Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
bc89e66f | 3 | * Copyright (c) 1991-2001, Larry Wall |
79072805 LW |
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 | * | |
a0d0e21e LW |
8 | */ |
9 | ||
10 | /* | |
11 | * "'So that was the job I felt I had to do when I started,' thought Sam." | |
79072805 LW |
12 | */ |
13 | ||
14 | #include "EXTERN.h" | |
864dbfa3 | 15 | #define PERL_IN_DOOP_C |
79072805 LW |
16 | #include "perl.h" |
17 | ||
64ca3a65 | 18 | #ifndef PERL_MICRO |
79072805 LW |
19 | #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) |
20 | #include <signal.h> | |
21 | #endif | |
64ca3a65 | 22 | #endif |
79072805 | 23 | |
942e002e | 24 | STATIC I32 |
b250498f | 25 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 | 26 | { |
4757a243 | 27 | U8 *s; |
b250498f | 28 | U8 *d; |
4757a243 | 29 | U8 *send; |
b250498f | 30 | U8 *dstart; |
4757a243 | 31 | I32 matches = 0; |
9b877dbb | 32 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 | 33 | STRLEN len; |
4757a243 LW |
34 | short *tbl; |
35 | I32 ch; | |
79072805 | 36 | |
4757a243 LW |
37 | tbl = (short*)cPVOP->op_pv; |
38 | if (!tbl) | |
2269b42e | 39 | Perl_croak(aTHX_ "panic: do_trans_simple"); |
a0ed51b3 | 40 | |
4757a243 LW |
41 | s = (U8*)SvPV(sv, len); |
42 | send = s + len; | |
43 | ||
b250498f | 44 | /* First, take care of non-UTF8 input strings, because they're easy */ |
1aa99e6b | 45 | if (!SvUTF8(sv)) { |
01ec43d0 | 46 | while (s < send) { |
036b4402 | 47 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 GS |
48 | matches++; |
49 | *s++ = ch; | |
50 | } | |
51 | else | |
52 | s++; | |
53 | } | |
54 | SvSETMAGIC(sv); | |
b250498f GS |
55 | return matches; |
56 | } | |
4757a243 | 57 | |
b250498f | 58 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb IH |
59 | if (grows) |
60 | New(0, d, len*2+1, U8); | |
61 | else | |
62 | d = s; | |
b250498f GS |
63 | dstart = d; |
64 | while (s < send) { | |
ba210ebe | 65 | STRLEN ulen; |
9b877dbb | 66 | UV c; |
b250498f | 67 | |
b250498f | 68 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 | 69 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb | 70 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f | 71 | matches++; |
9041c2e3 | 72 | d = uvchr_to_utf8(d, ch); |
b250498f | 73 | s += ulen; |
01ec43d0 GS |
74 | } |
75 | else { /* No match -> copy */ | |
9b877dbb IH |
76 | Copy(s, d, ulen, U8); |
77 | d += ulen; | |
78 | s += ulen; | |
b250498f GS |
79 | } |
80 | } | |
9b877dbb IH |
81 | if (grows) { |
82 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
83 | Safefree(dstart); | |
84 | } | |
85 | else { | |
86 | *d = '\0'; | |
87 | SvCUR_set(sv, d - dstart); | |
88 | } | |
b250498f | 89 | SvUTF8_on(sv); |
b250498f | 90 | SvSETMAGIC(sv); |
4757a243 LW |
91 | return matches; |
92 | } | |
93 | ||
942e002e | 94 | STATIC I32 |
036b4402 | 95 | S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 | 96 | { |
4757a243 LW |
97 | U8 *s; |
98 | U8 *send; | |
99 | I32 matches = 0; | |
100 | STRLEN len; | |
101 | short *tbl; | |
8973db79 | 102 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 LW |
103 | |
104 | tbl = (short*)cPVOP->op_pv; | |
105 | if (!tbl) | |
2269b42e | 106 | Perl_croak(aTHX_ "panic: do_trans_count"); |
4757a243 LW |
107 | |
108 | s = (U8*)SvPV(sv, len); | |
109 | send = s + len; | |
110 | ||
1aa99e6b IH |
111 | if (!SvUTF8(sv)) |
112 | while (s < send) { | |
113 | if (tbl[*s++] >= 0) | |
036b4402 | 114 | matches++; |
1aa99e6b IH |
115 | } |
116 | else | |
117 | while (s < send) { | |
118 | UV c; | |
119 | STRLEN ulen; | |
9041c2e3 | 120 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 JH |
121 | if (c < 0x100) { |
122 | if (tbl[c] >= 0) | |
123 | matches++; | |
124 | } else if (complement) | |
1aa99e6b IH |
125 | matches++; |
126 | s += ulen; | |
127 | } | |
4757a243 LW |
128 | |
129 | return matches; | |
130 | } | |
131 | ||
942e002e | 132 | STATIC I32 |
b250498f | 133 | S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ |
4757a243 | 134 | { |
4757a243 LW |
135 | U8 *s; |
136 | U8 *send; | |
137 | U8 *d; | |
5e44153e | 138 | U8 *dstart; |
1aa99e6b | 139 | I32 isutf8; |
4757a243 | 140 | I32 matches = 0; |
9b877dbb | 141 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 JH |
142 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
143 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
144 | STRLEN len, rlen; | |
4757a243 LW |
145 | short *tbl; |
146 | I32 ch; | |
147 | ||
148 | tbl = (short*)cPVOP->op_pv; | |
149 | if (!tbl) | |
2269b42e | 150 | Perl_croak(aTHX_ "panic: do_trans_complex"); |
4757a243 LW |
151 | |
152 | s = (U8*)SvPV(sv, len); | |
1aa99e6b | 153 | isutf8 = SvUTF8(sv); |
4757a243 LW |
154 | send = s + len; |
155 | ||
1aa99e6b IH |
156 | if (!isutf8) { |
157 | dstart = d = s; | |
158 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
159 | U8* p = send; | |
160 | while (s < send) { | |
161 | if ((ch = tbl[*s]) >= 0) { | |
036b4402 GS |
162 | *d = ch; |
163 | matches++; | |
1aa99e6b IH |
164 | if (p != d - 1 || *p != *d) |
165 | p = d++; | |
166 | } | |
167 | else if (ch == -1) /* -1 is unmapped character */ | |
168 | *d++ = *s; | |
169 | else if (ch == -2) /* -2 is delete character */ | |
170 | matches++; | |
171 | s++; | |
172 | } | |
a0ed51b3 | 173 | } |
1aa99e6b IH |
174 | else { |
175 | while (s < send) { | |
176 | if ((ch = tbl[*s]) >= 0) { | |
177 | matches++; | |
178 | *d++ = ch; | |
