Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
4bb101f2 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
b5f8cc5c | 4 | * 2000, 2001, 2002, 2004, 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 | /* | |
12 | * "'So that was the job I felt I had to do when I started,' thought Sam." | |
79072805 LW |
13 | */ |
14 | ||
15 | #include "EXTERN.h" | |
864dbfa3 | 16 | #define PERL_IN_DOOP_C |
79072805 LW |
17 | #include "perl.h" |
18 | ||
64ca3a65 | 19 | #ifndef PERL_MICRO |
79072805 LW |
20 | #include <signal.h> |
21 | #endif | |
22 | ||
942e002e | 23 | STATIC I32 |
b250498f | 24 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 | 25 | { |
4757a243 | 26 | U8 *s; |
b250498f | 27 | U8 *d; |
4757a243 | 28 | U8 *send; |
b250498f | 29 | U8 *dstart; |
4757a243 | 30 | I32 matches = 0; |
9b877dbb | 31 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 | 32 | STRLEN len; |
4757a243 LW |
33 | short *tbl; |
34 | I32 ch; | |
79072805 | 35 | |
4757a243 LW |
36 | tbl = (short*)cPVOP->op_pv; |
37 | if (!tbl) | |
7d85a32c | 38 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 39 | |
4757a243 LW |
40 | s = (U8*)SvPV(sv, len); |
41 | send = s + len; | |
42 | ||
1e54db1a | 43 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 44 | if (!SvUTF8(sv)) { |
01ec43d0 | 45 | while (s < send) { |
036b4402 | 46 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 | 47 | matches++; |
eb160463 | 48 | *s++ = (U8)ch; |
01ec43d0 GS |
49 | } |
50 | else | |
51 | s++; | |
52 | } | |
53 | SvSETMAGIC(sv); | |
b250498f GS |
54 | return matches; |
55 | } | |
4757a243 | 56 | |
b250498f | 57 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb IH |
58 | if (grows) |
59 | New(0, d, len*2+1, U8); | |
60 | else | |
61 | d = s; | |
b250498f GS |
62 | dstart = d; |
63 | while (s < send) { | |
ba210ebe | 64 | STRLEN ulen; |
9b877dbb | 65 | UV c; |
b250498f | 66 | |
b250498f | 67 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 | 68 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb | 69 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f | 70 | matches++; |
9041c2e3 | 71 | d = uvchr_to_utf8(d, ch); |
b250498f | 72 | s += ulen; |
01ec43d0 GS |
73 | } |
74 | else { /* No match -> copy */ | |
0376ff32 | 75 | Move(s, d, ulen, U8); |
9b877dbb IH |
76 | d += ulen; |
77 | s += ulen; | |
b250498f GS |
78 | } |
79 | } | |
9b877dbb IH |
80 | if (grows) { |
81 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
82 | Safefree(dstart); | |
83 | } | |
84 | else { | |
85 | *d = '\0'; | |
86 | SvCUR_set(sv, d - dstart); | |
87 | } | |
b250498f | 88 | SvUTF8_on(sv); |
b250498f | 89 | SvSETMAGIC(sv); |
4757a243 LW |
90 | return matches; |
91 | } | |
92 | ||
942e002e | 93 | STATIC I32 |
21304a3b | 94 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 | 95 | { |
4757a243 LW |
96 | U8 *s; |
97 | U8 *send; | |
98 | I32 matches = 0; | |
99 | STRLEN len; | |
100 | short *tbl; | |
8973db79 | 101 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 LW |
102 | |
103 | tbl = (short*)cPVOP->op_pv; | |
104 | if (!tbl) | |
7d85a32c | 105 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 LW |
106 | |
107 | s = (U8*)SvPV(sv, len); | |
108 | send = s + len; | |
109 | ||
1aa99e6b IH |
110 | if (!SvUTF8(sv)) |
111 | while (s < send) { | |
112 | if (tbl[*s++] >= 0) | |
036b4402 | 113 | matches++; |
1aa99e6b IH |
114 | } |
115 | else | |
116 | while (s < send) { | |
117 | UV c; | |
118 | STRLEN ulen; | |
9041c2e3 | 119 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 JH |
120 | if (c < 0x100) { |
121 | if (tbl[c] >= 0) | |
122 | matches++; | |
123 | } else if (complement) | |
1aa99e6b IH |
124 | matches++; |
125 | s += ulen; | |
126 | } | |
4757a243 LW |
127 | |
128 | return matches; | |
129 | } | |
130 | ||
942e002e | 131 | STATIC I32 |
21304a3b | 132 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 | 133 | { |
4757a243 LW |
134 | U8 *s; |
135 | U8 *send; | |
136 | U8 *d; | |
5e44153e | 137 | U8 *dstart; |
1aa99e6b | 138 | I32 isutf8; |
4757a243 | 139 | I32 matches = 0; |
9b877dbb | 140 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 JH |
141 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
142 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
b7953727 | 143 | STRLEN len, rlen = 0; |
4757a243 LW |
144 | short *tbl; |
145 | I32 ch; | |
146 | ||
147 | tbl = (short*)cPVOP->op_pv; | |
148 | if (!tbl) | |
7d85a32c | 149 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 LW |
150 | |
151 | s = (U8*)SvPV(sv, len); | |
1aa99e6b | 152 | isutf8 = SvUTF8(sv); |
4757a243 LW |
153 | send = s + len; |
154 | ||
1aa99e6b IH |
155 | if (!isutf8) { |
156 | dstart = d = s; | |
157 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
158 | U8* p = send; | |
159 | while (s < send) { | |
160 | if ((ch = tbl[*s]) >= 0) { | |
eb160463 | 161 | *d = (U8)ch; |
036b4402 | 162 | matches++; |
1aa99e6b IH |
163 | if (p != d - 1 || *p != *d) |
164 | p = d++; | |
165 | } | |
166 | else if (ch == -1) /* -1 is unmapped character */ | |
167 | *d++ = *s; | |
168 | else if (ch == -2) /* -2 is delete character */ | |
169 | matches++; | |
170 | s++; | |
171 | } | |
a0ed51b3 | 172 | } |
1aa99e6b IH |
173 | else { |
174 | while (s < send) { | |
175 | if ((ch = tbl[*s]) >= 0) { | |
176 | matches++; | |
eb160463 | 177 | *d++ = (U8)ch; |
1aa99e6b IH |
178 | } |
179 | else if (ch == -1) /* -1 is unmapped character */ | |
180 | *d++ = *s; | |
181 | else if (ch == -2) /* -2 is delete character */ | |
182 | matches++; | |
183 | s++; | |
184 | } | |
185 | } | |
76ef7183 | 186 | *d = '\0'; |
