; /* never reaches but be safe */ |
219 | } |
220 | if (check & ENCODE_WARN_ON_ERR){ |
221 | Perl_warner( |
222 | aTHX_ packWARN(WARN_UTF8), |
223 | ERR_DECODE_NOMAP, |
436c6dd3 |
224 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
225 | } |
226 | if (check & ENCODE_RETURN_ON_ERR){ |
227 | goto ENCODE_SET_SRC; |
228 | } |
229 | if (check & |
230 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
231 | SV* subchar = |
370462a2 |
232 | (fallback_cb != &PL_sv_undef) |
233 | ? do_fallback_cb(aTHX_ (UV)s[slen], fallback_cb) |
234 | : newSVpvf("\\x%02" UVXf, (UV)s[slen]); |
d1256cb1 |
235 | sdone += slen + 1; |
236 | ddone += dlen + SvCUR(subchar); |
237 | sv_catsv(dst, subchar); |
238 | SvREFCNT_dec(subchar); |
239 | } else { |
240 | sdone += slen + 1; |
241 | ddone += dlen + strlen(FBCHAR_UTF8); |
242 | sv_catpv(dst, FBCHAR_UTF8); |
243 | } |
244 | } |
245 | /* settle variables when fallback */ |
246 | d = (U8 *)SvEND(dst); |
b0b300a3 |
247 | dlen = SvLEN(dst) - ddone - 1; |
d1256cb1 |
248 | s = (U8*)SvPVX(src) + sdone; |
249 | slen = tlen - sdone; |
250 | break; |
251 | |
252 | default: |
253 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
254 | code, (dir == enc->f_utf8) ? "to" : "from", |
255 | enc->name[0]); |
256 | return &PL_sv_undef; |
257 | } |
85982a32 |
258 | } |
259 | ENCODE_SET_SRC: |
ca777f1c |
260 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
261 | sdone = SvCUR(src) - (slen+sdone); |
262 | if (sdone) { |
263 | sv_setpvn(src, (char*)s+slen, sdone); |
264 | } |
265 | SvCUR_set(src, sdone); |
2f2b4ff2 |
266 | } |
85982a32 |
267 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
268 | |
85982a32 |
269 | SvCUR_set(dst, dlen+ddone); |
270 | SvPOK_only(dst); |
c6a7db43 |
271 | |
39cf9a5e |
272 | #if ENCODE_XS_PROFILE |
273 | if (SvCUR(dst) > SvCUR(src)){ |
d1256cb1 |
274 | Perl_warn(aTHX_ |
275 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
276 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
277 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
278 | } |
3c49ab08 |
279 | #endif |
c6a7db43 |
280 | |
220e2d4e |
281 | if (offset) |
282 | *offset += sdone + slen; |
283 | |
85982a32 |
284 | ENCODE_END: |
0b3236bb |
285 | *SvEND(dst) = '\0'; |
220e2d4e |
286 | if (retcode) *retcode = code; |
aa0053b7 |
287 | return dst; |
2f2b4ff2 |
288 | } |
289 | |
7f0d54d7 |
290 | static bool |
291 | strict_utf8(pTHX_ SV* sv) |
292 | { |
293 | HV* hv; |
294 | SV** svp; |
295 | sv = SvRV(sv); |
296 | if (!sv || SvTYPE(sv) != SVt_PVHV) |
297 | return 0; |
298 | hv = (HV*)sv; |
299 | svp = hv_fetch(hv, "strict_utf8", 11, 0); |
300 | if (!svp) |
301 | return 0; |
302 | return SvTRUE(*svp); |
303 | } |
304 | |
305 | static U8* |
306 | process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check, |
307 | bool encode, bool strict, bool stop_at_partial) |
308 | { |
309 | UV uv; |
310 | STRLEN ulen; |
311 | |
312 | SvPOK_only(dst); |
313 | SvCUR_set(dst,0); |
314 | |
315 | while (s < e) { |
316 | if (UTF8_IS_INVARIANT(*s)) { |
317 | sv_catpvn(dst, (char *)s, 1); |
318 | s++; |
319 | continue; |
320 | } |
321 | |
322 | if (UTF8_IS_START(*s)) { |
323 | U8 skip = UTF8SKIP(s); |
324 | if ((s + skip) > e) { |
325 | /* Partial character */ |
326 | /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */ |
56ff7374 |
327 | if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL)) |