179 | } | |
180 | else if (ch == -1) /* -1 is unmapped character */ | |
181 | *d++ = *s; | |
182 | else if (ch == -2) /* -2 is delete character */ | |
183 | matches++; | |
184 | s++; | |
185 | } | |
186 | } | |
187 | SvCUR_set(sv, d - dstart); | |
4757a243 | 188 | } |
1aa99e6b | 189 | else { /* isutf8 */ |
9b877dbb IH |
190 | if (grows) |
191 | New(0, d, len*2+1, U8); | |
192 | else | |
193 | d = s; | |
1aa99e6b | 194 | dstart = d; |
8973db79 JH |
195 | if (complement && !del) |
196 | rlen = tbl[0x100]; | |
1aa99e6b | 197 | |
4aecb5b5 JH |
198 | #ifdef MACOS_TRADITIONAL |
199 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
200 | #endif | |
201 | ||
1aa99e6b | 202 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
203 | UV pch = 0xfeedface; |
204 | while (s < send) { | |
205 | STRLEN len; | |
9041c2e3 | 206 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b | 207 | |
8973db79 JH |
208 | if (comp > 0xff) { |
209 | if (!complement) { | |
210 | Copy(s, d, len, U8); | |
211 | d += len; | |
212 | } | |
213 | else { | |
214 | matches++; | |
215 | if (!del) { | |
216 | ch = (comp - 0x100 < rlen) ? | |
217 | tbl[comp+1] : tbl[0x100+rlen]; | |
218 | if (ch != pch) { | |
9041c2e3 | 219 | d = uvchr_to_utf8(d, ch); |
8973db79 JH |
220 | pch = ch; |
221 | } | |
222 | s += len; | |
223 | continue; | |
224 | } | |
225 | } | |
9b877dbb | 226 | } |
1aa99e6b IH |
227 | else if ((ch = tbl[comp]) >= 0) { |
228 | matches++; | |
229 | if (ch != pch) { | |
9041c2e3 | 230 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
231 | pch = ch; |
232 | } | |
233 | s += len; | |
234 | continue; | |
235 | } | |
9b877dbb IH |
236 | else if (ch == -1) { /* -1 is unmapped character */ |
237 | Copy(s, d, len, U8); | |
238 | d += len; | |
239 | } | |
1aa99e6b IH |
240 | else if (ch == -2) /* -2 is delete character */ |
241 | matches++; | |
242 | s += len; | |
243 | pch = 0xfeedface; | |
244 | } | |
5d06d08e | 245 | } |
1aa99e6b IH |
246 | else { |
247 | while (s < send) { | |
248 | STRLEN len; | |
9041c2e3 | 249 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
250 | if (comp > 0xff) { |
251 | if (!complement) { | |
252 | Copy(s, d, len, U8); | |
253 | d += len; | |
254 | } | |
255 | else { | |
256 | matches++; | |
257 | if (!del) { | |
9041c2e3 NIS |
258 | if (comp - 0x100 < rlen) |
259 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 260 | else |
9041c2e3 | 261 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
262 | } |
263 | } | |
9b877dbb | 264 | } |
1aa99e6b | 265 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 266 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
267 | matches++; |
268 | } | |
269 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
270 | Copy(s, d, len, U8); |
271 | d += len; | |
1aa99e6b IH |
272 | } |
273 | else if (ch == -2) /* -2 is delete character */ | |
274 | matches++; | |
275 | s += len; | |
276 | } | |
277 | } | |
9b877dbb IH |
278 | if (grows) { |
279 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
280 | Safefree(dstart); | |
281 | } | |
282 | else { | |
283 | *d = '\0'; | |
284 | SvCUR_set(sv, d - dstart); | |
285 | } | |
1aa99e6b | 286 | SvUTF8_on(sv); |
4757a243 | 287 | } |
5e44153e | 288 | SvSETMAGIC(sv); |
4757a243 LW |
289 | return matches; |
290 | } | |
291 | ||
942e002e | 292 | STATIC I32 |
036b4402 | 293 | S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 | 294 | { |
4757a243 LW |
295 | U8 *s; |
296 | U8 *send; | |
297 | U8 *d; | |
036b4402 | 298 | U8 *start; |
1aa99e6b | 299 | U8 *dstart, *dend; |
4757a243 | 300 | I32 matches = 0; |
9b877dbb | 301 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
302 | STRLEN len; |
303 | ||
304 | SV* rv = (SV*)cSVOP->op_sv; | |
305 | HV* hv = (HV*)SvRV(rv); | |
306 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
307 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
308 | UV extra = none + 1; | |
309 | UV final; | |
310 | UV uv; | |
1aa99e6b IH |
311 | I32 isutf8; |
312 | U8 hibit = 0; | |
4757a243 LW |
313 | |
314 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
315 | isutf8 = SvUTF8(sv); |
316 | if (!isutf8) { | |
317 | U8 *t = s, *e = s + len; | |
318 | while (t < e) | |
fd400ab9 | 319 | if ((hibit = UTF8_IS_CONTINUED(*t++))) |
1aa99e6b IH |
320 | break; |
321 | if (hibit) | |
322 | s = bytes_to_utf8(s, &len); | |
323 | } | |
4757a243 | 324 | send = s + len; |
036b4402 | 325 | start = s; |
4757a243 LW |
326 | |
327 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
328 | if (svp) | |
329 | final = SvUV(*svp); | |
330 | ||
9b877dbb IH |
331 | if (grows) { |
332 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
333 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
334 | dend = d + len * 3; | |
335 | dstart = d; | |
336 | } | |
337 | else { | |
338 | dstart = d = s; | |
339 | dend = d + len; | |
340 | } | |
1aa99e6b | 341 | |
4757a243 LW |
342 | while (s < send) { |
343 | if ((uv = swash_fetch(rv, s)) < none) { | |
344 | s += UTF8SKIP(s); | |
345 | matches++; | |
9041c2e3 | 346 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
347 | } |
348 | else if (uv == none) { | |
1aa99e6b | 349 | int i = UTF8SKIP(s); |
9b877dbb IH |
350 | Copy(s, d, i, U8); |
351 | d += i; | |
352 | s += i; | |
4757a243 LW |
353 | } |
354 | else if (uv == extra) { | |
1aa99e6b | 355 | int i = UTF8SKIP(s); |
036b4402 | 356 | s += i; |
4757a243 | 357 | matches++; |
