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) | |
7d85a32c | 39 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
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) | |
7d85a32c | 106 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
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; | |
b7953727 | 144 | STRLEN len, rlen = 0; |
4757a243 LW |
145 | short *tbl; |
146 | I32 ch; | |
147 | ||
148 | tbl = (short*)cPVOP->op_pv; | |
149 | if (!tbl) | |
7d85a32c | 150 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
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 | } | |
76ef7183 | 187 | *d = '\0'; |
1aa99e6b | 188 | SvCUR_set(sv, d - dstart); |
4757a243 | 189 | } |
1aa99e6b | 190 | else { /* isutf8 */ |
9b877dbb IH |
191 | if (grows) |
192 | New(0, d, len*2+1, U8); | |
193 | else | |
194 | d = s; | |
1aa99e6b | 195 | dstart = d; |
8973db79 JH |
196 | if (complement && !del) |
197 | rlen = tbl[0x100]; | |
1aa99e6b | 198 | |
4aecb5b5 JH |
199 | #ifdef MACOS_TRADITIONAL |
200 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
201 | #endif | |
202 | ||
1aa99e6b | 203 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
204 | UV pch = 0xfeedface; |
205 | while (s < send) { | |
206 | STRLEN len; | |
9041c2e3 | 207 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b | 208 | |
8973db79 JH |
209 | if (comp > 0xff) { |
210 | if (!complement) { | |
211 | Copy(s, d, len, U8); | |
212 | d += len; | |
213 | } | |
214 | else { | |
215 | matches++; | |
216 | if (!del) { | |
05d340b8 JH |
217 | ch = (rlen == 0) ? comp : |
218 | (comp - 0x100 < rlen) ? | |
8973db79 JH |
219 | tbl[comp+1] : tbl[0x100+rlen]; |
220 | if (ch != pch) { | |
9041c2e3 | 221 | d = uvchr_to_utf8(d, ch); |
8973db79 JH |
222 | pch = ch; |
223 | } | |
224 | s += len; | |
225 | continue; | |
226 | } | |
227 | } | |
9b877dbb | 228 | } |
1aa99e6b IH |
229 | else if ((ch = tbl[comp]) >= 0) { |
230 | matches++; | |
231 | if (ch != pch) { | |
9041c2e3 | 232 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
233 | pch = ch; |
234 | } | |
235 | s += len; | |
236 | continue; | |
237 | } | |
9b877dbb IH |
238 | else if (ch == -1) { /* -1 is unmapped character */ |
239 | Copy(s, d, len, U8); | |
240 | d += len; | |
241 | } | |
1aa99e6b IH |
242 | else if (ch == -2) /* -2 is delete character */ |
243 | matches++; | |
244 | s += len; | |
245 | pch = 0xfeedface; | |
246 | } | |
5d06d08e | 247 | } |
1aa99e6b IH |
248 | else { |
249 | while (s < send) { | |
250 | STRLEN len; | |
9041c2e3 | 251 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
252 | if (comp > 0xff) { |
253 | if (!complement) { | |
254 | Copy(s, d, len, U8); | |
255 | d += len; | |
256 | } | |
257 | else { | |
258 | matches++; | |
259 | if (!del) { | |
9041c2e3 NIS |
260 | if (comp - 0x100 < rlen) |
261 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 262 | else |
9041c2e3 | 263 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
264 | } |
265 | } | |
9b877dbb | 266 | } |
1aa99e6b | 267 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 268 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
269 | matches++; |
270 | } | |
271 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
272 | Copy(s, d, len, U8); |
273 | d += len; | |
1aa99e6b IH |
274 | } |
275 | else if (ch == -2) /* -2 is delete character */ | |
276 | matches++; | |
277 | s += len; | |
278 | } | |
279 | } | |
9b877dbb IH |
280 | if (grows) { |
281 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
282 | Safefree(dstart); | |
283 | } | |
284 | else { | |
285 | *d = '\0'; | |
286 | SvCUR_set(sv, d - dstart); | |
287 | } | |
1aa99e6b | 288 | SvUTF8_on(sv); |
4757a243 | 289 | } |
5e44153e | 290 | SvSETMAGIC(sv); |
4757a243 LW |
291 | return matches; |
292 | } | |
293 | ||
942e002e | 294 | STATIC I32 |
036b4402 | 295 | S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 | 296 | { |
4757a243 LW |
297 | U8 *s; |
298 | U8 *send; | |
299 | U8 *d; | |
036b4402 | 300 | U8 *start; |
1aa99e6b | 301 | U8 *dstart, *dend; |
4757a243 | 302 | I32 matches = 0; |
9b877dbb | 303 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
304 | STRLEN len; |
305 | ||
306 | SV* rv = (SV*)cSVOP->op_sv; | |
307 | HV* hv = (HV*)SvRV(rv); | |
308 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
309 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
310 | UV extra = none + 1; | |
b7953727 | 311 | UV final = 0; |
4757a243 | 312 | UV uv; |
1aa99e6b IH |
313 | I32 isutf8; |
314 | U8 hibit = 0; | |
4757a243 LW |
315 | |
316 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
317 | isutf8 = SvUTF8(sv); |
318 | if (!isutf8) { | |
319 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
320 | while (t < e) { |
321 | U8 ch = *t++; | |
322 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 323 | break; |
c4d5f83a | 324 | } |
1aa99e6b IH |
325 | if (hibit) |
326 | s = bytes_to_utf8(s, &len); | |
327 | } | |
4757a243 | 328 | send = s + len; |
