Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
4bb101f2 JH |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
4 | * 2000, 2001, 2002, 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 | #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) |
21 | #include <signal.h> | |
22 | #endif | |
64ca3a65 | 23 | #endif |
79072805 | 24 | |
942e002e | 25 | STATIC I32 |
b250498f | 26 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 | 27 | { |
4757a243 | 28 | U8 *s; |
b250498f | 29 | U8 *d; |
4757a243 | 30 | U8 *send; |
b250498f | 31 | U8 *dstart; |
4757a243 | 32 | I32 matches = 0; |
9b877dbb | 33 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 | 34 | STRLEN len; |
4757a243 LW |
35 | short *tbl; |
36 | I32 ch; | |
79072805 | 37 | |
4757a243 LW |
38 | tbl = (short*)cPVOP->op_pv; |
39 | if (!tbl) | |
7d85a32c | 40 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 41 | |
4757a243 LW |
42 | s = (U8*)SvPV(sv, len); |
43 | send = s + len; | |
44 | ||
b250498f | 45 | /* First, take care of non-UTF8 input strings, because they're easy */ |
1aa99e6b | 46 | if (!SvUTF8(sv)) { |
01ec43d0 | 47 | while (s < send) { |
036b4402 | 48 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 | 49 | matches++; |
eb160463 | 50 | *s++ = (U8)ch; |
01ec43d0 GS |
51 | } |
52 | else | |
53 | s++; | |
54 | } | |
55 | SvSETMAGIC(sv); | |
b250498f GS |
56 | return matches; |
57 | } | |
4757a243 | 58 | |
b250498f | 59 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb IH |
60 | if (grows) |
61 | New(0, d, len*2+1, U8); | |
62 | else | |
63 | d = s; | |
b250498f GS |
64 | dstart = d; |
65 | while (s < send) { | |
ba210ebe | 66 | STRLEN ulen; |
9b877dbb | 67 | UV c; |
b250498f | 68 | |
b250498f | 69 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 | 70 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb | 71 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f | 72 | matches++; |
9041c2e3 | 73 | d = uvchr_to_utf8(d, ch); |
b250498f | 74 | s += ulen; |
01ec43d0 GS |
75 | } |
76 | else { /* No match -> copy */ | |
9b877dbb IH |
77 | Copy(s, d, ulen, U8); |
78 | d += ulen; | |
79 | s += ulen; | |
b250498f GS |
80 | } |
81 | } | |
9b877dbb IH |
82 | if (grows) { |
83 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
84 | Safefree(dstart); | |
85 | } | |
86 | else { | |
87 | *d = '\0'; | |
88 | SvCUR_set(sv, d - dstart); | |
89 | } | |
b250498f | 90 | SvUTF8_on(sv); |
b250498f | 91 | SvSETMAGIC(sv); |
4757a243 LW |
92 | return matches; |
93 | } | |
94 | ||
942e002e | 95 | STATIC I32 |
21304a3b | 96 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 | 97 | { |
4757a243 LW |
98 | U8 *s; |
99 | U8 *send; | |
100 | I32 matches = 0; | |
101 | STRLEN len; | |
102 | short *tbl; | |
8973db79 | 103 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 LW |
104 | |
105 | tbl = (short*)cPVOP->op_pv; | |
106 | if (!tbl) | |
7d85a32c | 107 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 LW |
108 | |
109 | s = (U8*)SvPV(sv, len); | |
110 | send = s + len; | |
111 | ||
1aa99e6b IH |
112 | if (!SvUTF8(sv)) |
113 | while (s < send) { | |
114 | if (tbl[*s++] >= 0) | |
036b4402 | 115 | matches++; |
1aa99e6b IH |
116 | } |
117 | else | |
118 | while (s < send) { | |
119 | UV c; | |
120 | STRLEN ulen; | |
9041c2e3 | 121 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 JH |
122 | if (c < 0x100) { |
123 | if (tbl[c] >= 0) | |
124 | matches++; | |
125 | } else if (complement) | |
1aa99e6b IH |
126 | matches++; |
127 | s += ulen; | |
128 | } | |
4757a243 LW |
129 | |
130 | return matches; | |
131 | } | |
132 | ||
942e002e | 133 | STATIC I32 |
21304a3b | 134 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 | 135 | { |
