Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
1129b882 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, |
f4b6e4b3 | 4 | * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2009 by Larry Wall and others |
79072805 LW |
5 | * |
6 | * You may distribute under the terms of either the GNU General Public | |
7 | * License or the Artistic License, as specified in the README file. | |
8 | * | |
a0d0e21e LW |
9 | */ |
10 | ||
11 | /* | |
4ac71550 TC |
12 | * 'So that was the job I felt I had to do when I started,' thought Sam. |
13 | * | |
14 | * [p.934 of _The Lord of the Rings_, VI/iii: "Mount Doom"] | |
79072805 LW |
15 | */ |
16 | ||
166f8a29 | 17 | /* This file contains some common functions needed to carry out certain |
347f3823 | 18 | * ops. For example, both pp_sprintf() and pp_prtf() call the function |
fc7d284e | 19 | * do_sprintf() found in this file. |
166f8a29 DM |
20 | */ |
21 | ||
79072805 | 22 | #include "EXTERN.h" |
864dbfa3 | 23 | #define PERL_IN_DOOP_C |
79072805 | 24 | #include "perl.h" |
f34acfec | 25 | #include "invlist_inline.h" |
79072805 | 26 | |
64ca3a65 | 27 | #ifndef PERL_MICRO |
79072805 LW |
28 | #include <signal.h> |
29 | #endif | |
30 | ||
334c6444 DM |
31 | |
32 | /* Helper function for do_trans(). | |
b8e8e0fc KW |
33 | * Handles cases where the search and replacement charlists aren't UTF-8, |
34 | * aren't identical, and neither the /d nor /s flag is present. | |
35 | * | |
36 | * sv may or may not be utf8. Note that no code point above 255 can possibly | |
37 | * be in the to-translate set | |
334c6444 DM |
38 | */ |
39 | ||
f0fd0980 | 40 | STATIC Size_t |
f534d546 | 41 | S_do_trans_simple(pTHX_ SV * const sv, const OPtrans_map * const tbl) |
79072805 | 42 | { |
f0fd0980 | 43 | Size_t matches = 0; |
463ee0b2 | 44 | STRLEN len; |
9138d6ca | 45 | U8 *s = (U8*)SvPV_nomg(sv,len); |
c395bd6c | 46 | U8 * const send = s+len; |
7918f24d NC |
47 | |
48 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE; | |
5d7580af KW |
49 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_simple:" |
50 | " input sv:\n", | |
51 | __FILE__, __LINE__)); | |
52 | DEBUG_y(sv_dump(sv)); | |
7918f24d | 53 | |
1e54db1a | 54 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 55 | if (!SvUTF8(sv)) { |
01ec43d0 | 56 | while (s < send) { |
e4ee4c1b | 57 | const short ch = tbl->map[*s]; |
f54cb97a | 58 | if (ch >= 0) { |
01ec43d0 | 59 | matches++; |
c4420975 | 60 | *s = (U8)ch; |
01ec43d0 | 61 | } |
c4420975 | 62 | s++; |
01ec43d0 GS |
63 | } |
64 | SvSETMAGIC(sv); | |
9b877dbb IH |
65 | } |
66 | else { | |
e4ee4c1b | 67 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); |
c395bd6c AL |
68 | U8 *d; |
69 | U8 *dstart; | |
70 | ||
b8e8e0fc KW |
71 | /* Allow for worst-case expansion: Each input byte can become 2. For a |
72 | * given input character, this happens when it occupies a single byte | |
73 | * under UTF-8, but is to be translated to something that occupies two: | |
74 | * $_="a".chr(400); tr/a/\xFE/, FE needs encoding. */ | |
c395bd6c AL |
75 | if (grows) |
76 | Newx(d, len*2+1, U8); | |
77 | else | |
78 | d = s; | |
79 | dstart = d; | |
80 | while (s < send) { | |
81 | STRLEN ulen; | |
e4ee4c1b | 82 | short ch; |
c395bd6c AL |
83 | |
84 | /* Need to check this, otherwise 128..255 won't match */ | |
85 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); | |
c1048fcf | 86 | if (c < 0x100 && (ch = tbl->map[c]) >= 0) { |
c395bd6c | 87 | matches++; |
e4ee4c1b | 88 | d = uvchr_to_utf8(d, (UV)ch); |
c395bd6c AL |
89 | s += ulen; |
90 | } | |
91 | else { /* No match -> copy */ | |
92 | Move(s, d, ulen, U8); | |
93 | d += ulen; | |
94 | s += ulen; | |
95 | } | |
96 | } | |
97 | if (grows) { | |
98 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
99 | Safefree(dstart); | |
100 | } | |
101 | else { | |
102 | *d = '\0'; | |
103 | SvCUR_set(sv, d - dstart); | |
104 | } | |
105 | SvUTF8_on(sv); | |
106 | SvSETMAGIC(sv); | |
9b877dbb | 107 | } |
f1d561b5 | 108 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n", |
5d7580af KW |
109 | __FILE__, __LINE__, matches)); |
110 | DEBUG_y(sv_dump(sv)); | |
4757a243 LW |
111 | return matches; |
112 | } | |
113 | ||
334c6444 DM |
114 | |
115 | /* Helper function for do_trans(). | |
b8e8e0fc KW |
116 | * Handles cases where the search and replacement charlists are identical and |
117 | * non-utf8: so the string isn't modified, and only a count of modifiable | |
334c6444 | 118 | * chars is needed. |
b8e8e0fc KW |
119 | * |
120 | * Note that it doesn't handle /d or /s, since these modify the string even if | |
121 | * the replacement list is empty. | |
122 | * | |
123 | * sv may or may not be utf8. Note that no code point above 255 can possibly | |
124 | * be in the to-translate set | |
334c6444 DM |
125 | */ |
126 | ||
f0fd0980 | 127 | STATIC Size_t |
f534d546 | 128 | S_do_trans_count(pTHX_ SV * const sv, const OPtrans_map * const tbl) |
4757a243 | 129 | { |
4757a243 | 130 | STRLEN len; |
9138d6ca | 131 | const U8 *s = (const U8*)SvPV_nomg_const(sv, len); |
c395bd6c | 132 | const U8 * const send = s + len; |
f0fd0980 | 133 | Size_t matches = 0; |
7918f24d NC |
134 | |
135 | PERL_ARGS_ASSERT_DO_TRANS_COUNT; | |
136 | ||
5d7580af KW |
137 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_count:" |
138 | " input sv:\n", | |
139 | __FILE__, __LINE__)); | |
140 | DEBUG_y(sv_dump(sv)); | |
141 | ||
c395bd6c | 142 | if (!SvUTF8(sv)) { |
1aa99e6b | 143 | while (s < send) { |
c1048fcf | 144 | if (tbl->map[*s++] >= 0) |
036b4402 | 145 | matches++; |
1aa99e6b | 146 | } |
c395bd6c | 147 | } |
fabdb6c0 | 148 | else { |
e4ee4c1b | 149 | const bool complement = cBOOL(PL_op->op_private & OPpTRANS_COMPLEMENT); |
1aa99e6b | 150 | while (s < send) { |
1aa99e6b | 151 | STRLEN ulen; |
9f7f3913 | 152 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 | 153 | if (c < 0x100) { |
c1048fcf | 154 | if (tbl->map[c] >= 0) |
8973db79 JH |
155 | matches++; |
156 | } else if (complement) | |
1aa99e6b IH |
157 | matches++; |
158 | s += ulen; | |
159 | } | |
fabdb6c0 | 160 | } |
4757a243 | 161 | |