036b4402 | 329 | start = s; |
4757a243 LW |
330 | |
331 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
332 | if (svp) | |
333 | final = SvUV(*svp); | |
334 | ||
9b877dbb IH |
335 | if (grows) { |
336 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
337 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
338 | dend = d + len * 3; | |
339 | dstart = d; | |
340 | } | |
341 | else { | |
342 | dstart = d = s; | |
343 | dend = d + len; | |
344 | } | |
1aa99e6b | 345 | |
4757a243 | 346 | while (s < send) { |
3568d838 | 347 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 LW |
348 | s += UTF8SKIP(s); |
349 | matches++; | |
7d85a32c | 350 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
351 | } |
352 | else if (uv == none) { | |
1aa99e6b | 353 | int i = UTF8SKIP(s); |
9b877dbb IH |
354 | Copy(s, d, i, U8); |
355 | d += i; | |
356 | s += i; | |
4757a243 LW |
357 | } |
358 | else if (uv == extra) { | |
1aa99e6b | 359 | int i = UTF8SKIP(s); |
036b4402 | 360 | s += i; |
4757a243 | 361 | matches++; |
7d85a32c | 362 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
363 | } |
364 | else | |
365 | s += UTF8SKIP(s); | |
1aa99e6b | 366 | |
9b877dbb | 367 | if (d > dend) { |
1aa99e6b IH |
368 | STRLEN clen = d - dstart; |
369 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
9b877dbb | 370 | if (!grows) |
7d85a32c | 371 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b IH |
372 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
373 | d = dstart + clen; | |
374 | dend = dstart + nlen; | |
375 | } | |
4757a243 | 376 | } |
16ec844d | 377 | if (grows || hibit) { |
9b877dbb IH |
378 | sv_setpvn(sv, (char*)dstart, d - dstart); |
379 | Safefree(dstart); | |
16ec844d IH |
380 | if (grows && hibit) |
381 | Safefree(start); | |
9b877dbb IH |
382 | } |
383 | else { | |
384 | *d = '\0'; | |
385 | SvCUR_set(sv, d - dstart); | |
386 | } | |
4757a243 | 387 | SvSETMAGIC(sv); |
1aa99e6b | 388 | SvUTF8_on(sv); |
9041c2e3 | 389 | /* Downgrading just 'cos it will is suspect - NI-S */ |
1aa99e6b IH |
390 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
391 | sv_utf8_downgrade(sv, TRUE); | |
4757a243 LW |
392 | |
393 | return matches; | |
394 | } | |
395 | ||
942e002e | 396 | STATIC I32 |
036b4402 | 397 | S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 | 398 | { |
4757a243 | 399 | U8 *s; |
b7953727 | 400 | U8 *start = 0, *send; |
4757a243 LW |
401 | I32 matches = 0; |
402 | STRLEN len; | |
403 | ||
404 | SV* rv = (SV*)cSVOP->op_sv; | |
405 | HV* hv = (HV*)SvRV(rv); | |
406 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
407 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
45005bfb | 408 | UV extra = none + 1; |
4757a243 | 409 | UV uv; |
1aa99e6b | 410 | U8 hibit = 0; |
4757a243 LW |
411 | |
412 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
413 | if (!SvUTF8(sv)) { |
414 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
415 | while (t < e) { |
416 | U8 ch = *t++; | |
417 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 418 | break; |
c4d5f83a | 419 | } |
1aa99e6b IH |
420 | if (hibit) |
421 | start = s = bytes_to_utf8(s, &len); | |
422 | } | |
4757a243 LW |
423 | send = s + len; |
424 | ||
425 | while (s < send) { | |
3568d838 | 426 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 | 427 | matches++; |
834a4ddd | 428 | s += UTF8SKIP(s); |
4757a243 | 429 | } |
1aa99e6b IH |
430 | if (hibit) |
431 | Safefree(start); | |
4757a243 LW |
432 | |
433 | return matches; | |
434 | } | |
435 | ||
942e002e | 436 | STATIC I32 |
036b4402 | 437 | S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ |
4757a243 | 438 | { |
4757a243 | 439 | U8 *s; |
1aa99e6b | 440 | U8 *start, *send; |
4757a243 LW |
441 | U8 *d; |
442 | I32 matches = 0; | |
443 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 444 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 445 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
446 | SV* rv = (SV*)cSVOP->op_sv; |
447 | HV* hv = (HV*)SvRV(rv); | |
448 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
449 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
450 | UV extra = none + 1; | |
b7953727 | 451 | UV final = 0; |
45005bfb | 452 | bool havefinal = FALSE; |
4757a243 LW |
453 | UV uv; |
454 | STRLEN len; | |
1aa99e6b IH |
455 | U8 *dstart, *dend; |
456 | I32 isutf8; | |
457 | U8 hibit = 0; | |
4757a243 LW |
458 | |
459 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
460 | isutf8 = SvUTF8(sv); |
461 | if (!isutf8) { | |
462 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
463 | while (t < e) { |
464 | U8 ch = *t++; | |
465 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 466 | break; |
c4d5f83a | 467 | } |
1aa99e6b IH |
468 | if (hibit) |
469 | s = bytes_to_utf8(s, &len); | |
470 | } | |
4757a243 | 471 | send = s + len; |
1aa99e6b | 472 | start = s; |
4757a243 LW |
473 | |
474 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