4757a243 LW |
136 | U8 *s; |
137 | U8 *send; | |
138 | U8 *d; | |
5e44153e | 139 | U8 *dstart; |
1aa99e6b | 140 | I32 isutf8; |
4757a243 | 141 | I32 matches = 0; |
9b877dbb | 142 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 JH |
143 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
144 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
b7953727 | 145 | STRLEN len, rlen = 0; |
4757a243 LW |
146 | short *tbl; |
147 | I32 ch; | |
148 | ||
149 | tbl = (short*)cPVOP->op_pv; | |
150 | if (!tbl) | |
7d85a32c | 151 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 LW |
152 | |
153 | s = (U8*)SvPV(sv, len); | |
1aa99e6b | 154 | isutf8 = SvUTF8(sv); |
4757a243 LW |
155 | send = s + len; |
156 | ||
1aa99e6b IH |
157 | if (!isutf8) { |
158 | dstart = d = s; | |
159 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
160 | U8* p = send; | |
161 | while (s < send) { | |
162 | if ((ch = tbl[*s]) >= 0) { | |
eb160463 | 163 | *d = (U8)ch; |
036b4402 | 164 | matches++; |
1aa99e6b IH |
165 | if (p != d - 1 || *p != *d) |
166 | p = d++; | |
167 | } | |
168 | else if (ch == -1) /* -1 is unmapped character */ | |
169 | *d++ = *s; | |
170 | else if (ch == -2) /* -2 is delete character */ | |
171 | matches++; | |
172 | s++; | |
173 | } | |
a0ed51b3 | 174 | } |
1aa99e6b IH |
175 | else { |
176 | while (s < send) { | |
177 | if ((ch = tbl[*s]) >= 0) { | |
178 | matches++; | |
eb160463 | 179 | *d++ = (U8)ch; |
1aa99e6b IH |
180 | } |
181 | else if (ch == -1) /* -1 is unmapped character */ | |
182 | *d++ = *s; | |
183 | else if (ch == -2) /* -2 is delete character */ | |
184 | matches++; | |
185 | s++; | |
186 | } | |
187 | } | |
76ef7183 | 188 | *d = '\0'; |
1aa99e6b | 189 | SvCUR_set(sv, d - dstart); |
4757a243 | 190 | } |
1aa99e6b | 191 | else { /* isutf8 */ |
9b877dbb IH |
192 | if (grows) |
193 | New(0, d, len*2+1, U8); | |
194 | else | |
195 | d = s; | |
1aa99e6b | 196 | dstart = d; |
8973db79 JH |
197 | if (complement && !del) |
198 | rlen = tbl[0x100]; | |
1aa99e6b | 199 | |
4aecb5b5 JH |
200 | #ifdef MACOS_TRADITIONAL |
201 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
202 | #endif | |
203 | ||
1aa99e6b | 204 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
205 | UV pch = 0xfeedface; |
206 | while (s < send) { | |
207 | STRLEN len; | |
9041c2e3 | 208 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b | 209 | |
8973db79 JH |
210 | if (comp > 0xff) { |
211 | if (!complement) { | |
212 | Copy(s, d, len, U8); | |
213 | d += len; | |
214 | } | |
215 | else { | |
216 | matches++; | |
217 | if (!del) { | |
05d340b8 JH |
218 | ch = (rlen == 0) ? comp : |
219 | (comp - 0x100 < rlen) ? | |
8973db79 | 220 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 221 | if ((UV)ch != pch) { |
9041c2e3 | 222 | d = uvchr_to_utf8(d, ch); |
eb160463 | 223 | pch = (UV)ch; |
8973db79 JH |
224 | } |
225 | s += len; | |
226 | continue; | |
227 | } | |
228 | } | |
9b877dbb | 229 | } |
1aa99e6b IH |
230 | else if ((ch = tbl[comp]) >= 0) { |
231 | matches++; | |
eb160463 | 232 | if ((UV)ch != pch) { |
9041c2e3 | 233 | d = uvchr_to_utf8(d, ch); |
eb160463 | 234 | pch = (UV)ch; |
1aa99e6b IH |
235 | } |
236 | s += len; | |
237 | continue; | |
238 | } | |
9b877dbb IH |
239 | else if (ch == -1) { /* -1 is unmapped character */ |
240 | Copy(s, d, len, U8); | |
241 | d += len; | |
242 | } | |
1aa99e6b IH |
243 | else if (ch == -2) /* -2 is delete character */ |
244 | matches++; | |
245 | s += len; | |
246 | pch = 0xfeedface; | |
247 | } | |
5d06d08e | 248 | } |
1aa99e6b IH |
249 | else { |
250 | while (s < send) { | |
251 | STRLEN len; | |
9041c2e3 | 252 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
253 | if (comp > 0xff) { |
