Commit | Line | Data |
85982a32 |
1 | /* |
656ebd29 |
2 | $Id: Encode.xs,v 2.10 2006/06/03 20:28:48 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 | |
8e180e82 |
38 | static SV* fallback_cb = (SV*)NULL ; |
39 | |
b0b300a3 |
40 | void |
aa0053b7 |
41 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
42 | { |
aa0053b7 |
43 | dSP; |
44 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
45 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
46 | int i = 0; |
47 | PUSHMARK(sp); |
48 | XPUSHs(sv); |
49 | while (enc->name[i]) { |
d1256cb1 |
50 | const char *name = enc->name[i++]; |
51 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
aa0053b7 |
52 | } |
53 | PUTBACK; |
54 | call_pv("Encode::define_encoding", G_DISCARD); |
55 | SvREFCNT_dec(sv); |
2f2b4ff2 |
56 | } |
57 | |
aa0053b7 |
58 | void |
59 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
60 | { |
85982a32 |
61 | /* Exists for breakpointing */ |
aa0053b7 |
62 | } |
67e989fb |
63 | |
85982a32 |
64 | |
2fc614e0 |
65 | #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s" |
66 | #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode" |
67 | |
2f2b4ff2 |
68 | static SV * |
8e180e82 |
69 | do_fallback_cb(pTHX_ UV ch) |
70 | { |
71 | dSP; |
72 | int argc; |
73 | SV* retval; |
74 | ENTER; |
75 | SAVETMPS; |
76 | PUSHMARK(sp); |
77 | XPUSHs(sv_2mortal(newSVnv((UV)ch))); |
78 | PUTBACK; |
79 | argc = call_sv(fallback_cb, G_SCALAR); |
80 | SPAGAIN; |
81 | if (argc != 1){ |
d1256cb1 |
82 | croak("fallback sub must return scalar!"); |
8e180e82 |
83 | } |
84 | retval = newSVsv(POPs); |
85 | PUTBACK; |
86 | FREETMPS; |
87 | LEAVE; |
88 | return retval; |
89 | } |
90 | |
91 | static SV * |
0629a5b3 |
92 | encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src, |
d1256cb1 |
93 | int check, STRLEN * offset, SV * term, int * retcode) |
2f2b4ff2 |
94 | { |
aa0053b7 |
95 | STRLEN slen; |
96 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
97 | STRLEN tlen = slen; |
98 | STRLEN ddone = 0; |
99 | STRLEN sdone = 0; |
39cf9a5e |
100 | |
3c49ab08 |
101 | /* We allocate slen+1. |
85982a32 |
102 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
103 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
104 | U8 *d = (U8 *)SvPVX(dst); |
105 | STRLEN dlen = SvLEN(dst)-1; |
220e2d4e |
106 | int code = 0; |
107 | STRLEN trmlen = 0; |
cc7dbc11 |
108 | U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL; |
220e2d4e |
109 | |
110 | if (offset) { |
111 | s += *offset; |
6be7c101 |
112 | if (slen > *offset){ /* safeguard against slen overflow */ |
d1256cb1 |
113 | slen -= *offset; |
6be7c101 |
114 | }else{ |
d1256cb1 |
115 | slen = 0; |
6be7c101 |
116 | } |
220e2d4e |
117 | tlen = slen; |
118 | } |
85982a32 |
119 | |
6be7c101 |
120 | if (slen == 0){ |
d1256cb1 |
121 | SvCUR_set(dst, 0); |
122 | SvPOK_only(dst); |
123 | goto ENCODE_END; |
85982a32 |
124 | } |
125 | |
220e2d4e |
126 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
d1256cb1 |
127 | trm, trmlen)) ) |
85982a32 |
128 | { |
d1256cb1 |
129 | SvCUR_set(dst, dlen+ddone); |
130 | SvPOK_only(dst); |
131 | |
132 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
133 | code == ENCODE_FOUND_TERM) { |
134 | break; |
135 | } |
