Commit | Line | Data |
a0ed51b3 |
1 | /* utf8.c |
2 | * |
4eb8286e |
3 | * Copyright (c) 1998-1999, 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 | } |
71 | #ifdef Quad_t |
72 | if (uv < 0x2000000000) |
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 | } |
84 | #ifdef Quad_t |
85 | { |
86 | *d++ = 0xff; /* Can't match U+FFFE! */ |
87 | *d++ = (((uv >> 36) & 0x3f) | 0x80); |
88 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
89 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
90 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
91 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
92 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
93 | *d++ = (( uv & 0x3f) | 0x80); |
94 | return d; |
95 | } |
96 | #endif |
97 | } |
98 | |
99 | UV |
864dbfa3 |
100 | Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen) |
a0ed51b3 |
101 | { |
102 | UV uv = *s; |
103 | int len; |
104 | if (!(uv & 0x80)) { |
105 | if (retlen) |
106 | *retlen = 1; |
107 | return *s; |
108 | } |
109 | if (!(uv & 0x40)) { |
cea2e8a9 |
110 | Perl_warn(aTHX_ "Malformed UTF-8 character"); |
a0ed51b3 |
111 | if (retlen) |
112 | *retlen = 1; |
113 | return *s; |
114 | } |
115 | |
116 | if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } |
117 | else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } |
118 | else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } |
119 | else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } |
120 | else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } |
121 | else if (!(uv & 0x01)) { len = 7; uv &= 0x00; } |
122 | else len = 8; /* whoa! */ |
123 | |
124 | if (retlen) |
125 | *retlen = len; |
126 | --len; |
127 | s++; |
128 | while (len--) { |
129 | if ((*s & 0xc0) != 0x80) { |
cea2e8a9 |
130 | Perl_warn(aTHX_ "Malformed UTF-8 character"); |
a0ed51b3 |
131 | if (retlen) |
132 | *retlen -= len + 1; |
133 | return 0xfffd; |
134 | } |
135 | else |
136 | uv = (uv << 6) | (*s++ & 0x3f); |
137 | } |
138 | return uv; |
139 | } |
140 | |
141 | /* utf8_distance(a,b) is intended to be a - b in pointer arithmetic */ |
142 | |
143 | I32 |
864dbfa3 |
144 | Perl_utf8_distance(pTHX_ U8 *a, U8 *b) |
a0ed51b3 |
145 | { |
146 | I32 off = 0; |
147 | if (a < b) { |
148 | while (a < b) { |
149 | a += UTF8SKIP(a); |
150 | off--; |
151 | } |
152 | } |
153 | else { |
154 | while (b < a) { |
155 | b += UTF8SKIP(b); |
156 | off++; |
157 | } |
158 | } |
159 | return off; |
160 | } |
161 | |
162 | /* WARNING: do not use the following unless you *know* off is within bounds */ |
163 | |
164 | U8 * |
864dbfa3 |
165 | Perl_utf8_hop(pTHX_ U8 *s, I32 off) |
a0ed51b3 |
166 | { |
167 | if (off >= 0) { |
168 | while (off--) |
169 | s += UTF8SKIP(s); |
170 | } |
171 | else { |
172 | while (off++) { |
173 | s--; |
174 | if (*s & 0x80) { |
175 | while ((*s & 0xc0) == 0x80) |
176 | s--; |
177 | } |
178 | } |
179 | } |
180 | return s; |
181 | } |
182 | |
183 | /* XXX NOTHING CALLS THE FOLLOWING TWO ROUTINES YET!!! */ |
184 | /* |
185 | * Convert native or reversed UTF-16 to UTF-8. |
186 | * |
187 | * Destination must be pre-extended to 3/2 source. Do not use in-place. |
188 | * We optimize for native, for obvious reasons. */ |
189 | |
190 | U8* |
864dbfa3 |
191 | Perl_utf16_to_utf8(pTHX_ U16* p, U8* d, I32 bytelen) |
a0ed51b3 |
192 | { |
193 | U16* pend = p + bytelen / 2; |
194 | while (p < pend) { |
195 | UV uv = *p++; |
196 | if (uv < 0x80) { |
197 | *d++ = uv; |
198 | continue; |
199 | } |
200 | if (uv < 0x800) { |
201 | *d++ = (( uv >> 6) | 0xc0); |