45005bfb | 475 | if (svp) { |
4757a243 | 476 | final = SvUV(*svp); |
45005bfb JH |
477 | havefinal = TRUE; |
478 | } | |
4757a243 | 479 | |
9b877dbb IH |
480 | if (grows) { |
481 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
482 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
483 | dend = d + len * 3; | |
484 | dstart = d; | |
485 | } | |
486 | else { | |
487 | dstart = d = s; | |
488 | dend = d + len; | |
489 | } | |
4757a243 LW |
490 | |
491 | if (squash) { | |
492 | UV puv = 0xfeedface; | |
493 | while (s < send) { | |
3568d838 | 494 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 495 | |
9b877dbb IH |
496 | if (d > dend) { |
497 | STRLEN clen = d - dstart; | |
498 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
499 | if (!grows) | |
7d85a32c | 500 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
501 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
502 | d = dstart + clen; | |
503 | dend = dstart + nlen; | |
4757a243 LW |
504 | } |
505 | if (uv < none) { | |
506 | matches++; | |
94472101 | 507 | s += UTF8SKIP(s); |
4757a243 | 508 | if (uv != puv) { |
7d85a32c | 509 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
510 | puv = uv; |
511 | } | |
4757a243 LW |
512 | continue; |
513 | } | |
514 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 515 | int i = UTF8SKIP(s); |
9b877dbb IH |
516 | Copy(s, d, i, U8); |
517 | d += i; | |
518 | s += i; | |
4757a243 LW |
519 | puv = 0xfeedface; |
520 | continue; | |
a0ed51b3 | 521 | } |
4757a243 LW |
522 | else if (uv == extra && !del) { |
523 | matches++; | |
45005bfb JH |
524 | if (havefinal) { |
525 | s += UTF8SKIP(s); | |
526 | if (puv != final) { | |
7d85a32c | 527 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
528 | puv = final; |
529 | } | |
530 | } | |
531 | else { | |
532 | STRLEN len; | |
7d85a32c | 533 | uv = utf8_to_uvuni(s, &len); |
45005bfb JH |
534 | if (uv != puv) { |
535 | Copy(s, d, len, U8); | |
536 | d += len; | |
537 | puv = uv; | |
538 | } | |
539 | s += len; | |
4757a243 | 540 | } |
4757a243 LW |
541 | continue; |
542 | } | |
01ec43d0 GS |
543 | matches++; /* "none+1" is delete character */ |
544 | s += UTF8SKIP(s); | |
a0ed51b3 | 545 | } |
79072805 LW |
546 | } |
547 | else { | |
4757a243 | 548 | while (s < send) { |
3568d838 | 549 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb IH |
550 | if (d > dend) { |
551 | STRLEN clen = d - dstart; | |
552 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
553 | if (!grows) | |
7d85a32c | 554 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
555 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
556 | d = dstart + clen; | |
557 | dend = dstart + nlen; | |
a0ed51b3 | 558 | } |
4757a243 LW |
559 | if (uv < none) { |
560 | matches++; | |
01ec43d0 | 561 | s += UTF8SKIP(s); |
7d85a32c | 562 | d = uvuni_to_utf8(d, uv); |
4757a243 | 563 | continue; |
a0ed51b3 | 564 | } |
4757a243 | 565 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 566 | int i = UTF8SKIP(s); |
9b877dbb IH |
567 | Copy(s, d, i, U8); |
568 | d += i; | |
569 | s += i; | |
4757a243 | 570 | continue; |
79072805 | 571 | } |
4757a243 LW |
572 | else if (uv == extra && !del) { |
573 | matches++; | |
01ec43d0 | 574 | s += UTF8SKIP(s); |
7d85a32c | 575 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
576 | continue; |
577 | } | |
01ec43d0 GS |
578 | matches++; /* "none+1" is delete character */ |
579 | s += UTF8SKIP(s); | |
79072805 | 580 | } |
4757a243 | 581 | } |
16ec844d | 582 | if (grows || hibit) { |
9b877dbb IH |
583 | sv_setpvn(sv, (char*)dstart, d - dstart); |
584 | Safefree(dstart); | |
16ec844d IH |
585 | if (grows && hibit) |
586 | Safefree(start); | |
9b877dbb IH |
587 | } |
588 | else { | |
589 | *d = '\0'; | |
590 | SvCUR_set(sv, d - dstart); | |
591 | } | |
1aa99e6b | 592 | SvUTF8_on(sv); |
1aa99e6b IH |
593 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
594 | sv_utf8_downgrade(sv, TRUE); | |
4757a243 LW |
595 | SvSETMAGIC(sv); |
596 | ||
597 | return matches; | |
598 | } | |
599 | ||
600 | I32 | |
864dbfa3 | 601 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
602 | { |
603 | STRLEN len; | |
8e84507e | 604 | I32 hasutf = (PL_op->op_private & |
036b4402 | 605 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 LW |
606 | |
607 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) | |
cea2e8a9 | 608 | Perl_croak(aTHX_ PL_no_modify); |
4757a243 LW |
609 | |
610 | (void)SvPV(sv, len); | |
611 | if (!len) | |
612 | return 0; | |
613 | if (!SvPOKp(sv)) | |
614 | (void)SvPV_force(sv, len); | |
2de7b02f GS |
615 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) |
616 | (void)SvPOK_only_UTF8(sv); | |
4757a243 | 617 | |
cea2e8a9 | 618 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 619 | |
036b4402 | 620 | switch (PL_op->op_private & ~hasutf & 63) { |
4757a243 | 621 | case 0: |
01ec43d0 GS |
622 | if (hasutf) |
623 | return do_trans_simple_utf8(sv); | |
624 | else | |