f1d561b5 | 162 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: count returning %zu\n", |
5d7580af | 163 | __FILE__, __LINE__, matches)); |
4757a243 LW |
164 | return matches; |
165 | } | |
166 | ||
334c6444 DM |
167 | |
168 | /* Helper function for do_trans(). | |
b8e8e0fc KW |
169 | * Handles cases where the search and replacement charlists aren't identical |
170 | * and both are non-utf8, and one or both of /d, /s is specified. | |
171 | * | |
172 | * sv may or may not be utf8. Note that no code point above 255 can possibly | |
173 | * be in the to-translate set | |
334c6444 DM |
174 | */ |
175 | ||
f0fd0980 | 176 | STATIC Size_t |
f534d546 | 177 | S_do_trans_complex(pTHX_ SV * const sv, const OPtrans_map * const tbl) |
4757a243 | 178 | { |
c395bd6c | 179 | STRLEN len; |
9138d6ca | 180 | U8 *s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 181 | U8 * const send = s+len; |
f0fd0980 | 182 | Size_t matches = 0; |
00bd451d | 183 | const bool complement = cBOOL(PL_op->op_private & OPpTRANS_COMPLEMENT); |
7918f24d NC |
184 | |
185 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX; | |
186 | ||
5d7580af KW |
187 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_complex:" |
188 | " input sv:\n", | |
189 | __FILE__, __LINE__)); | |
190 | DEBUG_y(sv_dump(sv)); | |
191 | ||
c395bd6c AL |
192 | if (!SvUTF8(sv)) { |
193 | U8 *d = s; | |
194 | U8 * const dstart = d; | |
4757a243 | 195 | |
1aa99e6b | 196 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
00bd451d KW |
197 | |
198 | /* What the mapping of the previous character was to. If the new | |
199 | * character has the same mapping, it is squashed from the output | |
200 | * (but still is included in the count) */ | |
201 | short previous_map = (short) TR_OOB; | |
202 | ||
1aa99e6b | 203 | while (s < send) { |
5e874c42 KW |
204 | const short this_map = tbl->map[*s]; |
205 | if (this_map >= 0) { | |
00bd451d KW |
206 | matches++; |
207 | if (this_map != previous_map) { | |
208 | *d++ = (U8)this_map; | |
209 | previous_map = this_map; | |
210 | } | |
1aa99e6b | 211 | } |
00bd451d | 212 | else { |
a68ea463 | 213 | if (this_map == (short) TR_UNMAPPED) { |
00bd451d | 214 | *d++ = *s; |
a68ea463 KW |
215 | previous_map = (short) TR_OOB; |
216 | } | |
00bd451d KW |
217 | else { |
218 | assert(this_map == (short) TR_DELETE); | |
219 | matches++; | |
220 | } | |
00bd451d KW |
221 | } |
222 | ||
1aa99e6b IH |
223 | s++; |
224 | } | |
a0ed51b3 | 225 | } |
b8e8e0fc | 226 | else { /* Not to squash */ |
1aa99e6b | 227 | while (s < send) { |
5e874c42 KW |
228 | const short this_map = tbl->map[*s]; |
229 | if (this_map >= 0) { | |
1aa99e6b | 230 | matches++; |
5e874c42 | 231 | *d++ = (U8)this_map; |
1aa99e6b | 232 | } |
5e874c42 | 233 | else if (this_map == (short) TR_UNMAPPED) |
1aa99e6b | 234 | *d++ = *s; |
5e874c42 | 235 | else if (this_map == (short) TR_DELETE) |
1aa99e6b IH |
236 | matches++; |
237 | s++; | |
238 | } | |
239 | } | |
76ef7183 | 240 | *d = '\0'; |
1aa99e6b | 241 | SvCUR_set(sv, d - dstart); |
4757a243 | 242 | } |
c395bd6c | 243 | else { /* is utf8 */ |
e4ee4c1b DM |
244 | const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH); |
245 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); | |
c395bd6c AL |
246 | U8 *d; |
247 | U8 *dstart; | |
0b9a13c3 | 248 | Size_t size = tbl->size; |
b8e8e0fc KW |
249 | |
250 | /* What the mapping of the previous character was to. If the new | |
251 | * character has the same mapping, it is squashed from the output (but | |
252 | * still is included in the count) */ | |
47279991 | 253 | UV pch = TR_OOB; |
fabdb6c0 | 254 | |
9b877dbb | 255 | if (grows) |
b8e8e0fc KW |
256 | /* Allow for worst-case expansion: Each input byte can become 2. |
257 | * For a given input character, this happens when it occupies a | |
258 | * single byte under UTF-8, but is to be translated to something | |
259 | * that occupies two: */ | |
a02a5408 | 260 | Newx(d, len*2+1, U8); |
9b877dbb IH |
261 | else |
262 | d = s; | |
1aa99e6b IH |
263 | dstart = d; |
264 | ||
4eaf16e8 DM |
265 | while (s < send) { |
266 | STRLEN len; | |
267 | const UV comp = utf8n_to_uvchr(s, send - s, &len, | |
268 | UTF8_ALLOW_DEFAULT); | |
269 | UV ch; | |
270 | short sch; | |
271 | ||
00bd451d KW |
272 | sch = (comp < size) |
273 | ? tbl->map[comp] | |
274 | : (! complement) | |
275 | ? (short) TR_UNMAPPED | |
276 | : tbl->map[size]; | |
4eaf16e8 DM |
277 | |
278 | if (sch >= 0) { | |
279 | ch = (UV)sch; | |
280 | replace: | |
281 | matches++; | |
282 | if (LIKELY(!squash || ch != pch)) { | |
283 | d = uvchr_to_utf8(d, ch); | |
284 | pch = ch; | |
0b9a13c3 | 285 | } |
4eaf16e8 DM |
286 | s += len; |
287 | continue; | |
288 | } | |
482bf615 | 289 | else if (sch == (short) TR_UNMAPPED) { |
4eaf16e8 DM |
290 | Move(s, d, len, U8); |
291 | d += len; | |
a68ea463 | 292 | pch = TR_OOB; |
4eaf16e8 | 293 | } |
482bf615 | 294 | else if (sch == (short) TR_DELETE) |
4eaf16e8 DM |
295 | matches++; |
296 | else { | |
482bf615 | 297 | assert(sch == (short) TR_R_EMPTY); /* empty replacement */ |
4eaf16e8 DM |
298 | ch = comp; |
299 | goto replace; | |
300 | } | |
301 | ||
302 | s += len; | |
4eaf16e8 | 303 | } |
0b9a13c3 | 304 | |
9b877dbb IH |
305 | if (grows) { |
306 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
307 | Safefree(dstart); | |
308 | } | |
309 | else { | |
310 | *d = '\0'; | |
311 | SvCUR_set(sv, d - dstart); | |
312 | } | |
1aa99e6b | 313 | SvUTF8_on(sv); |
4757a243 | 314 | } |
5e44153e | 315 | SvSETMAGIC(sv); |
f1d561b5 | 316 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n", |
5d7580af KW |
317 | __FILE__, __LINE__, matches)); |
318 | DEBUG_y(sv_dump(sv)); | |
4757a243 LW |
319 | return matches; |
320 | } | |
321 | ||
334c6444 DM |
322 | |
323 | /* Helper function for do_trans(). | |
f34acfec KW |
324 | * Handles cases where an inversion map implementation is to be used and the |
325 | * search and replacement charlists are identical: so the string isn't | |
326 | * modified, and only a count of modifiable chars is needed. | |
327 | * | |