202 | *d++ = (( uv & 0x3f) | 0x80); |
203 | continue; |
204 | } |
205 | if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */ |
206 | int low = *p++; |
207 | if (low < 0xdc00 || low >= 0xdfff) { |
cea2e8a9 |
208 | Perl_warn(aTHX_ "Malformed UTF-16 surrogate"); |
a0ed51b3 |
209 | p--; |
210 | uv = 0xfffd; |
211 | } |
212 | uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000; |
213 | } |
214 | if (uv < 0x10000) { |
215 | *d++ = (( uv >> 12) | 0xe0); |
216 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
217 | *d++ = (( uv & 0x3f) | 0x80); |
218 | continue; |
219 | } |
220 | else { |
221 | *d++ = (( uv >> 18) | 0xf0); |
222 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
223 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
224 | *d++ = (( uv & 0x3f) | 0x80); |
225 | continue; |
226 | } |
227 | } |
228 | return d; |
229 | } |
230 | |
231 | /* Note: this one is slightly destructive of the source. */ |
232 | |
233 | U8* |
864dbfa3 |
234 | Perl_utf16_to_utf8_reversed(pTHX_ U16* p, U8* d, I32 bytelen) |
a0ed51b3 |
235 | { |
236 | U8* s = (U8*)p; |
237 | U8* send = s + bytelen; |
238 | while (s < send) { |
239 | U8 tmp = s[0]; |
240 | s[0] = s[1]; |
241 | s[1] = tmp; |
242 | s += 2; |
243 | } |
244 | return utf16_to_utf8(p, d, bytelen); |
245 | } |
246 | |
247 | /* for now these are all defined (inefficiently) in terms of the utf8 versions */ |
248 | |
249 | bool |
864dbfa3 |
250 | Perl_is_uni_alnum(pTHX_ U32 c) |
a0ed51b3 |
251 | { |
dfe13c55 |
252 | U8 tmpbuf[10]; |
a0ed51b3 |
253 | uv_to_utf8(tmpbuf, (UV)c); |
254 | return is_utf8_alnum(tmpbuf); |
255 | } |
256 | |
257 | bool |
864dbfa3 |
258 | Perl_is_uni_idfirst(pTHX_ U32 c) |
a0ed51b3 |
259 | { |
dfe13c55 |
260 | U8 tmpbuf[10]; |
a0ed51b3 |
261 | uv_to_utf8(tmpbuf, (UV)c); |
262 | return is_utf8_idfirst(tmpbuf); |
263 | } |
264 | |
265 | bool |
864dbfa3 |
266 | Perl_is_uni_alpha(pTHX_ U32 c) |
a0ed51b3 |
267 | { |
dfe13c55 |
268 | U8 tmpbuf[10]; |
a0ed51b3 |
269 | uv_to_utf8(tmpbuf, (UV)c); |
270 | return is_utf8_alpha(tmpbuf); |
271 | } |
272 | |
273 | bool |
864dbfa3 |
274 | Perl_is_uni_space(pTHX_ U32 c) |
a0ed51b3 |
275 | { |
dfe13c55 |
276 | U8 tmpbuf[10]; |
a0ed51b3 |
277 | uv_to_utf8(tmpbuf, (UV)c); |
278 | return is_utf8_space(tmpbuf); |
279 | } |
280 | |
281 | bool |
864dbfa3 |
282 | Perl_is_uni_digit(pTHX_ U32 c) |
a0ed51b3 |
283 | { |
dfe13c55 |
284 | U8 tmpbuf[10]; |
a0ed51b3 |
285 | uv_to_utf8(tmpbuf, (UV)c); |
286 | return is_utf8_digit(tmpbuf); |
287 | } |
288 | |
289 | bool |
864dbfa3 |
290 | Perl_is_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
291 | { |
dfe13c55 |
292 | U8 tmpbuf[10]; |
a0ed51b3 |
293 | uv_to_utf8(tmpbuf, (UV)c); |
294 | return is_utf8_upper(tmpbuf); |
295 | } |
296 | |
297 | bool |
864dbfa3 |
298 | Perl_is_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
299 | { |
dfe13c55 |
300 | U8 tmpbuf[10]; |
a0ed51b3 |
301 | uv_to_utf8(tmpbuf, (UV)c); |
302 | return is_utf8_lower(tmpbuf); |
303 | } |
304 | |
305 | bool |
864dbfa3 |
306 | Perl_is_uni_print(pTHX_ U32 c) |
a0ed51b3 |
307 | { |
dfe13c55 |
308 | U8 tmpbuf[10]; |
a0ed51b3 |
309 | uv_to_utf8(tmpbuf, (UV)c); |
310 | return is_utf8_print(tmpbuf); |
311 | } |
312 | |
313 | U32 |
864dbfa3 |
314 | Perl_to_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
315 | { |
dfe13c55 |
316 | U8 tmpbuf[10]; |
a0ed51b3 |
317 | uv_to_utf8(tmpbuf, (UV)c); |
318 | return to_utf8_upper(tmpbuf); |
319 | } |
320 | |
321 | U32 |
864dbfa3 |
322 | Perl_to_uni_title(pTHX_ U32 c) |