625 | return do_trans_simple(sv); | |
4757a243 LW |
626 | |
627 | case OPpTRANS_IDENTICAL: | |
05d340b8 | 628 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 GS |
629 | if (hasutf) |
630 | return do_trans_count_utf8(sv); | |
631 | else | |
632 | return do_trans_count(sv); | |
4757a243 LW |
633 | |
634 | default: | |
01ec43d0 | 635 | if (hasutf) |
036b4402 | 636 | return do_trans_complex_utf8(sv); |
4757a243 | 637 | else |
036b4402 | 638 | return do_trans_complex(sv); |
79072805 | 639 | } |
79072805 LW |
640 | } |
641 | ||
642 | void | |
864dbfa3 | 643 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
644 | { |
645 | SV **oldmark = mark; | |
646 | register I32 items = sp - mark; | |
79072805 | 647 | register STRLEN len; |
463ee0b2 | 648 | STRLEN delimlen; |
463ee0b2 | 649 | STRLEN tmplen; |
79072805 | 650 | |
5ba99574 NC |
651 | (void) SvPV(del, delimlen); /* stringify and get the delimlen */ |
652 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ | |
516a5887 | 653 | |
79072805 LW |
654 | mark++; |
655 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 656 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
657 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
658 | while (items-- > 0) { | |
1426bbf4 | 659 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 LW |
660 | SvPV(*mark, tmplen); |
661 | len += tmplen; | |
79072805 LW |
662 | } |
663 | mark++; | |
664 | } | |
665 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
666 | ||
667 | mark = oldmark; | |
db7c17d7 | 668 | items = sp - mark; |
79072805 LW |
669 | ++mark; |
670 | } | |
671 | ||
8d6d96c1 HS |
672 | sv_setpv(sv, ""); |
673 | if (PL_tainting && SvMAGICAL(sv)) | |
674 | SvTAINTED_off(sv); | |
675 | ||
463ee0b2 | 676 | if (items-- > 0) { |
92d29cee JH |
677 | if (*mark) |
678 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
679 | mark++; |
680 | } | |
8d6d96c1 | 681 | |
c512ce4f | 682 | if (delimlen) { |
79072805 | 683 | for (; items > 0; items--,mark++) { |
c512ce4f | 684 | sv_catsv(sv,del); |
79072805 LW |
685 | sv_catsv(sv,*mark); |
686 | } | |
687 | } | |
688 | else { | |
689 | for (; items > 0; items--,mark++) | |
690 | sv_catsv(sv,*mark); | |
691 | } | |
692 | SvSETMAGIC(sv); | |
693 | } | |
694 | ||
695 | void | |
864dbfa3 | 696 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 697 | { |
46fc3d4c | 698 | STRLEN patlen; |
699 | char *pat = SvPV(*sarg, patlen); | |
700 | bool do_taint = FALSE; | |
701 | ||
702 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); | |
79072805 | 703 | SvSETMAGIC(sv); |
46fc3d4c | 704 | if (do_taint) |
705 | SvTAINTED_on(sv); | |
79072805 LW |
706 | } |
707 | ||
33b45480 | 708 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
709 | UV |
710 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
711 | { | |
712 | STRLEN srclen, len; | |
713 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
714 | UV retnum = 0; | |
715 | ||
a50d7633 | 716 | if (offset < 0) |
81e118e0 | 717 | return retnum; |
8e84507e | 718 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 719 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 720 | |
dcad2880 | 721 | if (SvUTF8(sv)) |
33b45480 | 722 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 723 | |
81e118e0 JH |
724 | offset *= size; /* turn into bit offset */ |
725 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
726 | if (len > srclen) { | |
727 | if (size <= 8) | |
728 | retnum = 0; | |
729 | else { | |
730 | offset >>= 3; /* turn into byte offset */ | |
731 | if (size == 16) { | |
732 | if (offset >= srclen) | |
733 | retnum = 0; | |
734 | else | |
628e1a40 | 735 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
736 | } |
737 | else if (size == 32) { | |
738 | if (offset >= srclen) | |
739 | retnum = 0; | |
740 | else if (offset + 1 >= srclen) | |
741 | retnum = | |
742 | ((UV) s[offset ] << 24); | |
743 | else if (offset + 2 >= srclen) | |
744 | retnum = | |
745 | ((UV) s[offset ] << 24) + | |
746 | ((UV) s[offset + 1] << 16); | |
747 | else | |
748 | retnum = | |
749 | ((UV) s[offset ] << 24) + | |
750 | ((UV) s[offset + 1] << 16) + | |
751 | ( s[offset + 2] << 8); | |
752 | } | |
d7d93a81 | 753 | #ifdef UV_IS_QUAD |
c5a0f51a | 754 | else if (size == 64) { |
c5a0f51a JH |
755 | if (ckWARN(WARN_PORTABLE)) |
756 | Perl_warner(aTHX_ WARN_PORTABLE, | |
757 | "Bit vector size > 32 non-portable"); | |
758 | if (offset >= srclen) | |
759 | retnum = 0; | |
760 | else if (offset + 1 >= srclen) | |
761 | retnum = | |
762 | (UV) s[offset ] << 56; | |
763 | else if (offset + 2 >= srclen) | |
764 | retnum = | |
765 | ((UV) s[offset ] << 56) + | |
766 | ((UV) s[offset + 1] << 48); | |
767 | else if (offset + 3 >= srclen) | |
768 | retnum = | |
769 | ((UV) s[offset ] << 56) + | |
770 | ((UV) s[offset + 1] << 48) + | |
771 | ((UV) s[offset + 2] << 40); | |
772 | else if (offset + 4 >= srclen) | |
773 | retnum = | |
774 | ((UV) s[offset ] << 56) + | |