9041c2e3 | 358 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
359 | } |
360 | else | |
361 | s += UTF8SKIP(s); | |
1aa99e6b | 362 | |
9b877dbb | 363 | if (d > dend) { |
1aa99e6b IH |
364 | STRLEN clen = d - dstart; |
365 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
9b877dbb IH |
366 | if (!grows) |
367 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); | |
1aa99e6b IH |
368 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
369 | d = dstart + clen; | |
370 | dend = dstart + nlen; | |
371 | } | |
4757a243 | 372 | } |
16ec844d | 373 | if (grows || hibit) { |
9b877dbb IH |
374 | sv_setpvn(sv, (char*)dstart, d - dstart); |
375 | Safefree(dstart); | |
16ec844d IH |
376 | if (grows && hibit) |
377 | Safefree(start); | |
9b877dbb IH |
378 | } |
379 | else { | |
380 | *d = '\0'; | |
381 | SvCUR_set(sv, d - dstart); | |
382 | } | |
4757a243 | 383 | SvSETMAGIC(sv); |
1aa99e6b | 384 | SvUTF8_on(sv); |
9041c2e3 | 385 | /* Downgrading just 'cos it will is suspect - NI-S */ |
1aa99e6b IH |
386 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
387 | sv_utf8_downgrade(sv, TRUE); | |
4757a243 LW |
388 | |
389 | return matches; | |
390 | } | |
391 | ||
942e002e | 392 | STATIC I32 |
036b4402 | 393 | S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 | 394 | { |
4757a243 | 395 | U8 *s; |
1aa99e6b | 396 | U8 *start, *send; |
4757a243 LW |
397 | I32 matches = 0; |
398 | STRLEN len; | |
399 | ||
400 | SV* rv = (SV*)cSVOP->op_sv; | |
401 | HV* hv = (HV*)SvRV(rv); | |
402 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
403 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
404 | UV uv; | |
1aa99e6b | 405 | U8 hibit = 0; |
4757a243 LW |
406 | |
407 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
408 | if (!SvUTF8(sv)) { |
409 | U8 *t = s, *e = s + len; | |
410 | while (t < e) | |
d742c382 | 411 | if ((hibit = !UTF8_IS_INVARIANT(*t++))) |
1aa99e6b IH |
412 | break; |
413 | if (hibit) | |
414 | start = s = bytes_to_utf8(s, &len); | |
415 | } | |
4757a243 LW |
416 | send = s + len; |
417 | ||
418 | while (s < send) { | |
834a4ddd | 419 | if ((uv = swash_fetch(rv, s)) < none) |
4757a243 | 420 | matches++; |
834a4ddd | 421 | s += UTF8SKIP(s); |
4757a243 | 422 | } |
1aa99e6b IH |
423 | if (hibit) |
424 | Safefree(start); | |
4757a243 LW |
425 | |
426 | return matches; | |
427 | } | |
428 | ||
942e002e | 429 | STATIC I32 |
036b4402 | 430 | S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ |
4757a243 | 431 | { |
4757a243 | 432 | U8 *s; |
1aa99e6b | 433 | U8 *start, *send; |
4757a243 LW |
434 | U8 *d; |
435 | I32 matches = 0; | |
436 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 437 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 438 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
439 | SV* rv = (SV*)cSVOP->op_sv; |
440 | HV* hv = (HV*)SvRV(rv); | |
441 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
442 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
443 | UV extra = none + 1; | |
444 | UV final; | |
445 | UV uv; | |
446 | STRLEN len; | |
1aa99e6b IH |
447 | U8 *dstart, *dend; |
448 | I32 isutf8; | |
449 | U8 hibit = 0; | |
4757a243 LW |
450 | |
451 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
452 | isutf8 = SvUTF8(sv); |
453 | if (!isutf8) { | |
454 | U8 *t = s, *e = s + len; | |
455 | while (t < e) | |
d742c382 | 456 | if ((hibit = !UTF8_IS_INVARIANT(*t++))) |
1aa99e6b IH |
457 | break; |
458 | if (hibit) | |
459 | s = bytes_to_utf8(s, &len); | |
460 | } | |
4757a243 | 461 | send = s + len; |
1aa99e6b | 462 | start = s; |
4757a243 LW |
463 | |
464 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
465 | if (svp) | |
466 | final = SvUV(*svp); | |
467 | ||
9b877dbb IH |
468 | if (grows) { |
469 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
470 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
471 | dend = d + len * 3; | |
472 | dstart = d; | |
473 | } | |
474 | else { | |
475 | dstart = d = s; | |
476 | dend = d + len; | |
477 | } | |
4757a243 LW |
478 | |
479 | if (squash) { | |
480 | UV puv = 0xfeedface; | |
481 | while (s < send) { | |
1aa99e6b | 482 | uv = swash_fetch(rv, s); |
9041c2e3 | 483 | |
9b877dbb IH |
484 | if (d > dend) { |
485 | STRLEN clen = d - dstart; | |
486 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
487 | if (!grows) | |
488 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); | |
1aa99e6b IH |
489 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
490 | d = dstart + clen; | |
491 | dend = dstart + nlen; | |
4757a243 LW |
492 | } |
493 | if (uv < none) { | |
494 | matches++; | |
94472101 | 495 | s += UTF8SKIP(s); |
4757a243 | 496 | if (uv != puv) { |
9041c2e3 | 497 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
498 | puv = uv; |
499 | } | |
4757a243 LW |
500 | continue; |
501 | } | |
502 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 503 | int i = UTF8SKIP(s); |
9b877dbb IH |
504 | Copy(s, d, i, U8); |
505 | d += i; | |
506 | s += i; | |
4757a243 LW |
507 | puv = 0xfeedface; |
508 | continue; | |
a0ed51b3 | 509 | } |
4757a243 LW |
510 | else if (uv == extra && !del) { |
511 | matches++; | |
94472101 | 512 | s += UTF8SKIP(s); |
4757a243 | 513 | if (uv != puv) { |
9041c2e3 | 514 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
515 | puv = final; |
516 | } | |
4757a243 LW |
517 | continue; |
518 | } | |
01ec43d0 GS |
519 | matches++; /* "none+1" is delete character */ |
520 | s += UTF8SKIP(s); | |