136 | switch (code) { |
137 | case ENCODE_NOSPACE: |
138 | { |
139 | STRLEN more = 0; /* make sure you initialize! */ |
140 | STRLEN sleft; |
141 | sdone += slen; |
142 | ddone += dlen; |
143 | sleft = tlen - sdone; |
fcb875d4 |
144 | #if ENCODE_XS_PROFILE >= 2 |
d1256cb1 |
145 | Perl_warn(aTHX_ |
146 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
147 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
148 | #endif |
d1256cb1 |
149 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
150 | #if ENCODE_XS_USEFP == 2 |
d1256cb1 |
151 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
152 | - SvLEN(dst); |
39cf9a5e |
153 | #elif ENCODE_XS_USEFP |
d1256cb1 |
154 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
155 | #else |
d1256cb1 |
156 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
157 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
158 | #endif |
d1256cb1 |
159 | } |
160 | more += UTF8_MAXLEN; /* insurance policy */ |
161 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
162 | /* dst need to grow need MORE bytes! */ |
163 | if (ddone >= SvLEN(dst)) { |
164 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
165 | } |
166 | dlen = SvLEN(dst)-ddone-1; |
167 | d += ddone; |
168 | s += slen; |
169 | slen = tlen-sdone; |
170 | continue; |
171 | } |
172 | case ENCODE_NOREP: |
173 | /* encoding */ |
174 | if (dir == enc->f_utf8) { |
175 | STRLEN clen; |
176 | UV ch = |
177 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
178 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
179 | /* if non-representable multibyte prefix at end of current buffer - break*/ |
180 | if (clen > tlen - sdone) break; |
181 | if (check & ENCODE_DIE_ON_ERR) { |
182 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
183 | (UV)ch, enc->name[0]); |
184 | return &PL_sv_undef; /* never reaches but be safe */ |
185 | } |
186 | if (check & ENCODE_WARN_ON_ERR){ |
187 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
188 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
189 | } |
190 | if (check & ENCODE_RETURN_ON_ERR){ |
191 | goto ENCODE_SET_SRC; |
192 | } |
193 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
194 | SV* subchar = |
195 | (fallback_cb != (SV*)NULL) ? do_fallback_cb(aTHX_ ch) : |
196 | newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" : |
197 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
198 | "&#x%" UVxf ";", (UV)ch); |
199 | sdone += slen + clen; |
200 | ddone += dlen + SvCUR(subchar); |
201 | sv_catsv(dst, subchar); |
202 | SvREFCNT_dec(subchar); |
203 | } else { |
204 | /* fallback char */ |
205 | sdone += slen + clen; |
206 | ddone += dlen + enc->replen; |
207 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
208 | } |
209 | } |
210 | /* decoding */ |
211 | else { |
212 | if (check & ENCODE_DIE_ON_ERR){ |
213 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
214 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
215 | return &PL_sv_undef; /* never reaches but be safe */ |
216 | } |
217 | if (check & ENCODE_WARN_ON_ERR){ |
218 | Perl_warner( |
219 | aTHX_ packWARN(WARN_UTF8), |
220 | ERR_DECODE_NOMAP, |
436c6dd3 |
221 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
222 | } |
