Commit | Line | Data |
a0ed51b3 |
1 | /* utf8.c |
2 | * |
3818b22b |
3 | * Copyright (c) 1998-2000, Larry Wall |
a0ed51b3 |
4 | * |
5 | * You may distribute under the terms of either the GNU General Public |
6 | * License or the Artistic License, as specified in the README file. |
7 | * |
8 | */ |
9 | |
10 | /* |
11 | * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever |
12 | * heard of that we don't want to see any closer; and that's the one place |
13 | * we're trying to get to! And that's just where we can't get, nohow.' |
14 | * |
15 | * 'Well do I understand your speech,' he answered in the same language; |
16 | * 'yet few strangers do so. Why then do you not speak in the Common Tongue, |
17 | * as is the custom in the West, if you wish to be answered?' |
18 | * |
19 | * ...the travellers perceived that the floor was paved with stones of many |
20 | * hues; branching runes and strange devices intertwined beneath their feet. |
21 | */ |
22 | |
23 | #include "EXTERN.h" |
864dbfa3 |
24 | #define PERL_IN_UTF8_C |
a0ed51b3 |
25 | #include "perl.h" |
26 | |
27 | /* Unicode support */ |
28 | |
dfe13c55 |
29 | U8 * |
864dbfa3 |
30 | Perl_uv_to_utf8(pTHX_ U8 *d, UV uv) |
a0ed51b3 |
31 | { |
32 | if (uv < 0x80) { |
33 | *d++ = uv; |
34 | return d; |
35 | } |
36 | if (uv < 0x800) { |
37 | *d++ = (( uv >> 6) | 0xc0); |
38 | *d++ = (( uv & 0x3f) | 0x80); |
39 | return d; |
40 | } |
41 | if (uv < 0x10000) { |
42 | *d++ = (( uv >> 12) | 0xe0); |
43 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
44 | *d++ = (( uv & 0x3f) | 0x80); |
45 | return d; |
46 | } |
47 | if (uv < 0x200000) { |
48 | *d++ = (( uv >> 18) | 0xf0); |
49 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
50 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
51 | *d++ = (( uv & 0x3f) | 0x80); |
52 | return d; |
53 | } |
54 | if (uv < 0x4000000) { |
55 | *d++ = (( uv >> 24) | 0xf8); |
56 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
57 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
58 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
59 | *d++ = (( uv & 0x3f) | 0x80); |
60 | return d; |
61 | } |
62 | if (uv < 0x80000000) { |
63 | *d++ = (( uv >> 30) | 0xfc); |
64 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
65 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
66 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
67 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
68 | *d++ = (( uv & 0x3f) | 0x80); |
69 | return d; |
70 | } |
6b8eaf93 |
71 | #ifdef HAS_QUAD |
628e1a40 |
72 | if (uv < 0x1000000000LL) |
a0ed51b3 |
73 | #endif |
74 | { |
75 | *d++ = 0xfe; /* Can't match U+FEFF! */ |
76 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
77 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
78 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
79 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
80 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
81 | *d++ = (( uv & 0x3f) | 0x80); |
82 | return d; |
83 | } |
6b8eaf93 |
84 | #ifdef HAS_QUAD |
a0ed51b3 |
85 | { |
86 | *d++ = 0xff; /* Can't match U+FFFE! */ |
3c77ea2b |
87 | *d++ = 0x80; /* 6 Reserved bits */ |
88 | *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */ |
89 | *d++ = (((uv >> 54) & 0x3f) | 0x80); |
90 | *d++ = (((uv >> 48) & 0x3f) | 0x80); |
91 | *d++ = (((uv >> 42) & 0x3f) | 0x80); |
a0ed51b3 |
92 | *d++ = (((uv >> 36) & 0x3f) | 0x80); |
93 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
94 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
95 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
96 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
97 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
98 | *d++ = (( uv & 0x3f) | 0x80); |
99 | return d; |
100 | } |
101 | #endif |
102 | } |
103 | |
386d01d6 |
104 | /* Tests if some arbitrary number of bytes begins in a valid UTF-8 character. |
105 | * The actual number of bytes in the UTF-8 character will be returned if it |
106 | * is valid, otherwise 0. */ |
107 | int |
108 | Perl_is_utf8_char(pTHX_ U8 *s) |
109 | { |
110 | U8 u = *s; |
111 | int slen, len; |
112 | |
113 | if (!(u & 0x80)) |
114 | return 1; |
115 | |
116 | if (!(u & 0x40)) |
117 | return 0; |
118 | |
9f07fdcd |
119 | len = UTF8SKIP(s); |
386d01d6 |
120 | |
121 | slen = len - 1; |
122 | s++; |
123 | while (slen--) { |
124 | if ((*s & 0xc0) != 0x80) |