a0ed51b3 | 521 | } |
79072805 LW |
522 | } |
523 | else { | |
4757a243 | 524 | while (s < send) { |
1aa99e6b | 525 | uv = swash_fetch(rv, s); |
9b877dbb IH |
526 | if (d > dend) { |
527 | STRLEN clen = d - dstart; | |
528 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
529 | if (!grows) | |
530 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); | |
1aa99e6b IH |
531 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
532 | d = dstart + clen; | |
533 | dend = dstart + nlen; | |
a0ed51b3 | 534 | } |
4757a243 LW |
535 | if (uv < none) { |
536 | matches++; | |
01ec43d0 | 537 | s += UTF8SKIP(s); |
9041c2e3 | 538 | d = uvchr_to_utf8(d, uv); |
4757a243 | 539 | continue; |
a0ed51b3 | 540 | } |
4757a243 | 541 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 542 | int i = UTF8SKIP(s); |
9b877dbb IH |
543 | Copy(s, d, i, U8); |
544 | d += i; | |
545 | s += i; | |
4757a243 | 546 | continue; |
79072805 | 547 | } |
4757a243 LW |
548 | else if (uv == extra && !del) { |
549 | matches++; | |
01ec43d0 | 550 | s += UTF8SKIP(s); |
9041c2e3 | 551 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
552 | continue; |
553 | } | |
01ec43d0 GS |
554 | matches++; /* "none+1" is delete character */ |
555 | s += UTF8SKIP(s); | |
79072805 | 556 | } |
4757a243 | 557 | } |
16ec844d | 558 | if (grows || hibit) { |
9b877dbb IH |
559 | sv_setpvn(sv, (char*)dstart, d - dstart); |
560 | Safefree(dstart); | |
16ec844d IH |
561 | if (grows && hibit) |
562 | Safefree(start); | |
9b877dbb IH |
563 | } |
564 | else { | |
565 | *d = '\0'; | |
566 | SvCUR_set(sv, d - dstart); | |
567 | } | |
1aa99e6b | 568 | SvUTF8_on(sv); |
1aa99e6b IH |
569 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
570 | sv_utf8_downgrade(sv, TRUE); | |
4757a243 LW |
571 | SvSETMAGIC(sv); |
572 | ||
573 | return matches; | |
574 | } | |
575 | ||
576 | I32 | |
864dbfa3 | 577 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
578 | { |
579 | STRLEN len; | |
8e84507e | 580 | I32 hasutf = (PL_op->op_private & |
036b4402 | 581 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 LW |
582 | |
583 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) | |
cea2e8a9 | 584 | Perl_croak(aTHX_ PL_no_modify); |
4757a243 LW |
585 | |
586 | (void)SvPV(sv, len); | |
587 | if (!len) | |
588 | return 0; | |
589 | if (!SvPOKp(sv)) | |
590 | (void)SvPV_force(sv, len); | |
2de7b02f GS |
591 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) |
592 | (void)SvPOK_only_UTF8(sv); | |
4757a243 | 593 | |
cea2e8a9 | 594 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 595 | |
036b4402 | 596 | switch (PL_op->op_private & ~hasutf & 63) { |
4757a243 | 597 | case 0: |
01ec43d0 GS |
598 | if (hasutf) |
599 | return do_trans_simple_utf8(sv); | |
600 | else | |
601 | return do_trans_simple(sv); | |
4757a243 LW |
602 | |
603 | case OPpTRANS_IDENTICAL: | |
01ec43d0 GS |
604 | if (hasutf) |
605 | return do_trans_count_utf8(sv); | |
606 | else | |
607 | return do_trans_count(sv); | |
4757a243 LW |
608 | |
609 | default: | |
01ec43d0 | 610 | if (hasutf) |
036b4402 | 611 | return do_trans_complex_utf8(sv); |
4757a243 | 612 | else |
036b4402 | 613 | return do_trans_complex(sv); |
79072805 | 614 | } |
79072805 LW |
615 | } |
616 | ||
617 | void | |
864dbfa3 | 618 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
619 | { |
620 | SV **oldmark = mark; | |
621 | register I32 items = sp - mark; | |
79072805 | 622 | register STRLEN len; |
463ee0b2 LW |
623 | STRLEN delimlen; |
624 | register char *delim = SvPV(del, delimlen); | |
625 | STRLEN tmplen; | |
79072805 LW |
626 | |
627 | mark++; | |
628 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 629 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
630 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
631 | while (items-- > 0) { | |
1426bbf4 | 632 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 LW |
633 | SvPV(*mark, tmplen); |
634 | len += tmplen; | |
79072805 LW |
635 | } |
636 | mark++; | |
637 | } | |
638 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
639 | ||
640 | mark = oldmark; | |
db7c17d7 | 641 | items = sp - mark; |
79072805 LW |
642 | ++mark; |
643 | } | |
644 | ||
463ee0b2 | 645 | if (items-- > 0) { |
92d29cee JH |
646 | sv_setpv(sv, ""); |
647 | if (*mark) | |
648 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
649 | mark++; |
650 | } | |
79072805 LW |
651 | else |
652 | sv_setpv(sv,""); | |
c512ce4f | 653 | if (delimlen) { |
79072805 | 654 | for (; items > 0; items--,mark++) { |
c512ce4f | 655 | sv_catsv(sv,del); |
79072805 LW |
656 | sv_catsv(sv,*mark); |
657 | } | |
658 | } | |
659 | else { | |
660 | for (; items > 0; items--,mark++) | |
661 | sv_catsv(sv,*mark); | |
662 | } | |
663 | SvSETMAGIC(sv); | |
664 | } | |
665 | ||
666 | void | |
864dbfa3 | 667 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 668 | { |
46fc3d4c | 669 | STRLEN patlen; |
670 | char *pat = SvPV(*sarg, patlen); | |
671 | bool do_taint = FALSE; | |
672 | ||
673 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); | |
79072805 | 674 | SvSETMAGIC(sv); |
46fc3d4c | 675 | if (do_taint) |
676 | SvTAINTED_on(sv); | |
79072805 LW |
677 | } |
678 | ||
33b45480 | 679 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
680 | UV |
681 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
682 | { | |
683 | STRLEN srclen, len; | |
684 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