223 | if (check & ENCODE_RETURN_ON_ERR){ |
224 | goto ENCODE_SET_SRC; |
225 | } |
226 | if (check & |
227 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
228 | SV* subchar = |
229 | (fallback_cb != (SV*)NULL) ? |
230 | do_fallback_cb(aTHX_ (UV)s[slen]) : |
231 | newSVpvf("\\x%02" UVXf, (UV)s[slen]); |
232 | sdone += slen + 1; |
233 | ddone += dlen + SvCUR(subchar); |
234 | sv_catsv(dst, subchar); |
235 | SvREFCNT_dec(subchar); |
236 | } else { |
237 | sdone += slen + 1; |
238 | ddone += dlen + strlen(FBCHAR_UTF8); |
239 | sv_catpv(dst, FBCHAR_UTF8); |
240 | } |
241 | } |
242 | /* settle variables when fallback */ |
243 | d = (U8 *)SvEND(dst); |
b0b300a3 |
244 | dlen = SvLEN(dst) - ddone - 1; |
d1256cb1 |
245 | s = (U8*)SvPVX(src) + sdone; |
246 | slen = tlen - sdone; |
247 | break; |
248 | |
249 | default: |
250 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
251 | code, (dir == enc->f_utf8) ? "to" : "from", |
252 | enc->name[0]); |
253 | return &PL_sv_undef; |
254 | } |
85982a32 |
255 | } |
256 | ENCODE_SET_SRC: |
ca777f1c |
257 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
258 | sdone = SvCUR(src) - (slen+sdone); |
259 | if (sdone) { |
260 | sv_setpvn(src, (char*)s+slen, sdone); |
261 | } |
262 | SvCUR_set(src, sdone); |
2f2b4ff2 |
263 | } |
85982a32 |
264 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
265 | |
85982a32 |
266 | SvCUR_set(dst, dlen+ddone); |
267 | SvPOK_only(dst); |
c6a7db43 |
268 | |
39cf9a5e |
269 | #if ENCODE_XS_PROFILE |
270 | if (SvCUR(dst) > SvCUR(src)){ |
d1256cb1 |
271 | Perl_warn(aTHX_ |
272 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
273 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
274 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
275 | } |
3c49ab08 |
276 | #endif |
c6a7db43 |
277 | |
220e2d4e |
278 | if (offset) |
279 | *offset += sdone + slen; |
280 | |
85982a32 |
281 | ENCODE_END: |
0b3236bb |
282 | *SvEND(dst) = '\0'; |
220e2d4e |
283 | if (retcode) *retcode = code; |
aa0053b7 |
284 | return dst; |
2f2b4ff2 |
285 | } |
286 | |
7f0d54d7 |
287 | static bool |
288 | strict_utf8(pTHX_ SV* sv) |
289 | { |
290 | HV* hv; |
291 | SV** svp; |
292 | sv = SvRV(sv); |
293 | if (!sv || SvTYPE(sv) != SVt_PVHV) |
294 | return 0; |
295 | hv = (HV*)sv; |
296 | svp = hv_fetch(hv, "strict_utf8", 11, 0); |
297 | if (!svp) |
298 | return 0; |
299 | return SvTRUE(*svp); |
300 | } |
301 | |
302 | static U8* |
303 | process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check, |
304 | bool encode, bool strict, bool stop_at_partial) |
305 | { |
306 | UV uv; |
307 | STRLEN ulen; |
308 | |
309 | SvPOK_only(dst); |
310 | SvCUR_set(dst,0); |
311 | |
312 | while (s < e) { |
313 | if (UTF8_IS_INVARIANT(*s)) { |
314 | sv_catpvn(dst, (char *)s, 1); |
315 | s++; |
316 | continue; |
317 | } |
318 | |
319 | if (UTF8_IS_START(*s)) { |
320 | U8 skip = UTF8SKIP(s); |
321 | if ((s + skip) > e) { |
322 | /* Partial character */ |
323 | /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */ |
56ff7374 |
324 | if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL)) |
7f0d54d7 |
325 | break; |
326 | |
327 | goto malformed_byte; |
328 | } |
329 | |
330 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