1aa99e6b | 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) { | |
05d340b8 JH |
216 | ch = (rlen == 0) ? comp : |
217 | (comp - 0x100 < rlen) ? | |
8973db79 | 218 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 219 | if ((UV)ch != pch) { |
9041c2e3 | 220 | d = uvchr_to_utf8(d, ch); |
eb160463 | 221 | pch = (UV)ch; |
8973db79 JH |
222 | } |
223 | s += len; | |
224 | continue; | |
225 | } | |
226 | } | |
9b877dbb | 227 | } |
1aa99e6b IH |
228 | else if ((ch = tbl[comp]) >= 0) { |
229 | matches++; | |
eb160463 | 230 | if ((UV)ch != pch) { |
9041c2e3 | 231 | d = uvchr_to_utf8(d, ch); |
eb160463 | 232 | pch = (UV)ch; |
1aa99e6b IH |
233 | } |
234 | s += len; | |
235 | continue; | |
236 | } | |
9b877dbb IH |
237 | else if (ch == -1) { /* -1 is unmapped character */ |
238 | Copy(s, d, len, U8); | |
239 | d += len; | |
240 | } | |
1aa99e6b IH |
241 | else if (ch == -2) /* -2 is delete character */ |
242 | matches++; | |
243 | s += len; | |
244 | pch = 0xfeedface; | |
245 | } | |
5d06d08e | 246 | } |
1aa99e6b IH |
247 | else { |
248 | while (s < send) { | |
249 | STRLEN len; | |
9041c2e3 | 250 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
251 | if (comp > 0xff) { |
252 | if (!complement) { | |
0376ff32 | 253 | Move(s, d, len, U8); |
8973db79 JH |
254 | d += len; |
255 | } | |
256 | else { | |
257 | matches++; | |
258 | if (!del) { | |
9041c2e3 NIS |
259 | if (comp - 0x100 < rlen) |
260 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 261 | else |
9041c2e3 | 262 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
263 | } |
264 | } | |
9b877dbb | 265 | } |
1aa99e6b | 266 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 267 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
268 | matches++; |
269 | } | |
270 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
271 | Copy(s, d, len, U8); |
272 | d += len; | |
1aa99e6b IH |
273 | } |
274 | else if (ch == -2) /* -2 is delete character */ | |
275 | matches++; | |
276 | s += len; | |
277 | } | |
278 | } | |
9b877dbb IH |
279 | if (grows) { |
280 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
281 | Safefree(dstart); | |
282 | } | |
283 | else { | |
284 | *d = '\0'; | |
285 | SvCUR_set(sv, d - dstart); | |
286 | } | |
1aa99e6b | 287 | SvUTF8_on(sv); |
4757a243 | 288 | } |
5e44153e | 289 | SvSETMAGIC(sv); |
4757a243 LW |
290 | return matches; |
291 | } | |
292 | ||
942e002e | 293 | STATIC I32 |
21304a3b | 294 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 | 295 | { |
4757a243 LW |
296 | U8 *s; |
297 | U8 *send; | |
298 | U8 *d; | |
036b4402 | 299 | U8 *start; |
1aa99e6b | 300 | U8 *dstart, *dend; |
4757a243 | 301 | I32 matches = 0; |
9b877dbb | 302 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
303 | STRLEN len; |
304 | ||
305 | SV* rv = (SV*)cSVOP->op_sv; | |
306 | HV* hv = (HV*)SvRV(rv); | |
307 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
308 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
309 | UV extra = none + 1; | |
b7953727 | 310 | UV final = 0; |
4757a243 | 311 | UV uv; |
1aa99e6b IH |
312 | I32 isutf8; |
313 | U8 hibit = 0; | |
4757a243 LW |
314 | |
315 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
316 | isutf8 = SvUTF8(sv); |
317 | if (!isutf8) { | |
318 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
319 | while (t < e) { |
320 | U8 ch = *t++; | |
321 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 322 | break; |
c4d5f83a | 323 | } |
1aa99e6b IH |
324 | if (hibit) |
325 | s = bytes_to_utf8(s, &len); | |
326 | } | |
4757a243 | 327 | send = s + len; |
036b4402 | 328 | start = s; |
4757a243 LW |
329 | |
330 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
331 | if (svp) | |
332 | final = SvUV(*svp); | |
333 | ||
9b877dbb IH |
334 | if (grows) { |
335 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
336 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
337 | dend = d + len * 3; | |
338 | dstart = d; | |
339 | } | |
340 | else { | |
341 | dstart = d = s; | |
342 | dend = d + len; | |
343 | } | |
1aa99e6b | 344 | |
4757a243 | 345 | while (s < send) { |
3568d838 | 346 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 LW |
347 | s += UTF8SKIP(s); |
348 | matches++; | |
7d85a32c | 349 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
350 | } |
351 | else if (uv == none) { | |
1aa99e6b | 352 | int i = UTF8SKIP(s); |
0376ff32 | 353 | Move(s, d, i, U8); |
9b877dbb IH |
354 | d += i; |
355 | s += i; | |
4757a243 LW |
356 | } |
357 | else if (uv == extra) { | |
1aa99e6b | 358 | int i = UTF8SKIP(s); |
036b4402 | 359 | s += i; |
4757a243 | 360 | matches++; |
7d85a32c | 361 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
362 | } |
363 | else | |
364 | s += UTF8SKIP(s); | |
1aa99e6b | 365 | |
9b877dbb | 366 | if (d > dend) { |
1aa99e6b IH |
367 | STRLEN clen = d - dstart; |
368 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
9b877dbb | 369 | if (!grows) |
7d85a32c | 370 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b IH |
371 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
372 | d = dstart + clen; | |
373 | dend = dstart + nlen; | |
374 | } | |
4757a243 | 375 | } |
16ec844d | 376 | if (grows || hibit) { |
9b877dbb IH |
377 | sv_setpvn(sv, (char*)dstart, d - dstart); |
378 | Safefree(dstart); | |
16ec844d IH |
379 | if (grows && hibit) |
380 | Safefree(start); | |
9b877dbb IH |
381 | } |
382 | else { | |
383 | *d = '\0'; | |
384 | SvCUR_set(sv, d - dstart); | |
385 | } | |
4757a243 | 386 | SvSETMAGIC(sv); |
1aa99e6b | 387 | SvUTF8_on(sv); |
4757a243 LW |
388 | |
389 | return matches; | |