125 | return 0; |
126 | s++; |
127 | } |
128 | return len; |
129 | } |
130 | |
6662521e |
131 | /* |
b2a2e44b |
132 | =for apidoc Am|is_utf8_string|U8 *s|STRLEN len |
6662521e |
133 | |
134 | Returns true if first C<len> bytes of the given string form valid a UTF8 |
135 | string, false otherwise. |
136 | |
137 | =cut |
138 | */ |
139 | |
8e84507e |
140 | bool |
6662521e |
141 | Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len) |
142 | { |
143 | U8* x=s; |
144 | U8* send=s+len; |
145 | int c; |
146 | while (x < send) { |
147 | c = is_utf8_char(x); |
148 | x += c; |
149 | if (!c || x > send) |
150 | return 0; |
151 | } |
152 | return 1; |
153 | } |
154 | |
67e989fb |
155 | /* |
ba210ebe |
156 | =for apidoc Am|U8* s|utf8_to_uv_chk|STRLEN curlen|I32 *retlen|I32 checking |
67e989fb |
157 | |
158 | Returns the character value of the first character in the string C<s> |
ba210ebe |
159 | which is assumed to be in UTF8 encoding and no longer than C<curlen>; |
160 | C<retlen> will be set to the length, in bytes, of that character, |
161 | and the pointer C<s> will be advanced to the end of the character. |
67e989fb |
162 | |
163 | If C<s> does not point to a well-formed UTF8 character, the behaviour |
164 | is dependent on the value of C<checking>: if this is true, it is |
165 | assumed that the caller will raise a warning, and this function will |
166 | set C<retlen> to C<-1> and return. If C<checking> is not true, an optional UTF8 |
167 | warning is produced. |
168 | |
169 | =cut |
170 | */ |
171 | |
a0ed51b3 |
172 | UV |
ba210ebe |
173 | Perl_utf8_to_uv_chk(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, bool checking) |
a0ed51b3 |
174 | { |
ba210ebe |
175 | dTHR; |
176 | UV uv = *s, ouv; |
177 | STRLEN len = 1; |
178 | bool dowarn = ckWARN_d(WARN_UTF8); |
179 | STRLEN expectlen = 0; |
180 | |
181 | if (uv <= 0x7f) { /* Pure ASCII. */ |
a0ed51b3 |
182 | if (retlen) |
183 | *retlen = 1; |
184 | return *s; |
185 | } |
67e989fb |
186 | |
ba210ebe |
187 | if (uv >= 0x80 && uv <= 0xbf) { |
188 | if (dowarn) |
189 | Perl_warner(aTHX_ WARN_UTF8, |
190 | "Malformed UTF-8 character (unexpected continuation byte 0x%02x)", |
191 | uv); |
192 | goto malformed; |
193 | } |
194 | |
195 | if (uv >= 0xc0 && uv <= 0xfd && curlen > 1 && s[1] < 0x80) { |
196 | if (dowarn) |
197 | Perl_warner(aTHX_ WARN_UTF8, |
198 | "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x after byte 0x%02x)", |
199 | s[1], uv); |
200 | goto malformed; |
201 | } |
202 | |
203 | if ((uv == 0xfe || uv == 0xff) && IN_UTF8){ |
204 | if (dowarn) |
205 | Perl_warner(aTHX_ WARN_UTF8, |
206 | "Malformed UTF-8 character (impossible byte 0x%02x)", |
207 | uv); |
208 | goto malformed; |
a0ed51b3 |
209 | } |
210 | |
ba210ebe |
211 | if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } |
212 | else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } |
213 | else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } |
214 | else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } |
215 | else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } |
216 | else if (!(uv & 0x01)) { len = 7; uv = 0; } |
3c77ea2b |
217 | else { len = 13; uv = 0; } /* whoa! */ |
a0ed51b3 |
218 | |
219 | if (retlen) |
220 | *retlen = len; |
ba210ebe |
221 | |
222 | expectlen = len; |
223 | |
224 | if (curlen < expectlen) { |
225 | if (dowarn) |
226 | Perl_warner(aTHX_ WARN_UTF8, |
227 | "Malformed UTF-8 character (%d byte%s, need %d)", |
228 | curlen, curlen > 1 ? "s" : "", expectlen); |
229 | goto malformed; |
230 | } |
231 | |
232 | len--; |
a0ed51b3 |
233 | s++; |
ba210ebe |
234 | ouv = uv; |
235 | |
a0ed51b3 |
236 | while (len--) { |
237 | if ((*s & 0xc0) != 0x80) { |
ba210ebe |
238 | if (dowarn) |
239 | Perl_warner(aTHX_ WARN_UTF8, |
240 | "Malformed UTF-8 character (unexpected continuation byte 0x%02x)", |
241 | *s); |
242 | goto malformed; |
a0ed51b3 |
243 | } |
244 | else |
ba210ebe |
245 | uv = (uv << 6) | (*s & 0x3f); |
246 | if (uv < ouv) { |
247 | if (dowarn) |
248 | Perl_warner(aTHX_ WARN_UTF8, |
249 | "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)", |
250 | ouv, *s); |
251 | goto malformed; |
252 | } |
253 | s++; |
254 | ouv = uv; |
255 | } |
256 | |
257 | if (uv >= 0xd800 && uv <= 0xdfff) { |
258 | if (dowarn) |
259 | Perl_warner(aTHX_ WARN_UTF8, |
260 | "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")", |