775 | ((UV) s[offset + 1] << 48) + | |
776 | ((UV) s[offset + 2] << 40) + | |
777 | ((UV) s[offset + 3] << 32); | |
778 | else if (offset + 5 >= srclen) | |
779 | retnum = | |
780 | ((UV) s[offset ] << 56) + | |
781 | ((UV) s[offset + 1] << 48) + | |
782 | ((UV) s[offset + 2] << 40) + | |
783 | ((UV) s[offset + 3] << 32) + | |
784 | ( s[offset + 4] << 24); | |
785 | else if (offset + 6 >= srclen) | |
786 | retnum = | |
787 | ((UV) s[offset ] << 56) + | |
788 | ((UV) s[offset + 1] << 48) + | |
789 | ((UV) s[offset + 2] << 40) + | |
790 | ((UV) s[offset + 3] << 32) + | |
791 | ((UV) s[offset + 4] << 24) + | |
792 | ((UV) s[offset + 5] << 16); | |
793 | else | |
8e84507e | 794 | retnum = |
c5a0f51a JH |
795 | ((UV) s[offset ] << 56) + |
796 | ((UV) s[offset + 1] << 48) + | |
797 | ((UV) s[offset + 2] << 40) + | |
798 | ((UV) s[offset + 3] << 32) + | |
799 | ((UV) s[offset + 4] << 24) + | |
800 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 801 | ( s[offset + 6] << 8); |
c5a0f51a JH |
802 | } |
803 | #endif | |
81e118e0 JH |
804 | } |
805 | } | |
806 | else if (size < 8) | |
807 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
808 | else { | |
809 | offset >>= 3; /* turn into byte offset */ | |
810 | if (size == 8) | |
811 | retnum = s[offset]; | |
812 | else if (size == 16) | |
813 | retnum = | |
628e1a40 | 814 | ((UV) s[offset] << 8) + |
81e118e0 JH |
815 | s[offset + 1]; |
816 | else if (size == 32) | |
817 | retnum = | |
818 | ((UV) s[offset ] << 24) + | |
819 | ((UV) s[offset + 1] << 16) + | |
820 | ( s[offset + 2] << 8) + | |
821 | s[offset + 3]; | |
d7d93a81 | 822 | #ifdef UV_IS_QUAD |
c5a0f51a | 823 | else if (size == 64) { |
c5a0f51a JH |
824 | if (ckWARN(WARN_PORTABLE)) |
825 | Perl_warner(aTHX_ WARN_PORTABLE, | |
826 | "Bit vector size > 32 non-portable"); | |
827 | retnum = | |
828 | ((UV) s[offset ] << 56) + | |
829 | ((UV) s[offset + 1] << 48) + | |
830 | ((UV) s[offset + 2] << 40) + | |
831 | ((UV) s[offset + 3] << 32) + | |
832 | ((UV) s[offset + 4] << 24) + | |
833 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 834 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
835 | s[offset + 7]; |
836 | } | |
837 | #endif | |
81e118e0 JH |
838 | } |
839 | ||
840 | return retnum; | |
841 | } | |
842 | ||
33b45480 SB |
843 | /* currently converts input to bytes if possible but doesn't sweat failures, |
844 | * although it does ensure that the string it clobbers is not marked as | |
845 | * utf8-valid any more | |
846 | */ | |
79072805 | 847 | void |
864dbfa3 | 848 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
849 | { |
850 | SV *targ = LvTARG(sv); | |
851 | register I32 offset; | |
852 | register I32 size; | |
8990e307 | 853 | register unsigned char *s; |
81e118e0 | 854 | register UV lval; |
79072805 | 855 | I32 mask; |
a0d0e21e LW |
856 | STRLEN targlen; |
857 | STRLEN len; | |
79072805 | 858 | |
8990e307 LW |
859 | if (!targ) |
860 | return; | |
a0d0e21e | 861 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 862 | if (SvUTF8(targ)) { |
33b45480 SB |
863 | /* This is handled by the SvPOK_only below... |
864 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
865 | SvUTF8_off(targ); | |
866 | */ | |
867 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
868 | } |
869 | ||
4ebbc975 | 870 | (void)SvPOK_only(targ); |
81e118e0 | 871 | lval = SvUV(sv); |
79072805 | 872 | offset = LvTARGOFF(sv); |
fe58ced6 MG |
873 | if (offset < 0) |
874 | Perl_croak(aTHX_ "Assigning to negative offset in vec"); | |
79072805 | 875 | size = LvTARGLEN(sv); |
8e84507e | 876 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 877 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 878 | |
81e118e0 JH |
879 | offset *= size; /* turn into bit offset */ |
880 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
881 | if (len > targlen) { |
882 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 883 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
884 | SvCUR_set(targ, len); |
885 | } | |
8e84507e | 886 | |
79072805 LW |
887 | if (size < 8) { |
888 | mask = (1 << size) - 1; | |
889 | size = offset & 7; | |
890 | lval &= mask; | |
81e118e0 | 891 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
892 | s[offset] &= ~(mask << size); |
893 | s[offset] |= lval << size; | |
894 | } | |
895 | else { | |
81e118e0 | 896 | offset >>= 3; /* turn into byte offset */ |
79072805 | 897 | if (size == 8) |
c5a0f51a | 898 | s[offset ] = lval & 0xff; |
79072805 | 899 | else if (size == 16) { |
c5a0f51a JH |
900 | s[offset ] = (lval >> 8) & 0xff; |
901 | s[offset+1] = lval & 0xff; | |
79072805 LW |
902 | } |
903 | else if (size == 32) { | |
c5a0f51a JH |
904 | s[offset ] = (lval >> 24) & 0xff; |
905 | s[offset+1] = (lval >> 16) & 0xff; | |
906 | s[offset+2] = (lval >> 8) & 0xff; | |
907 | s[offset+3] = lval & 0xff; | |
908 | } | |
d7d93a81 | 909 | #ifdef UV_IS_QUAD |
c5a0f51a | 910 | else if (size == 64) { |