328 | * Note that it doesn't handle /d nor /s, since these modify the string | |
329 | * even if the replacement charlist is empty. | |
330 | * | |
331 | * sv may or may not be utf8. | |
334c6444 DM |
332 | */ |
333 | ||
f0fd0980 | 334 | STATIC Size_t |
f34acfec | 335 | S_do_trans_count_invmap(pTHX_ SV * const sv, AV * const invmap) |
4757a243 | 336 | { |
4757a243 LW |
337 | U8 *s; |
338 | U8 *send; | |
f0fd0980 | 339 | Size_t matches = 0; |
4757a243 | 340 | STRLEN len; |
f34acfec KW |
341 | SV** const from_invlist_ptr = av_fetch(invmap, 0, TRUE); |
342 | SV** const to_invmap_ptr = av_fetch(invmap, 1, TRUE); | |
343 | SV* from_invlist = *from_invlist_ptr; | |
344 | SV* to_invmap_sv = *to_invmap_ptr; | |
345 | UV* map = (UV *) SvPVX(to_invmap_sv); | |
4757a243 | 346 | |
f34acfec | 347 | PERL_ARGS_ASSERT_DO_TRANS_COUNT_INVMAP; |
7918f24d | 348 | |
5d7580af KW |
349 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d:" |
350 | "entering do_trans_count_invmap:" | |
351 | " input sv:\n", | |
352 | __FILE__, __LINE__)); | |
353 | DEBUG_y(sv_dump(sv)); | |
354 | DEBUG_y(PerlIO_printf(Perl_debug_log, "mapping:\n")); | |
355 | DEBUG_y(invmap_dump(from_invlist, (UV *) SvPVX(to_invmap_sv))); | |
356 | ||
9138d6ca | 357 | s = (U8*)SvPV_nomg(sv, len); |
f34acfec | 358 | |
4757a243 LW |
359 | send = s + len; |
360 | ||
f34acfec KW |
361 | while (s < send) { |
362 | UV from; | |
363 | SSize_t i; | |
364 | STRLEN s_len; | |
365 | ||
366 | /* Get the code point of the next character in the string */ | |
367 | if (! SvUTF8(sv) || UTF8_IS_INVARIANT(*s)) { | |
368 | from = *s; | |
369 | s_len = 1; | |
370 | } | |
371 | else { | |
372 | from = utf8_to_uvchr_buf(s, send, &s_len); | |
373 | if (from == 0 && *s != '\0') { | |
374 | _force_out_malformed_utf8_message(s, send, 0, /*die*/TRUE); | |
375 | } | |
376 | } | |
4757a243 | 377 | |
f34acfec KW |
378 | /* Look the code point up in the data structure for this tr/// to get |
379 | * what it maps to */ | |
380 | i = _invlist_search(from_invlist, from); | |
381 | assert(i >= 0); | |
1aa99e6b | 382 | |
f34acfec KW |
383 | if (map[i] != (UV) TR_UNLISTED) { |
384 | matches++; | |
385 | } | |
386 | ||
387 | s += s_len; | |
9b877dbb | 388 | } |
4757a243 | 389 | |
f1d561b5 | 390 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n", |
5d7580af | 391 | __FILE__, __LINE__, matches)); |
4757a243 LW |
392 | return matches; |
393 | } | |
394 | ||
334c6444 | 395 | /* Helper function for do_trans(). |
f34acfec KW |
396 | * Handles cases where an inversion map implementation is to be used and the |
397 | * search and replacement charlists are either not identical or flags are | |
398 | * present. | |
399 | * | |
400 | * sv may or may not be utf8. | |
334c6444 DM |
401 | */ |
402 | ||
f0fd0980 | 403 | STATIC Size_t |
f34acfec | 404 | S_do_trans_invmap(pTHX_ SV * const sv, AV * const invmap) |
4757a243 | 405 | { |
f34acfec KW |
406 | U8 *s; |
407 | U8 *send; | |
408 | U8 *d; | |
409 | U8 *s0; | |
410 | U8 *d0; | |
f0fd0980 | 411 | Size_t matches = 0; |
4757a243 | 412 | STRLEN len; |
f34acfec KW |
413 | SV** const from_invlist_ptr = av_fetch(invmap, 0, TRUE); |
414 | SV** const to_invmap_ptr = av_fetch(invmap, 1, TRUE); | |
415 | SV** const to_expansion_ptr = av_fetch(invmap, 2, TRUE); | |
416 | NV max_expansion = SvNV(*to_expansion_ptr); | |
417 | SV* from_invlist = *from_invlist_ptr; | |
418 | SV* to_invmap_sv = *to_invmap_ptr; | |
419 | UV* map = (UV *) SvPVX(to_invmap_sv); | |
420 | UV previous_map = TR_OOB; | |
421 | const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH); | |
422 | const bool delete_unfound = cBOOL(PL_op->op_private & OPpTRANS_DELETE); | |
423 | bool inplace = ! cBOOL(PL_op->op_private & OPpTRANS_GROWS); | |
424 | const UV* from_array = invlist_array(from_invlist); | |
8b4eae17 | 425 | UV final_map = TR_OOB; |
58aa6738 | 426 | bool out_is_utf8 = cBOOL(SvUTF8(sv)); |
f34acfec KW |
427 | STRLEN s_len; |
428 | ||
429 | PERL_ARGS_ASSERT_DO_TRANS_INVMAP; | |
430 | ||
431 | /* A third element in the array indicates that the replacement list was | |
432 | * shorter than the search list, and this element contains the value to use | |
433 | * for the items that don't correspond */ | |
434 | if (av_top_index(invmap) >= 3) { | |
435 | SV** const final_map_ptr = av_fetch(invmap, 3, TRUE); | |
436 | SV* const final_map_sv = *final_map_ptr; | |
437 | final_map = SvUV(final_map_sv); | |
438 | } | |
7918f24d | 439 | |
f34acfec KW |
440 | /* If there is something in the transliteration that could force the input |
441 | * to be changed to UTF-8, we don't know if we can do it in place, so | |
442 | * assume cannot */ | |
443 | if (! out_is_utf8 && (PL_op->op_private & OPpTRANS_CAN_FORCE_UTF8)) { | |
444 | inplace = FALSE; | |
1aa99e6b | 445 | } |
f34acfec KW |
446 | |
447 | s = (U8*)SvPV_nomg(sv, len); | |
5d7580af KW |
448 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_invmap:" |
449 | " input sv:\n", | |
450 | __FILE__, __LINE__)); | |
451 | DEBUG_y(sv_dump(sv)); | |
452 | DEBUG_y(PerlIO_printf(Perl_debug_log, "mapping:\n")); | |
453 | DEBUG_y(invmap_dump(from_invlist, map)); | |
454 | ||
4757a243 | 455 | send = s + len; |
f34acfec | 456 | s0 = s; |
4757a243 | 457 | |
f34acfec KW |
458 | /* We know by now if there are some possible input strings whose |
459 | * transliterations are longer than the input. If none can, we just edit | |
460 | * in place. */ | |
461 | if (inplace) { | |
462 | d0 = d = s; | |
463 | } | |
464 | else { | |
465 | /* Here, we can't edit in place. We have no idea how much, if any, | |
466 | * this particular input string will grow. However, the compilation | |
8bbdcb37 KW |
467 | * calculated the maximum expansion possible. Use that to allocate |
468 | * based on the worst case scenario. (First +1 is to round up; 2nd is | |
469 | * for \0) */ | |
2077aa94 | 470 | Newx(d, (STRLEN) (len * max_expansion + 1 + 1), U8); |
f34acfec | 471 | d0 = d; |
4757a243 LW |
472 | } |
473 | ||
f34acfec | 474 | restart: |
4757a243 | 475 | |
f34acfec KW |
476 | /* Do the actual transliteration */ |