261 | uv); |
262 | goto malformed; |
263 | } else if (uv == 0xfffe) { |
264 | if (dowarn) |
265 | Perl_warner(aTHX_ WARN_UTF8, |
266 | "Malformed UTF-8 character (byte order mark 0x%04"UVxf")", |
267 | uv); |
268 | goto malformed; |
269 | } else if (uv == 0xffff && IN_UTF8) { |
270 | if (dowarn) |
271 | Perl_warner(aTHX_ WARN_UTF8, |
272 | "Malformed UTF-8 character (impossible character 0x%04"UVxf")", |
273 | uv); |
274 | goto malformed; |
275 | } else if (expectlen > UTF8LEN(uv)) { |
276 | if (dowarn) |
277 | Perl_warner(aTHX_ WARN_UTF8, |
278 | "Malformed UTF-8 character (%d byte%s, need %d)", |
279 | expectlen, expectlen > 1 ? "s": "", UTF8LEN(uv)); |
280 | goto malformed; |
a0ed51b3 |
281 | } |
ba210ebe |
282 | |
a0ed51b3 |
283 | return uv; |
ba210ebe |
284 | |
285 | malformed: |
286 | |
287 | if (checking) { |
288 | if (retlen) |
289 | *retlen = len; |
290 | return 0; |
291 | } |
292 | |
293 | if (retlen) |
294 | *retlen = -1; |
295 | |
296 | return UNICODE_REPLACEMENT_CHARACTER; |
a0ed51b3 |
297 | } |
298 | |
8e84507e |
299 | /* |
ba210ebe |
300 | =for apidoc Am|U8* s|utf8_to_uv|STRLEN *retlen |
8e84507e |
301 | |
302 | Returns the character value of the first character in the string C<s> |
303 | which is assumed to be in UTF8 encoding; C<retlen> will be set to the |
304 | length, in bytes, of that character, and the pointer C<s> will be |
305 | advanced to the end of the character. |
306 | |
ba210ebe |
307 | If C<s> does not point to a well-formed UTF8 character, zero is |
308 | returned and retlen is set, if possible, to -1. |
8e84507e |
309 | |
310 | =cut |
311 | */ |
312 | |
313 | UV |
ba210ebe |
314 | Perl_utf8_to_uv(pTHX_ U8* s, STRLEN* retlen) |
8e84507e |
315 | { |
ba210ebe |
316 | return Perl_utf8_to_uv_chk(aTHX_ s, (STRLEN)-1, retlen, 0); |
8e84507e |
317 | } |
318 | |
246fae53 |
319 | /* utf8_distance(a,b) returns the number of UTF8 characters between |
320 | the pointers a and b */ |
a0ed51b3 |
321 | |
322 | I32 |
864dbfa3 |
323 | Perl_utf8_distance(pTHX_ U8 *a, U8 *b) |
a0ed51b3 |
324 | { |
325 | I32 off = 0; |
326 | if (a < b) { |
327 | while (a < b) { |
328 | a += UTF8SKIP(a); |
329 | off--; |
330 | } |
331 | } |
332 | else { |
333 | while (b < a) { |
334 | b += UTF8SKIP(b); |
335 | off++; |
336 | } |
337 | } |
338 | return off; |
339 | } |
340 | |
341 | /* WARNING: do not use the following unless you *know* off is within bounds */ |
342 | |
343 | U8 * |
864dbfa3 |
344 | Perl_utf8_hop(pTHX_ U8 *s, I32 off) |
a0ed51b3 |
345 | { |
346 | if (off >= 0) { |
347 | while (off--) |
348 | s += UTF8SKIP(s); |
349 | } |
350 | else { |
351 | while (off++) { |
352 | s--; |
353 | if (*s & 0x80) { |
354 | while ((*s & 0xc0) == 0x80) |
355 | s--; |
356 | } |
357 | } |
358 | } |
359 | return s; |
360 | } |
361 | |
6940069f |
362 | /* |
246fae53 |
363 | =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len |
6940069f |
364 | |
246fae53 |
365 | Converts a string C<s> of length C<len> from UTF8 into byte encoding. |
366 | Unlike C<bytes_to_utf8>, this over-writes the original string, and |
367 | updates len to contain the new length. |
67e989fb |
368 | Returns zero on failure, setting C<len> to -1. |
6940069f |
369 | |
370 | =cut |
371 | */ |
372 | |
373 | U8 * |
246fae53 |
374 | Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len) |
6940069f |
375 | { |
376 | dTHR; |
377 | U8 *send; |
378 | U8 *d; |
379 | U8 *save; |
380 | |
246fae53 |
381 | send = s + *len; |
6940069f |
382 | d = save = s; |
246fae53 |
383 | |
384 | /* ensure valid UTF8 and chars < 256 before updating string */ |
385 | while (s < send) { |
386 | U8 c = *s++; |
9f9ab905 |
387 | if (c >= 0x80 && |
388 | ( (s >= send) || ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) { |
67e989fb |
389 | *len = -1; |
8e84507e |
390 | return 0; |
67e989fb |
391 | } |
246fae53 |
392 | } |
393 | s = save; |
6940069f |
394 | while (s < send) { |
395 | if (*s < 0x80) |
396 | *d++ = *s++; |
397 | else { |
ba210ebe |
398 | STRLEN ulen; |
8e84507e |
399 | *d++ = (U8)utf8_to_uv(s, &ulen); |
6940069f |
400 | s += ulen; |
6940069f |
401 | } |
402 | } |
403 | *d = '\0'; |
246fae53 |
404 | *len = d - save; |
6940069f |
405 | return save; |
406 | } |
407 | |
408 | /* |
6662521e |
409 | =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len |
6940069f |
410 | |