7f0d54d7 |
328 | break; |
329 | |
330 | goto malformed_byte; |
331 | } |
332 | |
333 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
334 | UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT : |
335 | UTF8_ALLOW_NONSTRICT) |
336 | ); |
337 | #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */ |
d1256cb1 |
338 | if (strict && uv > PERL_UNICODE_MAX) |
1b9a72a9 |
339 | ulen = (STRLEN) -1; |
7f0d54d7 |
340 | #endif |
341 | if (ulen == -1) { |
342 | if (strict) { |
343 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
344 | UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT); |
345 | if (ulen == -1) |
346 | goto malformed_byte; |
347 | goto malformed; |
348 | } |
349 | goto malformed_byte; |
350 | } |
351 | |
352 | |
353 | /* Whole char is good */ |
354 | sv_catpvn(dst,(char *)s,skip); |
355 | s += skip; |
356 | continue; |
357 | } |
358 | |
359 | /* If we get here there is something wrong with alleged UTF-8 */ |
360 | malformed_byte: |
361 | uv = (UV)*s; |
362 | ulen = 1; |
363 | |
364 | malformed: |
365 | if (check & ENCODE_DIE_ON_ERR){ |
366 | if (encode) |
367 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8"); |
368 | else |
369 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv); |
370 | } |
371 | if (check & ENCODE_WARN_ON_ERR){ |
372 | if (encode) |
373 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
374 | ERR_ENCODE_NOMAP, uv, "utf8"); |
375 | else |
376 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
377 | ERR_DECODE_NOMAP, "utf8", uv); |
378 | } |
379 | if (check & ENCODE_RETURN_ON_ERR) { |
380 | break; |
381 | } |
382 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
383 | SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"): |
384 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
385 | "&#x%" UVxf ";", uv); |
386 | sv_catsv(dst, subchar); |
387 | SvREFCNT_dec(subchar); |
388 | } else { |
389 | sv_catpv(dst, FBCHAR_UTF8); |
390 | } |
391 | s += ulen; |
392 | } |
393 | *SvEND(dst) = '\0'; |
394 | |
395 | return s; |
396 | } |
397 | |
398 | |
ab3374e4 |
399 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
400 | |
a0d8a30e |
401 | PROTOTYPES: DISABLE |
402 | |
403 | void |
b536bf57 |
404 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
405 | SV * obj |
406 | SV * src |
407 | int check |
408 | CODE: |
409 | { |
410 | STRLEN slen; |
411 | U8 *s = (U8 *) SvPV(src, slen); |
412 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
413 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
414 | |
415 | /* |
7f0d54d7 |
416 | * PerlIO check -- we assume the object is of PerlIO if renewed |
cc836e95 |
417 | */ |
f5a22bee |
418 | bool renewed = 0; |
cc836e95 |
419 | dSP; ENTER; SAVETMPS; |
420 | PUSHMARK(sp); |
421 | XPUSHs(obj); |
422 | PUTBACK; |
423 | if (call_method("renewed",G_SCALAR) == 1) { |
d1256cb1 |
424 | SPAGAIN; |
425 | renewed = (bool)POPi; |
426 | PUTBACK; |
cc836e95 |
427 | #if 0 |
d1256cb1 |
428 | fprintf(stderr, "renewed == %d\n", renewed); |
cc836e95 |
429 | #endif |
cc836e95 |
430 | } |
431 | FREETMPS; LEAVE; |
432 | /* end PerlIO check */ |
433 | |
ab3374e4 |
434 | if (SvUTF8(src)) { |
d1256cb1 |
435 | s = utf8_to_bytes(s,&slen); |
436 | if (s) { |
437 | SvCUR_set(src,slen); |
438 | SvUTF8_off(src); |
439 | e = s+slen; |
440 | } |
441 | else { |
442 | croak("Cannot decode string with wide characters"); |
443 | } |