254 | if (!complement) { | |
255 | Copy(s, d, len, U8); | |
256 | d += len; | |
257 | } | |
258 | else { | |
259 | matches++; | |
260 | if (!del) { | |
9041c2e3 NIS |
261 | if (comp - 0x100 < rlen) |
262 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 263 | else |
9041c2e3 | 264 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
265 | } |
266 | } | |
9b877dbb | 267 | } |
1aa99e6b | 268 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 269 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
270 | matches++; |
271 | } | |
272 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
273 | Copy(s, d, len, U8); |
274 | d += len; | |
1aa99e6b IH |
275 | } |
276 | else if (ch == -2) /* -2 is delete character */ | |
277 | matches++; | |
278 | s += len; | |
279 | } | |
280 | } | |
9b877dbb IH |
281 | if (grows) { |
282 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
283 | Safefree(dstart); | |
284 | } | |
285 | else { | |
286 | *d = '\0'; | |
287 | SvCUR_set(sv, d - dstart); | |
288 | } | |
1aa99e6b | 289 | SvUTF8_on(sv); |
4757a243 | 290 | } |
5e44153e | 291 | SvSETMAGIC(sv); |
4757a243 LW |
292 | return matches; |
293 | } | |
294 | ||
942e002e | 295 | STATIC I32 |
21304a3b | 296 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 | 297 | { |
4757a243 LW |
298 | U8 *s; |
299 | U8 *send; | |
300 | U8 *d; | |
036b4402 | 301 | U8 *start; |
1aa99e6b | 302 | U8 *dstart, *dend; |
4757a243 | 303 | I32 matches = 0; |
9b877dbb | 304 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
305 | STRLEN len; |
306 | ||
307 | SV* rv = (SV*)cSVOP->op_sv; | |
308 | HV* hv = (HV*)SvRV(rv); | |
309 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
310 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
311 | UV extra = none + 1; | |
b7953727 | 312 | UV final = 0; |
4757a243 | 313 | UV uv; |
1aa99e6b IH |
314 | I32 isutf8; |
315 | U8 hibit = 0; | |
4757a243 LW |
316 | |
317 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
318 | isutf8 = SvUTF8(sv); |
319 | if (!isutf8) { | |
320 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
321 | while (t < e) { |
322 | U8 ch = *t++; | |
323 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 324 | break; |
c4d5f83a | 325 | } |
1aa99e6b IH |
326 | if (hibit) |
327 | s = bytes_to_utf8(s, &len); | |
328 | } | |
4757a243 | 329 | send = s + len; |
036b4402 | 330 | start = s; |
4757a243 LW |
331 | |
332 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
333 | if (svp) | |
334 | final = SvUV(*svp); | |
335 | ||
9b877dbb IH |
336 | if (grows) { |
337 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
338 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
339 | dend = d + len * 3; | |
340 | dstart = d; | |
341 | } | |
342 | else { | |
343 | dstart = d = s; | |
344 | dend = d + len; | |
345 | } | |
1aa99e6b | 346 | |
4757a243 | 347 | while (s < send) { |
3568d838 | 348 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 LW |
349 | s += UTF8SKIP(s); |
350 | matches++; | |
7d85a32c | 351 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
352 | } |
353 | else if (uv == none) { | |
1aa99e6b | 354 | int i = UTF8SKIP(s); |
9b877dbb IH |
355 | Copy(s, d, i, U8); |
356 | d += i; | |
357 | s += i; | |
4757a243 LW |
358 | } |
359 | else if (uv == extra) { | |
1aa99e6b | 360 | int i = UTF8SKIP(s); |
036b4402 | 361 | s += i; |
4757a243 | 362 | matches++; |
7d85a32c | 363 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
364 | } |
365 | else | |
366 | s += UTF8SKIP(s); | |
1aa99e6b | 367 | |
9b877dbb | 368 | if (d > dend) { |
1aa99e6b IH |
369 | STRLEN clen = d - dstart; |
370 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
9b877dbb | 371 | if (!grows) |