411 | Converts a string C<s> of length C<len> from ASCII into UTF8 encoding. |
6662521e |
412 | Returns a pointer to the newly-created string, and sets C<len> to |
413 | reflect the new length. |
6940069f |
414 | |
497711e7 |
415 | =cut |
6940069f |
416 | */ |
417 | |
418 | U8* |
6662521e |
419 | Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len) |
6940069f |
420 | { |
421 | dTHR; |
422 | U8 *send; |
423 | U8 *d; |
424 | U8 *dst; |
6662521e |
425 | send = s + (*len); |
6940069f |
426 | |
6662521e |
427 | Newz(801, d, (*len) * 2 + 1, U8); |
6940069f |
428 | dst = d; |
429 | |
430 | while (s < send) { |
431 | if (*s < 0x80) |
432 | *d++ = *s++; |
433 | else { |
434 | UV uv = *s++; |
435 | *d++ = (( uv >> 6) | 0xc0); |
436 | *d++ = (( uv & 0x3f) | 0x80); |
437 | } |
438 | } |
439 | *d = '\0'; |
6662521e |
440 | *len = d-dst; |
6940069f |
441 | return dst; |
442 | } |
443 | |
a0ed51b3 |
444 | /* |
dea0fc0b |
445 | * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8. |
a0ed51b3 |
446 | * |
447 | * Destination must be pre-extended to 3/2 source. Do not use in-place. |
448 | * We optimize for native, for obvious reasons. */ |
449 | |
450 | U8* |
dea0fc0b |
451 | Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
452 | { |
dea0fc0b |
453 | U8* pend; |
454 | U8* dstart = d; |
455 | |
456 | if (bytelen & 1) |
a7867d0a |
457 | Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen"); |
dea0fc0b |
458 | |
459 | pend = p + bytelen; |
460 | |
a0ed51b3 |
461 | while (p < pend) { |
dea0fc0b |
462 | UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */ |
463 | p += 2; |
a0ed51b3 |
464 | if (uv < 0x80) { |
465 | *d++ = uv; |
466 | continue; |
467 | } |
468 | if (uv < 0x800) { |
469 | *d++ = (( uv >> 6) | 0xc0); |
470 | *d++ = (( uv & 0x3f) | 0x80); |
471 | continue; |
472 | } |
473 | if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */ |
0453d815 |
474 | dTHR; |
dea0fc0b |
475 | UV low = *p++; |
476 | if (low < 0xdc00 || low >= 0xdfff) |
477 | Perl_croak(aTHX_ "Malformed UTF-16 surrogate"); |
a0ed51b3 |
478 | uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000; |
479 | } |
480 | if (uv < 0x10000) { |
481 | *d++ = (( uv >> 12) | 0xe0); |
482 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
483 | *d++ = (( uv & 0x3f) | 0x80); |
484 | continue; |
485 | } |
486 | else { |
487 | *d++ = (( uv >> 18) | 0xf0); |
488 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
489 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
490 | *d++ = (( uv & 0x3f) | 0x80); |
491 | continue; |
492 | } |
493 | } |
dea0fc0b |
494 | *newlen = d - dstart; |
a0ed51b3 |
495 | return d; |
496 | } |
497 | |
498 | /* Note: this one is slightly destructive of the source. */ |
499 | |
500 | U8* |
dea0fc0b |
501 | Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
502 | { |
503 | U8* s = (U8*)p; |
504 | U8* send = s + bytelen; |
505 | while (s < send) { |
506 | U8 tmp = s[0]; |
507 | s[0] = s[1]; |
508 | s[1] = tmp; |
509 | s += 2; |
510 | } |
dea0fc0b |
511 | return utf16_to_utf8(p, d, bytelen, newlen); |
a0ed51b3 |
512 | } |
513 | |
514 | /* for now these are all defined (inefficiently) in terms of the utf8 versions */ |
515 | |
516 | bool |
864dbfa3 |
517 | Perl_is_uni_alnum(pTHX_ U32 c) |
a0ed51b3 |
518 | { |
aa6ffa16 |
519 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
520 | uv_to_utf8(tmpbuf, (UV)c); |
521 | return is_utf8_alnum(tmpbuf); |
522 | } |
523 | |
524 | bool |
b8c5462f |
525 | Perl_is_uni_alnumc(pTHX_ U32 c) |
526 | { |
aa6ffa16 |
527 | U8 tmpbuf[UTF8_MAXLEN]; |
b8c5462f |
528 | uv_to_utf8(tmpbuf, (UV)c); |
529 | return is_utf8_alnumc(tmpbuf); |
530 | } |
531 | |
532 | bool |
864dbfa3 |
533 | Perl_is_uni_idfirst(pTHX_ U32 c) |
a0ed51b3 |
534 | { |
aa6ffa16 |
535 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
536 | uv_to_utf8(tmpbuf, (UV)c); |
537 | return is_utf8_idfirst(tmpbuf); |
538 | } |
539 | |
540 | bool |
864dbfa3 |
541 | Perl_is_uni_alpha(pTHX_ U32 c) |
a0ed51b3 |
542 | { |
aa6ffa16 |
543 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
544 | uv_to_utf8(tmpbuf, (UV)c); |
545 | return is_utf8_alpha(tmpbuf); |
546 | } |
547 | |
548 | bool |
4d61ec05 |
549 | Perl_is_uni_ascii(pTHX_ U32 c) |
550 | { |
aa6ffa16 |
551 | U8 tmpbuf[UTF8_MAXLEN]; |
4d61ec05 |
552 | uv_to_utf8(tmpbuf, (UV)c); |
553 | return is_utf8_ascii(tmpbuf); |
554 | } |
555 | |
556 | bool |
