Commit | Line | Data |
85982a32 |
1 | /* |
e83d9f5f |
2 | $Id: Encode.xs,v 2.16 2009/09/06 14:32:21 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 |
e83d9f5f |
408 | PREINIT: |
409 | STRLEN slen; |
410 | U8 *s; |
411 | U8 *e; |
412 | SV *dst; |
413 | bool renewed = 0; |
ab3374e4 |
414 | CODE: |
415 | { |
e83d9f5f |
416 | dSP; ENTER; SAVETMPS; |
417 | if (src == &PL_sv_undef) src = newSV(0); |
418 | s = (U8 *) SvPV(src, slen); |
419 | e = (U8 *) SvEND(src); |
420 | dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
421 | |
422 | /* |
7f0d54d7 |
423 | * PerlIO check -- we assume the object is of PerlIO if renewed |
cc836e95 |
424 | */ |
cc836e95 |
425 | PUSHMARK(sp); |
426 | XPUSHs(obj); |
427 | PUTBACK; |
428 | if (call_method("renewed",G_SCALAR) == 1) { |
d1256cb1 |
429 | SPAGAIN; |
430 | renewed = (bool)POPi; |
431 | PUTBACK; |
cc836e95 |
432 | #if 0 |
d1256cb1 |
433 | fprintf(stderr, "renewed == %d\n", renewed); |
cc836e95 |
434 | #endif |
cc836e95 |
435 | } |
436 | FREETMPS; LEAVE; |
437 | /* end PerlIO check */ |
438 | |
ab3374e4 |
439 | if (SvUTF8(src)) { |
d1256cb1 |
440 | s = utf8_to_bytes(s,&slen); |
441 | if (s) { |
442 | SvCUR_set(src,slen); |
443 | SvUTF8_off(src); |
444 | e = s+slen; |
445 | } |
446 | else { |
447 | croak("Cannot decode string with wide characters"); |
448 | } |
ab3374e4 |
449 | } |
7f0d54d7 |
450 | |
451 | s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed); |
ab3374e4 |
452 | |
453 | /* Clear out translated part of source unless asked not to */ |
454 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
455 | slen = e-s; |
456 | if (slen) { |
457 | sv_setpvn(src, (char*)s, slen); |
458 | } |
459 | SvCUR_set(src, slen); |
ab3374e4 |
460 | } |
461 | SvUTF8_on(dst); |
462 | ST(0) = sv_2mortal(dst); |
463 | XSRETURN(1); |
464 | } |
465 | |
466 | void |
b536bf57 |
467 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
468 | SV * obj |
469 | SV * src |
470 | int check |
e83d9f5f |
471 | PREINIT: |
472 | STRLEN slen; |
473 | U8 *s; |
474 | U8 *e; |
475 | SV *dst; |
476 | bool renewed = 0; |
ab3374e4 |
477 | CODE: |
478 | { |
e83d9f5f |
479 | if (src == &PL_sv_undef) src = newSV(0); |
480 | s = (U8 *) SvPV(src, slen); |
481 | e = (U8 *) SvEND(src); |
482 | dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
483 | if (SvUTF8(src)) { |
d1256cb1 |
484 | /* Already encoded */ |
485 | if (strict_utf8(aTHX_ obj)) { |
486 | s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0); |
487 | } |
7f0d54d7 |
488 | else { |
489 | /* trust it and just copy the octets */ |
490 | sv_setpvn(dst,(char *)s,(e-s)); |
d1256cb1 |
491 | s = e; |
7f0d54d7 |
492 | } |
ab3374e4 |
493 | } |
494 | else { |
495 | /* Native bytes - can always encode */ |
d1256cb1 |
496 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
497 | while (s < e) { |
cf0d1c66 |
498 | UV uv = NATIVE_TO_UNI((UV) *s); |
499 | s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */ |
ab3374e4 |
500 | if (UNI_IS_INVARIANT(uv)) |