7d85a32c | 372 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b IH |
373 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
374 | d = dstart + clen; | |
375 | dend = dstart + nlen; | |
376 | } | |
4757a243 | 377 | } |
16ec844d | 378 | if (grows || hibit) { |
9b877dbb IH |
379 | sv_setpvn(sv, (char*)dstart, d - dstart); |
380 | Safefree(dstart); | |
16ec844d IH |
381 | if (grows && hibit) |
382 | Safefree(start); | |
9b877dbb IH |
383 | } |
384 | else { | |
385 | *d = '\0'; | |
386 | SvCUR_set(sv, d - dstart); | |
387 | } | |
4757a243 | 388 | SvSETMAGIC(sv); |
1aa99e6b | 389 | SvUTF8_on(sv); |
4757a243 LW |
390 | |
391 | return matches; | |
392 | } | |
393 | ||
942e002e | 394 | STATIC I32 |
21304a3b | 395 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 | 396 | { |
4757a243 | 397 | U8 *s; |
b7953727 | 398 | U8 *start = 0, *send; |
4757a243 LW |
399 | I32 matches = 0; |
400 | STRLEN len; | |
401 | ||
402 | SV* rv = (SV*)cSVOP->op_sv; | |
403 | HV* hv = (HV*)SvRV(rv); | |
404 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
405 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
45005bfb | 406 | UV extra = none + 1; |
4757a243 | 407 | UV uv; |
1aa99e6b | 408 | U8 hibit = 0; |
4757a243 LW |
409 | |
410 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
411 | if (!SvUTF8(sv)) { |
412 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
413 | while (t < e) { |
414 | U8 ch = *t++; | |
415 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 416 | break; |
c4d5f83a | 417 | } |
1aa99e6b IH |
418 | if (hibit) |
419 | start = s = bytes_to_utf8(s, &len); | |
420 | } | |
4757a243 LW |
421 | send = s + len; |
422 | ||
423 | while (s < send) { | |
3568d838 | 424 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 | 425 | matches++; |
834a4ddd | 426 | s += UTF8SKIP(s); |
4757a243 | 427 | } |
1aa99e6b IH |
428 | if (hibit) |
429 | Safefree(start); | |
4757a243 LW |
430 | |
431 | return matches; | |
432 | } | |
433 | ||
942e002e | 434 | STATIC I32 |
21304a3b | 435 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 | 436 | { |
4757a243 | 437 | U8 *s; |
1aa99e6b | 438 | U8 *start, *send; |
4757a243 LW |
439 | U8 *d; |
440 | I32 matches = 0; | |
441 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 442 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 443 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
444 | SV* rv = (SV*)cSVOP->op_sv; |
445 | HV* hv = (HV*)SvRV(rv); | |
446 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
447 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
448 | UV extra = none + 1; | |
b7953727 | 449 | UV final = 0; |
45005bfb | 450 | bool havefinal = FALSE; |
4757a243 LW |
451 | UV uv; |
452 | STRLEN len; | |
1aa99e6b IH |
453 | U8 *dstart, *dend; |
454 | I32 isutf8; | |
455 | U8 hibit = 0; | |
4757a243 LW |
456 | |
457 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
458 | isutf8 = SvUTF8(sv); |
459 | if (!isutf8) { | |
460 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
461 | while (t < e) { |
462 | U8 ch = *t++; | |
463 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 464 | break; |
c4d5f83a | 465 | } |
1aa99e6b IH |
466 | if (hibit) |
467 | s = bytes_to_utf8(s, &len); | |
468 | } | |
4757a243 | 469 | send = s + len; |
1aa99e6b | 470 | start = s; |
4757a243 LW |
471 | |
472 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
45005bfb | 473 | if (svp) { |
4757a243 | 474 | final = SvUV(*svp); |
45005bfb JH |
475 | havefinal = TRUE; |
476 | } | |
4757a243 | 477 | |
9b877dbb IH |
478 | if (grows) { |
479 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
480 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
481 | dend = d + len * 3; | |