685 | UV retnum = 0; | |
686 | ||
a50d7633 | 687 | if (offset < 0) |
81e118e0 | 688 | return retnum; |
8e84507e | 689 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 690 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 691 | |
dcad2880 | 692 | if (SvUTF8(sv)) |
33b45480 | 693 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 694 | |
81e118e0 JH |
695 | offset *= size; /* turn into bit offset */ |
696 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
697 | if (len > srclen) { | |
698 | if (size <= 8) | |
699 | retnum = 0; | |
700 | else { | |
701 | offset >>= 3; /* turn into byte offset */ | |
702 | if (size == 16) { | |
703 | if (offset >= srclen) | |
704 | retnum = 0; | |
705 | else | |
628e1a40 | 706 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
707 | } |
708 | else if (size == 32) { | |
709 | if (offset >= srclen) | |
710 | retnum = 0; | |
711 | else if (offset + 1 >= srclen) | |
712 | retnum = | |
713 | ((UV) s[offset ] << 24); | |
714 | else if (offset + 2 >= srclen) | |
715 | retnum = | |
716 | ((UV) s[offset ] << 24) + | |
717 | ((UV) s[offset + 1] << 16); | |
718 | else | |
719 | retnum = | |
720 | ((UV) s[offset ] << 24) + | |
721 | ((UV) s[offset + 1] << 16) + | |
722 | ( s[offset + 2] << 8); | |
723 | } | |
d7d93a81 | 724 | #ifdef UV_IS_QUAD |
c5a0f51a | 725 | else if (size == 64) { |
c5a0f51a JH |
726 | if (ckWARN(WARN_PORTABLE)) |
727 | Perl_warner(aTHX_ WARN_PORTABLE, | |
728 | "Bit vector size > 32 non-portable"); | |
729 | if (offset >= srclen) | |
730 | retnum = 0; | |
731 | else if (offset + 1 >= srclen) | |
732 | retnum = | |
733 | (UV) s[offset ] << 56; | |
734 | else if (offset + 2 >= srclen) | |
735 | retnum = | |
736 | ((UV) s[offset ] << 56) + | |
737 | ((UV) s[offset + 1] << 48); | |
738 | else if (offset + 3 >= srclen) | |
739 | retnum = | |
740 | ((UV) s[offset ] << 56) + | |
741 | ((UV) s[offset + 1] << 48) + | |
742 | ((UV) s[offset + 2] << 40); | |
743 | else if (offset + 4 >= srclen) | |
744 | retnum = | |
745 | ((UV) s[offset ] << 56) + | |
746 | ((UV) s[offset + 1] << 48) + | |
747 | ((UV) s[offset + 2] << 40) + | |
748 | ((UV) s[offset + 3] << 32); | |
749 | else if (offset + 5 >= srclen) | |
750 | retnum = | |
751 | ((UV) s[offset ] << 56) + | |
752 | ((UV) s[offset + 1] << 48) + | |
753 | ((UV) s[offset + 2] << 40) + | |
754 | ((UV) s[offset + 3] << 32) + | |
755 | ( s[offset + 4] << 24); | |
756 | else if (offset + 6 >= srclen) | |
757 | retnum = | |
758 | ((UV) s[offset ] << 56) + | |
759 | ((UV) s[offset + 1] << 48) + | |
760 | ((UV) s[offset + 2] << 40) + | |
761 | ((UV) s[offset + 3] << 32) + | |
762 | ((UV) s[offset + 4] << 24) + | |
763 | ((UV) s[offset + 5] << 16); | |
764 | else | |
8e84507e | 765 | retnum = |
c5a0f51a JH |
766 | ((UV) s[offset ] << 56) + |
767 | ((UV) s[offset + 1] << 48) + | |
768 | ((UV) s[offset + 2] << 40) + | |
769 | ((UV) s[offset + 3] << 32) + | |
770 | ((UV) s[offset + 4] << 24) + | |
771 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 772 | ( s[offset + 6] << 8); |
c5a0f51a JH |
773 | } |
774 | #endif | |
81e118e0 JH |
775 | } |
776 | } | |
777 | else if (size < 8) | |
778 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
779 | else { | |
780 | offset >>= 3; /* turn into byte offset */ | |
781 | if (size == 8) | |
782 | retnum = s[offset]; | |
783 | else if (size == 16) | |
784 | retnum = | |
628e1a40 | 785 | ((UV) s[offset] << 8) + |
81e118e0 JH |
786 | s[offset + 1]; |
787 | else if (size == 32) | |
788 | retnum = | |
789 | ((UV) s[offset ] << 24) + | |
790 | ((UV) s[offset + 1] << 16) + | |
791 | ( s[offset + 2] << 8) + | |
792 | s[offset + 3]; | |
d7d93a81 | 793 | #ifdef UV_IS_QUAD |
c5a0f51a | 794 | else if (size == 64) { |
c5a0f51a JH |
795 | if (ckWARN(WARN_PORTABLE)) |
796 | Perl_warner(aTHX_ WARN_PORTABLE, | |
797 | "Bit vector size > 32 non-portable"); | |
798 | retnum = | |
799 | ((UV) s[offset ] << 56) + | |
800 | ((UV) s[offset + 1] << 48) + | |
801 | ((UV) s[offset + 2] << 40) + | |
802 | ((UV) s[offset + 3] << 32) + | |
803 | ((UV) s[offset + 4] << 24) + | |
804 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 805 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
806 | s[offset + 7]; |
807 | } | |
808 | #endif | |
81e118e0 JH |
809 | } |
810 | ||
811 | return retnum; | |
812 | } | |
813 | ||
33b45480 SB |
814 | /* currently converts input to bytes if possible but doesn't sweat failures, |
815 | * although it does ensure that the string it clobbers is not marked as | |
816 | * utf8-valid any more | |
817 | */ | |
79072805 | 818 | void |
864dbfa3 | 819 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
820 | { |
821 | SV *targ = LvTARG(sv); | |
822 | register I32 offset; | |
823 | register I32 size; | |
8990e307 | 824 | register unsigned char *s; |
81e118e0 | 825 | register UV lval; |
79072805 | 826 | I32 mask; |
a0d0e21e LW |
827 | STRLEN targlen; |
828 | STRLEN len; | |
79072805 | 829 | |
8990e307 LW |
830 | if (!targ) |
831 | return; | |
a0d0e21e | 832 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 833 | if (SvUTF8(targ)) { |
33b45480 SB |
834 | /* This is handled by the SvPOK_only below... |
835 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
836 | SvUTF8_off(targ); | |
837 | */ | |
838 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
839 | } |
840 | ||
4ebbc975 | 841 | (void)SvPOK_only(targ); |
81e118e0 | 842 | lval = SvUV(sv); |