ab3374e4 |
444 | } |
7f0d54d7 |
445 | |
446 | s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed); |
ab3374e4 |
447 | |
448 | /* Clear out translated part of source unless asked not to */ |
449 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
450 | slen = e-s; |
451 | if (slen) { |
452 | sv_setpvn(src, (char*)s, slen); |
453 | } |
454 | SvCUR_set(src, slen); |
ab3374e4 |
455 | } |
456 | SvUTF8_on(dst); |
457 | ST(0) = sv_2mortal(dst); |
458 | XSRETURN(1); |
459 | } |
460 | |
461 | void |
b536bf57 |
462 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
463 | SV * obj |
464 | SV * src |
465 | int check |
466 | CODE: |
467 | { |
468 | STRLEN slen; |
469 | U8 *s = (U8 *) SvPV(src, slen); |
470 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
471 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
472 | if (SvUTF8(src)) { |
d1256cb1 |
473 | /* Already encoded */ |
474 | if (strict_utf8(aTHX_ obj)) { |
475 | s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0); |
476 | } |
7f0d54d7 |
477 | else { |
478 | /* trust it and just copy the octets */ |
479 | sv_setpvn(dst,(char *)s,(e-s)); |
d1256cb1 |
480 | s = e; |
7f0d54d7 |
481 | } |
ab3374e4 |
482 | } |
483 | else { |
484 | /* Native bytes - can always encode */ |
d1256cb1 |
485 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
486 | while (s < e) { |
cf0d1c66 |
487 | UV uv = NATIVE_TO_UNI((UV) *s); |
488 | s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */ |
ab3374e4 |
489 | if (UNI_IS_INVARIANT(uv)) |
490 | *d++ = (U8)UTF_TO_NATIVE(uv); |
491 | else { |
492 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
493 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
494 | } |
d1256cb1 |
495 | } |
ab3374e4 |
496 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
497 | *SvEND(dst) = '\0'; |
498 | } |
499 | |
500 | /* Clear out translated part of source unless asked not to */ |
501 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
502 | slen = e-s; |
503 | if (slen) { |
504 | sv_setpvn(src, (char*)s, slen); |
505 | } |
506 | SvCUR_set(src, slen); |
ab3374e4 |
507 | } |
508 | SvPOK_only(dst); |
509 | SvUTF8_off(dst); |
510 | ST(0) = sv_2mortal(dst); |
511 | XSRETURN(1); |
512 | } |
513 | |
50d26985 |
514 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
515 | |
516 | PROTOTYPES: ENABLE |
517 | |
518 | void |
a0d8a30e |
519 | Method_renew(obj) |
520 | SV * obj |
521 | CODE: |
522 | { |
523 | XSRETURN(1); |
524 | } |
525 | |
cc836e95 |
526 | int |
527 | Method_renewed(obj) |
528 | SV * obj |
529 | CODE: |
530 | RETVAL = 0; |
531 | OUTPUT: |
532 | RETVAL |
533 | |
a0d8a30e |
534 | void |
0a95303c |
535 | Method_name(obj) |
536 | SV * obj |
537 | CODE: |
85982a32 |
538 | { |
539 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
540 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
541 | XSRETURN(1); |
542 | } |
0a95303c |
543 | |
544 | void |
370462a2 |
545 | Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no) |
220e2d4e |
546 | SV * obj |
547 | SV * dst |
548 | SV * src |
549 | SV * off |
550 | SV * term |
370462a2 |
551 | SV * check_sv |
220e2d4e |
552 | CODE: |
553 | { |
370462a2 |
554 | int check; |
555 | SV *fallback_cb = &PL_sv_undef; |
220e2d4e |
556 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
557 | STRLEN offset = (STRLEN)SvIV(off); |
558 | int code = 0; |
559 | if (SvUTF8(src)) { |