864dbfa3 |
557 | Perl_is_uni_space(pTHX_ U32 c) |
a0ed51b3 |
558 | { |
aa6ffa16 |
559 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
560 | uv_to_utf8(tmpbuf, (UV)c); |
561 | return is_utf8_space(tmpbuf); |
562 | } |
563 | |
564 | bool |
864dbfa3 |
565 | Perl_is_uni_digit(pTHX_ U32 c) |
a0ed51b3 |
566 | { |
aa6ffa16 |
567 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
568 | uv_to_utf8(tmpbuf, (UV)c); |
569 | return is_utf8_digit(tmpbuf); |
570 | } |
571 | |
572 | bool |
864dbfa3 |
573 | Perl_is_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
574 | { |
aa6ffa16 |
575 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
576 | uv_to_utf8(tmpbuf, (UV)c); |
577 | return is_utf8_upper(tmpbuf); |
578 | } |
579 | |
580 | bool |
864dbfa3 |
581 | Perl_is_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
582 | { |
aa6ffa16 |
583 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
584 | uv_to_utf8(tmpbuf, (UV)c); |
585 | return is_utf8_lower(tmpbuf); |
586 | } |
587 | |
588 | bool |
b8c5462f |
589 | Perl_is_uni_cntrl(pTHX_ U32 c) |
590 | { |
aa6ffa16 |
591 | U8 tmpbuf[UTF8_MAXLEN]; |
b8c5462f |
592 | uv_to_utf8(tmpbuf, (UV)c); |
593 | return is_utf8_cntrl(tmpbuf); |
594 | } |
595 | |
596 | bool |
597 | Perl_is_uni_graph(pTHX_ U32 c) |
598 | { |
aa6ffa16 |
599 | U8 tmpbuf[UTF8_MAXLEN]; |
b8c5462f |
600 | uv_to_utf8(tmpbuf, (UV)c); |
601 | return is_utf8_graph(tmpbuf); |
602 | } |
603 | |
604 | bool |
864dbfa3 |
605 | Perl_is_uni_print(pTHX_ U32 c) |
a0ed51b3 |
606 | { |
aa6ffa16 |
607 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
608 | uv_to_utf8(tmpbuf, (UV)c); |
609 | return is_utf8_print(tmpbuf); |
610 | } |
611 | |
b8c5462f |
612 | bool |
f248d071 |
613 | Perl_is_uni_punct(pTHX_ U32 c) |
b8c5462f |
614 | { |
aa6ffa16 |
615 | U8 tmpbuf[UTF8_MAXLEN]; |
b8c5462f |
616 | uv_to_utf8(tmpbuf, (UV)c); |
617 | return is_utf8_punct(tmpbuf); |
618 | } |
619 | |
4d61ec05 |
620 | bool |
621 | Perl_is_uni_xdigit(pTHX_ U32 c) |
622 | { |
aa6ffa16 |
623 | U8 tmpbuf[UTF8_MAXLEN]; |
4d61ec05 |
624 | uv_to_utf8(tmpbuf, (UV)c); |
625 | return is_utf8_xdigit(tmpbuf); |
626 | } |
627 | |
a0ed51b3 |
628 | U32 |
864dbfa3 |
629 | Perl_to_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
630 | { |
aa6ffa16 |
631 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
632 | uv_to_utf8(tmpbuf, (UV)c); |
633 | return to_utf8_upper(tmpbuf); |
634 | } |
635 | |
636 | U32 |
864dbfa3 |
637 | Perl_to_uni_title(pTHX_ U32 c) |
a0ed51b3 |
638 | { |
aa6ffa16 |
639 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
640 | uv_to_utf8(tmpbuf, (UV)c); |
641 | return to_utf8_title(tmpbuf); |
642 | } |
643 | |
644 | U32 |
864dbfa3 |
645 | Perl_to_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
646 | { |
aa6ffa16 |
647 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 |
648 | uv_to_utf8(tmpbuf, (UV)c); |
649 | return to_utf8_lower(tmpbuf); |
650 | } |
651 | |
652 | /* for now these all assume no locale info available for Unicode > 255 */ |
653 | |
654 | bool |
864dbfa3 |
655 | Perl_is_uni_alnum_lc(pTHX_ U32 c) |
a0ed51b3 |
656 | { |
657 | return is_uni_alnum(c); /* XXX no locale support yet */ |
658 | } |
659 | |
660 | bool |
b8c5462f |
661 | Perl_is_uni_alnumc_lc(pTHX_ U32 c) |
662 | { |
663 | return is_uni_alnumc(c); /* XXX no locale support yet */ |
664 | } |
665 | |
666 | bool |
864dbfa3 |
667 | Perl_is_uni_idfirst_lc(pTHX_ U32 c) |
a0ed51b3 |
668 | { |
669 | return is_uni_idfirst(c); /* XXX no locale support yet */ |
670 | } |
671 | |
672 | bool |
864dbfa3 |
673 | Perl_is_uni_alpha_lc(pTHX_ U32 c) |
a0ed51b3 |
674 | { |
675 | return is_uni_alpha(c); /* XXX no locale support yet */ |
676 | } |
677 | |
678 | bool |
4d61ec05 |
679 | Perl_is_uni_ascii_lc(pTHX_ U32 c) |
680 | { |
681 | return is_uni_ascii(c); /* XXX no locale support yet */ |
682 | } |
683 | |
684 | bool |
864dbfa3 |
685 | Perl_is_uni_space_lc(pTHX_ U32 c) |
a0ed51b3 |
686 | { |
687 | return is_uni_space(c); /* XXX no locale support yet */ |
688 | } |
689 | |
690 | bool |
864dbfa3 |
691 | Perl_is_uni_digit_lc(pTHX_ U32 c) |
a0ed51b3 |
692 | { |
693 | return is_uni_digit(c); /* XXX no locale support yet */ |
694 | } |
695 | |
696 | bool |
864dbfa3 |
697 | Perl_is_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
698 | { |
699 | return is_uni_upper(c); /* XXX no locale support yet */ |
700 | } |
701 | |