331 | UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT : |
332 | UTF8_ALLOW_NONSTRICT) |
333 | ); |
334 | #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */ |
d1256cb1 |
335 | if (strict && uv > PERL_UNICODE_MAX) |
336 | ulen = -1; |
7f0d54d7 |
337 | #endif |
338 | if (ulen == -1) { |
339 | if (strict) { |
340 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
341 | UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT); |
342 | if (ulen == -1) |
343 | goto malformed_byte; |
344 | goto malformed; |
345 | } |
346 | goto malformed_byte; |
347 | } |
348 | |
349 | |
350 | /* Whole char is good */ |
351 | sv_catpvn(dst,(char *)s,skip); |
352 | s += skip; |
353 | continue; |
354 | } |
355 | |
356 | /* If we get here there is something wrong with alleged UTF-8 */ |
357 | malformed_byte: |
358 | uv = (UV)*s; |
359 | ulen = 1; |
360 | |
361 | malformed: |
362 | if (check & ENCODE_DIE_ON_ERR){ |
363 | if (encode) |
364 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8"); |
365 | else |
366 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv); |
367 | } |
368 | if (check & ENCODE_WARN_ON_ERR){ |
369 | if (encode) |
370 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
371 | ERR_ENCODE_NOMAP, uv, "utf8"); |
372 | else |
373 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
374 | ERR_DECODE_NOMAP, "utf8", uv); |
375 | } |
376 | if (check & ENCODE_RETURN_ON_ERR) { |
377 | break; |
378 | } |
379 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
380 | SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"): |
381 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
382 | "&#x%" UVxf ";", uv); |
383 | sv_catsv(dst, subchar); |
384 | SvREFCNT_dec(subchar); |
385 | } else { |
386 | sv_catpv(dst, FBCHAR_UTF8); |
387 | } |
388 | s += ulen; |
389 | } |
390 | *SvEND(dst) = '\0'; |
391 | |
392 | return s; |
393 | } |
394 | |
395 | |
ab3374e4 |
396 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
397 | |
a0d8a30e |
398 | PROTOTYPES: DISABLE |
399 | |
400 | void |
b536bf57 |
401 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
402 | SV * obj |
403 | SV * src |
404 | int check |
405 | CODE: |
406 | { |
407 | STRLEN slen; |
408 | U8 *s = (U8 *) SvPV(src, slen); |
409 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
410 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
411 | |
412 | /* |
7f0d54d7 |
413 | * PerlIO check -- we assume the object is of PerlIO if renewed |
cc836e95 |
414 | */ |
f5a22bee |
415 | bool renewed = 0; |
cc836e95 |
416 | dSP; ENTER; SAVETMPS; |
417 | PUSHMARK(sp); |
418 | XPUSHs(obj); |
419 | PUTBACK; |
420 | if (call_method("renewed",G_SCALAR) == 1) { |
d1256cb1 |
421 | SPAGAIN; |
422 | renewed = (bool)POPi; |
423 | PUTBACK; |
cc836e95 |
424 | #if 0 |
d1256cb1 |
425 | fprintf(stderr, "renewed == %d\n", renewed); |
cc836e95 |
426 | #endif |
cc836e95 |
427 | } |
428 | FREETMPS; LEAVE; |
429 | /* end PerlIO check */ |
430 | |
ab3374e4 |
431 | if (SvUTF8(src)) { |
d1256cb1 |
432 | s = utf8_to_bytes(s,&slen); |
433 | if (s) { |
434 | SvCUR_set(src,slen); |
435 | SvUTF8_off(src); |
436 | e = s+slen; |
437 | } |
438 | else { |
439 | croak("Cannot decode string with wide characters"); |
440 | } |