390 | } | |
391 | ||
942e002e | 392 | STATIC I32 |
21304a3b | 393 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 | 394 | { |
4757a243 | 395 | U8 *s; |
b7953727 | 396 | U8 *start = 0, *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; | |
45005bfb | 404 | UV extra = none + 1; |
4757a243 | 405 | UV uv; |
1aa99e6b | 406 | U8 hibit = 0; |
4757a243 LW |
407 | |
408 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
409 | if (!SvUTF8(sv)) { |
410 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
411 | while (t < e) { |
412 | U8 ch = *t++; | |
413 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 414 | break; |
c4d5f83a | 415 | } |
1aa99e6b IH |
416 | if (hibit) |
417 | start = s = bytes_to_utf8(s, &len); | |
418 | } | |
4757a243 LW |
419 | send = s + len; |
420 | ||
421 | while (s < send) { | |
3568d838 | 422 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 | 423 | matches++; |
834a4ddd | 424 | s += UTF8SKIP(s); |
4757a243 | 425 | } |
1aa99e6b IH |
426 | if (hibit) |
427 | Safefree(start); | |
4757a243 LW |
428 | |
429 | return matches; | |
430 | } | |
431 | ||
942e002e | 432 | STATIC I32 |
21304a3b | 433 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 | 434 | { |
4757a243 | 435 | U8 *s; |
1aa99e6b | 436 | U8 *start, *send; |
4757a243 LW |
437 | U8 *d; |
438 | I32 matches = 0; | |
439 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 440 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 441 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
442 | SV* rv = (SV*)cSVOP->op_sv; |
443 | HV* hv = (HV*)SvRV(rv); | |
444 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
445 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
446 | UV extra = none + 1; | |
b7953727 | 447 | UV final = 0; |
45005bfb | 448 | bool havefinal = FALSE; |
4757a243 LW |
449 | UV uv; |
450 | STRLEN len; | |
1aa99e6b IH |
451 | U8 *dstart, *dend; |
452 | I32 isutf8; | |
453 | U8 hibit = 0; | |
4757a243 LW |
454 | |
455 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
456 | isutf8 = SvUTF8(sv); |
457 | if (!isutf8) { | |
458 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
459 | while (t < e) { |
460 | U8 ch = *t++; | |
461 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 462 | break; |
c4d5f83a | 463 | } |
1aa99e6b IH |
464 | if (hibit) |
465 | s = bytes_to_utf8(s, &len); | |
466 | } | |
4757a243 | 467 | send = s + len; |
1aa99e6b | 468 | start = s; |
4757a243 LW |
469 | |
470 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
45005bfb | 471 | if (svp) { |
4757a243 | 472 | final = SvUV(*svp); |
45005bfb JH |
473 | havefinal = TRUE; |
474 | } | |
4757a243 | 475 | |
9b877dbb IH |
476 | if (grows) { |
477 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
478 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
479 | dend = d + len * 3; | |
480 | dstart = d; | |
481 | } | |
482 | else { | |
483 | dstart = d = s; | |
484 | dend = d + len; | |
485 | } | |
4757a243 LW |
486 | |
487 | if (squash) { | |
488 | UV puv = 0xfeedface; | |
489 | while (s < send) { | |
3568d838 | 490 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 491 | |
9b877dbb IH |
492 | if (d > dend) { |
493 | STRLEN clen = d - dstart; | |
494 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
495 | if (!grows) | |
7d85a32c | 496 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
497 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
498 | d = dstart + clen; | |
499 | dend = dstart + nlen; | |
4757a243 LW |
500 | } |
501 | if (uv < none) { | |
502 | matches++; | |
94472101 | 503 | s += UTF8SKIP(s); |
4757a243 | 504 | if (uv != puv) { |
7d85a32c | 505 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
506 | puv = uv; |
507 | } | |
4757a243 LW |
508 | continue; |
509 | } | |
510 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 511 | int i = UTF8SKIP(s); |
0376ff32 | 512 | Move(s, d, i, U8); |
9b877dbb IH |
513 | d += i; |
514 | s += i; | |
4757a243 LW |
515 | puv = 0xfeedface; |
516 | continue; | |
a0ed51b3 | 517 | } |
4757a243 LW |
518 | else if (uv == extra && !del) { |
519 | matches++; | |
45005bfb JH |
520 | if (havefinal) { |
521 | s += UTF8SKIP(s); | |
522 | if (puv != final) { | |
7d85a32c | 523 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
524 | puv = final; |
525 | } | |
526 | } | |
527 | else { | |
528 | STRLEN len; | |
7d85a32c | 529 | uv = utf8_to_uvuni(s, &len); |
45005bfb | 530 | if (uv != puv) { |
0376ff32 | 531 | Move(s, d, len, U8); |
45005bfb JH |
532 | d += len; |
533 | puv = uv; | |
534 | } | |
535 | s += len; | |
4757a243 | 536 | } |
4757a243 LW |
537 | continue; |
538 | } | |
01ec43d0 GS |
539 | matches++; /* "none+1" is delete character */ |
540 | s += UTF8SKIP(s); | |
a0ed51b3 | 541 | } |
79072805 LW |
542 | } |
543 | else { | |
4757a243 | 544 | while (s < send) { |
3568d838 | 545 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb IH |
546 | if (d > dend) { |
547 | STRLEN clen = d - dstart; | |
548 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
549 | if (!grows) | |
7d85a32c | 550 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
551 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
552 | d = dstart + clen; | |
553 | dend = dstart + nlen; | |
a0ed51b3 | 554 | } |
4757a243 LW |
555 | if (uv < none) { |
556 | matches++; | |
01ec43d0 | 557 | s += UTF8SKIP(s); |
7d85a32c | 558 | d = uvuni_to_utf8(d, uv); |
4757a243 | 559 | continue; |
a0ed51b3 | 560 | } |
4757a243 | 561 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 562 | int i = UTF8SKIP(s); |
0376ff32 | 563 | Move(s, d, i, U8); |