79072805 | 843 | offset = LvTARGOFF(sv); |
fe58ced6 MG |
844 | if (offset < 0) |
845 | Perl_croak(aTHX_ "Assigning to negative offset in vec"); | |
79072805 | 846 | size = LvTARGLEN(sv); |
8e84507e | 847 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 848 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 849 | |
81e118e0 JH |
850 | offset *= size; /* turn into bit offset */ |
851 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
852 | if (len > targlen) { |
853 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 854 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
855 | SvCUR_set(targ, len); |
856 | } | |
8e84507e | 857 | |
79072805 LW |
858 | if (size < 8) { |
859 | mask = (1 << size) - 1; | |
860 | size = offset & 7; | |
861 | lval &= mask; | |
81e118e0 | 862 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
863 | s[offset] &= ~(mask << size); |
864 | s[offset] |= lval << size; | |
865 | } | |
866 | else { | |
81e118e0 | 867 | offset >>= 3; /* turn into byte offset */ |
79072805 | 868 | if (size == 8) |
c5a0f51a | 869 | s[offset ] = lval & 0xff; |
79072805 | 870 | else if (size == 16) { |
c5a0f51a JH |
871 | s[offset ] = (lval >> 8) & 0xff; |
872 | s[offset+1] = lval & 0xff; | |
79072805 LW |
873 | } |
874 | else if (size == 32) { | |
c5a0f51a JH |
875 | s[offset ] = (lval >> 24) & 0xff; |
876 | s[offset+1] = (lval >> 16) & 0xff; | |
877 | s[offset+2] = (lval >> 8) & 0xff; | |
878 | s[offset+3] = lval & 0xff; | |
879 | } | |
d7d93a81 | 880 | #ifdef UV_IS_QUAD |
c5a0f51a | 881 | else if (size == 64) { |
c5a0f51a JH |
882 | if (ckWARN(WARN_PORTABLE)) |
883 | Perl_warner(aTHX_ WARN_PORTABLE, | |
884 | "Bit vector size > 32 non-portable"); | |
885 | s[offset ] = (lval >> 56) & 0xff; | |
886 | s[offset+1] = (lval >> 48) & 0xff; | |
887 | s[offset+2] = (lval >> 40) & 0xff; | |
888 | s[offset+3] = (lval >> 32) & 0xff; | |
889 | s[offset+4] = (lval >> 24) & 0xff; | |
890 | s[offset+5] = (lval >> 16) & 0xff; | |
891 | s[offset+6] = (lval >> 8) & 0xff; | |
892 | s[offset+7] = lval & 0xff; | |
79072805 | 893 | } |
dc1e3f56 | 894 | #endif |
79072805 | 895 | } |
7bb043c3 | 896 | SvSETMAGIC(targ); |
79072805 LW |
897 | } |
898 | ||
899 | void | |
864dbfa3 | 900 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 901 | { |
463ee0b2 | 902 | STRLEN len; |
a0d0e21e | 903 | char *s; |
8e84507e | 904 | |
79072805 | 905 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
906 | register I32 i; |
907 | I32 max; | |
908 | AV* av = (AV*)sv; | |
909 | max = AvFILL(av); | |
910 | for (i = 0; i <= max; i++) { | |
911 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 912 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
913 | do_chop(astr, sv); |
914 | } | |
915 | return; | |
79072805 | 916 | } |
aa854799 | 917 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
918 | HV* hv = (HV*)sv; |
919 | HE* entry; | |
920 | (void)hv_iterinit(hv); | |
921 | /*SUPPRESS 560*/ | |
155aba94 | 922 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
923 | do_chop(astr,hv_iterval(hv,entry)); |
924 | return; | |
79072805 | 925 | } |
aa854799 | 926 | else if (SvREADONLY(sv)) |
cea2e8a9 | 927 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e | 928 | s = SvPV(sv, len); |
748a9306 | 929 | if (len && !SvPOK(sv)) |
a0d0e21e | 930 | s = SvPV_force(sv, len); |
7e2040f0 | 931 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
932 | if (s && len) { |
933 | char *send = s + len; | |
934 | char *start = s; | |
935 | s = send - 1; | |
a0dbb045 JH |
936 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
937 | s--; | |
9041c2e3 | 938 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
939 | sv_setpvn(astr, s, send - s); |
940 | *s = '\0'; | |
941 | SvCUR_set(sv, s - start); | |
942 | SvNIOK_off(sv); | |
943 | SvUTF8_on(astr); | |
944 | } | |
a0ed51b3 LW |
945 | } |
946 | else | |
947 | sv_setpvn(astr, "", 0); | |
948 | } | |
7e2040f0 | 949 | else if (s && len) { |
a0d0e21e LW |
950 | s += --len; |
951 | sv_setpvn(astr, s, 1); | |
952 | *s = '\0'; | |
953 | SvCUR_set(sv, len); | |
2c19a612 | 954 | SvUTF8_off(sv); |
a0d0e21e | 955 | SvNIOK_off(sv); |
79072805 LW |
956 | } |
957 | else | |
a0d0e21e LW |
958 | sv_setpvn(astr, "", 0); |
959 | SvSETMAGIC(sv); | |
7e2040f0 | 960 | } |
a0d0e21e LW |
961 | |
962 | I32 | |
864dbfa3 | 963 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 964 | { |
c07a80fd | 965 | register I32 count; |
a0d0e21e LW |
966 | STRLEN len; |
967 | char *s; | |
c07a80fd | 968 | |
3280af22 | 969 | if (RsSNARF(PL_rs)) |
c07a80fd | 970 | return 0; |
4c5a6083 GS |
971 | if (RsRECORD(PL_rs)) |
972 | return 0; | |
c07a80fd | 973 | count = 0; |
a0d0e21e LW |
974 | if (SvTYPE(sv) == SVt_PVAV) { |
975 | register I32 i; | |
976 | I32 max; | |
977 | AV* av = (AV*)sv; | |
978 | max = AvFILL(av); | |
979 | for (i = 0; i <= max; i++) { | |
980 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 981 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
982 | count += do_chomp(sv); |
983 | } | |
984 | return count; | |
985 | } | |
aa854799 | 986 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
987 | HV* hv = (HV*)sv; |
988 | HE* entry; | |
989 | (void)hv_iterinit(hv); | |
990 | /*SUPPRESS 560*/ | |
155aba94 | 991 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
992 | count += do_chomp(hv_iterval(hv,entry)); |
993 | return count; | |
994 | } | |
aa854799 | 995 | else if (SvREADONLY(sv)) |