482 | dstart = d; | |
483 | } | |
484 | else { | |
485 | dstart = d = s; | |
486 | dend = d + len; | |
487 | } | |
4757a243 LW |
488 | |
489 | if (squash) { | |
490 | UV puv = 0xfeedface; | |
491 | while (s < send) { | |
3568d838 | 492 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 493 | |
9b877dbb IH |
494 | if (d > dend) { |
495 | STRLEN clen = d - dstart; | |
496 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
497 | if (!grows) | |
7d85a32c | 498 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
499 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
500 | d = dstart + clen; | |
501 | dend = dstart + nlen; | |
4757a243 LW |
502 | } |
503 | if (uv < none) { | |
504 | matches++; | |
94472101 | 505 | s += UTF8SKIP(s); |
4757a243 | 506 | if (uv != puv) { |
7d85a32c | 507 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
508 | puv = uv; |
509 | } | |
4757a243 LW |
510 | continue; |
511 | } | |
512 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 513 | int i = UTF8SKIP(s); |
9b877dbb IH |
514 | Copy(s, d, i, U8); |
515 | d += i; | |
516 | s += i; | |
4757a243 LW |
517 | puv = 0xfeedface; |
518 | continue; | |
a0ed51b3 | 519 | } |
4757a243 LW |
520 | else if (uv == extra && !del) { |
521 | matches++; | |
45005bfb JH |
522 | if (havefinal) { |
523 | s += UTF8SKIP(s); | |
524 | if (puv != final) { | |
7d85a32c | 525 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
526 | puv = final; |
527 | } | |
528 | } | |
529 | else { | |
530 | STRLEN len; | |
7d85a32c | 531 | uv = utf8_to_uvuni(s, &len); |
45005bfb JH |
532 | if (uv != puv) { |
533 | Copy(s, d, len, U8); | |
534 | d += len; | |
535 | puv = uv; | |
536 | } | |
537 | s += len; | |
4757a243 | 538 | } |
4757a243 LW |
539 | continue; |
540 | } | |
01ec43d0 GS |
541 | matches++; /* "none+1" is delete character */ |
542 | s += UTF8SKIP(s); | |
a0ed51b3 | 543 | } |
79072805 LW |
544 | } |
545 | else { | |
4757a243 | 546 | while (s < send) { |
3568d838 | 547 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb IH |
548 | if (d > dend) { |
549 | STRLEN clen = d - dstart; | |
550 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
551 | if (!grows) | |
7d85a32c | 552 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
553 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
554 | d = dstart + clen; | |
555 | dend = dstart + nlen; | |
a0ed51b3 | 556 | } |
4757a243 LW |
557 | if (uv < none) { |
558 | matches++; | |
01ec43d0 | 559 | s += UTF8SKIP(s); |
7d85a32c | 560 | d = uvuni_to_utf8(d, uv); |
4757a243 | 561 | continue; |
a0ed51b3 | 562 | } |
4757a243 | 563 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 564 | int i = UTF8SKIP(s); |
9b877dbb IH |
565 | Copy(s, d, i, U8); |
566 | d += i; | |
567 | s += i; | |
4757a243 | 568 | continue; |
79072805 | 569 | } |
4757a243 LW |
570 | else if (uv == extra && !del) { |
571 | matches++; | |
01ec43d0 | 572 | s += UTF8SKIP(s); |
7d85a32c | 573 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
574 | continue; |
575 | } | |
01ec43d0 GS |
576 | matches++; /* "none+1" is delete character */ |
577 | s += UTF8SKIP(s); | |
79072805 | 578 | } |
4757a243 | 579 | } |
16ec844d | 580 | if (grows || hibit) { |
9b877dbb IH |
581 | sv_setpvn(sv, (char*)dstart, d - dstart); |
582 | Safefree(dstart); | |
16ec844d IH |
583 | if (grows && hibit) |
584 | Safefree(start); | |
9b877dbb IH |
585 | } |
586 | else { | |
587 | *d = '\0'; | |
588 | SvCUR_set(sv, d - dstart); | |
589 | } | |
1aa99e6b | 590 | SvUTF8_on(sv); |
4757a243 LW |
591 | SvSETMAGIC(sv); |
592 | ||
593 | return matches; | |
594 | } | |
595 | ||
596 | I32 | |
864dbfa3 | 597 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
598 | { |