9b877dbb IH |
564 | d += i; |
565 | s += i; | |
4757a243 | 566 | continue; |
79072805 | 567 | } |
4757a243 LW |
568 | else if (uv == extra && !del) { |
569 | matches++; | |
01ec43d0 | 570 | s += UTF8SKIP(s); |
7d85a32c | 571 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
572 | continue; |
573 | } | |
01ec43d0 GS |
574 | matches++; /* "none+1" is delete character */ |
575 | s += UTF8SKIP(s); | |
79072805 | 576 | } |
4757a243 | 577 | } |
16ec844d | 578 | if (grows || hibit) { |
9b877dbb IH |
579 | sv_setpvn(sv, (char*)dstart, d - dstart); |
580 | Safefree(dstart); | |
16ec844d IH |
581 | if (grows && hibit) |
582 | Safefree(start); | |
9b877dbb IH |
583 | } |
584 | else { | |
585 | *d = '\0'; | |
586 | SvCUR_set(sv, d - dstart); | |
587 | } | |
1aa99e6b | 588 | SvUTF8_on(sv); |
4757a243 LW |
589 | SvSETMAGIC(sv); |
590 | ||
591 | return matches; | |
592 | } | |
593 | ||
594 | I32 | |
864dbfa3 | 595 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
596 | { |
597 | STRLEN len; | |
8e84507e | 598 | I32 hasutf = (PL_op->op_private & |
036b4402 | 599 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 600 | |
2233f375 | 601 | if (SvREADONLY(sv)) { |
765f542d NC |
602 | if (SvIsCOW(sv)) |
603 | sv_force_normal_flags(sv, 0); | |
2233f375 NC |
604 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
605 | Perl_croak(aTHX_ PL_no_modify); | |
606 | } | |
4757a243 LW |
607 | (void)SvPV(sv, len); |
608 | if (!len) | |
609 | return 0; | |
d59e14db RGS |
610 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
611 | if (!SvPOKp(sv)) | |
612 | (void)SvPV_force(sv, len); | |
2de7b02f | 613 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 614 | } |
4757a243 | 615 | |
cea2e8a9 | 616 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 617 | |
a77e643a RGS |
618 | switch (PL_op->op_private & ~hasutf & ( |
619 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| | |
620 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
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 | ||
fc0199cd | 672 | sv_setpvn(sv, "", 0); |
e4803c42 TS |
673 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
674 | if (SvUTF8(sv)) | |
675 | SvUTF8_off(sv); | |
676 | ||
8d6d96c1 HS |
677 | if (PL_tainting && SvMAGICAL(sv)) |
678 | SvTAINTED_off(sv); | |
679 | ||
463ee0b2 | 680 | if (items-- > 0) { |
92d29cee JH |
681 | if (*mark) |
682 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
683 | mark++; |
684 | } | |
8d6d96c1 | 685 | |
c512ce4f | 686 | if (delimlen) { |
79072805 | 687 | for (; items > 0; items--,mark++) { |
c512ce4f | 688 | sv_catsv(sv,del); |
79072805 LW |
689 | sv_catsv(sv,*mark); |
690 | } | |
691 | } | |
692 | else { | |
693 | for (; items > 0; items--,mark++) | |
694 | sv_catsv(sv,*mark); | |
695 | } | |
696 | SvSETMAGIC(sv); | |
697 | } | |
698 | ||
699 | void | |
864dbfa3 | 700 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 701 | { |
46fc3d4c | 702 | STRLEN patlen; |
703 | char *pat = SvPV(*sarg, patlen); | |
704 | bool do_taint = FALSE; | |
705 | ||
5b781b5b | 706 | SvUTF8_off(sv); |
2cf2cfc6 A |
707 | if (DO_UTF8(*sarg)) |
708 | SvUTF8_on(sv); | |
46fc3d4c | 709 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 | 710 | SvSETMAGIC(sv); |
46fc3d4c | 711 | if (do_taint) |
712 | SvTAINTED_on(sv); | |
79072805 LW |
713 | } |
714 | ||
33b45480 | 715 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
716 | UV |
717 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
718 | { | |
719 | STRLEN srclen, len; | |
720 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
721 | UV retnum = 0; | |
722 | ||
a50d7633 | 723 | if (offset < 0) |
81e118e0 | 724 | return retnum; |
8e84507e | 725 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 726 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 727 | |
dcad2880 | 728 | if (SvUTF8(sv)) |
33b45480 | 729 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 730 | |
81e118e0 JH |
731 | offset *= size; /* turn into bit offset */ |
732 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
733 | if (len > srclen) { | |
734 | if (size <= 8) | |
735 | retnum = 0; | |
736 | else { | |
737 | offset >>= 3; /* turn into byte offset */ | |
738 | if (size == 16) { | |
eb160463 | 739 | if ((STRLEN)offset >= srclen) |
81e118e0 JH |
740 | retnum = 0; |
741 | else | |
628e1a40 | 742 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
743 | } |
744 | else if (size == 32) { | |
eb160463 | 745 | if ((STRLEN)offset >= srclen) |
81e118e0 | 746 | retnum = 0; |
eb160463 | 747 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 JH |
748 | retnum = |
749 | ((UV) s[offset ] << 24); | |
eb160463 | 750 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 JH |
751 | retnum = |
752 | ((UV) s[offset ] << 24) + | |
753 | ((UV) s[offset + 1] << 16); | |
754 | else | |
755 | retnum = | |
756 | ((UV) s[offset ] << 24) + | |
757 | ((UV) s[offset + 1] << 16) + | |
758 | ( s[offset + 2] << 8); | |
759 | } | |
d7d93a81 | 760 | #ifdef UV_IS_QUAD |
c5a0f51a | 761 | else if (size == 64) { |
c5a0f51a | 762 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 763 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
764 | "Bit vector size > 32 non-portable"); |
765 | if (offset >= srclen) | |
766 | retnum = 0; | |
767 | else if (offset + 1 >= srclen) | |
768 | retnum = | |
769 | (UV) s[offset ] << 56; | |
770 | else if (offset + 2 >= srclen) | |
771 | retnum = | |
772 | ((UV) s[offset ] << 56) + | |
773 | ((UV) s[offset + 1] << 48); | |
774 | else if (offset + 3 >= srclen) | |
775 | retnum = | |
776 | ((UV) s[offset ] << 56) + | |