a0ed51b3 |
323 | { |
dfe13c55 |
324 | U8 tmpbuf[10]; |
a0ed51b3 |
325 | uv_to_utf8(tmpbuf, (UV)c); |
326 | return to_utf8_title(tmpbuf); |
327 | } |
328 | |
329 | U32 |
864dbfa3 |
330 | Perl_to_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
331 | { |
dfe13c55 |
332 | U8 tmpbuf[10]; |
a0ed51b3 |
333 | uv_to_utf8(tmpbuf, (UV)c); |
334 | return to_utf8_lower(tmpbuf); |
335 | } |
336 | |
337 | /* for now these all assume no locale info available for Unicode > 255 */ |
338 | |
339 | bool |
864dbfa3 |
340 | Perl_is_uni_alnum_lc(pTHX_ U32 c) |
a0ed51b3 |
341 | { |
342 | return is_uni_alnum(c); /* XXX no locale support yet */ |
343 | } |
344 | |
345 | bool |
864dbfa3 |
346 | Perl_is_uni_idfirst_lc(pTHX_ U32 c) |
a0ed51b3 |
347 | { |
348 | return is_uni_idfirst(c); /* XXX no locale support yet */ |
349 | } |
350 | |
351 | bool |
864dbfa3 |
352 | Perl_is_uni_alpha_lc(pTHX_ U32 c) |
a0ed51b3 |
353 | { |
354 | return is_uni_alpha(c); /* XXX no locale support yet */ |
355 | } |
356 | |
357 | bool |
864dbfa3 |
358 | Perl_is_uni_space_lc(pTHX_ U32 c) |
a0ed51b3 |
359 | { |
360 | return is_uni_space(c); /* XXX no locale support yet */ |
361 | } |
362 | |
363 | bool |
864dbfa3 |
364 | Perl_is_uni_digit_lc(pTHX_ U32 c) |
a0ed51b3 |
365 | { |
366 | return is_uni_digit(c); /* XXX no locale support yet */ |
367 | } |
368 | |
369 | bool |
864dbfa3 |
370 | Perl_is_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
371 | { |
372 | return is_uni_upper(c); /* XXX no locale support yet */ |
373 | } |
374 | |
375 | bool |
864dbfa3 |
376 | Perl_is_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
377 | { |
378 | return is_uni_lower(c); /* XXX no locale support yet */ |
379 | } |
380 | |
381 | bool |
864dbfa3 |
382 | Perl_is_uni_print_lc(pTHX_ U32 c) |
a0ed51b3 |
383 | { |
384 | return is_uni_print(c); /* XXX no locale support yet */ |
385 | } |
386 | |
387 | U32 |
864dbfa3 |
388 | Perl_to_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
389 | { |
390 | return to_uni_upper(c); /* XXX no locale support yet */ |
391 | } |
392 | |
393 | U32 |
864dbfa3 |
394 | Perl_to_uni_title_lc(pTHX_ U32 c) |
a0ed51b3 |
395 | { |
396 | return to_uni_title(c); /* XXX no locale support yet */ |
397 | } |
398 | |
399 | U32 |
864dbfa3 |
400 | Perl_to_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
401 | { |
402 | return to_uni_lower(c); /* XXX no locale support yet */ |
403 | } |
404 | |
405 | |
406 | bool |
864dbfa3 |
407 | Perl_is_utf8_alnum(pTHX_ U8 *p) |
a0ed51b3 |
408 | { |
409 | if (!PL_utf8_alnum) |
e24b16f9 |
410 | PL_utf8_alnum = swash_init("utf8", "IsAlnum", &PL_sv_undef, 0, 0); |
a0ed51b3 |
411 | return swash_fetch(PL_utf8_alnum, p); |
412 | /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */ |
413 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
414 | if (!PL_utf8_alnum) |
415 | PL_utf8_alnum = swash_init("utf8", "", |
416 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
417 | return swash_fetch(PL_utf8_alnum, p); |
418 | #endif |
419 | } |
420 | |
421 | bool |
864dbfa3 |
422 | Perl_is_utf8_idfirst(pTHX_ U8 *p) |
a0ed51b3 |
423 | { |
424 | return *p == '_' || is_utf8_alpha(p); |
425 | } |
426 | |
427 | bool |
864dbfa3 |
428 | Perl_is_utf8_alpha(pTHX_ U8 *p) |
a0ed51b3 |
429 | { |
430 | if (!PL_utf8_alpha) |
e24b16f9 |
431 | PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0); |
a0ed51b3 |
432 | return swash_fetch(PL_utf8_alpha, p); |
433 | } |
434 | |
435 | bool |
864dbfa3 |
436 | Perl_is_utf8_space(pTHX_ U8 *p) |
a0ed51b3 |
437 | { |