477 | while (s < send) { | |
478 | UV from; | |
479 | UV to; | |
480 | SSize_t i; | |
481 | STRLEN s_len; | |
482 | ||
483 | /* Get the code point of the next character in the string */ | |
484 | if (! SvUTF8(sv) || UTF8_IS_INVARIANT(*s)) { | |
485 | from = *s; | |
486 | s_len = 1; | |
487 | } | |
488 | else { | |
489 | from = utf8_to_uvchr_buf(s, send, &s_len); | |
490 | if (from == 0 && *s != '\0') { | |
491 | _force_out_malformed_utf8_message(s, send, 0, /*die*/TRUE); | |
492 | } | |
493 | } | |
334c6444 | 494 | |
f34acfec KW |
495 | /* Look the code point up in the data structure for this tr/// to get |
496 | * what it maps to */ | |
497 | i = _invlist_search(from_invlist, from); | |
498 | assert(i >= 0); | |
334c6444 | 499 | |
f34acfec | 500 | to = map[i]; |
7918f24d | 501 | |
f34acfec KW |
502 | if (to == (UV) TR_UNLISTED) { /* Just copy the unreplaced character */ |
503 | if (UVCHR_IS_INVARIANT(from) || ! out_is_utf8) { | |
2077aa94 | 504 | *d++ = (U8) from; |
f34acfec KW |
505 | } |
506 | else if (SvUTF8(sv)) { | |
507 | Move(s, d, s_len, U8); | |
508 | d += s_len; | |
509 | } | |
510 | else { /* Convert to UTF-8 */ | |
511 | append_utf8_from_native_byte(*s, &d); | |
512 | } | |
7918f24d | 513 | |
f34acfec KW |
514 | previous_map = to; |
515 | s += s_len; | |
516 | continue; | |
c4d5f83a | 517 | } |
4757a243 | 518 | |
f34acfec KW |
519 | /* Everything else is counted as a match */ |
520 | matches++; | |
4757a243 | 521 | |
f34acfec KW |
522 | if (to == (UV) TR_SPECIAL_HANDLING) { |
523 | if (delete_unfound) { | |
f34acfec KW |
524 | s += s_len; |
525 | continue; | |
526 | } | |
4757a243 | 527 | |
f34acfec KW |
528 | /* Use the final character in the replacement list */ |
529 | to = final_map; | |
530 | } | |
531 | else { /* Here the input code point is to be remapped. The actual | |
532 | value is offset from the base of this entry */ | |
533 | to += from - from_array[i]; | |
534 | } | |
535 | ||
536 | /* If copying all occurrences, or this is the first occurrence, copy it | |
537 | * to the output */ | |
538 | if (! squash || to != previous_map) { | |
539 | if (out_is_utf8) { | |
540 | d = uvchr_to_utf8(d, to); | |
541 | } | |
542 | else { | |
543 | if (to >= 256) { /* If need to convert to UTF-8, restart */ | |
544 | out_is_utf8 = TRUE; | |
545 | s = s0; | |
546 | d = d0; | |
547 | matches = 0; | |
548 | goto restart; | |
549 | } | |
2077aa94 | 550 | *d++ = (U8) to; |
f34acfec KW |
551 | } |
552 | } | |
553 | ||
554 | previous_map = to; | |
555 | s += s_len; | |
4757a243 | 556 | } |
f34acfec KW |
557 | |
558 | s_len = 0; | |
559 | s += s_len; | |
560 | if (! inplace) { | |
561 | sv_setpvn(sv, (char*)d0, d - d0); | |
ecfd8588 | 562 | Safefree(d0); |
9b877dbb IH |
563 | } |
564 | else { | |
565 | *d = '\0'; | |
f34acfec KW |
566 | SvCUR_set(sv, d - d0); |
567 | } | |
568 | ||
569 | if (! SvUTF8(sv) && out_is_utf8) { | |
570 | SvUTF8_on(sv); | |
9b877dbb | 571 | } |
4757a243 LW |
572 | SvSETMAGIC(sv); |
573 | ||
f1d561b5 | 574 | DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n", |
5d7580af KW |
575 | __FILE__, __LINE__, matches)); |
576 | DEBUG_y(sv_dump(sv)); | |
4757a243 LW |
577 | return matches; |
578 | } | |
579 | ||
334c6444 DM |
580 | /* Execute a tr//. sv is the value to be translated, while PL_op |
581 | * should be an OP_TRANS or OP_TRANSR op, whose op_pv field contains a | |
f34acfec KW |
582 | * translation table or whose op_sv field contains an inversion map. |
583 | * | |
334c6444 DM |
584 | * Returns a count of number of characters translated |
585 | */ | |
586 | ||
f0fd0980 | 587 | Size_t |
864dbfa3 | 588 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
589 | { |
590 | STRLEN len; | |
e4ee4c1b | 591 | const U8 flags = PL_op->op_private; |
f34acfec KW |
592 | bool use_utf8_fcns = cBOOL(flags & OPpTRANS_USE_SVOP); |
593 | bool identical = cBOOL(flags & OPpTRANS_IDENTICAL); | |
4757a243 | 594 | |
7918f24d NC |
595 | PERL_ARGS_ASSERT_DO_TRANS; |
596 | ||
f34acfec | 597 | if (SvREADONLY(sv) && ! identical) { |
a53bfdae | 598 | Perl_croak_no_modify(); |
2233f375 | 599 | } |
10516c54 | 600 | (void)SvPV_const(sv, len); |
4757a243 LW |
601 | if (!len) |
602 | return 0; | |
f34acfec | 603 | if (! identical) { |
4499db73 | 604 | if (!SvPOKp(sv) || SvTHINKFIRST(sv)) |
b4cc4d79 | 605 | (void)SvPV_force_nomg(sv, len); |
2de7b02f | 606 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 607 | } |
4757a243 | 608 | |
f34acfec KW |
609 | if (use_utf8_fcns) { |
610 | SV* const map = | |
611 | #ifdef USE_ITHREADS | |
612 | PAD_SVl(cPADOP->op_padix); | |
613 | #else | |
614 | MUTABLE_SV(cSVOP->op_sv); | |
615 | #endif | |
616 | ||
617 | if (identical) { | |
618 | return do_trans_count_invmap(sv, (AV *) map); | |
619 | } | |
620 | else { | |
621 | return do_trans_invmap(sv, (AV *) map); | |
622 | } | |
623 | } | |
624 | else { | |
625 | const OPtrans_map * const map = (OPtrans_map*)cPVOP->op_pv; | |
626 | ||
627 | if (identical) { | |
628 | return do_trans_count(sv, map); | |
629 | } | |
630 | else if (flags & (OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
631 | return do_trans_complex(sv, map); | |
632 | } | |
633 | else | |
634 | return do_trans_simple(sv, map); | |
79072805 | 635 | } |
79072805 LW |
636 | } |
637 | ||
638 | void | |
5aaab254 | 639 | Perl_do_join(pTHX_ SV *sv, SV *delim, SV **mark, SV **sp) |
79072805 | 640 | { |
53c1dcc0 | 641 | SV ** const oldmark = mark; |
eb578fdb KW |
642 | I32 items = sp - mark; |
643 | STRLEN len; | |
463ee0b2 | 644 | STRLEN delimlen; |
810a07df | 645 | const char * const delims = SvPV_const(delim, delimlen); |
79072805 | 646 | |
7918f24d NC |
647 | PERL_ARGS_ASSERT_DO_JOIN; |
648 | ||
79072805 LW |
649 | mark++; |
650 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
862a34c6 | 651 | SvUPGRADE(sv, SVt_PV); |
79072805 LW |
652 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
653 | while (items-- > 0) { | |
1426bbf4 | 654 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a | 655 | STRLEN tmplen; |
5c144d81 | 656 | SvPV_const(*mark, tmplen); |
463ee0b2 | 657 | len += tmplen; |