777 | ((UV) s[offset + 1] << 48) + | |
778 | ((UV) s[offset + 2] << 40); | |
779 | else if (offset + 4 >= srclen) | |
780 | retnum = | |
781 | ((UV) s[offset ] << 56) + | |
782 | ((UV) s[offset + 1] << 48) + | |
783 | ((UV) s[offset + 2] << 40) + | |
784 | ((UV) s[offset + 3] << 32); | |
785 | else if (offset + 5 >= 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 | ( s[offset + 4] << 24); | |
792 | else if (offset + 6 >= srclen) | |
793 | retnum = | |
794 | ((UV) s[offset ] << 56) + | |
795 | ((UV) s[offset + 1] << 48) + | |
796 | ((UV) s[offset + 2] << 40) + | |
797 | ((UV) s[offset + 3] << 32) + | |
798 | ((UV) s[offset + 4] << 24) + | |
799 | ((UV) s[offset + 5] << 16); | |
800 | else | |
8e84507e | 801 | retnum = |
c5a0f51a JH |
802 | ((UV) s[offset ] << 56) + |
803 | ((UV) s[offset + 1] << 48) + | |
804 | ((UV) s[offset + 2] << 40) + | |
805 | ((UV) s[offset + 3] << 32) + | |
806 | ((UV) s[offset + 4] << 24) + | |
807 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 808 | ( s[offset + 6] << 8); |
c5a0f51a JH |
809 | } |
810 | #endif | |
81e118e0 JH |
811 | } |
812 | } | |
813 | else if (size < 8) | |
814 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
815 | else { | |
816 | offset >>= 3; /* turn into byte offset */ | |
817 | if (size == 8) | |
818 | retnum = s[offset]; | |
819 | else if (size == 16) | |
820 | retnum = | |
628e1a40 | 821 | ((UV) s[offset] << 8) + |
81e118e0 JH |
822 | s[offset + 1]; |
823 | else if (size == 32) | |
824 | retnum = | |
825 | ((UV) s[offset ] << 24) + | |
826 | ((UV) s[offset + 1] << 16) + | |
827 | ( s[offset + 2] << 8) + | |
828 | s[offset + 3]; | |
d7d93a81 | 829 | #ifdef UV_IS_QUAD |
c5a0f51a | 830 | else if (size == 64) { |
c5a0f51a | 831 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 832 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
833 | "Bit vector size > 32 non-portable"); |
834 | retnum = | |
835 | ((UV) s[offset ] << 56) + | |
836 | ((UV) s[offset + 1] << 48) + | |
837 | ((UV) s[offset + 2] << 40) + | |
838 | ((UV) s[offset + 3] << 32) + | |
839 | ((UV) s[offset + 4] << 24) + | |
840 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 841 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
842 | s[offset + 7]; |
843 | } | |
844 | #endif | |
81e118e0 JH |
845 | } |
846 | ||
847 | return retnum; | |
848 | } | |
849 | ||
33b45480 SB |
850 | /* currently converts input to bytes if possible but doesn't sweat failures, |
851 | * although it does ensure that the string it clobbers is not marked as | |
852 | * utf8-valid any more | |
853 | */ | |
79072805 | 854 | void |
864dbfa3 | 855 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
856 | { |
857 | SV *targ = LvTARG(sv); | |
858 | register I32 offset; | |
859 | register I32 size; | |
8990e307 | 860 | register unsigned char *s; |
81e118e0 | 861 | register UV lval; |
79072805 | 862 | I32 mask; |
a0d0e21e LW |
863 | STRLEN targlen; |
864 | STRLEN len; | |
79072805 | 865 | |
8990e307 LW |
866 | if (!targ) |
867 | return; | |
a0d0e21e | 868 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 869 | if (SvUTF8(targ)) { |
33b45480 SB |
870 | /* This is handled by the SvPOK_only below... |
871 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
872 | SvUTF8_off(targ); | |
873 | */ | |
874 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
875 | } |
876 | ||
4ebbc975 | 877 | (void)SvPOK_only(targ); |
81e118e0 | 878 | lval = SvUV(sv); |
79072805 | 879 | offset = LvTARGOFF(sv); |
fe58ced6 | 880 | if (offset < 0) |
ed9aa3b7 | 881 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 882 | size = LvTARGLEN(sv); |
8e84507e | 883 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 884 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 885 | |
81e118e0 JH |
886 | offset *= size; /* turn into bit offset */ |
887 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
888 | if (len > targlen) { |
889 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 890 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
891 | SvCUR_set(targ, len); |
892 | } | |
8e84507e | 893 | |
79072805 LW |
894 | if (size < 8) { |
895 | mask = (1 << size) - 1; | |
896 | size = offset & 7; | |
897 | lval &= mask; | |
81e118e0 | 898 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
899 | s[offset] &= ~(mask << size); |
900 | s[offset] |= lval << size; | |
901 | } | |
902 | else { | |
81e118e0 | 903 | offset >>= 3; /* turn into byte offset */ |
79072805 | 904 | if (size == 8) |
eb160463 | 905 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 906 | else if (size == 16) { |
eb160463 GS |
907 | s[offset ] = (U8)((lval >> 8) & 0xff); |
908 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
909 | } |
910 | else if (size == 32) { | |
eb160463 GS |
911 | s[offset ] = (U8)((lval >> 24) & 0xff); |
912 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
913 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
914 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 915 | } |
d7d93a81 | 916 | #ifdef UV_IS_QUAD |
c5a0f51a | 917 | else if (size == 64) { |
c5a0f51a | 918 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 919 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 920 | "Bit vector size > 32 non-portable"); |
eb160463 GS |
921 | s[offset ] = (U8)((lval >> 56) & 0xff); |
922 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
923 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
924 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
925 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