702 | bool |
864dbfa3 |
703 | Perl_is_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
704 | { |
705 | return is_uni_lower(c); /* XXX no locale support yet */ |
706 | } |
707 | |
708 | bool |
b8c5462f |
709 | Perl_is_uni_cntrl_lc(pTHX_ U32 c) |
710 | { |
711 | return is_uni_cntrl(c); /* XXX no locale support yet */ |
712 | } |
713 | |
714 | bool |
715 | Perl_is_uni_graph_lc(pTHX_ U32 c) |
716 | { |
717 | return is_uni_graph(c); /* XXX no locale support yet */ |
718 | } |
719 | |
720 | bool |
864dbfa3 |
721 | Perl_is_uni_print_lc(pTHX_ U32 c) |
a0ed51b3 |
722 | { |
723 | return is_uni_print(c); /* XXX no locale support yet */ |
724 | } |
725 | |
b8c5462f |
726 | bool |
727 | Perl_is_uni_punct_lc(pTHX_ U32 c) |
728 | { |
729 | return is_uni_punct(c); /* XXX no locale support yet */ |
730 | } |
731 | |
4d61ec05 |
732 | bool |
733 | Perl_is_uni_xdigit_lc(pTHX_ U32 c) |
734 | { |
735 | return is_uni_xdigit(c); /* XXX no locale support yet */ |
736 | } |
737 | |
a0ed51b3 |
738 | U32 |
864dbfa3 |
739 | Perl_to_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
740 | { |
741 | return to_uni_upper(c); /* XXX no locale support yet */ |
742 | } |
743 | |
744 | U32 |
864dbfa3 |
745 | Perl_to_uni_title_lc(pTHX_ U32 c) |
a0ed51b3 |
746 | { |
747 | return to_uni_title(c); /* XXX no locale support yet */ |
748 | } |
749 | |
750 | U32 |
864dbfa3 |
751 | Perl_to_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
752 | { |
753 | return to_uni_lower(c); /* XXX no locale support yet */ |
754 | } |
755 | |
a0ed51b3 |
756 | bool |
864dbfa3 |
757 | Perl_is_utf8_alnum(pTHX_ U8 *p) |
a0ed51b3 |
758 | { |
386d01d6 |
759 | if (!is_utf8_char(p)) |
760 | return FALSE; |
a0ed51b3 |
761 | if (!PL_utf8_alnum) |
289d4f09 |
762 | /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true |
763 | * descendant of isalnum(3), in other words, it doesn't |
764 | * contain the '_'. --jhi */ |
765 | PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0); |
a0ed51b3 |
766 | return swash_fetch(PL_utf8_alnum, p); |
767 | /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */ |
768 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
769 | if (!PL_utf8_alnum) |
770 | PL_utf8_alnum = swash_init("utf8", "", |
771 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
772 | return swash_fetch(PL_utf8_alnum, p); |
773 | #endif |
774 | } |
775 | |
776 | bool |
b8c5462f |
777 | Perl_is_utf8_alnumc(pTHX_ U8 *p) |
778 | { |
386d01d6 |
779 | if (!is_utf8_char(p)) |
780 | return FALSE; |
b8c5462f |
781 | if (!PL_utf8_alnum) |
782 | PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0); |
783 | return swash_fetch(PL_utf8_alnum, p); |
784 | /* return is_utf8_alpha(p) || is_utf8_digit(p); */ |
785 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
786 | if (!PL_utf8_alnum) |
787 | PL_utf8_alnum = swash_init("utf8", "", |
788 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
789 | return swash_fetch(PL_utf8_alnum, p); |
790 | #endif |
791 | } |
792 | |
793 | bool |
864dbfa3 |
794 | Perl_is_utf8_idfirst(pTHX_ U8 *p) |
a0ed51b3 |
795 | { |
796 | return *p == '_' || is_utf8_alpha(p); |
797 | } |
798 | |
799 | bool |
864dbfa3 |
800 | Perl_is_utf8_alpha(pTHX_ U8 *p) |
a0ed51b3 |
801 | { |
386d01d6 |
802 | if (!is_utf8_char(p)) |
803 | return FALSE; |
a0ed51b3 |
804 | if (!PL_utf8_alpha) |
e24b16f9 |
805 | PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0); |
a0ed51b3 |
806 | return swash_fetch(PL_utf8_alpha, p); |
807 | } |
808 | |
809 | bool |
b8c5462f |
810 | Perl_is_utf8_ascii(pTHX_ U8 *p) |
811 | { |
386d01d6 |
812 | if (!is_utf8_char(p)) |
813 | return FALSE; |
b8c5462f |
814 | if (!PL_utf8_ascii) |
815 | PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0); |
816 | return swash_fetch(PL_utf8_ascii, p); |
817 | } |
818 | |
819 | bool |
864dbfa3 |
820 | Perl_is_utf8_space(pTHX_ U8 *p) |
a0ed51b3 |
821 | { |
386d01d6 |
822 | if (!is_utf8_char(p)) |
823 | return FALSE; |
a0ed51b3 |
824 | if (!PL_utf8_space) |
e24b16f9 |
825 | PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0); |
a0ed51b3 |
826 | return swash_fetch(PL_utf8_space, p); |
827 | } |
828 | |
829 | bool |
864dbfa3 |
830 | Perl_is_utf8_digit(pTHX_ U8 *p) |
a0ed51b3 |
831 | { |
386d01d6 |
832 | if (!is_utf8_char(p)) |
833 | return FALSE; |
a0ed51b3 |
834 | if (!PL_utf8_digit) |
e24b16f9 |
835 | PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0); |