ab3374e4 |
441 | } |
7f0d54d7 |
442 | |
443 | s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed); |
ab3374e4 |
444 | |
445 | /* Clear out translated part of source unless asked not to */ |
446 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
447 | slen = e-s; |
448 | if (slen) { |
449 | sv_setpvn(src, (char*)s, slen); |
450 | } |
451 | SvCUR_set(src, slen); |
ab3374e4 |
452 | } |
453 | SvUTF8_on(dst); |
454 | ST(0) = sv_2mortal(dst); |
455 | XSRETURN(1); |
456 | } |
457 | |
458 | void |
b536bf57 |
459 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
460 | SV * obj |
461 | SV * src |
462 | int check |
463 | CODE: |
464 | { |
465 | STRLEN slen; |
466 | U8 *s = (U8 *) SvPV(src, slen); |
467 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
468 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
469 | if (SvUTF8(src)) { |
d1256cb1 |
470 | /* Already encoded */ |
471 | if (strict_utf8(aTHX_ obj)) { |
472 | s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0); |
473 | } |
7f0d54d7 |
474 | else { |
475 | /* trust it and just copy the octets */ |
476 | sv_setpvn(dst,(char *)s,(e-s)); |
d1256cb1 |
477 | s = e; |
7f0d54d7 |
478 | } |
ab3374e4 |
479 | } |
480 | else { |
481 | /* Native bytes - can always encode */ |
d1256cb1 |
482 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
483 | while (s < e) { |
cf0d1c66 |
484 | UV uv = NATIVE_TO_UNI((UV) *s); |
485 | s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */ |
ab3374e4 |
486 | if (UNI_IS_INVARIANT(uv)) |
487 | *d++ = (U8)UTF_TO_NATIVE(uv); |
488 | else { |
489 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
490 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
491 | } |
d1256cb1 |
492 | } |
ab3374e4 |
493 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
494 | *SvEND(dst) = '\0'; |
495 | } |
496 | |
497 | /* Clear out translated part of source unless asked not to */ |
498 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
499 | slen = e-s; |
500 | if (slen) { |
501 | sv_setpvn(src, (char*)s, slen); |
502 | } |
503 | SvCUR_set(src, slen); |
ab3374e4 |
504 | } |
505 | SvPOK_only(dst); |
506 | SvUTF8_off(dst); |
507 | ST(0) = sv_2mortal(dst); |
508 | XSRETURN(1); |
509 | } |
510 | |
50d26985 |
511 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
512 | |
513 | PROTOTYPES: ENABLE |
514 | |
515 | void |
a0d8a30e |
516 | Method_renew(obj) |
517 | SV * obj |
518 | CODE: |
519 | { |
520 | XSRETURN(1); |
521 | } |
522 | |
cc836e95 |
523 | int |
524 | Method_renewed(obj) |
525 | SV * obj |
526 | CODE: |
527 | RETVAL = 0; |
528 | OUTPUT: |
529 | RETVAL |
530 | |
a0d8a30e |
531 | void |
0a95303c |
532 | Method_name(obj) |
533 | SV * obj |
534 | CODE: |
85982a32 |
535 | { |
536 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
537 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
538 | XSRETURN(1); |
539 | } |
0a95303c |
540 | |
541 | void |
220e2d4e |
542 | Method_cat_decode(obj, dst, src, off, term, check = 0) |
543 | SV * obj |
544 | SV * dst |
545 | SV * src |
546 | SV * off |
547 | SV * term |
548 | int check |
549 | CODE: |
550 | { |
551 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
552 | STRLEN offset = (STRLEN)SvIV(off); |
553 | int code = 0; |