926 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
927 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
928 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 929 | } |
dc1e3f56 | 930 | #endif |
79072805 | 931 | } |
7bb043c3 | 932 | SvSETMAGIC(targ); |
79072805 LW |
933 | } |
934 | ||
935 | void | |
864dbfa3 | 936 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 937 | { |
463ee0b2 | 938 | STRLEN len; |
a0d0e21e | 939 | char *s; |
8e84507e | 940 | |
79072805 | 941 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
942 | register I32 i; |
943 | I32 max; | |
944 | AV* av = (AV*)sv; | |
945 | max = AvFILL(av); | |
946 | for (i = 0; i <= max; i++) { | |
947 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 948 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
949 | do_chop(astr, sv); |
950 | } | |
951 | return; | |
79072805 | 952 | } |
aa854799 | 953 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
954 | HV* hv = (HV*)sv; |
955 | HE* entry; | |
956 | (void)hv_iterinit(hv); | |
957 | /*SUPPRESS 560*/ | |
155aba94 | 958 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
959 | do_chop(astr,hv_iterval(hv,entry)); |
960 | return; | |
79072805 | 961 | } |
991e6d41 NC |
962 | else if (SvREADONLY(sv)) { |
963 | if (SvFAKE(sv)) { | |
964 | /* SV is copy-on-write */ | |
965 | sv_force_normal_flags(sv, 0); | |
966 | } | |
967 | if (SvREADONLY(sv)) | |
968 | Perl_croak(aTHX_ PL_no_modify); | |
969 | } | |
a0d0e21e | 970 | s = SvPV(sv, len); |
748a9306 | 971 | if (len && !SvPOK(sv)) |
a0d0e21e | 972 | s = SvPV_force(sv, len); |
7e2040f0 | 973 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
974 | if (s && len) { |
975 | char *send = s + len; | |
976 | char *start = s; | |
977 | s = send - 1; | |
a0dbb045 JH |
978 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
979 | s--; | |
9041c2e3 | 980 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
981 | sv_setpvn(astr, s, send - s); |
982 | *s = '\0'; | |
983 | SvCUR_set(sv, s - start); | |
984 | SvNIOK_off(sv); | |
985 | SvUTF8_on(astr); | |
986 | } | |
a0ed51b3 LW |
987 | } |
988 | else | |
989 | sv_setpvn(astr, "", 0); | |
990 | } | |
7e2040f0 | 991 | else if (s && len) { |
a0d0e21e LW |
992 | s += --len; |
993 | sv_setpvn(astr, s, 1); | |
994 | *s = '\0'; | |
995 | SvCUR_set(sv, len); | |
2c19a612 | 996 | SvUTF8_off(sv); |
a0d0e21e | 997 | SvNIOK_off(sv); |
79072805 LW |
998 | } |
999 | else | |
a0d0e21e LW |
1000 | sv_setpvn(astr, "", 0); |
1001 | SvSETMAGIC(sv); | |
7e2040f0 | 1002 | } |
a0d0e21e LW |
1003 | |
1004 | I32 | |
864dbfa3 | 1005 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 1006 | { |
c07a80fd | 1007 | register I32 count; |
a0d0e21e | 1008 | STRLEN len; |
9b33ce3b | 1009 | STRLEN n_a; |
a0d0e21e | 1010 | char *s; |
c4c87a06 | 1011 | char *temp_buffer = NULL; |
a6aa349d | 1012 | SV* svrecode = Nullsv; |
c07a80fd | 1013 | |
3280af22 | 1014 | if (RsSNARF(PL_rs)) |
c07a80fd | 1015 | return 0; |
4c5a6083 GS |
1016 | if (RsRECORD(PL_rs)) |
1017 | return 0; | |
c07a80fd | 1018 | count = 0; |
a0d0e21e LW |
1019 | if (SvTYPE(sv) == SVt_PVAV) { |
1020 | register I32 i; | |
1021 | I32 max; | |
1022 | AV* av = (AV*)sv; | |
1023 | max = AvFILL(av); | |
1024 | for (i = 0; i <= max; i++) { | |
1025 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 1026 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1027 | count += do_chomp(sv); |
1028 | } | |
1029 | return count; | |
1030 | } | |
aa854799 | 1031 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
1032 | HV* hv = (HV*)sv; |
1033 | HE* entry; | |
1034 | (void)hv_iterinit(hv); | |
1035 | /*SUPPRESS 560*/ | |
155aba94 | 1036 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1037 | count += do_chomp(hv_iterval(hv,entry)); |
1038 | return count; | |
1039 | } | |
991e6d41 NC |
1040 | else if (SvREADONLY(sv)) { |
1041 | if (SvFAKE(sv)) { | |
1042 | /* SV is copy-on-write */ | |
1043 | sv_force_normal_flags(sv, 0); | |
1044 | } | |
1045 | if (SvREADONLY(sv)) | |
1046 | Perl_croak(aTHX_ PL_no_modify); | |
1047 | } | |
a6aa349d TS |
1048 | |
1049 | if (PL_encoding) { | |
1050 | if (!SvUTF8(sv)) { | |
1051 | /* XXX, here sv is utf8-ized as a side-effect! | |
1052 | If encoding.pm is used properly, almost string-generating | |
1053 | operations, including literal strings, chr(), input data, etc. | |
1054 | should have been utf8-ized already, right? | |
1055 | */ | |
1056 | sv_recode_to_utf8(sv, PL_encoding); | |
1057 | } | |
1058 | } | |
1059 | ||
a0d0e21e | 1060 | s = SvPV(sv, len); |
a0d0e21e LW |
1061 | if (s && len) { |
1062 | s += --len; | |
3280af22 | 1063 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1064 | if (*s != '\n') |
1065 | goto nope; | |
1066 | ++count; | |
1067 | while (len && s[-1] == '\n') { | |
1068 | --len; | |
1069 | --s; | |
1070 | ++count; | |
1071 | } | |
1072 | } | |
a0d0e21e | 1073 | else { |
a6aa349d | 1074 | STRLEN rslen, rs_charlen; |
3280af22 | 1075 | char *rsptr = SvPV(PL_rs, rslen); |
a6aa349d TS |
1076 | |
1077 | rs_charlen = SvUTF8(PL_rs) | |
1078 | ? sv_len_utf8(PL_rs) | |
1079 | : rslen; | |
1080 | ||
c4c87a06 NC |
1081 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1082 | /* Assumption is that rs is shorter than the scalar. */ | |
1083 | if (SvUTF8(PL_rs)) { | |
1084 | /* RS is utf8, scalar is 8 bit. */ | |
1085 | bool is_utf8 = TRUE; | |
1086 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, | |
1087 | &rslen, &is_utf8); | |
1088 | if (is_utf8) { | |
1089 | /* Cannot downgrade, therefore cannot possibly match | |
1090 | */ | |
1091 | assert (temp_buffer == rsptr); | |