a0ed51b3 |
836 | return swash_fetch(PL_utf8_digit, p); |
837 | } |
838 | |
839 | bool |
864dbfa3 |
840 | Perl_is_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
841 | { |
386d01d6 |
842 | if (!is_utf8_char(p)) |
843 | return FALSE; |
a0ed51b3 |
844 | if (!PL_utf8_upper) |
e24b16f9 |
845 | PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0); |
a0ed51b3 |
846 | return swash_fetch(PL_utf8_upper, p); |
847 | } |
848 | |
849 | bool |
864dbfa3 |
850 | Perl_is_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
851 | { |
386d01d6 |
852 | if (!is_utf8_char(p)) |
853 | return FALSE; |
a0ed51b3 |
854 | if (!PL_utf8_lower) |
e24b16f9 |
855 | PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0); |
a0ed51b3 |
856 | return swash_fetch(PL_utf8_lower, p); |
857 | } |
858 | |
859 | bool |
b8c5462f |
860 | Perl_is_utf8_cntrl(pTHX_ U8 *p) |
861 | { |
386d01d6 |
862 | if (!is_utf8_char(p)) |
863 | return FALSE; |
b8c5462f |
864 | if (!PL_utf8_cntrl) |
865 | PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0); |
866 | return swash_fetch(PL_utf8_cntrl, p); |
867 | } |
868 | |
869 | bool |
870 | Perl_is_utf8_graph(pTHX_ U8 *p) |
871 | { |
386d01d6 |
872 | if (!is_utf8_char(p)) |
873 | return FALSE; |
b8c5462f |
874 | if (!PL_utf8_graph) |
875 | PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0); |
876 | return swash_fetch(PL_utf8_graph, p); |
877 | } |
878 | |
879 | bool |
864dbfa3 |
880 | Perl_is_utf8_print(pTHX_ U8 *p) |
a0ed51b3 |
881 | { |
386d01d6 |
882 | if (!is_utf8_char(p)) |
883 | return FALSE; |
a0ed51b3 |
884 | if (!PL_utf8_print) |
e24b16f9 |
885 | PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0); |
a0ed51b3 |
886 | return swash_fetch(PL_utf8_print, p); |
887 | } |
888 | |
889 | bool |
b8c5462f |
890 | Perl_is_utf8_punct(pTHX_ U8 *p) |
891 | { |
386d01d6 |
892 | if (!is_utf8_char(p)) |
893 | return FALSE; |
b8c5462f |
894 | if (!PL_utf8_punct) |
895 | PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0); |
896 | return swash_fetch(PL_utf8_punct, p); |
897 | } |
898 | |
899 | bool |
900 | Perl_is_utf8_xdigit(pTHX_ U8 *p) |
901 | { |
386d01d6 |
902 | if (!is_utf8_char(p)) |
903 | return FALSE; |
b8c5462f |
904 | if (!PL_utf8_xdigit) |
905 | PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0); |
906 | return swash_fetch(PL_utf8_xdigit, p); |
907 | } |
908 | |
909 | bool |
864dbfa3 |
910 | Perl_is_utf8_mark(pTHX_ U8 *p) |
a0ed51b3 |
911 | { |
386d01d6 |
912 | if (!is_utf8_char(p)) |
913 | return FALSE; |
a0ed51b3 |
914 | if (!PL_utf8_mark) |
e24b16f9 |
915 | PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0); |
a0ed51b3 |
916 | return swash_fetch(PL_utf8_mark, p); |
917 | } |
918 | |
2104c8d9 |
919 | UV |
864dbfa3 |
920 | Perl_to_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
921 | { |
922 | UV uv; |
923 | |
924 | if (!PL_utf8_toupper) |
e24b16f9 |
925 | PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0); |
a0ed51b3 |
926 | uv = swash_fetch(PL_utf8_toupper, p); |
ba210ebe |
927 | return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0); |
a0ed51b3 |
928 | } |
929 | |
2104c8d9 |
930 | UV |
864dbfa3 |
931 | Perl_to_utf8_title(pTHX_ U8 *p) |
a0ed51b3 |
932 | { |
933 | UV uv; |
934 | |
935 | if (!PL_utf8_totitle) |
e24b16f9 |
936 | PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0); |
a0ed51b3 |
937 | uv = swash_fetch(PL_utf8_totitle, p); |
ba210ebe |
938 | return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0); |
a0ed51b3 |
939 | } |
940 | |
2104c8d9 |
941 | UV |
864dbfa3 |
942 | Perl_to_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
943 | { |
944 | UV uv; |
945 | |
946 | if (!PL_utf8_tolower) |
e24b16f9 |
947 | PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0); |
a0ed51b3 |
948 | uv = swash_fetch(PL_utf8_tolower, p); |
ba210ebe |
949 | return uv ? uv : utf8_to_uv_chk(p,STRLEN_MAX,0,0); |
a0ed51b3 |
950 | } |
951 | |
952 | /* a "swash" is a swatch hash */ |
953 | |
954 | SV* |
864dbfa3 |
955 | Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) |
a0ed51b3 |
956 | { |
957 | SV* retval; |
958 | char tmpbuf[256]; |
8e84507e |
959 | dSP; |
ce3b816e |
960 | |
961 | if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */ |
962 | ENTER; |
963 | Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv); |
964 | LEAVE; |
965 | } |
966 | SPAGAIN; |
a0ed51b3 |