560 | sv_utf8_downgrade(src, FALSE); |
561 | } |
370462a2 |
562 | if (SvROK(check_sv)){ |
563 | fallback_cb = check_sv; |
564 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
565 | }else{ |
566 | check = SvIV(check_sv); |
567 | } |
220e2d4e |
568 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
370462a2 |
569 | &offset, term, &code, fallback_cb)); |
b162af07 |
570 | SvIV_set(off, (IV)offset); |
220e2d4e |
571 | if (code == ENCODE_FOUND_TERM) { |
d1256cb1 |
572 | ST(0) = &PL_sv_yes; |
220e2d4e |
573 | }else{ |
d1256cb1 |
574 | ST(0) = &PL_sv_no; |
220e2d4e |
575 | } |
576 | XSRETURN(1); |
577 | } |
578 | |
579 | void |
8e180e82 |
580 | Method_decode(obj,src,check_sv = &PL_sv_no) |
2f2b4ff2 |
581 | SV * obj |
582 | SV * src |
8e180e82 |
583 | SV * check_sv |
2f2b4ff2 |
584 | CODE: |
aae85ceb |
585 | { |
8e180e82 |
586 | int check; |
370462a2 |
587 | SV *fallback_cb = &PL_sv_undef; |
85982a32 |
588 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
589 | if (SvUTF8(src)) { |
590 | sv_utf8_downgrade(src, FALSE); |
591 | } |
8e180e82 |
592 | if (SvROK(check_sv)){ |
370462a2 |
593 | fallback_cb = check_sv; |
594 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
595 | }else{ |
370462a2 |
596 | check = SvIV(check_sv); |
8e180e82 |
597 | } |
220e2d4e |
598 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
370462a2 |
599 | NULL, Nullsv, NULL, fallback_cb); |
85982a32 |
600 | SvUTF8_on(ST(0)); |
aae85ceb |
601 | XSRETURN(1); |
602 | } |
603 | |
604 | void |
8e180e82 |
605 | Method_encode(obj,src,check_sv = &PL_sv_no) |
aae85ceb |
606 | SV * obj |
85982a32 |
607 | SV * src |
8e180e82 |
608 | SV * check_sv |
aae85ceb |
609 | CODE: |
610 | { |
8e180e82 |
611 | int check; |
370462a2 |
612 | SV *fallback_cb = &PL_sv_undef; |
85982a32 |
613 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
614 | sv_utf8_upgrade(src); |
8e180e82 |
615 | if (SvROK(check_sv)){ |
370462a2 |
616 | fallback_cb = check_sv; |
617 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
618 | }else{ |
370462a2 |
619 | check = SvIV(check_sv); |
8e180e82 |
620 | } |
220e2d4e |
621 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
370462a2 |
622 | NULL, Nullsv, NULL, fallback_cb); |
aae85ceb |
623 | XSRETURN(1); |
624 | } |
625 | |
0ab8f81e |
626 | void |
627 | Method_needs_lines(obj) |
628 | SV * obj |
629 | CODE: |
630 | { |
b32afa7c |
631 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
632 | ST(0) = &PL_sv_no; |
633 | XSRETURN(1); |
634 | } |
635 | |
636 | void |
637 | Method_perlio_ok(obj) |
638 | SV * obj |
639 | CODE: |
640 | { |
b32afa7c |
641 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
642 | /* require_pv(PERLIO_FILENAME); */ |
643 | |
644 | eval_pv("require PerlIO::encoding", 0); |
645 | |
646 | if (SvTRUE(get_sv("@", 0))) { |
d1256cb1 |
647 | ST(0) = &PL_sv_no; |
10c5ecbb |
648 | }else{ |
d1256cb1 |
649 | ST(0) = &PL_sv_yes; |
0ab8f81e |
650 | } |
651 | XSRETURN(1); |
652 | } |
653 | |
742555bd |
654 | void |
655 | Method_mime_name(obj) |
656 | SV * obj |
657 | CODE: |
658 | { |
659 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
660 | SV *retval; |
661 | eval_pv("require Encode::MIME::Name", 0); |
662 | |
663 | if (SvTRUE(get_sv("@", 0))) { |
664 | ST(0) = &PL_sv_undef; |
665 | }else{ |
666 | ENTER; |
667 | SAVETMPS; |
668 | PUSHMARK(sp); |
669 | XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0])))); |