438 | if (!PL_utf8_space) |
e24b16f9 |
439 | PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0); |
a0ed51b3 |
440 | return swash_fetch(PL_utf8_space, p); |
441 | } |
442 | |
443 | bool |
864dbfa3 |
444 | Perl_is_utf8_digit(pTHX_ U8 *p) |
a0ed51b3 |
445 | { |
446 | if (!PL_utf8_digit) |
e24b16f9 |
447 | PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0); |
a0ed51b3 |
448 | return swash_fetch(PL_utf8_digit, p); |
449 | } |
450 | |
451 | bool |
864dbfa3 |
452 | Perl_is_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
453 | { |
454 | if (!PL_utf8_upper) |
e24b16f9 |
455 | PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0); |
a0ed51b3 |
456 | return swash_fetch(PL_utf8_upper, p); |
457 | } |
458 | |
459 | bool |
864dbfa3 |
460 | Perl_is_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
461 | { |
462 | if (!PL_utf8_lower) |
e24b16f9 |
463 | PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0); |
a0ed51b3 |
464 | return swash_fetch(PL_utf8_lower, p); |
465 | } |
466 | |
467 | bool |
864dbfa3 |
468 | Perl_is_utf8_print(pTHX_ U8 *p) |
a0ed51b3 |
469 | { |
470 | if (!PL_utf8_print) |
e24b16f9 |
471 | PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0); |
a0ed51b3 |
472 | return swash_fetch(PL_utf8_print, p); |
473 | } |
474 | |
475 | bool |
864dbfa3 |
476 | Perl_is_utf8_mark(pTHX_ U8 *p) |
a0ed51b3 |
477 | { |
478 | if (!PL_utf8_mark) |
e24b16f9 |
479 | PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0); |
a0ed51b3 |
480 | return swash_fetch(PL_utf8_mark, p); |
481 | } |
482 | |
2104c8d9 |
483 | UV |
864dbfa3 |
484 | Perl_to_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
485 | { |
486 | UV uv; |
487 | |
488 | if (!PL_utf8_toupper) |
e24b16f9 |
489 | PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0); |
a0ed51b3 |
490 | uv = swash_fetch(PL_utf8_toupper, p); |
491 | return uv ? uv : utf8_to_uv(p,0); |
492 | } |
493 | |
2104c8d9 |
494 | UV |
864dbfa3 |
495 | Perl_to_utf8_title(pTHX_ U8 *p) |
a0ed51b3 |
496 | { |
497 | UV uv; |
498 | |
499 | if (!PL_utf8_totitle) |
e24b16f9 |
500 | PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0); |
a0ed51b3 |
501 | uv = swash_fetch(PL_utf8_totitle, p); |
502 | return uv ? uv : utf8_to_uv(p,0); |
503 | } |
504 | |
2104c8d9 |
505 | UV |
864dbfa3 |
506 | Perl_to_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
507 | { |
508 | UV uv; |
509 | |
510 | if (!PL_utf8_tolower) |
e24b16f9 |
511 | PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0); |
a0ed51b3 |
512 | uv = swash_fetch(PL_utf8_tolower, p); |
513 | return uv ? uv : utf8_to_uv(p,0); |
514 | } |
515 | |
516 | /* a "swash" is a swatch hash */ |
517 | |
518 | SV* |
864dbfa3 |
519 | Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) |
a0ed51b3 |
520 | { |
521 | SV* retval; |
522 | char tmpbuf[256]; |
523 | dSP; |
524 | PUSHSTACKi(PERLSI_MAGIC); |
525 | PUSHMARK(SP); |
526 | EXTEND(SP,5); |
527 | PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg)))); |
528 | PUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
529 | PUSHs(listsv); |
530 | PUSHs(sv_2mortal(newSViv(minbits))); |
531 | PUSHs(sv_2mortal(newSViv(none))); |
532 | PUTBACK; |
533 | ENTER; |
534 | SAVEI32(PL_hints); |
535 | PL_hints = 0; |
536 | save_re_context(); |
e24b16f9 |
537 | if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */ |
a0ed51b3 |
538 | strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf); |
864dbfa3 |
539 | if (call_method("SWASHNEW", G_SCALAR)) |
e24b16f9 |
540 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