554 | if (SvUTF8(src)) { |
555 | sv_utf8_downgrade(src, FALSE); |
556 | } |
557 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
d1256cb1 |
558 | &offset, term, &code)); |
b162af07 |
559 | SvIV_set(off, (IV)offset); |
220e2d4e |
560 | if (code == ENCODE_FOUND_TERM) { |
d1256cb1 |
561 | ST(0) = &PL_sv_yes; |
220e2d4e |
562 | }else{ |
d1256cb1 |
563 | ST(0) = &PL_sv_no; |
220e2d4e |
564 | } |
565 | XSRETURN(1); |
566 | } |
567 | |
568 | void |
8e180e82 |
569 | Method_decode(obj,src,check_sv = &PL_sv_no) |
2f2b4ff2 |
570 | SV * obj |
571 | SV * src |
8e180e82 |
572 | SV * check_sv |
2f2b4ff2 |
573 | CODE: |
aae85ceb |
574 | { |
8e180e82 |
575 | int check; |
85982a32 |
576 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
577 | if (SvUTF8(src)) { |
578 | sv_utf8_downgrade(src, FALSE); |
579 | } |
8e180e82 |
580 | if (SvROK(check_sv)){ |
d1256cb1 |
581 | if (fallback_cb == (SV*)NULL){ |
8e180e82 |
582 | fallback_cb = newSVsv(check_sv); /* First time */ |
583 | }else{ |
584 | SvSetSV(fallback_cb, check_sv); /* Been here before */ |
d1256cb1 |
585 | } |
586 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
587 | }else{ |
d1256cb1 |
588 | fallback_cb = (SV*)NULL; |
589 | check = SvIV(check_sv); |
8e180e82 |
590 | } |
220e2d4e |
591 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
d1256cb1 |
592 | NULL, Nullsv, NULL); |
85982a32 |
593 | SvUTF8_on(ST(0)); |
aae85ceb |
594 | XSRETURN(1); |
595 | } |
596 | |
8e180e82 |
597 | |
598 | |
aae85ceb |
599 | void |
8e180e82 |
600 | Method_encode(obj,src,check_sv = &PL_sv_no) |
aae85ceb |
601 | SV * obj |
85982a32 |
602 | SV * src |
8e180e82 |
603 | SV * check_sv |
aae85ceb |
604 | CODE: |
605 | { |
8e180e82 |
606 | int check; |
85982a32 |
607 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
608 | sv_utf8_upgrade(src); |
8e180e82 |
609 | if (SvROK(check_sv)){ |
d1256cb1 |
610 | if (fallback_cb == (SV*)NULL){ |
8e180e82 |
611 | fallback_cb = newSVsv(check_sv); /* First time */ |
612 | }else{ |
613 | SvSetSV(fallback_cb, check_sv); /* Been here before */ |
d1256cb1 |
614 | } |
615 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
616 | }else{ |
d1256cb1 |
617 | fallback_cb = (SV*)NULL; |
618 | check = SvIV(check_sv); |
8e180e82 |
619 | } |
220e2d4e |
620 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
d1256cb1 |
621 | NULL, Nullsv, NULL); |
aae85ceb |
622 | XSRETURN(1); |
623 | } |
624 | |
0ab8f81e |
625 | void |
626 | Method_needs_lines(obj) |
627 | SV * obj |
628 | CODE: |
629 | { |
b32afa7c |
630 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
631 | ST(0) = &PL_sv_no; |
632 | XSRETURN(1); |
633 | } |
634 | |
635 | void |
636 | Method_perlio_ok(obj) |
637 | SV * obj |
638 | CODE: |
639 | { |
b32afa7c |
640 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
641 | /* require_pv(PERLIO_FILENAME); */ |
642 | |
643 | eval_pv("require PerlIO::encoding", 0); |
644 | |
645 | if (SvTRUE(get_sv("@", 0))) { |
d1256cb1 |
646 | ST(0) = &PL_sv_no; |
10c5ecbb |
647 | }else{ |
d1256cb1 |
648 | ST(0) = &PL_sv_yes; |
0ab8f81e |
649 | } |
650 | XSRETURN(1); |
651 | } |
652 | |
67e989fb |
653 | MODULE = Encode PACKAGE = Encode |
2c674647 |
654 | |
655 | PROTOTYPES: ENABLE |
656 | |