c5a0f51a JH |
911 | if (ckWARN(WARN_PORTABLE)) |
912 | Perl_warner(aTHX_ WARN_PORTABLE, | |
913 | "Bit vector size > 32 non-portable"); | |
914 | s[offset ] = (lval >> 56) & 0xff; | |
915 | s[offset+1] = (lval >> 48) & 0xff; | |
916 | s[offset+2] = (lval >> 40) & 0xff; | |
917 | s[offset+3] = (lval >> 32) & 0xff; | |
918 | s[offset+4] = (lval >> 24) & 0xff; | |
919 | s[offset+5] = (lval >> 16) & 0xff; | |
920 | s[offset+6] = (lval >> 8) & 0xff; | |
921 | s[offset+7] = lval & 0xff; | |
79072805 | 922 | } |
dc1e3f56 | 923 | #endif |
79072805 | 924 | } |
7bb043c3 | 925 | SvSETMAGIC(targ); |
79072805 LW |
926 | } |
927 | ||
928 | void | |
864dbfa3 | 929 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 930 | { |
463ee0b2 | 931 | STRLEN len; |
a0d0e21e | 932 | char *s; |
8e84507e | 933 | |
79072805 | 934 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
935 | register I32 i; |
936 | I32 max; | |
937 | AV* av = (AV*)sv; | |
938 | max = AvFILL(av); | |
939 | for (i = 0; i <= max; i++) { | |
940 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 941 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
942 | do_chop(astr, sv); |
943 | } | |
944 | return; | |
79072805 | 945 | } |
aa854799 | 946 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
947 | HV* hv = (HV*)sv; |
948 | HE* entry; | |
949 | (void)hv_iterinit(hv); | |
950 | /*SUPPRESS 560*/ | |
155aba94 | 951 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
952 | do_chop(astr,hv_iterval(hv,entry)); |
953 | return; | |
79072805 | 954 | } |
aa854799 | 955 | else if (SvREADONLY(sv)) |
cea2e8a9 | 956 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e | 957 | s = SvPV(sv, len); |
748a9306 | 958 | if (len && !SvPOK(sv)) |
a0d0e21e | 959 | s = SvPV_force(sv, len); |
7e2040f0 | 960 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
961 | if (s && len) { |
962 | char *send = s + len; | |
963 | char *start = s; | |
964 | s = send - 1; | |
a0dbb045 JH |
965 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
966 | s--; | |
9041c2e3 | 967 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
968 | sv_setpvn(astr, s, send - s); |
969 | *s = '\0'; | |
970 | SvCUR_set(sv, s - start); | |
971 | SvNIOK_off(sv); | |
972 | SvUTF8_on(astr); | |
973 | } | |
a0ed51b3 LW |
974 | } |
975 | else | |
976 | sv_setpvn(astr, "", 0); | |
977 | } | |
7e2040f0 | 978 | else if (s && len) { |
a0d0e21e LW |
979 | s += --len; |
980 | sv_setpvn(astr, s, 1); | |
981 | *s = '\0'; | |
982 | SvCUR_set(sv, len); | |
2c19a612 | 983 | SvUTF8_off(sv); |
a0d0e21e | 984 | SvNIOK_off(sv); |
79072805 LW |
985 | } |
986 | else | |
a0d0e21e LW |
987 | sv_setpvn(astr, "", 0); |
988 | SvSETMAGIC(sv); | |
7e2040f0 | 989 | } |
a0d0e21e LW |
990 | |
991 | I32 | |
864dbfa3 | 992 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 993 | { |
c07a80fd | 994 | register I32 count; |
a0d0e21e | 995 | STRLEN len; |
9b33ce3b | 996 | STRLEN n_a; |
a0d0e21e | 997 | char *s; |
c07a80fd | 998 | |
3280af22 | 999 | if (RsSNARF(PL_rs)) |
c07a80fd | 1000 | return 0; |
4c5a6083 GS |
1001 | if (RsRECORD(PL_rs)) |
1002 | return 0; | |
c07a80fd | 1003 | count = 0; |
a0d0e21e LW |
1004 | if (SvTYPE(sv) == SVt_PVAV) { |
1005 | register I32 i; | |
1006 | I32 max; | |
1007 | AV* av = (AV*)sv; | |
1008 | max = AvFILL(av); | |
1009 | for (i = 0; i <= max; i++) { | |
1010 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 1011 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1012 | count += do_chomp(sv); |
1013 | } | |
1014 | return count; | |
1015 | } | |
aa854799 | 1016 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
1017 | HV* hv = (HV*)sv; |
1018 | HE* entry; | |
1019 | (void)hv_iterinit(hv); | |
1020 | /*SUPPRESS 560*/ | |
155aba94 | 1021 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1022 | count += do_chomp(hv_iterval(hv,entry)); |
1023 | return count; | |
1024 | } | |
aa854799 | 1025 | else if (SvREADONLY(sv)) |
cea2e8a9 | 1026 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e | 1027 | s = SvPV(sv, len); |
a0d0e21e LW |
1028 | if (s && len) { |
1029 | s += --len; | |
3280af22 | 1030 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1031 | if (*s != '\n') |
1032 | goto nope; | |
1033 | ++count; | |
1034 | while (len && s[-1] == '\n') { | |
1035 | --len; | |
1036 | --s; | |
1037 | ++count; | |
1038 | } | |
1039 | } | |
a0d0e21e | 1040 | else { |
c07a80fd | 1041 | STRLEN rslen; |
3280af22 | 1042 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 1043 | if (rslen == 1) { |
1044 | if (*s != *rsptr) | |
1045 | goto nope; | |
1046 | ++count; | |
1047 | } | |
1048 | else { | |
8c2cee6f | 1049 | if (len < rslen - 1) |
c07a80fd | 1050 | goto nope; |
1051 | len -= rslen - 1; | |
1052 | s -= rslen - 1; | |
36477c24 | 1053 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1054 | goto nope; |
1055 | count += rslen; | |
1056 | } | |
a0d0e21e | 1057 | } |
9b33ce3b | 1058 | s = SvPV_force(sv, n_a); |