79072805 LW |
658 | } |
659 | mark++; | |
660 | } | |
661 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
662 | ||
663 | mark = oldmark; | |
db7c17d7 | 664 | items = sp - mark; |
79072805 LW |
665 | ++mark; |
666 | } | |
667 | ||
f607343f | 668 | SvPVCLEAR(sv); |
e4803c42 | 669 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 670 | SvUTF8_off(sv); |
e4803c42 | 671 | |
284167a5 | 672 | if (TAINTING_get && SvMAGICAL(sv)) |
8d6d96c1 HS |
673 | SvTAINTED_off(sv); |
674 | ||
463ee0b2 | 675 | if (items-- > 0) { |
92d29cee JH |
676 | if (*mark) |
677 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
678 | mark++; |
679 | } | |
8d6d96c1 | 680 | |
c512ce4f | 681 | if (delimlen) { |
810a07df | 682 | const U32 delimflag = DO_UTF8(delim) ? SV_CATUTF8 : SV_CATBYTES; |
79072805 | 683 | for (; items > 0; items--,mark++) { |
f0ee3863 FC |
684 | STRLEN len; |
685 | const char *s; | |
810a07df | 686 | sv_catpvn_flags(sv,delims,delimlen,delimflag); |
f0ee3863 FC |
687 | s = SvPV_const(*mark,len); |
688 | sv_catpvn_flags(sv,s,len, | |
689 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
79072805 LW |
690 | } |
691 | } | |
692 | else { | |
693 | for (; items > 0; items--,mark++) | |
f0ee3863 FC |
694 | { |
695 | STRLEN len; | |
696 | const char *s = SvPV_const(*mark,len); | |
697 | sv_catpvn_flags(sv,s,len, | |
698 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
699 | } | |
79072805 LW |
700 | } |
701 | SvSETMAGIC(sv); | |
702 | } | |
703 | ||
704 | void | |
03a22d83 | 705 | Perl_do_sprintf(pTHX_ SV *sv, SSize_t len, SV **sarg) |
79072805 | 706 | { |
46fc3d4c | 707 | STRLEN patlen; |
53c1dcc0 | 708 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c | 709 | bool do_taint = FALSE; |
710 | ||
7918f24d | 711 | PERL_ARGS_ASSERT_DO_SPRINTF; |
03a22d83 | 712 | assert(len >= 1); |
7918f24d | 713 | |
e06d98fb DM |
714 | if (SvTAINTED(*sarg)) |
715 | TAINT_PROPER( | |
716 | (PL_op && PL_op->op_type < OP_max) | |
717 | ? (PL_op->op_type == OP_PRTF) | |
718 | ? "printf" | |
719 | : PL_op_name[PL_op->op_type] | |
720 | : "(unknown)" | |
721 | ); | |
5b781b5b | 722 | SvUTF8_off(sv); |
2cf2cfc6 A |
723 | if (DO_UTF8(*sarg)) |
724 | SvUTF8_on(sv); | |
03a22d83 | 725 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, (Size_t)(len - 1), &do_taint); |
79072805 | 726 | SvSETMAGIC(sv); |
46fc3d4c | 727 | if (do_taint) |
728 | SvTAINTED_on(sv); | |
79072805 LW |
729 | } |
730 | ||
81e118e0 | 731 | UV |
d69c4304 | 732 | Perl_do_vecget(pTHX_ SV *sv, STRLEN offset, int size) |
81e118e0 | 733 | { |
67dd6f35 | 734 | STRLEN srclen, len, avail, uoffset, bitoffs = 0; |
fc9668ae DM |
735 | const I32 svpv_flags = ((PL_op->op_flags & OPf_MOD || LVRET) |
736 | ? SV_UNDEF_RETURNS_NULL : 0); | |
737 | unsigned char *s = (unsigned char *) | |
738 | SvPV_flags(sv, srclen, (svpv_flags|SV_GMAGIC)); | |
81e118e0 JH |
739 | UV retnum = 0; |
740 | ||
032061d2 | 741 | if (!s) { |
fc9668ae | 742 | s = (unsigned char *)""; |
032061d2 | 743 | } |
2f96a1b4 | 744 | |
7918f24d NC |
745 | PERL_ARGS_ASSERT_DO_VECGET; |
746 | ||
8e84507e | 747 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 748 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 749 | |
fc9668ae | 750 | if (SvUTF8(sv)) { |
27c41eac | 751 | if (Perl_sv_utf8_downgrade_flags(aTHX_ sv, TRUE, 0)) { |
315f3fc1 KW |
752 | /* PVX may have changed */ |
753 | s = (unsigned char *) SvPV_flags(sv, srclen, svpv_flags); | |
754 | } | |
755 | else { | |
d3597f89 KW |
756 | Perl_croak(aTHX_ "Use of strings with code points over 0xFF" |
757 | " as arguments to vec is forbidden"); | |
315f3fc1 | 758 | } |
fc9668ae | 759 | } |
246fae53 | 760 | |
bbb8a7e0 MHM |
761 | if (size < 8) { |
762 | bitoffs = ((offset%8)*size)%8; | |
763 | uoffset = offset/(8/size); | |
764 | } | |
67dd6f35 DM |
765 | else if (size > 8) { |
766 | int n = size/8; | |
767 | if (offset > Size_t_MAX / n - 1) /* would overflow */ | |
768 | return 0; | |
769 | uoffset = offset*n; | |
770 | } | |
bbb8a7e0 MHM |
771 | else |
772 | uoffset = offset; | |
773 | ||
67dd6f35 DM |
774 | if (uoffset >= srclen) |
775 | return 0; | |
776 | ||
777 | len = (bitoffs + size + 7)/8; /* required number of bytes */ | |
778 | avail = srclen - uoffset; /* available number of bytes */ | |
779 | ||
780 | /* Does the byte range overlap the end of the string? If so, | |
781 | * handle specially. */ | |
782 | if (avail < len) { | |
81e118e0 JH |
783 | if (size <= 8) |
784 | retnum = 0; | |
785 | else { | |
81e118e0 | 786 | if (size == 16) { |
67dd6f35 DM |
787 | assert(avail == 1); |
788 | retnum = (UV) s[uoffset] << 8; | |
81e118e0 JH |
789 | } |
790 | else if (size == 32) { | |
67dd6f35 DM |
791 | assert(avail >= 1 && avail <= 3); |
792 | if (avail == 1) | |
81e118e0 | 793 | retnum = |
bb7a0f54 | 794 | ((UV) s[uoffset ] << 24); |
67dd6f35 | 795 | else if (avail == 2) |
81e118e0 | 796 | retnum = |
bb7a0f54 MHM |
797 | ((UV) s[uoffset ] << 24) + |
798 | ((UV) s[uoffset + 1] << 16); | |
81e118e0 JH |
799 | else |
800 | retnum = | |
bb7a0f54 MHM |
801 | ((UV) s[uoffset ] << 24) + |
802 | ((UV) s[uoffset + 1] << 16) + | |
803 | ( s[uoffset + 2] << 8); | |
81e118e0 | 804 | } |
d7d93a81 | 805 | #ifdef UV_IS_QUAD |
c5a0f51a | 806 | else if (size == 64) { |
a2a5de95 NC |
807 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
808 | "Bit vector size > 32 non-portable"); | |
67dd6f35 DM |
809 | assert(avail >= 1 && avail <= 7); |
810 | if (avail == 1) | |
c5a0f51a | 811 | retnum = |
bb7a0f54 | 812 | (UV) s[uoffset ] << 56; |
67dd6f35 | 813 | else if (avail == 2) |
c5a0f51a | 814 | retnum = |
bb7a0f54 MHM |
815 | ((UV) s[uoffset ] << 56) + |
816 | ((UV) s[uoffset + 1] << 48); | |
67dd6f35 | 817 | else if (avail == 3) |
c5a0f51a | 818 | retnum = |
bb7a0f54 MHM |
819 | ((UV) s[uoffset ] << 56) + |
820 | ((UV) s[uoffset + 1] << 48) + | |
821 | ((UV) s[uoffset + 2] << 40); | |
67dd6f35 | 822 | else if (avail == 4) |
c5a0f51a | 823 | retnum = |
bb7a0f54 MHM |
824 | ((UV) s[uoffset ] << 56) + |
825 | ((UV) s[uoffset + 1] << 48) + | |
826 | ((UV) s[uoffset + 2] << 40) + | |