67e989fb |
657 | I32 |
2c674647 |
658 | _bytes_to_utf8(sv, ...) |
85982a32 |
659 | SV * sv |
660 | CODE: |
661 | { |
662 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
663 | |
85982a32 |
664 | if (encoding) |
665 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
666 | else { |
d1256cb1 |
667 | STRLEN len; |
668 | U8* s = (U8*)SvPV(sv, len); |
669 | U8* converted; |
670 | |
671 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
672 | sv_setpvn(sv, (char *)converted, len); |
673 | SvUTF8_on(sv); /* XXX Should we? */ |
674 | Safefree(converted); /* ... so free it */ |
675 | RETVAL = len; |
85982a32 |
676 | } |
677 | } |
678 | OUTPUT: |
679 | RETVAL |
2c674647 |
680 | |
67e989fb |
681 | I32 |
2c674647 |
682 | _utf8_to_bytes(sv, ...) |
85982a32 |
683 | SV * sv |
684 | CODE: |
685 | { |
686 | SV * to = items > 1 ? ST(1) : Nullsv; |
687 | SV * check = items > 2 ? ST(2) : Nullsv; |
688 | |
689 | if (to) { |
d1256cb1 |
690 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
85982a32 |
691 | } else { |
d1256cb1 |
692 | STRLEN len; |
693 | U8 *s = (U8*)SvPV(sv, len); |
85982a32 |
694 | |
d1256cb1 |
695 | RETVAL = 0; |
696 | if (SvTRUE(check)) { |
697 | /* Must do things the slow way */ |
698 | U8 *dest; |
85982a32 |
699 | /* We need a copy to pass to check() */ |
2cdf8b6c |
700 | U8 *src = s; |
d1256cb1 |
701 | U8 *send = s + len; |
2cdf8b6c |
702 | U8 *d0; |
85982a32 |
703 | |
d1256cb1 |
704 | New(83, dest, len, U8); /* I think */ |
2cdf8b6c |
705 | d0 = dest; |
85982a32 |
706 | |
d1256cb1 |
707 | while (s < send) { |
85982a32 |
708 | if (*s < 0x80){ |
d1256cb1 |
709 | *dest++ = *s++; |
85982a32 |
710 | } else { |
d1256cb1 |
711 | STRLEN ulen; |
712 | UV uv = *s++; |
713 | |
714 | /* Have to do it all ourselves because of error routine, |
715 | aargh. */ |
716 | if (!(uv & 0x40)){ goto failure; } |
717 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
718 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
719 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
720 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
721 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
722 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
723 | else { ulen = 13; uv = 0; } |
724 | |
725 | /* Note change to utf8.c variable naming, for variety */ |
726 | while (ulen--) { |
727 | if ((*s & 0xc0) != 0x80){ |
728 | goto failure; |
729 | } else { |
730 | uv = (uv << 6) | (*s++ & 0x3f); |
731 | } |
732 | } |
733 | if (uv > 256) { |
734 | failure: |
735 | call_failure(check, s, dest, src); |
736 | /* Now what happens? */ |
737 | } |
738 | *dest++ = (U8)uv; |
739 | } |
740 | } |
2cdf8b6c |
741 | RETVAL = dest - d0; |
742 | sv_usepvn(sv, (char *)dest, RETVAL); |
743 | SvUTF8_off(sv); |
d1256cb1 |
744 | } else { |
745 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
746 | } |
85982a32 |
747 | } |
748 | } |
749 | OUTPUT: |
750 | RETVAL |
2c674647 |
751 | |
2c674647 |
752 | bool |
b2704119 |
753 | is_utf8(sv, check = 0) |
4411f3b6 |
754 | SV * sv |
b2704119 |
755 | int check |
85982a32 |
756 | CODE: |
757 | { |
758 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
d1256cb1 |
759 | sv = newSVsv(sv); /* GMAGIG will be done */ |
85982a32 |