541 | else |
e24b16f9 |
542 | retval = &PL_sv_undef; |
a0ed51b3 |
543 | LEAVE; |
544 | POPSTACK; |
e24b16f9 |
545 | if (PL_curcop == &PL_compiling) { |
a0ed51b3 |
546 | strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf); |
e24b16f9 |
547 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
548 | } |
549 | if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) |
cea2e8a9 |
550 | Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref"); |
a0ed51b3 |
551 | return retval; |
552 | } |
553 | |
554 | UV |
864dbfa3 |
555 | Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) |
a0ed51b3 |
556 | { |
557 | HV* hv = (HV*)SvRV(sv); |
558 | U32 klen = UTF8SKIP(ptr) - 1; |
559 | U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */ |
560 | STRLEN slen; |
561 | STRLEN needents = (klen ? 64 : 128); |
dfe13c55 |
562 | U8 *tmps; |
a0ed51b3 |
563 | U32 bit; |
564 | SV *retval; |
565 | |
566 | /* |
567 | * This single-entry cache saves about 1/3 of the utf8 overhead in test |
568 | * suite. (That is, only 7-8% overall over just a hash cache. Still, |
569 | * it's nothing to sniff at.) Pity we usually come through at least |
570 | * two function calls to get here... |
571 | * |
572 | * NB: this code assumes that swatches are never modified, once generated! |
573 | */ |
574 | |
575 | if (hv == PL_last_swash_hv && |
576 | klen == PL_last_swash_klen && |
577 | (!klen || memEQ(ptr,PL_last_swash_key,klen)) ) |
578 | { |
579 | tmps = PL_last_swash_tmps; |
580 | slen = PL_last_swash_slen; |
581 | } |
582 | else { |
583 | /* Try our second-level swatch cache, kept in a hash. */ |
dfe13c55 |
584 | SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE); |
a0ed51b3 |
585 | |
586 | /* If not cached, generate it via utf8::SWASHGET */ |
dfe13c55 |
587 | if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { |
a0ed51b3 |
588 | dSP; |
589 | ENTER; |
590 | SAVETMPS; |
591 | save_re_context(); |
592 | PUSHSTACKi(PERLSI_MAGIC); |
593 | PUSHMARK(SP); |
594 | EXTEND(SP,3); |
595 | PUSHs((SV*)sv); |
596 | PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, 0) & ~(needents - 1)))); |
597 | PUSHs(sv_2mortal(newSViv(needents))); |
598 | PUTBACK; |
864dbfa3 |
599 | if (call_method("SWASHGET", G_SCALAR)) |
e24b16f9 |
600 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
601 | else |
e24b16f9 |
602 | retval = &PL_sv_undef; |
a0ed51b3 |
603 | POPSTACK; |
604 | FREETMPS; |
605 | LEAVE; |
e24b16f9 |
606 | if (PL_curcop == &PL_compiling) |
607 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
608 | |
dfe13c55 |
609 | svp = hv_store(hv, (char*)ptr, klen, retval, 0); |
a0ed51b3 |
610 | |
dfe13c55 |
611 | if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8) |
cea2e8a9 |
612 | Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); |
a0ed51b3 |
613 | } |
614 | |
615 | PL_last_swash_hv = hv; |
616 | PL_last_swash_klen = klen; |
617 | PL_last_swash_tmps = tmps; |
618 | PL_last_swash_slen = slen; |
619 | if (klen) |
620 | Copy(ptr, PL_last_swash_key, klen, U8); |
621 | } |
622 | |
623 | switch ((slen << 3) / needents) { |
624 | case 1: |
625 | bit = 1 << (off & 7); |
626 | off >>= 3; |
627 | return (tmps[off] & bit) != 0; |
628 | case 8: |
629 | return tmps[off]; |
630 | case 16: |
631 | off <<= 1; |
632 | return (tmps[off] << 8) + tmps[off + 1] ; |
633 | case 32: |
634 | off <<= 2; |
635 | return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ; |
636 | } |
cea2e8a9 |
637 | Perl_croak(aTHX_ "panic: swash_fetch"); |
a0ed51b3 |
638 | return 0; |
639 | } |