670 | PUTBACK; |
671 | call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR); |
672 | SPAGAIN; |
673 | retval = newSVsv(POPs); |
674 | PUTBACK; |
675 | FREETMPS; |
676 | LEAVE; |
677 | /* enc->name[0] */ |
678 | ST(0) = retval; |
679 | } |
680 | XSRETURN(1); |
681 | } |
682 | |
67e989fb |
683 | MODULE = Encode PACKAGE = Encode |
2c674647 |
684 | |
685 | PROTOTYPES: ENABLE |
686 | |
67e989fb |
687 | I32 |
2c674647 |
688 | _bytes_to_utf8(sv, ...) |
85982a32 |
689 | SV * sv |
690 | CODE: |
691 | { |
692 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
693 | |
85982a32 |
694 | if (encoding) |
695 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
696 | else { |
d1256cb1 |
697 | STRLEN len; |
698 | U8* s = (U8*)SvPV(sv, len); |
699 | U8* converted; |
700 | |
701 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
702 | sv_setpvn(sv, (char *)converted, len); |
703 | SvUTF8_on(sv); /* XXX Should we? */ |
704 | Safefree(converted); /* ... so free it */ |
705 | RETVAL = len; |
85982a32 |
706 | } |
707 | } |
708 | OUTPUT: |
709 | RETVAL |
2c674647 |
710 | |
67e989fb |
711 | I32 |
2c674647 |
712 | _utf8_to_bytes(sv, ...) |
85982a32 |
713 | SV * sv |
714 | CODE: |
715 | { |
716 | SV * to = items > 1 ? ST(1) : Nullsv; |
717 | SV * check = items > 2 ? ST(2) : Nullsv; |
718 | |
719 | if (to) { |
d1256cb1 |
720 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
85982a32 |
721 | } else { |
d1256cb1 |
722 | STRLEN len; |
723 | U8 *s = (U8*)SvPV(sv, len); |
85982a32 |
724 | |
d1256cb1 |
725 | RETVAL = 0; |
726 | if (SvTRUE(check)) { |
727 | /* Must do things the slow way */ |
728 | U8 *dest; |
85982a32 |
729 | /* We need a copy to pass to check() */ |
2cdf8b6c |
730 | U8 *src = s; |
d1256cb1 |
731 | U8 *send = s + len; |
2cdf8b6c |
732 | U8 *d0; |
85982a32 |
733 | |
d1256cb1 |
734 | New(83, dest, len, U8); /* I think */ |
2cdf8b6c |
735 | d0 = dest; |
85982a32 |
736 | |
d1256cb1 |
737 | while (s < send) { |
85982a32 |
738 | if (*s < 0x80){ |
d1256cb1 |
739 | *dest++ = *s++; |
85982a32 |
740 | } else { |
d1256cb1 |
741 | STRLEN ulen; |
742 | UV uv = *s++; |
743 | |
744 | /* Have to do it all ourselves because of error routine, |
745 | aargh. */ |
746 | if (!(uv & 0x40)){ goto failure; } |
747 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
748 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
749 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
750 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
751 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
752 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
753 | else { ulen = 13; uv = 0; } |
754 | |
755 | /* Note change to utf8.c variable naming, for variety */ |
756 | while (ulen--) { |
757 | if ((*s & 0xc0) != 0x80){ |
758 | goto failure; |
759 | } else { |
760 | uv = (uv << 6) | (*s++ & 0x3f); |
761 | } |
762 | } |
763 | if (uv > 256) { |
764 | failure: |
765 | call_failure(check, s, dest, src); |
766 | /* Now what happens? */ |
767 | } |
768 | *dest++ = (U8)uv; |
769 | } |
770 | } |
2cdf8b6c |
771 | RETVAL = dest - d0; |
772 | sv_usepvn(sv, (char *)dest, RETVAL); |
773 | SvUTF8_off(sv); |
d1256cb1 |
774 | } else { |
775 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
776 | } |
85982a32 |
777 | } |
778 | } |
779 | OUTPUT: |
780 | RETVAL |
2c674647 |
781 | |
2c674647 |
782 | bool |
b2704119 |