cea2e8a9 | 996 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e LW |
997 | s = SvPV(sv, len); |
998 | if (len && !SvPOKp(sv)) | |
999 | s = SvPV_force(sv, len); | |
1000 | if (s && len) { | |
1001 | s += --len; | |
3280af22 | 1002 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1003 | if (*s != '\n') |
1004 | goto nope; | |
1005 | ++count; | |
1006 | while (len && s[-1] == '\n') { | |
1007 | --len; | |
1008 | --s; | |
1009 | ++count; | |
1010 | } | |
1011 | } | |
a0d0e21e | 1012 | else { |
c07a80fd | 1013 | STRLEN rslen; |
3280af22 | 1014 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 1015 | if (rslen == 1) { |
1016 | if (*s != *rsptr) | |
1017 | goto nope; | |
1018 | ++count; | |
1019 | } | |
1020 | else { | |
8c2cee6f | 1021 | if (len < rslen - 1) |
c07a80fd | 1022 | goto nope; |
1023 | len -= rslen - 1; | |
1024 | s -= rslen - 1; | |
36477c24 | 1025 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1026 | goto nope; |
1027 | count += rslen; | |
1028 | } | |
a0d0e21e | 1029 | } |
a0d0e21e LW |
1030 | *s = '\0'; |
1031 | SvCUR_set(sv, len); | |
1032 | SvNIOK_off(sv); | |
1033 | } | |
1034 | nope: | |
1035 | SvSETMAGIC(sv); | |
1036 | return count; | |
8e84507e | 1037 | } |
79072805 LW |
1038 | |
1039 | void | |
864dbfa3 | 1040 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1041 | { |
1042 | #ifdef LIBERAL | |
1043 | register long *dl; | |
1044 | register long *ll; | |
1045 | register long *rl; | |
1046 | #endif | |
1047 | register char *dc; | |
463ee0b2 LW |
1048 | STRLEN leftlen; |
1049 | STRLEN rightlen; | |
7a4c00b4 | 1050 | register char *lc; |
1051 | register char *rc; | |
79072805 | 1052 | register I32 len; |
a0d0e21e | 1053 | I32 lensave; |
7a4c00b4 | 1054 | char *lsave; |
1055 | char *rsave; | |
0c57e439 GS |
1056 | bool left_utf = DO_UTF8(left); |
1057 | bool right_utf = DO_UTF8(right); | |
c9b3c8d0 | 1058 | I32 needlen; |
0c57e439 GS |
1059 | |
1060 | if (left_utf && !right_utf) | |
1061 | sv_utf8_upgrade(right); | |
a1ca4561 | 1062 | else if (!left_utf && right_utf) |
0c57e439 | 1063 | sv_utf8_upgrade(left); |
79072805 | 1064 | |
1fbd88dc CS |
1065 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1066 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 1067 | lsave = lc = SvPV(left, leftlen); |
1068 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 1069 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1070 | lensave = len; |
c9b3c8d0 JH |
1071 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1072 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1073 | Newz(801, dc, needlen + 1, char); | |
1074 | } | |
1075 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d GS |
1076 | STRLEN n_a; |
1077 | dc = SvPV_force(sv, n_a); | |
ff68c719 | 1078 | if (SvCUR(sv) < len) { |
1079 | dc = SvGROW(sv, len + 1); | |
1080 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); | |
1081 | } | |
c9b3c8d0 JH |
1082 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1083 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1084 | } |
1085 | else { | |
c9b3c8d0 JH |
1086 | needlen = ((optype == OP_BIT_AND) |
1087 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1088 | Newz(801, dc, needlen + 1, char); |
1089 | (void)sv_usepvn(sv, dc, needlen); | |
1090 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1091 | } |
a0d0e21e LW |
1092 | SvCUR_set(sv, len); |
1093 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1094 | if (left_utf || right_utf) { |
1095 | UV duc, luc, ruc; | |
c9b3c8d0 | 1096 | char *dcsave = dc; |
0c57e439 GS |
1097 | STRLEN lulen = leftlen; |
1098 | STRLEN rulen = rightlen; | |
ba210ebe | 1099 | STRLEN ulen; |
0c57e439 GS |
1100 | |
1101 | switch (optype) { | |
1102 | case OP_BIT_AND: | |
1103 | while (lulen && rulen) { | |
9041c2e3 | 1104 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1105 | lc += ulen; |
1106 | lulen -= ulen; | |
9041c2e3 | 1107 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1108 | rc += ulen; |
1109 | rulen -= ulen; | |
1110 | duc = luc & ruc; | |
9041c2e3 | 1111 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1112 | } |
c9b3c8d0 JH |
1113 | if (sv == left || sv == right) |
1114 | (void)sv_usepvn(sv, dcsave, needlen); | |
1115 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1116 | break; |
1117 | case OP_BIT_XOR: | |
1118 | while (lulen && rulen) { | |
9041c2e3 | 1119 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1120 | lc += ulen; |
1121 | lulen -= ulen; | |
9041c2e3 | 1122 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1123 | rc += ulen; |
1124 | rulen -= ulen; | |
1125 | duc = luc ^ ruc; | |
9041c2e3 | 1126 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1127 | } |
1128 | goto mop_up_utf; | |
1129 | case OP_BIT_OR: | |
1130 | while (lulen && rulen) { | |
9041c2e3 | 1131 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1132 | lc += ulen; |
1133 | lulen -= ulen; | |
9041c2e3 | 1134 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1135 | rc += ulen; |
1136 | rulen -= ulen; | |
1137 | duc = luc | ruc; | |
9041c2e3 | 1138 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1139 | } |
1140 | mop_up_utf: | |
c9b3c8d0 JH |
1141 | if (sv == left || sv == right) |
1142 | (void)sv_usepvn(sv, dcsave, needlen); | |
1143 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1144 | if (rulen) |
1145 | sv_catpvn(sv, rc, rulen); | |
1146 | else if (lulen) | |