1092 | temp_buffer = NULL; | |
1093 | goto nope; | |
1094 | } | |
1095 | rsptr = temp_buffer; | |
a6aa349d TS |
1096 | } |
1097 | else if (PL_encoding) { | |
1098 | /* RS is 8 bit, encoding.pm is used. | |
1099 | * Do not recode PL_rs as a side-effect. */ | |
1100 | svrecode = newSVpvn(rsptr, rslen); | |
1101 | sv_recode_to_utf8(svrecode, PL_encoding); | |
1102 | rsptr = SvPV(svrecode, rslen); | |
1103 | rs_charlen = sv_len_utf8(svrecode); | |
1104 | } | |
1105 | else { | |
c4c87a06 NC |
1106 | /* RS is 8 bit, scalar is utf8. */ |
1107 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); | |
1108 | rsptr = temp_buffer; | |
1109 | } | |
1110 | } | |
c07a80fd | 1111 | if (rslen == 1) { |
1112 | if (*s != *rsptr) | |
1113 | goto nope; | |
1114 | ++count; | |
1115 | } | |
1116 | else { | |
8c2cee6f | 1117 | if (len < rslen - 1) |
c07a80fd | 1118 | goto nope; |
1119 | len -= rslen - 1; | |
1120 | s -= rslen - 1; | |
36477c24 | 1121 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1122 | goto nope; |
a6aa349d | 1123 | count += rs_charlen; |
c07a80fd | 1124 | } |
a0d0e21e | 1125 | } |
9b33ce3b | 1126 | s = SvPV_force(sv, n_a); |
a0d0e21e | 1127 | SvCUR_set(sv, len); |
9b33ce3b | 1128 | *SvEND(sv) = '\0'; |
a0d0e21e | 1129 | SvNIOK_off(sv); |
9b33ce3b | 1130 | SvSETMAGIC(sv); |
a0d0e21e LW |
1131 | } |
1132 | nope: | |
a6aa349d TS |
1133 | |
1134 | if (svrecode) | |
1135 | SvREFCNT_dec(svrecode); | |
1136 | ||
c4c87a06 | 1137 | Safefree(temp_buffer); |
a0d0e21e | 1138 | return count; |
8e84507e | 1139 | } |
79072805 LW |
1140 | |
1141 | void | |
864dbfa3 | 1142 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1143 | { |
1144 | #ifdef LIBERAL | |
1145 | register long *dl; | |
1146 | register long *ll; | |
1147 | register long *rl; | |
1148 | #endif | |
1149 | register char *dc; | |
463ee0b2 LW |
1150 | STRLEN leftlen; |
1151 | STRLEN rightlen; | |
7a4c00b4 | 1152 | register char *lc; |
1153 | register char *rc; | |
79072805 | 1154 | register I32 len; |
a0d0e21e | 1155 | I32 lensave; |
7a4c00b4 | 1156 | char *lsave; |
1157 | char *rsave; | |
0c57e439 GS |
1158 | bool left_utf = DO_UTF8(left); |
1159 | bool right_utf = DO_UTF8(right); | |
b7953727 | 1160 | I32 needlen = 0; |
0c57e439 GS |
1161 | |
1162 | if (left_utf && !right_utf) | |
1163 | sv_utf8_upgrade(right); | |
a1ca4561 | 1164 | else if (!left_utf && right_utf) |
0c57e439 | 1165 | sv_utf8_upgrade(left); |
79072805 | 1166 | |
1fbd88dc CS |
1167 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1168 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
891f9566 YST |
1169 | lsave = lc = SvPV_nomg(left, leftlen); |
1170 | rsave = rc = SvPV_nomg(right, rightlen); | |
93a17b20 | 1171 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1172 | lensave = len; |
c9b3c8d0 JH |
1173 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1174 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1175 | Newz(801, dc, needlen + 1, char); | |
1176 | } | |
1177 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d | 1178 | STRLEN n_a; |
891f9566 | 1179 | dc = SvPV_force_nomg(sv, n_a); |
eb160463 GS |
1180 | if (SvCUR(sv) < (STRLEN)len) { |
1181 | dc = SvGROW(sv, (STRLEN)(len + 1)); | |
ff68c719 | 1182 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1183 | } | |
c9b3c8d0 JH |
1184 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1185 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1186 | } |
1187 | else { | |
c9b3c8d0 JH |
1188 | needlen = ((optype == OP_BIT_AND) |
1189 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1190 | Newz(801, dc, needlen + 1, char); |
1191 | (void)sv_usepvn(sv, dc, needlen); | |
1192 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1193 | } |
a0d0e21e LW |
1194 | SvCUR_set(sv, len); |
1195 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1196 | if (left_utf || right_utf) { |
1197 | UV duc, luc, ruc; | |
c9b3c8d0 | 1198 | char *dcsave = dc; |
0c57e439 GS |
1199 | STRLEN lulen = leftlen; |
1200 | STRLEN rulen = rightlen; | |
ba210ebe | 1201 | STRLEN ulen; |
0c57e439 GS |
1202 | |
1203 | switch (optype) { | |
1204 | case OP_BIT_AND: | |
1205 | while (lulen && rulen) { | |
9041c2e3 | 1206 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1207 | lc += ulen; |
1208 | lulen -= ulen; | |
9041c2e3 | 1209 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1210 | rc += ulen; |
1211 | rulen -= ulen; | |
1212 | duc = luc & ruc; | |
9041c2e3 | 1213 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1214 | } |
c9b3c8d0 JH |
1215 | if (sv == left || sv == right) |
1216 | (void)sv_usepvn(sv, dcsave, needlen); | |
1217 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1218 | break; |
1219 | case OP_BIT_XOR: | |
1220 | while (lulen && rulen) { | |
9041c2e3 | 1221 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1222 | lc += ulen; |
1223 | lulen -= ulen; | |
9041c2e3 | 1224 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1225 | rc += ulen; |
1226 | rulen -= ulen; | |
1227 | duc = luc ^ ruc; | |
9041c2e3 | 1228 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1229 | } |
1230 | goto mop_up_utf; | |
1231 | case OP_BIT_OR: | |
1232 | while (lulen && rulen) { | |
9041c2e3 | 1233 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1234 | lc += ulen; |
1235 | lulen -= ulen; | |
9041c2e3 | 1236 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1237 | rc += ulen; |
1238 | rulen -= ulen; | |
1239 | duc = luc | ruc; | |
9041c2e3 | 1240 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1241 | } |
1242 | mop_up_utf: | |
c9b3c8d0 JH |
1243 | if (sv == left || sv == right) |