a0d0e21e | 1059 | SvCUR_set(sv, len); |
9b33ce3b | 1060 | *SvEND(sv) = '\0'; |
a0d0e21e | 1061 | SvNIOK_off(sv); |
9b33ce3b | 1062 | SvSETMAGIC(sv); |
a0d0e21e LW |
1063 | } |
1064 | nope: | |
a0d0e21e | 1065 | return count; |
8e84507e | 1066 | } |
79072805 LW |
1067 | |
1068 | void | |
864dbfa3 | 1069 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1070 | { |
1071 | #ifdef LIBERAL | |
1072 | register long *dl; | |
1073 | register long *ll; | |
1074 | register long *rl; | |
1075 | #endif | |
1076 | register char *dc; | |
463ee0b2 LW |
1077 | STRLEN leftlen; |
1078 | STRLEN rightlen; | |
7a4c00b4 | 1079 | register char *lc; |
1080 | register char *rc; | |
79072805 | 1081 | register I32 len; |
a0d0e21e | 1082 | I32 lensave; |
7a4c00b4 | 1083 | char *lsave; |
1084 | char *rsave; | |
0c57e439 GS |
1085 | bool left_utf = DO_UTF8(left); |
1086 | bool right_utf = DO_UTF8(right); | |
b7953727 | 1087 | I32 needlen = 0; |
0c57e439 GS |
1088 | |
1089 | if (left_utf && !right_utf) | |
1090 | sv_utf8_upgrade(right); | |
a1ca4561 | 1091 | else if (!left_utf && right_utf) |
0c57e439 | 1092 | sv_utf8_upgrade(left); |
79072805 | 1093 | |
1fbd88dc CS |
1094 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1095 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 1096 | lsave = lc = SvPV(left, leftlen); |
1097 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 1098 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1099 | lensave = len; |
c9b3c8d0 JH |
1100 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1101 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1102 | Newz(801, dc, needlen + 1, char); | |
1103 | } | |
1104 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d GS |
1105 | STRLEN n_a; |
1106 | dc = SvPV_force(sv, n_a); | |
ff68c719 | 1107 | if (SvCUR(sv) < len) { |
1108 | dc = SvGROW(sv, len + 1); | |
1109 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); | |
1110 | } | |
c9b3c8d0 JH |
1111 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1112 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1113 | } |
1114 | else { | |
c9b3c8d0 JH |
1115 | needlen = ((optype == OP_BIT_AND) |
1116 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1117 | Newz(801, dc, needlen + 1, char); |
1118 | (void)sv_usepvn(sv, dc, needlen); | |
1119 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1120 | } |
a0d0e21e LW |
1121 | SvCUR_set(sv, len); |
1122 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1123 | if (left_utf || right_utf) { |
1124 | UV duc, luc, ruc; | |
c9b3c8d0 | 1125 | char *dcsave = dc; |
0c57e439 GS |
1126 | STRLEN lulen = leftlen; |
1127 | STRLEN rulen = rightlen; | |
ba210ebe | 1128 | STRLEN ulen; |
0c57e439 GS |
1129 | |
1130 | switch (optype) { | |
1131 | case OP_BIT_AND: | |
1132 | while (lulen && rulen) { | |
9041c2e3 | 1133 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1134 | lc += ulen; |
1135 | lulen -= ulen; | |
9041c2e3 | 1136 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1137 | rc += ulen; |
1138 | rulen -= ulen; | |
1139 | duc = luc & ruc; | |
9041c2e3 | 1140 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1141 | } |
c9b3c8d0 JH |
1142 | if (sv == left || sv == right) |
1143 | (void)sv_usepvn(sv, dcsave, needlen); | |
1144 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1145 | break; |
1146 | case OP_BIT_XOR: | |
1147 | while (lulen && rulen) { | |
9041c2e3 | 1148 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1149 | lc += ulen; |
1150 | lulen -= ulen; | |
9041c2e3 | 1151 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1152 | rc += ulen; |
1153 | rulen -= ulen; | |
1154 | duc = luc ^ ruc; | |
9041c2e3 | 1155 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1156 | } |
1157 | goto mop_up_utf; | |
1158 | case OP_BIT_OR: | |
1159 | while (lulen && rulen) { | |
9041c2e3 | 1160 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1161 | lc += ulen; |
1162 | lulen -= ulen; | |
9041c2e3 | 1163 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1164 | rc += ulen; |
1165 | rulen -= ulen; | |
1166 | duc = luc | ruc; | |
9041c2e3 | 1167 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1168 | } |
1169 | mop_up_utf: | |
c9b3c8d0 JH |
1170 | if (sv == left || sv == right) |
1171 | (void)sv_usepvn(sv, dcsave, needlen); | |
1172 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1173 | if (rulen) |
1174 | sv_catpvn(sv, rc, rulen); | |
1175 | else if (lulen) | |
1176 | sv_catpvn(sv, lc, lulen); | |
1177 | else | |
1178 | *SvEND(sv) = '\0'; | |
1179 | break; | |
1180 | } | |
1181 | SvUTF8_on(sv); | |
1182 | goto finish; | |
1183 | } | |
1184 | else | |
79072805 LW |
1185 | #ifdef LIBERAL |
1186 | if (len >= sizeof(long)*4 && | |
1187 | !((long)dc % sizeof(long)) && | |
1188 | !((long)lc % sizeof(long)) && | |
1189 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1190 | { | |
1191 | I32 remainder = len % (sizeof(long)*4); | |
1192 | len /= (sizeof(long)*4); | |