783 | is_utf8(sv, check = 0) |
4411f3b6 |
784 | SV * sv |
b2704119 |
785 | int check |
85982a32 |
786 | CODE: |
787 | { |
788 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
d1256cb1 |
789 | sv = newSVsv(sv); /* GMAGIG will be done */ |
d1256cb1 |
790 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
791 | if (RETVAL && |
792 | check && |
793 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
794 | RETVAL = FALSE; |
85982a32 |
795 | if (sv != ST(0)) |
d1256cb1 |
796 | SvREFCNT_dec(sv); /* it was a temp copy */ |
85982a32 |
797 | } |
798 | OUTPUT: |
799 | RETVAL |
2c674647 |
800 | |
801 | SV * |
4411f3b6 |
802 | _utf8_on(sv) |
85982a32 |
803 | SV * sv |
804 | CODE: |
805 | { |
806 | if (SvPOK(sv)) { |
d1256cb1 |
807 | SV *rsv = newSViv(SvUTF8(sv)); |
808 | RETVAL = rsv; |
809 | SvUTF8_on(sv); |
85982a32 |
810 | } else { |
d1256cb1 |
811 | RETVAL = &PL_sv_undef; |
85982a32 |
812 | } |
813 | } |
814 | OUTPUT: |
815 | RETVAL |
2c674647 |
816 | |
817 | SV * |
4411f3b6 |
818 | _utf8_off(sv) |
85982a32 |
819 | SV * sv |
820 | CODE: |
821 | { |
822 | if (SvPOK(sv)) { |
d1256cb1 |
823 | SV *rsv = newSViv(SvUTF8(sv)); |
824 | RETVAL = rsv; |
825 | SvUTF8_off(sv); |
85982a32 |
826 | } else { |
d1256cb1 |
827 | RETVAL = &PL_sv_undef; |
85982a32 |
828 | } |
829 | } |
830 | OUTPUT: |
831 | RETVAL |
832 | |
85982a32 |
833 | int |
834 | DIE_ON_ERR() |
835 | CODE: |
836 | RETVAL = ENCODE_DIE_ON_ERR; |
837 | OUTPUT: |
838 | RETVAL |
839 | |
c6a7db43 |
840 | int |
85982a32 |
841 | WARN_ON_ERR() |
842 | CODE: |
843 | RETVAL = ENCODE_WARN_ON_ERR; |
844 | OUTPUT: |
845 | RETVAL |
846 | |
847 | int |
848 | LEAVE_SRC() |
849 | CODE: |
850 | RETVAL = ENCODE_LEAVE_SRC; |
851 | OUTPUT: |
852 | RETVAL |
853 | |
854 | int |
855 | RETURN_ON_ERR() |
856 | CODE: |
857 | RETVAL = ENCODE_RETURN_ON_ERR; |
858 | OUTPUT: |
859 | RETVAL |
860 | |
861 | int |
862 | PERLQQ() |
863 | CODE: |
864 | RETVAL = ENCODE_PERLQQ; |
865 | OUTPUT: |
866 | RETVAL |
867 | |
868 | int |
af1f55d9 |
869 | HTMLCREF() |
870 | CODE: |
871 | RETVAL = ENCODE_HTMLCREF; |
872 | OUTPUT: |
873 | RETVAL |
874 | |
875 | int |
876 | XMLCREF() |
877 | CODE: |
878 | RETVAL = ENCODE_XMLCREF; |
879 | OUTPUT: |
880 | RETVAL |
881 | |
882 | int |
56ff7374 |
883 | STOP_AT_PARTIAL() |
884 | CODE: |
885 | RETVAL = ENCODE_STOP_AT_PARTIAL; |
886 | OUTPUT: |
887 | RETVAL |
888 | |
889 | int |
85982a32 |
890 | FB_DEFAULT() |
891 | CODE: |
892 | RETVAL = ENCODE_FB_DEFAULT; |
893 | OUTPUT: |
894 | RETVAL |
895 | |
896 | int |
897 | FB_CROAK() |
898 | CODE: |
899 | RETVAL = ENCODE_FB_CROAK; |
900 | OUTPUT: |
901 | RETVAL |
902 | |
903 | int |
904 | FB_QUIET() |
905 | CODE: |
906 | RETVAL = ENCODE_FB_QUIET; |
907 | OUTPUT: |
908 | RETVAL |
909 | |
910 | int |
911 | FB_WARN() |
912 | CODE: |
913 | RETVAL = ENCODE_FB_WARN; |
914 | OUTPUT: |
915 | RETVAL |
916 | |
917 | int |
918 | FB_PERLQQ() |
919 | CODE: |
920 | RETVAL = ENCODE_FB_PERLQQ; |
921 | OUTPUT: |
922 | RETVAL |
2c674647 |
923 | |
af1f55d9 |
924 | int |
925 | FB_HTMLCREF() |
926 | CODE: |
927 | RETVAL = ENCODE_FB_HTMLCREF; |
928 | OUTPUT: |
929 | RETVAL |
930 | |
931 | int |
932 | FB_XMLCREF() |
933 | CODE: |
934 | RETVAL = ENCODE_FB_XMLCREF; |
935 | OUTPUT: |
936 | RETVAL |
937 | |
33af2bc7 |
938 | BOOT: |
939 | { |
85982a32 |
940 | #include "def_t.h" |
e7cbefb8 |
941 | #include "def_t.exh" |
33af2bc7 |
942 | } |