501 | *d++ = (U8)UTF_TO_NATIVE(uv); |
502 | else { |
503 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
504 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
505 | } |
d1256cb1 |
506 | } |
ab3374e4 |
507 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
508 | *SvEND(dst) = '\0'; |
509 | } |
510 | |
511 | /* Clear out translated part of source unless asked not to */ |
512 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
513 | slen = e-s; |
514 | if (slen) { |
515 | sv_setpvn(src, (char*)s, slen); |
516 | } |
517 | SvCUR_set(src, slen); |
ab3374e4 |
518 | } |
519 | SvPOK_only(dst); |
520 | SvUTF8_off(dst); |
521 | ST(0) = sv_2mortal(dst); |
522 | XSRETURN(1); |
523 | } |
524 | |
50d26985 |
525 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
526 | |
527 | PROTOTYPES: ENABLE |
528 | |
529 | void |
a0d8a30e |
530 | Method_renew(obj) |
531 | SV * obj |
532 | CODE: |
533 | { |
534 | XSRETURN(1); |
535 | } |
536 | |
cc836e95 |
537 | int |
538 | Method_renewed(obj) |
539 | SV * obj |
540 | CODE: |
541 | RETVAL = 0; |
542 | OUTPUT: |
543 | RETVAL |
544 | |
a0d8a30e |
545 | void |
0a95303c |
546 | Method_name(obj) |
547 | SV * obj |
548 | CODE: |
85982a32 |
549 | { |
550 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
551 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
552 | XSRETURN(1); |
553 | } |
0a95303c |
554 | |
555 | void |
370462a2 |
556 | Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no) |
220e2d4e |
557 | SV * obj |
558 | SV * dst |
559 | SV * src |
560 | SV * off |
561 | SV * term |
370462a2 |
562 | SV * check_sv |
220e2d4e |
563 | CODE: |
564 | { |
370462a2 |
565 | int check; |
566 | SV *fallback_cb = &PL_sv_undef; |
220e2d4e |
567 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
568 | STRLEN offset = (STRLEN)SvIV(off); |
569 | int code = 0; |
570 | if (SvUTF8(src)) { |
571 | sv_utf8_downgrade(src, FALSE); |
572 | } |
370462a2 |
573 | if (SvROK(check_sv)){ |
574 | fallback_cb = check_sv; |
575 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
576 | }else{ |
577 | check = SvIV(check_sv); |
578 | } |
220e2d4e |
579 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
370462a2 |
580 | &offset, term, &code, fallback_cb)); |
b162af07 |
581 | SvIV_set(off, (IV)offset); |
220e2d4e |
582 | if (code == ENCODE_FOUND_TERM) { |
d1256cb1 |
583 | ST(0) = &PL_sv_yes; |
220e2d4e |
584 | }else{ |
d1256cb1 |
585 | ST(0) = &PL_sv_no; |
220e2d4e |
586 | } |
587 | XSRETURN(1); |
588 | } |
589 | |
590 | void |
8e180e82 |
591 | Method_decode(obj,src,check_sv = &PL_sv_no) |
2f2b4ff2 |
592 | SV * obj |
593 | SV * src |
8e180e82 |
594 | SV * check_sv |
2f2b4ff2 |
595 | CODE: |
aae85ceb |
596 | { |
8e180e82 |
597 | int check; |
370462a2 |
598 | SV *fallback_cb = &PL_sv_undef; |
85982a32 |
599 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
600 | if (SvUTF8(src)) { |
601 | sv_utf8_downgrade(src, FALSE); |
602 | } |
8e180e82 |
603 | if (SvROK(check_sv)){ |
370462a2 |
604 | fallback_cb = check_sv; |