1244 | (void)sv_usepvn(sv, dcsave, needlen); | |
1245 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1246 | if (rulen) |
1247 | sv_catpvn(sv, rc, rulen); | |
1248 | else if (lulen) | |
1249 | sv_catpvn(sv, lc, lulen); | |
1250 | else | |
1251 | *SvEND(sv) = '\0'; | |
1252 | break; | |
1253 | } | |
1254 | SvUTF8_on(sv); | |
1255 | goto finish; | |
1256 | } | |
1257 | else | |
79072805 LW |
1258 | #ifdef LIBERAL |
1259 | if (len >= sizeof(long)*4 && | |
1260 | !((long)dc % sizeof(long)) && | |
1261 | !((long)lc % sizeof(long)) && | |
1262 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1263 | { | |
1264 | I32 remainder = len % (sizeof(long)*4); | |
1265 | len /= (sizeof(long)*4); | |
1266 | ||
1267 | dl = (long*)dc; | |
1268 | ll = (long*)lc; | |
1269 | rl = (long*)rc; | |
1270 | ||
1271 | switch (optype) { | |
1272 | case OP_BIT_AND: | |
1273 | while (len--) { | |
1274 | *dl++ = *ll++ & *rl++; | |
1275 | *dl++ = *ll++ & *rl++; | |
1276 | *dl++ = *ll++ & *rl++; | |
1277 | *dl++ = *ll++ & *rl++; | |
1278 | } | |
1279 | break; | |
a0d0e21e | 1280 | case OP_BIT_XOR: |
79072805 LW |
1281 | while (len--) { |
1282 | *dl++ = *ll++ ^ *rl++; | |
1283 | *dl++ = *ll++ ^ *rl++; | |
1284 | *dl++ = *ll++ ^ *rl++; | |
1285 | *dl++ = *ll++ ^ *rl++; | |
1286 | } | |
1287 | break; | |
1288 | case OP_BIT_OR: | |
1289 | while (len--) { | |
1290 | *dl++ = *ll++ | *rl++; | |
1291 | *dl++ = *ll++ | *rl++; | |
1292 | *dl++ = *ll++ | *rl++; | |
1293 | *dl++ = *ll++ | *rl++; | |
1294 | } | |
1295 | } | |
1296 | ||
1297 | dc = (char*)dl; | |
1298 | lc = (char*)ll; | |
1299 | rc = (char*)rl; | |
1300 | ||
1301 | len = remainder; | |
1302 | } | |
1303 | #endif | |
a0d0e21e | 1304 | { |
a0d0e21e LW |
1305 | switch (optype) { |
1306 | case OP_BIT_AND: | |
1307 | while (len--) | |
1308 | *dc++ = *lc++ & *rc++; | |
1309 | break; | |
1310 | case OP_BIT_XOR: | |
1311 | while (len--) | |
1312 | *dc++ = *lc++ ^ *rc++; | |
1313 | goto mop_up; | |
1314 | case OP_BIT_OR: | |
1315 | while (len--) | |
1316 | *dc++ = *lc++ | *rc++; | |
1317 | mop_up: | |
1318 | len = lensave; | |
eb160463 | 1319 | if (rightlen > (STRLEN)len) |
a0d0e21e | 1320 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 | 1321 | else if (leftlen > (STRLEN)len) |
a0d0e21e | 1322 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1323 | else |
1324 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1325 | break; |
1326 | } | |
79072805 | 1327 | } |
0c57e439 | 1328 | finish: |
fb73857a | 1329 | SvTAINT(sv); |
79072805 | 1330 | } |
463ee0b2 LW |
1331 | |
1332 | OP * | |
cea2e8a9 | 1333 | Perl_do_kv(pTHX) |
463ee0b2 | 1334 | { |
39644a26 | 1335 | dSP; |
463ee0b2 | 1336 | HV *hv = (HV*)POPs; |
800e9ae0 | 1337 | HV *keys; |
463ee0b2 | 1338 | register HE *entry; |
463ee0b2 | 1339 | SV *tmpstr; |
54310121 | 1340 | I32 gimme = GIMME_V; |
533c011a NIS |
1341 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1342 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
8e84507e NIS |
1343 | |
1344 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1345 | dokeys = dovalues = TRUE; |
463ee0b2 | 1346 | |
85581909 | 1347 | if (!hv) { |
78f9721b | 1348 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1349 | dTARGET; /* make sure to clear its target here */ |
1350 | if (SvTYPE(TARG) == SVt_PVLV) | |
1351 | LvTARG(TARG) = Nullsv; | |
1352 | PUSHs(TARG); | |
1353 | } | |
463ee0b2 | 1354 | RETURN; |
85581909 | 1355 | } |
748a9306 | 1356 | |
6d822dc4 | 1357 | keys = hv; |
800e9ae0 | 1358 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1359 | |
54310121 | 1360 | if (gimme == G_VOID) |
aa689395 | 1361 | RETURN; |
1362 | ||
54310121 | 1363 | if (gimme == G_SCALAR) { |
6ee623d5 | 1364 | IV i; |
463ee0b2 LW |
1365 | dTARGET; |
1366 | ||
78f9721b | 1367 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1368 | if (SvTYPE(TARG) < SVt_PVLV) { |
1369 | sv_upgrade(TARG, SVt_PVLV); | |
14befaf4 | 1370 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 SB |
1371 | } |
1372 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1373 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1374 | if (LvTARG(TARG)) |
1375 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1376 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1377 | } |
85581909 SB |
1378 | PUSHs(TARG); |
1379 | RETURN; | |
1380 | } | |
1381 | ||
14befaf4 | 1382 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 | 1383 | i = HvKEYS(keys); |
463ee0b2 LW |
1384 | else { |
1385 | i = 0; | |
463ee0b2 | 1386 | /*SUPPRESS 560*/ |
800e9ae0 | 1387 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1388 | } |
1389 | PUSHi( i ); | |
1390 | RETURN; | |
1391 | } | |
1392 | ||
8ed4b672 | 1393 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1394 | |
463ee0b2 | 1395 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1396 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1397 | SPAGAIN; |
574c8022 JH |
1398 | if (dokeys) { |
1399 | SV* sv = hv_iterkeysv(entry); | |
574c8022 JH |
1400 | XPUSHs(sv); /* won't clobber stack_sp */ |
1401 | } | |
463ee0b2 | 1402 | if (dovalues) { |
463ee0b2 | 1403 | PUTBACK; |
6d822dc4 | 1404 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 | 1405 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1406 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1407 | HvMAX(keys)+1, |
1408 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1409 | SPAGAIN; |
46fc3d4c | 1410 | XPUSHs(tmpstr); |
463ee0b2 LW |
1411 | } |
1412 | PUTBACK; | |
1413 | } | |
1414 | return NORMAL; | |
1415 | } | |
4e35701f | 1416 |