827 | ((UV) s[uoffset + 3] << 32); | |
67dd6f35 | 828 | else if (avail == 5) |
c5a0f51a | 829 | retnum = |
bb7a0f54 MHM |
830 | ((UV) s[uoffset ] << 56) + |
831 | ((UV) s[uoffset + 1] << 48) + | |
832 | ((UV) s[uoffset + 2] << 40) + | |
833 | ((UV) s[uoffset + 3] << 32) + | |
e7aca353 | 834 | ((UV) s[uoffset + 4] << 24); |
67dd6f35 | 835 | else if (avail == 6) |
c5a0f51a | 836 | retnum = |
bb7a0f54 MHM |
837 | ((UV) s[uoffset ] << 56) + |
838 | ((UV) s[uoffset + 1] << 48) + | |
839 | ((UV) s[uoffset + 2] << 40) + | |
840 | ((UV) s[uoffset + 3] << 32) + | |
841 | ((UV) s[uoffset + 4] << 24) + | |
842 | ((UV) s[uoffset + 5] << 16); | |
c5a0f51a | 843 | else |
8e84507e | 844 | retnum = |
bb7a0f54 MHM |
845 | ((UV) s[uoffset ] << 56) + |
846 | ((UV) s[uoffset + 1] << 48) + | |
847 | ((UV) s[uoffset + 2] << 40) + | |
848 | ((UV) s[uoffset + 3] << 32) + | |
849 | ((UV) s[uoffset + 4] << 24) + | |
850 | ((UV) s[uoffset + 5] << 16) + | |
e7aca353 | 851 | ((UV) s[uoffset + 6] << 8); |
c5a0f51a JH |
852 | } |
853 | #endif | |
81e118e0 JH |
854 | } |
855 | } | |
856 | else if (size < 8) | |
d223e1ea | 857 | retnum = (s[uoffset] >> bitoffs) & nBIT_MASK(size); |
81e118e0 | 858 | else { |
81e118e0 | 859 | if (size == 8) |
bb7a0f54 | 860 | retnum = s[uoffset]; |
81e118e0 JH |
861 | else if (size == 16) |
862 | retnum = | |
bb7a0f54 MHM |
863 | ((UV) s[uoffset] << 8) + |
864 | s[uoffset + 1]; | |
81e118e0 JH |
865 | else if (size == 32) |
866 | retnum = | |
bb7a0f54 MHM |
867 | ((UV) s[uoffset ] << 24) + |
868 | ((UV) s[uoffset + 1] << 16) + | |
869 | ( s[uoffset + 2] << 8) + | |
870 | s[uoffset + 3]; | |
d7d93a81 | 871 | #ifdef UV_IS_QUAD |
c5a0f51a | 872 | else if (size == 64) { |
a2a5de95 NC |
873 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
874 | "Bit vector size > 32 non-portable"); | |
c5a0f51a | 875 | retnum = |
bb7a0f54 MHM |
876 | ((UV) s[uoffset ] << 56) + |
877 | ((UV) s[uoffset + 1] << 48) + | |
878 | ((UV) s[uoffset + 2] << 40) + | |
879 | ((UV) s[uoffset + 3] << 32) + | |
880 | ((UV) s[uoffset + 4] << 24) + | |
881 | ((UV) s[uoffset + 5] << 16) + | |
882 | ( s[uoffset + 6] << 8) + | |
883 | s[uoffset + 7]; | |
c5a0f51a JH |
884 | } |
885 | #endif | |
81e118e0 JH |
886 | } |
887 | ||
888 | return retnum; | |
889 | } | |
890 | ||
33b45480 SB |
891 | /* currently converts input to bytes if possible but doesn't sweat failures, |
892 | * although it does ensure that the string it clobbers is not marked as | |
893 | * utf8-valid any more | |
894 | */ | |
79072805 | 895 | void |
864dbfa3 | 896 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 | 897 | { |
67dd6f35 | 898 | STRLEN offset, bitoffs = 0; |
eb578fdb KW |
899 | int size; |
900 | unsigned char *s; | |
901 | UV lval; | |
79072805 | 902 | I32 mask; |
a0d0e21e LW |
903 | STRLEN targlen; |
904 | STRLEN len; | |
c4420975 | 905 | SV * const targ = LvTARG(sv); |
1b92e694 | 906 | char errflags = LvFLAGS(sv); |
79072805 | 907 | |
7918f24d NC |
908 | PERL_ARGS_ASSERT_DO_VECSET; |
909 | ||
1b92e694 DM |
910 | /* some out-of-range errors have been deferred if/until the LV is |
911 | * actually written to: f(vec($s,-1,8)) is not always fatal */ | |
912 | if (errflags) { | |
b063b0a8 DIM |
913 | assert(!(errflags & ~(LVf_NEG_OFF|LVf_OUT_OF_RANGE))); |
914 | if (errflags & LVf_NEG_OFF) | |
1b92e694 DM |
915 | Perl_croak_nocontext("Negative offset to vec in lvalue context"); |
916 | Perl_croak_nocontext("Out of memory!"); | |
917 | } | |
918 | ||
8990e307 LW |
919 | if (!targ) |
920 | return; | |
032061d2 BF |
921 | s = (unsigned char*)SvPV_force_flags(targ, targlen, |
922 | SV_GMAGIC | SV_UNDEF_RETURNS_NULL); | |
246fae53 | 923 | if (SvUTF8(targ)) { |
33b45480 | 924 | /* This is handled by the SvPOK_only below... |
27c41eac | 925 | if (!Perl_sv_utf8_downgrade_flags(aTHX_ targ, TRUE, 0)) |
33b45480 SB |
926 | SvUTF8_off(targ); |
927 | */ | |
27c41eac | 928 | (void) Perl_sv_utf8_downgrade_flags(aTHX_ targ, TRUE, 0); |
246fae53 MG |
929 | } |
930 | ||
4ebbc975 | 931 | (void)SvPOK_only(targ); |
81e118e0 | 932 | lval = SvUV(sv); |
79072805 LW |
933 | offset = LvTARGOFF(sv); |
934 | size = LvTARGLEN(sv); | |
67dd6f35 | 935 | |
8e84507e | 936 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 937 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 938 | |
bbb8a7e0 MHM |
939 | if (size < 8) { |
940 | bitoffs = ((offset%8)*size)%8; | |
941 | offset /= 8/size; | |
942 | } | |
67dd6f35 DM |
943 | else if (size > 8) { |
944 | int n = size/8; | |
945 | if (offset > Size_t_MAX / n - 1) /* would overflow */ | |
946 | Perl_croak_nocontext("Out of memory!"); | |
947 | offset *= n; | |
948 | } | |
bbb8a7e0 | 949 | |
67dd6f35 DM |
950 | len = (bitoffs + size + 7)/8; /* required number of bytes */ |
951 | if (targlen < offset || targlen - offset < len) { | |
952 | STRLEN newlen = offset > Size_t_MAX - len - 1 ? /* avoid overflow */ | |
953 | Size_t_MAX : offset + len + 1; | |
954 | s = (unsigned char*)SvGROW(targ, newlen); | |
955 | (void)memzero((char *)(s + targlen), newlen - targlen); | |
956 | SvCUR_set(targ, newlen - 1); | |
a0d0e21e | 957 | } |
8e84507e | 958 | |
79072805 | 959 | if (size < 8) { |
d223e1ea | 960 | mask = nBIT_MASK(size); |
79072805 | 961 | lval &= mask; |
bbb8a7e0 MHM |
962 | s[offset] &= ~(mask << bitoffs); |
963 | s[offset] |= lval << bitoffs; | |
79072805 LW |
964 | } |
965 | else { | |
966 | if (size == 8) | |
eb160463 | 967 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 968 | else if (size == 16) { |
eb160463 GS |
969 | s[offset ] = (U8)((lval >> 8) & 0xff); |
970 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
971 | } |
972 | else if (size == 32) { | |
eb160463 GS |
973 | s[offset ] = (U8)((lval >> 24) & 0xff); |
974 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
975 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
976 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 977 | } |
d7d93a81 | 978 | #ifdef UV_IS_QUAD |
c5a0f51a | 979 | else if (size == 64) { |
a2a5de95 NC |
980 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
981 | "Bit vector size > 32 non-portable"); | |