605 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
606 | }else{ |
370462a2 |
607 | check = SvIV(check_sv); |
8e180e82 |
608 | } |
220e2d4e |
609 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
370462a2 |
610 | NULL, Nullsv, NULL, fallback_cb); |
85982a32 |
611 | SvUTF8_on(ST(0)); |
aae85ceb |
612 | XSRETURN(1); |
613 | } |
614 | |
615 | void |
8e180e82 |
616 | Method_encode(obj,src,check_sv = &PL_sv_no) |
aae85ceb |
617 | SV * obj |
85982a32 |
618 | SV * src |
8e180e82 |
619 | SV * check_sv |
aae85ceb |
620 | CODE: |
621 | { |
8e180e82 |
622 | int check; |
370462a2 |
623 | SV *fallback_cb = &PL_sv_undef; |
85982a32 |
624 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
625 | sv_utf8_upgrade(src); |
8e180e82 |
626 | if (SvROK(check_sv)){ |
370462a2 |
627 | fallback_cb = check_sv; |
628 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
629 | }else{ |
370462a2 |
630 | check = SvIV(check_sv); |
8e180e82 |
631 | } |
220e2d4e |
632 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
370462a2 |
633 | NULL, Nullsv, NULL, fallback_cb); |
aae85ceb |
634 | XSRETURN(1); |
635 | } |
636 | |
0ab8f81e |
637 | void |
638 | Method_needs_lines(obj) |
639 | SV * obj |
640 | CODE: |
641 | { |
b32afa7c |
642 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
643 | ST(0) = &PL_sv_no; |
644 | XSRETURN(1); |
645 | } |
646 | |
647 | void |
648 | Method_perlio_ok(obj) |
649 | SV * obj |
650 | CODE: |
651 | { |
b32afa7c |
652 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
653 | /* require_pv(PERLIO_FILENAME); */ |
654 | |
655 | eval_pv("require PerlIO::encoding", 0); |
656 | |
657 | if (SvTRUE(get_sv("@", 0))) { |
d1256cb1 |
658 | ST(0) = &PL_sv_no; |
10c5ecbb |
659 | }else{ |
d1256cb1 |
660 | ST(0) = &PL_sv_yes; |
0ab8f81e |
661 | } |
662 | XSRETURN(1); |
663 | } |
664 | |
742555bd |
665 | void |
666 | Method_mime_name(obj) |
667 | SV * obj |
668 | CODE: |
669 | { |
670 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
671 | SV *retval; |
672 | eval_pv("require Encode::MIME::Name", 0); |
673 | |
674 | if (SvTRUE(get_sv("@", 0))) { |
675 | ST(0) = &PL_sv_undef; |
676 | }else{ |
677 | ENTER; |
678 | SAVETMPS; |
679 | PUSHMARK(sp); |
680 | XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0])))); |
681 | PUTBACK; |
682 | call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR); |
683 | SPAGAIN; |
684 | retval = newSVsv(POPs); |
685 | PUTBACK; |
686 | FREETMPS; |
687 | LEAVE; |
688 | /* enc->name[0] */ |
689 | ST(0) = retval; |
690 | } |
691 | XSRETURN(1); |
692 | } |
693 | |
67e989fb |
694 | MODULE = Encode PACKAGE = Encode |
2c674647 |
695 | |
696 | PROTOTYPES: ENABLE |
697 | |
67e989fb |
698 | I32 |
2c674647 |
699 | _bytes_to_utf8(sv, ...) |
85982a32 |
700 | SV * sv |
701 | CODE: |
702 | { |
703 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
704 | |
85982a32 |
705 | if (encoding) |
706 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
707 | else { |
d1256cb1 |
708 | STRLEN len; |
709 | U8* s = (U8*)SvPV(sv, len); |