967 | PUSHSTACKi(PERLSI_MAGIC); |
968 | PUSHMARK(SP); |
969 | EXTEND(SP,5); |
970 | PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg)))); |
971 | PUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
972 | PUSHs(listsv); |
973 | PUSHs(sv_2mortal(newSViv(minbits))); |
974 | PUSHs(sv_2mortal(newSViv(none))); |
975 | PUTBACK; |
976 | ENTER; |
977 | SAVEI32(PL_hints); |
978 | PL_hints = 0; |
979 | save_re_context(); |
e24b16f9 |
980 | if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */ |
a0ed51b3 |
981 | strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf); |
864dbfa3 |
982 | if (call_method("SWASHNEW", G_SCALAR)) |
8e84507e |
983 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
984 | else |
e24b16f9 |
985 | retval = &PL_sv_undef; |
a0ed51b3 |
986 | LEAVE; |
987 | POPSTACK; |
e24b16f9 |
988 | if (PL_curcop == &PL_compiling) { |
a0ed51b3 |
989 | strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf); |
e24b16f9 |
990 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
991 | } |
992 | if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) |
cea2e8a9 |
993 | Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref"); |
a0ed51b3 |
994 | return retval; |
995 | } |
996 | |
997 | UV |
864dbfa3 |
998 | Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) |
a0ed51b3 |
999 | { |
1000 | HV* hv = (HV*)SvRV(sv); |
1001 | U32 klen = UTF8SKIP(ptr) - 1; |
1002 | U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */ |
1003 | STRLEN slen; |
1004 | STRLEN needents = (klen ? 64 : 128); |
dfe13c55 |
1005 | U8 *tmps; |
a0ed51b3 |
1006 | U32 bit; |
1007 | SV *retval; |
1008 | |
1009 | /* |
1010 | * This single-entry cache saves about 1/3 of the utf8 overhead in test |
1011 | * suite. (That is, only 7-8% overall over just a hash cache. Still, |
1012 | * it's nothing to sniff at.) Pity we usually come through at least |
1013 | * two function calls to get here... |
1014 | * |
1015 | * NB: this code assumes that swatches are never modified, once generated! |
1016 | */ |
1017 | |
1018 | if (hv == PL_last_swash_hv && |
1019 | klen == PL_last_swash_klen && |
12ae5dfc |
1020 | (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) ) |
a0ed51b3 |
1021 | { |
1022 | tmps = PL_last_swash_tmps; |
1023 | slen = PL_last_swash_slen; |
1024 | } |
1025 | else { |
1026 | /* Try our second-level swatch cache, kept in a hash. */ |
dfe13c55 |
1027 | SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE); |
a0ed51b3 |
1028 | |
1029 | /* If not cached, generate it via utf8::SWASHGET */ |
dfe13c55 |
1030 | if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { |
a0ed51b3 |
1031 | dSP; |
1032 | ENTER; |
1033 | SAVETMPS; |
1034 | save_re_context(); |
1035 | PUSHSTACKi(PERLSI_MAGIC); |
1036 | PUSHMARK(SP); |
1037 | EXTEND(SP,3); |
1038 | PUSHs((SV*)sv); |
ba210ebe |
1039 | PUSHs(sv_2mortal(newSViv(utf8_to_uv_chk(ptr, STRLEN_MAX, 0, 0) & ~(needents - 1)))); |
a0ed51b3 |
1040 | PUSHs(sv_2mortal(newSViv(needents))); |
1041 | PUTBACK; |
864dbfa3 |
1042 | if (call_method("SWASHGET", G_SCALAR)) |
8e84507e |
1043 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
1044 | else |
e24b16f9 |
1045 | retval = &PL_sv_undef; |
a0ed51b3 |
1046 | POPSTACK; |
1047 | FREETMPS; |
1048 | LEAVE; |
e24b16f9 |
1049 | if (PL_curcop == &PL_compiling) |
1050 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
1051 | |
dfe13c55 |
1052 | svp = hv_store(hv, (char*)ptr, klen, retval, 0); |
a0ed51b3 |
1053 | |
dfe13c55 |
1054 | if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8) |
cea2e8a9 |
1055 | Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); |
a0ed51b3 |
1056 | } |
1057 | |
1058 | PL_last_swash_hv = hv; |
1059 | PL_last_swash_klen = klen; |
1060 | PL_last_swash_tmps = tmps; |
1061 | PL_last_swash_slen = slen; |
1062 | if (klen) |
1063 | Copy(ptr, PL_last_swash_key, klen, U8); |
1064 | } |
1065 | |
1066 | switch ((slen << 3) / needents) { |
1067 | case 1: |
1068 | bit = 1 << (off & 7); |
1069 | off >>= 3; |
1070 | return (tmps[off] & bit) != 0; |
1071 | case 8: |
1072 | return tmps[off]; |
1073 | case 16: |
1074 | off <<= 1; |
1075 | return (tmps[off] << 8) + tmps[off + 1] ; |
1076 | case 32: |
1077 | off <<= 2; |
1078 | return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ; |
1079 | } |
cea2e8a9 |
1080 | Perl_croak(aTHX_ "panic: swash_fetch"); |
a0ed51b3 |
1081 | return 0; |
1082 | } |