760 | if (SvPOK(sv)) { |
d1256cb1 |
761 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
762 | if (RETVAL && |
763 | check && |
764 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
765 | RETVAL = FALSE; |
85982a32 |
766 | } else { |
d1256cb1 |
767 | RETVAL = FALSE; |
85982a32 |
768 | } |
769 | if (sv != ST(0)) |
d1256cb1 |
770 | SvREFCNT_dec(sv); /* it was a temp copy */ |
85982a32 |
771 | } |
772 | OUTPUT: |
773 | RETVAL |
2c674647 |
774 | |
775 | SV * |
4411f3b6 |
776 | _utf8_on(sv) |
85982a32 |
777 | SV * sv |
778 | CODE: |
779 | { |
780 | if (SvPOK(sv)) { |
d1256cb1 |
781 | SV *rsv = newSViv(SvUTF8(sv)); |
782 | RETVAL = rsv; |
783 | SvUTF8_on(sv); |
85982a32 |
784 | } else { |
d1256cb1 |
785 | RETVAL = &PL_sv_undef; |
85982a32 |
786 | } |
787 | } |
788 | OUTPUT: |
789 | RETVAL |
2c674647 |
790 | |
791 | SV * |
4411f3b6 |
792 | _utf8_off(sv) |
85982a32 |
793 | SV * sv |
794 | CODE: |
795 | { |
796 | if (SvPOK(sv)) { |
d1256cb1 |
797 | SV *rsv = newSViv(SvUTF8(sv)); |
798 | RETVAL = rsv; |
799 | SvUTF8_off(sv); |
85982a32 |
800 | } else { |
d1256cb1 |
801 | RETVAL = &PL_sv_undef; |
85982a32 |
802 | } |
803 | } |
804 | OUTPUT: |
805 | RETVAL |
806 | |
85982a32 |
807 | int |
808 | DIE_ON_ERR() |
809 | CODE: |
810 | RETVAL = ENCODE_DIE_ON_ERR; |
811 | OUTPUT: |
812 | RETVAL |
813 | |
c6a7db43 |
814 | int |
85982a32 |
815 | WARN_ON_ERR() |
816 | CODE: |
817 | RETVAL = ENCODE_WARN_ON_ERR; |
818 | OUTPUT: |
819 | RETVAL |
820 | |
821 | int |
822 | LEAVE_SRC() |
823 | CODE: |
824 | RETVAL = ENCODE_LEAVE_SRC; |
825 | OUTPUT: |
826 | RETVAL |
827 | |
828 | int |
829 | RETURN_ON_ERR() |
830 | CODE: |
831 | RETVAL = ENCODE_RETURN_ON_ERR; |
832 | OUTPUT: |
833 | RETVAL |
834 | |
835 | int |
836 | PERLQQ() |
837 | CODE: |
838 | RETVAL = ENCODE_PERLQQ; |
839 | OUTPUT: |
840 | RETVAL |
841 | |
842 | int |
af1f55d9 |
843 | HTMLCREF() |
844 | CODE: |
845 | RETVAL = ENCODE_HTMLCREF; |
846 | OUTPUT: |
847 | RETVAL |
848 | |
849 | int |
850 | XMLCREF() |
851 | CODE: |
852 | RETVAL = ENCODE_XMLCREF; |
853 | OUTPUT: |
854 | RETVAL |
855 | |
856 | int |
56ff7374 |
857 | STOP_AT_PARTIAL() |
858 | CODE: |
859 | RETVAL = ENCODE_STOP_AT_PARTIAL; |
860 | OUTPUT: |
861 | RETVAL |
862 | |
863 | int |
85982a32 |
864 | FB_DEFAULT() |
865 | CODE: |
866 | RETVAL = ENCODE_FB_DEFAULT; |
867 | OUTPUT: |
868 | RETVAL |
869 | |
870 | int |
871 | FB_CROAK() |
872 | CODE: |
873 | RETVAL = ENCODE_FB_CROAK; |
874 | OUTPUT: |
875 | RETVAL |
876 | |
877 | int |
878 | FB_QUIET() |
879 | CODE: |
880 | RETVAL = ENCODE_FB_QUIET; |
881 | OUTPUT: |
882 | RETVAL |
883 | |
884 | int |
885 | FB_WARN() |
886 | CODE: |
887 | RETVAL = ENCODE_FB_WARN; |
888 | OUTPUT: |
889 | RETVAL |
890 | |
891 | int |
892 | FB_PERLQQ() |
893 | CODE: |
894 | RETVAL = ENCODE_FB_PERLQQ; |
895 | OUTPUT: |
896 | RETVAL |
2c674647 |
897 | |
af1f55d9 |
898 | int |
899 | FB_HTMLCREF() |
900 | CODE: |
901 | RETVAL = ENCODE_FB_HTMLCREF; |
902 | OUTPUT: |
903 | RETVAL |
904 | |
905 | int |
906 | FB_XMLCREF() |
907 | CODE: |
908 | RETVAL = ENCODE_FB_XMLCREF; |
909 | OUTPUT: |
910 | RETVAL |
911 | |
33af2bc7 |
912 | BOOT: |
913 | { |
85982a32 |
914 | #include "def_t.h" |
e7cbefb8 |
915 | #include "def_t.exh" |
33af2bc7 |
916 | } |