eb160463 GS |
982 | s[offset ] = (U8)((lval >> 56) & 0xff); |
983 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
984 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
985 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
986 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
987 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
988 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
989 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 990 | } |
dc1e3f56 | 991 | #endif |
79072805 | 992 | } |
7bb043c3 | 993 | SvSETMAGIC(targ); |
79072805 LW |
994 | } |
995 | ||
996 | void | |
864dbfa3 | 997 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 | 998 | { |
eb578fdb KW |
999 | long *dl; |
1000 | long *ll; | |
1001 | long *rl; | |
eb578fdb | 1002 | char *dc; |
463ee0b2 LW |
1003 | STRLEN leftlen; |
1004 | STRLEN rightlen; | |
eb578fdb KW |
1005 | const char *lc; |
1006 | const char *rc; | |
b404a7f5 | 1007 | STRLEN len = 0; |
bb7a0f54 | 1008 | STRLEN lensave; |
e62f0680 NC |
1009 | const char *lsave; |
1010 | const char *rsave; | |
bb7a0f54 | 1011 | STRLEN needlen = 0; |
08b6664b | 1012 | bool result_needs_to_be_utf8 = FALSE; |
b50535da KW |
1013 | bool left_utf8 = FALSE; |
1014 | bool right_utf8 = FALSE; | |
1015 | U8 * left_non_downgraded = NULL; | |
1016 | U8 * right_non_downgraded = NULL; | |
1017 | Size_t left_non_downgraded_len = 0; | |
1018 | Size_t right_non_downgraded_len = 0; | |
1019 | char * non_downgraded = NULL; | |
1020 | Size_t non_downgraded_len = 0; | |
0c57e439 | 1021 | |
7918f24d | 1022 | PERL_ARGS_ASSERT_DO_VOP; |
79072805 | 1023 | |
6b349a5c | 1024 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv))) |
f607343f | 1025 | SvPVCLEAR(sv); /* avoid undef warning on |= and ^= */ |
8c8eee82 | 1026 | if (sv == left) { |
08b6664b | 1027 | lc = SvPV_force_nomg(left, leftlen); |
8c8eee82 BM |
1028 | } |
1029 | else { | |
08b6664b | 1030 | lc = SvPV_nomg_const(left, leftlen); |
8c8eee82 BM |
1031 | SvPV_force_nomg_nolen(sv); |
1032 | } | |
08b6664b | 1033 | rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 | 1034 | |
0e6e7171 | 1035 | /* This needs to come after SvPV to ensure that string overloading has |
12abf4f0 NC |
1036 | fired off. */ |
1037 | ||
08b6664b KW |
1038 | /* Create downgraded temporaries of any UTF-8 encoded operands */ |
1039 | if (DO_UTF8(left)) { | |
b50535da | 1040 | const U8 * save_lc = (U8 *) lc; |
08b6664b | 1041 | |
b50535da | 1042 | left_utf8 = TRUE; |
08b6664b KW |
1043 | result_needs_to_be_utf8 = TRUE; |
1044 | ||
b50535da KW |
1045 | left_non_downgraded_len = leftlen; |
1046 | lc = (char *) bytes_from_utf8_loc((const U8 *) lc, &leftlen, | |
1047 | &left_utf8, | |
1048 | (const U8 **) &left_non_downgraded); | |
1049 | /* Calculate the number of trailing unconvertible bytes. This quantity | |
1050 | * is the original length minus the length of the converted portion. */ | |
1051 | left_non_downgraded_len -= left_non_downgraded - save_lc; | |
1052 | SAVEFREEPV(lc); | |
12abf4f0 | 1053 | } |
08b6664b | 1054 | if (DO_UTF8(right)) { |
b50535da | 1055 | const U8 * save_rc = (U8 *) rc; |
08b6664b | 1056 | |
b50535da | 1057 | right_utf8 = TRUE; |
08b6664b KW |
1058 | result_needs_to_be_utf8 = TRUE; |
1059 | ||
b50535da KW |
1060 | right_non_downgraded_len = rightlen; |
1061 | rc = (char *) bytes_from_utf8_loc((const U8 *) rc, &rightlen, | |
1062 | &right_utf8, | |
1063 | (const U8 **) &right_non_downgraded); | |
1064 | right_non_downgraded_len -= right_non_downgraded - save_rc; | |
1065 | SAVEFREEPV(rc); | |
1066 | } | |
1067 | ||
1068 | /* We set 'len' to the length that the operation actually operates on. The | |
1069 | * dangling part of the longer operand doesn't actually participate in the | |
1070 | * operation. What happens is that we pretend that the shorter operand has | |
1071 | * been extended to the right by enough imaginary zeros to match the length | |
1072 | * of the longer one. But we know in advance the result of the operation | |
1073 | * on zeros without having to do it. In the case of '&', the result is | |
1074 | * zero, and the dangling portion is simply discarded. For '|' and '^', the | |
1075 | * result is the same as the other operand, so the dangling part is just | |
c8b94fe0 JK |
1076 | * appended to the final result, unchanged. As of perl-5.32, we no longer |
1077 | * accept above-FF code points in the dangling portion. | |
1078 | */ | |
ba52ce15 | 1079 | if (left_utf8 || right_utf8) { |
c8b94fe0 | 1080 | Perl_croak(aTHX_ FATAL_ABOVE_FF_MSG, PL_op_desc[optype]); |
ba52ce15 | 1081 | } |
b50535da | 1082 | else { /* Neither is UTF-8 */ |
78ba9007 | 1083 | len = MIN(leftlen, rightlen); |
12abf4f0 NC |
1084 | } |
1085 | ||
b50535da | 1086 | lensave = len; |
08b6664b KW |
1087 | lsave = lc; |
1088 | rsave = rc; | |
b50535da | 1089 | |
9fdd7463 | 1090 | (void)SvPOK_only(sv); |
08b6664b | 1091 | if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe | 1092 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 MHM |
1093 | if (SvLEN(sv) < len + 1) { |
1094 | dc = SvGROW(sv, len + 1); | |
ff68c719 | 1095 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1096 | } | |
1097 | } | |
1098 | else { | |
bb7a0f54 MHM |
1099 | needlen = optype == OP_BIT_AND |
1100 | ? len : (leftlen > rightlen ? leftlen : rightlen); | |
a02a5408 | 1101 | Newxz(dc, needlen + 1, char); |
aa0a69cb | 1102 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 | 1103 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 | 1104 | } |
45f235c1 | 1105 | SvCUR_set(sv, len); |
0c57e439 | 1106 | |
79072805 | 1107 | if (len >= sizeof(long)*4 && |
d398c6bf TK |
1108 | !(PTR2nat(dc) % sizeof(long)) && |
1109 | !(PTR2nat(lc) % sizeof(long)) && | |
1110 | !(PTR2nat(rc) % sizeof(long))) /* It's almost always aligned... */ | |
79072805 | 1111 | { |
bb7a0f54 | 1112 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 LW |
1113 | len /= (sizeof(long)*4); |
1114 | ||
1115 | dl = (long*)dc; | |
1116 | ll = (long*)lc; | |
1117 | rl = (long*)rc; | |
1118 | ||
1119 | switch (optype) { | |