599 | STRLEN len; | |
8e84507e | 600 | I32 hasutf = (PL_op->op_private & |
036b4402 | 601 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 602 | |
2233f375 | 603 | if (SvREADONLY(sv)) { |
765f542d NC |
604 | if (SvIsCOW(sv)) |
605 | sv_force_normal_flags(sv, 0); | |
2233f375 NC |
606 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
607 | Perl_croak(aTHX_ PL_no_modify); | |
608 | } | |
4757a243 LW |
609 | (void)SvPV(sv, len); |
610 | if (!len) | |
611 | return 0; | |
d59e14db RGS |
612 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
613 | if (!SvPOKp(sv)) | |
614 | (void)SvPV_force(sv, len); | |
2de7b02f | 615 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 616 | } |
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 | ||
5b781b5b | 702 | SvUTF8_off(sv); |
2cf2cfc6 A |
703 | if (DO_UTF8(*sarg)) |
704 | SvUTF8_on(sv); | |
46fc3d4c | 705 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 | 706 | SvSETMAGIC(sv); |
46fc3d4c | 707 | if (do_taint) |
708 | SvTAINTED_on(sv); | |
79072805 LW |
709 | } |
710 | ||
33b45480 | 711 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
712 | UV |
713 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
714 | { | |
715 | STRLEN srclen, len; | |
716 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
717 | UV retnum = 0; | |
718 | ||
a50d7633 | 719 | if (offset < 0) |
81e118e0 | 720 | return retnum; |
8e84507e | 721 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 722 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 723 | |
dcad2880 | 724 | if (SvUTF8(sv)) |
33b45480 | 725 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 726 | |
81e118e0 JH |
727 | offset *= size; /* turn into bit offset */ |
728 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
729 | if (len > srclen) { | |
730 | if (size <= 8) | |
731 | retnum = 0; | |
732 | else { | |
733 | offset >>= 3; /* turn into byte offset */ | |
734 | if (size == 16) { | |
eb160463 | 735 | if ((STRLEN)offset >= srclen) |
81e118e0 JH |
736 | retnum = 0; |
737 | else | |
628e1a40 | 738 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
739 | } |
740 | else if (size == 32) { | |
eb160463 | 741 | if ((STRLEN)offset >= srclen) |
81e118e0 | 742 | retnum = 0; |
eb160463 | 743 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 JH |
744 | retnum = |
745 | ((UV) s[offset ] << 24); | |
eb160463 | 746 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 JH |
747 | retnum = |
748 | ((UV) s[offset ] << 24) + | |
749 | ((UV) s[offset + 1] << 16); | |
750 | else | |
751 | retnum = | |
752 | ((UV) s[offset ] << 24) + | |
753 | ((UV) s[offset + 1] << 16) + | |
754 | ( s[offset + 2] << 8); | |
755 | } | |
d7d93a81 | 756 | #ifdef UV_IS_QUAD |
c5a0f51a | 757 | else if (size == 64) { |
c5a0f51a | 758 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 759 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
760 | "Bit vector size > 32 non-portable"); |
761 | if (offset >= srclen) | |
762 | retnum = 0; | |
763 | else if (offset + 1 >= srclen) | |
764 | retnum = | |
765 | (UV) s[offset ] << 56; | |
766 | else if (offset + 2 >= srclen) | |
767 | retnum = | |
768 | ((UV) s[offset ] << 56) + | |
769 | ((UV) s[offset + 1] << 48); | |
770 | else if (offset + 3 >= srclen) | |
771 | retnum = | |
772 | ((UV) s[offset ] << 56) + | |
773 | ((UV) s[offset + 1] << 48) + | |
774 | ((UV) s[offset + 2] << 40); | |
775 | else if (offset + 4 >= srclen) | |
776 | retnum = | |
777 | ((UV) s[offset ] << 56) + | |
778 | ((UV) s[offset + 1] << 48) + | |
779 | ((UV) s[offset + 2] << 40) + | |
780 | ((UV) s[offset + 3] << 32); | |
781 | else if (offset + 5 >= srclen) | |
782 | retnum = | |
783 | ((UV) s[offset ] << 56) + | |