1147 | sv_catpvn(sv, lc, lulen); | |
1148 | else | |
1149 | *SvEND(sv) = '\0'; | |
1150 | break; | |
1151 | } | |
1152 | SvUTF8_on(sv); | |
1153 | goto finish; | |
1154 | } | |
1155 | else | |
79072805 LW |
1156 | #ifdef LIBERAL |
1157 | if (len >= sizeof(long)*4 && | |
1158 | !((long)dc % sizeof(long)) && | |
1159 | !((long)lc % sizeof(long)) && | |
1160 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1161 | { | |
1162 | I32 remainder = len % (sizeof(long)*4); | |
1163 | len /= (sizeof(long)*4); | |
1164 | ||
1165 | dl = (long*)dc; | |
1166 | ll = (long*)lc; | |
1167 | rl = (long*)rc; | |
1168 | ||
1169 | switch (optype) { | |
1170 | case OP_BIT_AND: | |
1171 | while (len--) { | |
1172 | *dl++ = *ll++ & *rl++; | |
1173 | *dl++ = *ll++ & *rl++; | |
1174 | *dl++ = *ll++ & *rl++; | |
1175 | *dl++ = *ll++ & *rl++; | |
1176 | } | |
1177 | break; | |
a0d0e21e | 1178 | case OP_BIT_XOR: |
79072805 LW |
1179 | while (len--) { |
1180 | *dl++ = *ll++ ^ *rl++; | |
1181 | *dl++ = *ll++ ^ *rl++; | |
1182 | *dl++ = *ll++ ^ *rl++; | |
1183 | *dl++ = *ll++ ^ *rl++; | |
1184 | } | |
1185 | break; | |
1186 | case OP_BIT_OR: | |
1187 | while (len--) { | |
1188 | *dl++ = *ll++ | *rl++; | |
1189 | *dl++ = *ll++ | *rl++; | |
1190 | *dl++ = *ll++ | *rl++; | |
1191 | *dl++ = *ll++ | *rl++; | |
1192 | } | |
1193 | } | |
1194 | ||
1195 | dc = (char*)dl; | |
1196 | lc = (char*)ll; | |
1197 | rc = (char*)rl; | |
1198 | ||
1199 | len = remainder; | |
1200 | } | |
1201 | #endif | |
a0d0e21e | 1202 | { |
a0d0e21e LW |
1203 | switch (optype) { |
1204 | case OP_BIT_AND: | |
1205 | while (len--) | |
1206 | *dc++ = *lc++ & *rc++; | |
1207 | break; | |
1208 | case OP_BIT_XOR: | |
1209 | while (len--) | |
1210 | *dc++ = *lc++ ^ *rc++; | |
1211 | goto mop_up; | |
1212 | case OP_BIT_OR: | |
1213 | while (len--) | |
1214 | *dc++ = *lc++ | *rc++; | |
1215 | mop_up: | |
1216 | len = lensave; | |
1217 | if (rightlen > len) | |
1218 | sv_catpvn(sv, rsave + len, rightlen - len); | |
1219 | else if (leftlen > len) | |
1220 | sv_catpvn(sv, lsave + len, leftlen - len); | |
4633a7c4 LW |
1221 | else |
1222 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1223 | break; |
1224 | } | |
79072805 | 1225 | } |
0c57e439 | 1226 | finish: |
fb73857a | 1227 | SvTAINT(sv); |
79072805 | 1228 | } |
463ee0b2 LW |
1229 | |
1230 | OP * | |
cea2e8a9 | 1231 | Perl_do_kv(pTHX) |
463ee0b2 | 1232 | { |
39644a26 | 1233 | dSP; |
463ee0b2 | 1234 | HV *hv = (HV*)POPs; |
800e9ae0 | 1235 | HV *keys; |
463ee0b2 | 1236 | register HE *entry; |
463ee0b2 | 1237 | SV *tmpstr; |
54310121 | 1238 | I32 gimme = GIMME_V; |
533c011a NIS |
1239 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1240 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
c750a3ec | 1241 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
8e84507e NIS |
1242 | |
1243 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1244 | dokeys = dovalues = TRUE; |
463ee0b2 | 1245 | |
85581909 | 1246 | if (!hv) { |
78f9721b | 1247 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1248 | dTARGET; /* make sure to clear its target here */ |
1249 | if (SvTYPE(TARG) == SVt_PVLV) | |
1250 | LvTARG(TARG) = Nullsv; | |
1251 | PUSHs(TARG); | |
1252 | } | |
463ee0b2 | 1253 | RETURN; |
85581909 | 1254 | } |
748a9306 | 1255 | |
800e9ae0 JP |
1256 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1257 | (void)hv_iterinit(keys); /* always reset iterator regardless */ | |
748a9306 | 1258 | |
54310121 | 1259 | if (gimme == G_VOID) |
aa689395 | 1260 | RETURN; |
1261 | ||
54310121 | 1262 | if (gimme == G_SCALAR) { |
6ee623d5 | 1263 | IV i; |
463ee0b2 LW |
1264 | dTARGET; |
1265 | ||
78f9721b | 1266 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1267 | if (SvTYPE(TARG) < SVt_PVLV) { |
1268 | sv_upgrade(TARG, SVt_PVLV); | |
1269 | sv_magic(TARG, Nullsv, 'k', Nullch, 0); | |
1270 | } | |
1271 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1272 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1273 | if (LvTARG(TARG)) |
1274 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1275 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1276 | } |
85581909 SB |
1277 | PUSHs(TARG); |
1278 | RETURN; | |
1279 | } | |
1280 | ||
33c27489 | 1281 | if (! SvTIED_mg((SV*)keys, 'P')) |
800e9ae0 | 1282 | i = HvKEYS(keys); |
463ee0b2 LW |
1283 | else { |
1284 | i = 0; | |
463ee0b2 | 1285 | /*SUPPRESS 560*/ |
800e9ae0 | 1286 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1287 | } |
1288 | PUSHi( i ); | |
1289 | RETURN; | |
1290 | } | |
1291 | ||
8ed4b672 | 1292 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1293 | |
463ee0b2 | 1294 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1295 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1296 | SPAGAIN; |
6bfc225d | 1297 | if (dokeys) |
8c2cee6f | 1298 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ |
463ee0b2 | 1299 | if (dovalues) { |
463ee0b2 | 1300 | PUTBACK; |
b6429b1b GS |
1301 | tmpstr = realhv ? |
1302 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); | |
cea2e8a9 | 1303 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1304 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1305 | HvMAX(keys)+1, |
1306 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1307 | SPAGAIN; |
46fc3d4c | 1308 | XPUSHs(tmpstr); |
463ee0b2 LW |
1309 | } |
1310 | PUTBACK; | |
1311 | } | |
1312 | return NORMAL; | |
1313 | } | |
4e35701f | 1314 |