1193 | ||
1194 | dl = (long*)dc; | |
1195 | ll = (long*)lc; | |
1196 | rl = (long*)rc; | |
1197 | ||
1198 | switch (optype) { | |
1199 | case OP_BIT_AND: | |
1200 | while (len--) { | |
1201 | *dl++ = *ll++ & *rl++; | |
1202 | *dl++ = *ll++ & *rl++; | |
1203 | *dl++ = *ll++ & *rl++; | |
1204 | *dl++ = *ll++ & *rl++; | |
1205 | } | |
1206 | break; | |
a0d0e21e | 1207 | case OP_BIT_XOR: |
79072805 LW |
1208 | while (len--) { |
1209 | *dl++ = *ll++ ^ *rl++; | |
1210 | *dl++ = *ll++ ^ *rl++; | |
1211 | *dl++ = *ll++ ^ *rl++; | |
1212 | *dl++ = *ll++ ^ *rl++; | |
1213 | } | |
1214 | break; | |
1215 | case OP_BIT_OR: | |
1216 | while (len--) { | |
1217 | *dl++ = *ll++ | *rl++; | |
1218 | *dl++ = *ll++ | *rl++; | |
1219 | *dl++ = *ll++ | *rl++; | |
1220 | *dl++ = *ll++ | *rl++; | |
1221 | } | |
1222 | } | |
1223 | ||
1224 | dc = (char*)dl; | |
1225 | lc = (char*)ll; | |
1226 | rc = (char*)rl; | |
1227 | ||
1228 | len = remainder; | |
1229 | } | |
1230 | #endif | |
a0d0e21e | 1231 | { |
a0d0e21e LW |
1232 | switch (optype) { |
1233 | case OP_BIT_AND: | |
1234 | while (len--) | |
1235 | *dc++ = *lc++ & *rc++; | |
1236 | break; | |
1237 | case OP_BIT_XOR: | |
1238 | while (len--) | |
1239 | *dc++ = *lc++ ^ *rc++; | |
1240 | goto mop_up; | |
1241 | case OP_BIT_OR: | |
1242 | while (len--) | |
1243 | *dc++ = *lc++ | *rc++; | |
1244 | mop_up: | |
1245 | len = lensave; | |
1246 | if (rightlen > len) | |
1247 | sv_catpvn(sv, rsave + len, rightlen - len); | |
1248 | else if (leftlen > len) | |
1249 | sv_catpvn(sv, lsave + len, leftlen - len); | |
4633a7c4 LW |
1250 | else |
1251 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1252 | break; |
1253 | } | |
79072805 | 1254 | } |
0c57e439 | 1255 | finish: |
fb73857a | 1256 | SvTAINT(sv); |
79072805 | 1257 | } |
463ee0b2 LW |
1258 | |
1259 | OP * | |
cea2e8a9 | 1260 | Perl_do_kv(pTHX) |
463ee0b2 | 1261 | { |
39644a26 | 1262 | dSP; |
463ee0b2 | 1263 | HV *hv = (HV*)POPs; |
800e9ae0 | 1264 | HV *keys; |
463ee0b2 | 1265 | register HE *entry; |
463ee0b2 | 1266 | SV *tmpstr; |
54310121 | 1267 | I32 gimme = GIMME_V; |
533c011a NIS |
1268 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1269 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
c750a3ec | 1270 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
8e84507e NIS |
1271 | |
1272 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1273 | dokeys = dovalues = TRUE; |
463ee0b2 | 1274 | |
85581909 | 1275 | if (!hv) { |
78f9721b | 1276 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1277 | dTARGET; /* make sure to clear its target here */ |
1278 | if (SvTYPE(TARG) == SVt_PVLV) | |
1279 | LvTARG(TARG) = Nullsv; | |
1280 | PUSHs(TARG); | |
1281 | } | |
463ee0b2 | 1282 | RETURN; |
85581909 | 1283 | } |
748a9306 | 1284 | |
800e9ae0 JP |
1285 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1286 | (void)hv_iterinit(keys); /* always reset iterator regardless */ | |
748a9306 | 1287 | |
54310121 | 1288 | if (gimme == G_VOID) |
aa689395 | 1289 | RETURN; |
1290 | ||
54310121 | 1291 | if (gimme == G_SCALAR) { |
6ee623d5 | 1292 | IV i; |
463ee0b2 LW |
1293 | dTARGET; |
1294 | ||
78f9721b | 1295 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1296 | if (SvTYPE(TARG) < SVt_PVLV) { |
1297 | sv_upgrade(TARG, SVt_PVLV); | |
14befaf4 | 1298 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 SB |
1299 | } |
1300 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1301 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1302 | if (LvTARG(TARG)) |
1303 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1304 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1305 | } |
85581909 SB |
1306 | PUSHs(TARG); |
1307 | RETURN; | |
1308 | } | |
1309 | ||
14befaf4 | 1310 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
800e9ae0 | 1311 | i = HvKEYS(keys); |
463ee0b2 LW |
1312 | else { |
1313 | i = 0; | |
463ee0b2 | 1314 | /*SUPPRESS 560*/ |
800e9ae0 | 1315 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1316 | } |
1317 | PUSHi( i ); | |
1318 | RETURN; | |
1319 | } | |
1320 | ||
8ed4b672 | 1321 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1322 | |
463ee0b2 | 1323 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1324 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1325 | SPAGAIN; |
6bfc225d | 1326 | if (dokeys) |
8c2cee6f | 1327 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ |
463ee0b2 | 1328 | if (dovalues) { |
463ee0b2 | 1329 | PUTBACK; |
b6429b1b GS |
1330 | tmpstr = realhv ? |
1331 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); | |
cea2e8a9 | 1332 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1333 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1334 | HvMAX(keys)+1, |
1335 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1336 | SPAGAIN; |
46fc3d4c | 1337 | XPUSHs(tmpstr); |
463ee0b2 LW |
1338 | } |
1339 | PUTBACK; | |
1340 | } | |
1341 | return NORMAL; | |
1342 | } | |
4e35701f | 1343 |