784 | ((UV) s[offset + 1] << 48) + | |
785 | ((UV) s[offset + 2] << 40) + | |
786 | ((UV) s[offset + 3] << 32) + | |
787 | ( s[offset + 4] << 24); | |
788 | else if (offset + 6 >= srclen) | |
789 | retnum = | |
790 | ((UV) s[offset ] << 56) + | |
791 | ((UV) s[offset + 1] << 48) + | |
792 | ((UV) s[offset + 2] << 40) + | |
793 | ((UV) s[offset + 3] << 32) + | |
794 | ((UV) s[offset + 4] << 24) + | |
795 | ((UV) s[offset + 5] << 16); | |
796 | else | |
8e84507e | 797 | retnum = |
c5a0f51a JH |
798 | ((UV) s[offset ] << 56) + |
799 | ((UV) s[offset + 1] << 48) + | |
800 | ((UV) s[offset + 2] << 40) + | |
801 | ((UV) s[offset + 3] << 32) + | |
802 | ((UV) s[offset + 4] << 24) + | |
803 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 804 | ( s[offset + 6] << 8); |
c5a0f51a JH |
805 | } |
806 | #endif | |
81e118e0 JH |
807 | } |
808 | } | |
809 | else if (size < 8) | |
810 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
811 | else { | |
812 | offset >>= 3; /* turn into byte offset */ | |
813 | if (size == 8) | |
814 | retnum = s[offset]; | |
815 | else if (size == 16) | |
816 | retnum = | |
628e1a40 | 817 | ((UV) s[offset] << 8) + |
81e118e0 JH |
818 | s[offset + 1]; |
819 | else if (size == 32) | |
820 | retnum = | |
821 | ((UV) s[offset ] << 24) + | |
822 | ((UV) s[offset + 1] << 16) + | |
823 | ( s[offset + 2] << 8) + | |
824 | s[offset + 3]; | |
d7d93a81 | 825 | #ifdef UV_IS_QUAD |
c5a0f51a | 826 | else if (size == 64) { |
c5a0f51a | 827 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 828 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
829 | "Bit vector size > 32 non-portable"); |
830 | retnum = | |
831 | ((UV) s[offset ] << 56) + | |
832 | ((UV) s[offset + 1] << 48) + | |
833 | ((UV) s[offset + 2] << 40) + | |
834 | ((UV) s[offset + 3] << 32) + | |
835 | ((UV) s[offset + 4] << 24) + | |
836 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 837 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
838 | s[offset + 7]; |
839 | } | |
840 | #endif | |
81e118e0 JH |
841 | } |
842 | ||
843 | return retnum; | |
844 | } | |
845 | ||
33b45480 SB |
846 | /* currently converts input to bytes if possible but doesn't sweat failures, |
847 | * although it does ensure that the string it clobbers is not marked as | |
848 | * utf8-valid any more | |
849 | */ | |
79072805 | 850 | void |
864dbfa3 | 851 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
852 | { |
853 | SV *targ = LvTARG(sv); | |
854 | register I32 offset; | |
855 | register I32 size; | |
8990e307 | 856 | register unsigned char *s; |
81e118e0 | 857 | register UV lval; |
79072805 | 858 | I32 mask; |
a0d0e21e LW |
859 | STRLEN targlen; |
860 | STRLEN len; | |
79072805 | 861 | |
8990e307 LW |
862 | if (!targ) |
863 | return; | |
a0d0e21e | 864 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 865 | if (SvUTF8(targ)) { |
33b45480 SB |
866 | /* This is handled by the SvPOK_only below... |
867 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
868 | SvUTF8_off(targ); | |
869 | */ | |
870 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
871 | } |
872 | ||
4ebbc975 | 873 | (void)SvPOK_only(targ); |
81e118e0 | 874 | lval = SvUV(sv); |
79072805 | 875 | offset = LvTARGOFF(sv); |
fe58ced6 | 876 | if (offset < 0) |
ed9aa3b7 | 877 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 878 | size = LvTARGLEN(sv); |
8e84507e | 879 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 880 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 881 | |
81e118e0 JH |
882 | offset *= size; /* turn into bit offset */ |
883 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
884 | if (len > targlen) { |