710 | U8* converted; |
711 | |
712 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
713 | sv_setpvn(sv, (char *)converted, len); |
714 | SvUTF8_on(sv); /* XXX Should we? */ |
715 | Safefree(converted); /* ... so free it */ |
716 | RETVAL = len; |
85982a32 |
717 | } |
718 | } |
719 | OUTPUT: |
720 | RETVAL |
2c674647 |
721 | |
67e989fb |
722 | I32 |
2c674647 |
723 | _utf8_to_bytes(sv, ...) |
85982a32 |
724 | SV * sv |
725 | CODE: |
726 | { |
727 | SV * to = items > 1 ? ST(1) : Nullsv; |
728 | SV * check = items > 2 ? ST(2) : Nullsv; |
729 | |
730 | if (to) { |
d1256cb1 |
731 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
85982a32 |
732 | } else { |
d1256cb1 |
733 | STRLEN len; |
734 | U8 *s = (U8*)SvPV(sv, len); |
85982a32 |
735 | |
d1256cb1 |
736 | RETVAL = 0; |
737 | if (SvTRUE(check)) { |
738 | /* Must do things the slow way */ |
739 | U8 *dest; |
85982a32 |
740 | /* We need a copy to pass to check() */ |
2cdf8b6c |
741 | U8 *src = s; |
d1256cb1 |
742 | U8 *send = s + len; |
2cdf8b6c |
743 | U8 *d0; |
85982a32 |
744 | |
d1256cb1 |
745 | New(83, dest, len, U8); /* I think */ |
2cdf8b6c |
746 | d0 = dest; |
85982a32 |
747 | |
d1256cb1 |
748 | while (s < send) { |
85982a32 |
749 | if (*s < 0x80){ |
d1256cb1 |
750 | *dest++ = *s++; |
85982a32 |
751 | } else { |
d1256cb1 |
752 | STRLEN ulen; |
753 | UV uv = *s++; |
754 | |
755 | /* Have to do it all ourselves because of error routine, |
756 | aargh. */ |
757 | if (!(uv & 0x40)){ goto failure; } |
758 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
759 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
760 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
761 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
762 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
763 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
764 | else { ulen = 13; uv = 0; } |
765 | |
766 | /* Note change to utf8.c variable naming, for variety */ |
767 | while (ulen--) { |
768 | if ((*s & 0xc0) != 0x80){ |
769 | goto failure; |
770 | } else { |
771 | uv = (uv << 6) | (*s++ & 0x3f); |
772 | } |
773 | } |
774 | if (uv > 256) { |
775 | failure: |
776 | call_failure(check, s, dest, src); |
777 | /* Now what happens? */ |
778 | } |
779 | *dest++ = (U8)uv; |
780 | } |
781 | } |
2cdf8b6c |
782 | RETVAL = dest - d0; |
783 | sv_usepvn(sv, (char *)dest, RETVAL); |
784 | SvUTF8_off(sv); |
d1256cb1 |
785 | } else { |
786 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
787 | } |
85982a32 |
788 | } |
789 | } |
790 | OUTPUT: |
791 | RETVAL |
2c674647 |
792 | |
2c674647 |
793 | bool |
b2704119 |
794 | is_utf8(sv, check = 0) |
4411f3b6 |
795 | SV * sv |
b2704119 |
796 | int check |
85982a32 |
797 | CODE: |
798 | { |
799 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
d1256cb1 |
800 | sv = newSVsv(sv); /* GMAGIG will be done */ |
d1256cb1 |
801 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
802 | if (RETVAL && |
803 | check && |
804 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