1120 | case OP_BIT_AND: | |
1121 | while (len--) { | |
1122 | *dl++ = *ll++ & *rl++; | |
1123 | *dl++ = *ll++ & *rl++; | |
1124 | *dl++ = *ll++ & *rl++; | |
1125 | *dl++ = *ll++ & *rl++; | |
1126 | } | |
1127 | break; | |
a0d0e21e | 1128 | case OP_BIT_XOR: |
79072805 LW |
1129 | while (len--) { |
1130 | *dl++ = *ll++ ^ *rl++; | |
1131 | *dl++ = *ll++ ^ *rl++; | |
1132 | *dl++ = *ll++ ^ *rl++; | |
1133 | *dl++ = *ll++ ^ *rl++; | |
1134 | } | |
1135 | break; | |
1136 | case OP_BIT_OR: | |
1137 | while (len--) { | |
1138 | *dl++ = *ll++ | *rl++; | |
1139 | *dl++ = *ll++ | *rl++; | |
1140 | *dl++ = *ll++ | *rl++; | |
1141 | *dl++ = *ll++ | *rl++; | |
1142 | } | |
1143 | } | |
1144 | ||
1145 | dc = (char*)dl; | |
1146 | lc = (char*)ll; | |
1147 | rc = (char*)rl; | |
1148 | ||
1149 | len = remainder; | |
1150 | } | |
17d44595 | 1151 | |
27a9d47d KW |
1152 | switch (optype) { |
1153 | case OP_BIT_AND: | |
1154 | while (len--) | |
1155 | *dc++ = *lc++ & *rc++; | |
1156 | *dc = '\0'; | |
1157 | break; | |
1158 | case OP_BIT_XOR: | |
1159 | while (len--) | |
1160 | *dc++ = *lc++ ^ *rc++; | |
1161 | goto mop_up; | |
1162 | case OP_BIT_OR: | |
1163 | while (len--) | |
1164 | *dc++ = *lc++ | *rc++; | |
1165 | mop_up: | |
1166 | len = lensave; | |
1167 | if (rightlen > len) { | |
1168 | if (dc == rc) | |
2324bdb9 | 1169 | SvCUR_set(sv, rightlen); |
27a9d47d KW |
1170 | else |
1171 | sv_catpvn_nomg(sv, rsave + len, rightlen - len); | |
1172 | } | |
1173 | else if (leftlen > len) { | |
1174 | if (dc == lc) | |
2324bdb9 | 1175 | SvCUR_set(sv, leftlen); |
27a9d47d KW |
1176 | else |
1177 | sv_catpvn_nomg(sv, lsave + len, leftlen - len); | |
1178 | } | |
1179 | *SvEND(sv) = '\0'; | |
392582f8 | 1180 | |
b50535da KW |
1181 | /* If there is trailing stuff that couldn't be converted from UTF-8, it |
1182 | * is appended as-is for the ^ and | operators. This preserves | |
1183 | * backwards compatibility */ | |
1184 | if (right_non_downgraded) { | |
1185 | non_downgraded = (char *) right_non_downgraded; | |
1186 | non_downgraded_len = right_non_downgraded_len; | |
1187 | } | |
1188 | else if (left_non_downgraded) { | |
1189 | non_downgraded = (char *) left_non_downgraded; | |
1190 | non_downgraded_len = left_non_downgraded_len; | |
1191 | } | |
1192 | ||
27a9d47d KW |
1193 | break; |
1194 | } | |
08b6664b KW |
1195 | |
1196 | if (result_needs_to_be_utf8) { | |
b50535da KW |
1197 | sv_utf8_upgrade_nomg(sv); |
1198 | ||
1199 | /* Append any trailing UTF-8 as-is. */ | |
1200 | if (non_downgraded) { | |
1201 | sv_catpvn_nomg(sv, non_downgraded, non_downgraded_len); | |
1202 | } | |
79072805 | 1203 | } |
08b6664b | 1204 | |
fb73857a | 1205 | SvTAINT(sv); |
79072805 | 1206 | } |
463ee0b2 | 1207 | |
b1c05ba5 | 1208 | |
a2232057 DM |
1209 | /* Perl_do_kv() may be: |
1210 | * * called directly as the pp function for pp_keys() and pp_values(); | |
a2232057 DM |
1211 | * * It may also be called directly when the op is OP_AVHVSWITCH, to |
1212 | * implement CORE::keys(), CORE::values(). | |
1213 | * | |
1214 | * In all cases it expects an HV on the stack and returns a list of keys, | |
1215 | * values, or key-value pairs, depending on PL_op. | |
1216 | */ | |
b1c05ba5 | 1217 | |
463ee0b2 | 1218 | OP * |
cea2e8a9 | 1219 | Perl_do_kv(pTHX) |
463ee0b2 | 1220 | { |
39644a26 | 1221 | dSP; |
73ff03e8 | 1222 | HV * const keys = MUTABLE_HV(POPs); |
1c23e2bd | 1223 | const U8 gimme = GIMME_V; |
a2232057 | 1224 | |
af3b1cba | 1225 | const I32 dokeys = (PL_op->op_type == OP_KEYS) |
a2232057 | 1226 | || ( PL_op->op_type == OP_AVHVSWITCH |
94184451 DM |
1227 | && (PL_op->op_private & OPpAVHVSWITCH_MASK) |
1228 | + OP_EACH == OP_KEYS); | |
a2232057 | 1229 | |
af3b1cba | 1230 | const I32 dovalues = (PL_op->op_type == OP_VALUES) |
a2232057 | 1231 | || ( PL_op->op_type == OP_AVHVSWITCH |
94184451 DM |
1232 | && (PL_op->op_private & OPpAVHVSWITCH_MASK) |
1233 | + OP_EACH == OP_VALUES); | |
a2232057 | 1234 | |
af3b1cba | 1235 | assert( PL_op->op_type == OP_KEYS |
a2232057 DM |
1236 | || PL_op->op_type == OP_VALUES |
1237 | || PL_op->op_type == OP_AVHVSWITCH); | |
463ee0b2 | 1238 | |
4fa080db DM |
1239 | assert(!( PL_op->op_type == OP_VALUES |
1240 | && (PL_op->op_private & OPpMAYBE_LVSUB))); | |
1241 | ||
800e9ae0 | 1242 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1243 | |
54310121 | 1244 | if (gimme == G_VOID) |
aa689395 | 1245 | RETURN; |
1246 | ||
54310121 | 1247 | if (gimme == G_SCALAR) { |
78f9721b | 1248 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
2154eca7 EB |
1249 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
1250 | sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0); | |
1251 | LvTYPE(ret) = 'k'; | |
1252 | LvTARG(ret) = SvREFCNT_inc_simple(keys); | |
1253 | PUSHs(ret); | |
81714fb9 | 1254 | } |
463ee0b2 | 1255 | else { |
2154eca7 EB |
1256 | IV i; |
1257 | dTARGET; | |
1258 | ||
af3b1cba DM |
1259 | /* note that in 'scalar(keys %h)' the OP_KEYS is usually |
1260 | * optimised away and the action is performed directly by the | |
1261 | * padhv or rv2hv op. We now only get here via OP_AVHVSWITCH | |
1262 | * and \&CORE::keys | |
1263 | */ | |
2154eca7 | 1264 | if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) { |
1b95d04f | 1265 | i = HvUSEDKEYS(keys); |
2154eca7 EB |
1266 | } |
1267 | else { | |
1268 | i = 0; | |
1269 | while (hv_iternext(keys)) i++; | |
1270 | } | |
1271 | PUSHi( i ); | |
463ee0b2 | 1272 | } |
463ee0b2 LW |
1273 | RETURN; |
1274 | } | |
1275 | ||
a061ab0b FC |
1276 | if (UNLIKELY(PL_op->op_private & OPpMAYBE_LVSUB)) { |
1277 | const I32 flags = is_lvalue_sub(); | |
1278 | if (flags && !(flags & OPpENTERSUB_INARGS)) | |
1279 | /* diag_listed_as: Can't modify %s in %s */ | |
1280 | Perl_croak(aTHX_ "Can't modify keys in list assignment"); | |
1281 | } | |
1282 | ||
8dc9003f DM |
1283 | PUTBACK; |
1284 | hv_pushkv(keys, (dokeys | (dovalues << 1))); | |
1285 | return NORMAL; | |
463ee0b2 | 1286 | } |
4e35701f | 1287 | |
af3babe4 | 1288 | /* |
14d04a33 | 1289 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 1290 | */ |