805 | RETVAL = FALSE; |
85982a32 |
806 | if (sv != ST(0)) |
d1256cb1 |
807 | SvREFCNT_dec(sv); /* it was a temp copy */ |
85982a32 |
808 | } |
809 | OUTPUT: |
810 | RETVAL |
2c674647 |
811 | |
812 | SV * |
4411f3b6 |
813 | _utf8_on(sv) |
85982a32 |
814 | SV * sv |
815 | CODE: |
816 | { |
817 | if (SvPOK(sv)) { |
d1256cb1 |
818 | SV *rsv = newSViv(SvUTF8(sv)); |
819 | RETVAL = rsv; |
820 | SvUTF8_on(sv); |
85982a32 |
821 | } else { |
d1256cb1 |
822 | RETVAL = &PL_sv_undef; |
85982a32 |
823 | } |
824 | } |
825 | OUTPUT: |
826 | RETVAL |
2c674647 |
827 | |
828 | SV * |
4411f3b6 |
829 | _utf8_off(sv) |
85982a32 |
830 | SV * sv |
831 | CODE: |
832 | { |
833 | if (SvPOK(sv)) { |
d1256cb1 |
834 | SV *rsv = newSViv(SvUTF8(sv)); |
835 | RETVAL = rsv; |
836 | SvUTF8_off(sv); |
85982a32 |
837 | } else { |
d1256cb1 |
838 | RETVAL = &PL_sv_undef; |
85982a32 |
839 | } |
840 | } |
841 | OUTPUT: |
842 | RETVAL |
843 | |
85982a32 |
844 | int |
845 | DIE_ON_ERR() |
846 | CODE: |
847 | RETVAL = ENCODE_DIE_ON_ERR; |
848 | OUTPUT: |
849 | RETVAL |
850 | |
c6a7db43 |
851 | int |
85982a32 |
852 | WARN_ON_ERR() |
853 | CODE: |
854 | RETVAL = ENCODE_WARN_ON_ERR; |
855 | OUTPUT: |
856 | RETVAL |
857 | |
858 | int |
859 | LEAVE_SRC() |
860 | CODE: |
861 | RETVAL = ENCODE_LEAVE_SRC; |
862 | OUTPUT: |
863 | RETVAL |
864 | |
865 | int |
866 | RETURN_ON_ERR() |
867 | CODE: |
868 | RETVAL = ENCODE_RETURN_ON_ERR; |
869 | OUTPUT: |
870 | RETVAL |
871 | |
872 | int |
873 | PERLQQ() |
874 | CODE: |
875 | RETVAL = ENCODE_PERLQQ; |
876 | OUTPUT: |
877 | RETVAL |
878 | |
879 | int |
af1f55d9 |
880 | HTMLCREF() |
881 | CODE: |
882 | RETVAL = ENCODE_HTMLCREF; |
883 | OUTPUT: |
884 | RETVAL |
885 | |
886 | int |
887 | XMLCREF() |
888 | CODE: |
889 | RETVAL = ENCODE_XMLCREF; |
890 | OUTPUT: |
891 | RETVAL |
892 | |
893 | int |
56ff7374 |
894 | STOP_AT_PARTIAL() |
895 | CODE: |
896 | RETVAL = ENCODE_STOP_AT_PARTIAL; |
897 | OUTPUT: |
898 | RETVAL |
899 | |
900 | int |
85982a32 |
901 | FB_DEFAULT() |
902 | CODE: |
903 | RETVAL = ENCODE_FB_DEFAULT; |
904 | OUTPUT: |
905 | RETVAL |
906 | |
907 | int |
908 | FB_CROAK() |
909 | CODE: |
910 | RETVAL = ENCODE_FB_CROAK; |
911 | OUTPUT: |
912 | RETVAL |
913 | |
914 | int |
915 | FB_QUIET() |
916 | CODE: |
917 | RETVAL = ENCODE_FB_QUIET; |
918 | OUTPUT: |
919 | RETVAL |
920 | |
921 | int |
922 | FB_WARN() |
923 | CODE: |
924 | RETVAL = ENCODE_FB_WARN; |
925 | OUTPUT: |
926 | RETVAL |
927 | |
928 | int |
929 | FB_PERLQQ() |
930 | CODE: |
931 | RETVAL = ENCODE_FB_PERLQQ; |
932 | OUTPUT: |
933 | RETVAL |
2c674647 |
934 | |
af1f55d9 |
935 | int |
936 | FB_HTMLCREF() |
937 | CODE: |
938 | RETVAL = ENCODE_FB_HTMLCREF; |
939 | OUTPUT: |
940 | RETVAL |
941 | |
942 | int |
943 | FB_XMLCREF() |
944 | CODE: |
945 | RETVAL = ENCODE_FB_XMLCREF; |
946 | OUTPUT: |
947 | RETVAL |
948 | |
33af2bc7 |
949 | BOOT: |
950 | { |
85982a32 |
951 | #include "def_t.h" |
e7cbefb8 |
952 | #include "def_t.exh" |
33af2bc7 |
953 | } |