885 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 886 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
887 | SvCUR_set(targ, len); |
888 | } | |
8e84507e | 889 | |
79072805 LW |
890 | if (size < 8) { |
891 | mask = (1 << size) - 1; | |
892 | size = offset & 7; | |
893 | lval &= mask; | |
81e118e0 | 894 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
895 | s[offset] &= ~(mask << size); |
896 | s[offset] |= lval << size; | |
897 | } | |
898 | else { | |
81e118e0 | 899 | offset >>= 3; /* turn into byte offset */ |
79072805 | 900 | if (size == 8) |
eb160463 | 901 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 902 | else if (size == 16) { |
eb160463 GS |
903 | s[offset ] = (U8)((lval >> 8) & 0xff); |
904 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
905 | } |
906 | else if (size == 32) { | |
eb160463 GS |
907 | s[offset ] = (U8)((lval >> 24) & 0xff); |
908 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
909 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
910 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 911 | } |
d7d93a81 | 912 | #ifdef UV_IS_QUAD |
c5a0f51a | 913 | else if (size == 64) { |
c5a0f51a | 914 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 915 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 916 | "Bit vector size > 32 non-portable"); |
eb160463 GS |
917 | s[offset ] = (U8)((lval >> 56) & 0xff); |
918 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
919 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
920 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
921 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
922 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
923 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
924 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 925 | } |
dc1e3f56 | 926 | #endif |
79072805 | 927 | } |
7bb043c3 | 928 | SvSETMAGIC(targ); |
79072805 LW |
929 | } |
930 | ||
931 | void | |
864dbfa3 | 932 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 933 | { |
463ee0b2 | 934 | STRLEN len; |
a0d0e21e | 935 | char *s; |
8e84507e | 936 | |
79072805 | 937 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
938 | register I32 i; |
939 | I32 max; | |
940 | AV* av = (AV*)sv; | |
941 | max = AvFILL(av); | |
942 | for (i = 0; i <= max; i++) { | |
943 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 944 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
945 | do_chop(astr, sv); |
946 | } | |
947 | return; | |
79072805 | 948 | } |
aa854799 | 949 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
950 | HV* hv = (HV*)sv; |
951 | HE* entry; | |
952 | (void)hv_iterinit(hv); | |
953 | /*SUPPRESS 560*/ | |
155aba94 | 954 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
955 | do_chop(astr,hv_iterval(hv,entry)); |
956 | return; | |
79072805 | 957 | } |
991e6d41 NC |
958 | else if (SvREADONLY(sv)) { |
959 | if (SvFAKE(sv)) { | |
960 | /* SV is copy-on-write */ | |
961 | sv_force_normal_flags(sv, 0); | |
962 | } | |
963 | if (SvREADONLY(sv)) | |
964 | Perl_croak(aTHX_ PL_no_modify); | |
965 | } | |
a0d0e21e | 966 | s = SvPV(sv, len); |
748a9306 | 967 | if (len && !SvPOK(sv)) |
a0d0e21e | 968 | s = SvPV_force(sv, len); |
7e2040f0 | 969 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
970 | if (s && len) { |
971 | char *send = s + len; | |
972 | char *start = s; | |
973 | s = send - 1; | |
a0dbb045 JH |
974 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
975 | s--; | |
9041c2e3 | 976 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
977 | sv_setpvn(astr, s, send - s); |
978 | *s = '\0'; | |
979 | SvCUR_set(sv, s - start); | |
980 | SvNIOK_off(sv); | |
981 | SvUTF8_on(astr); | |
982 | } | |
a0ed51b3 LW |
983 | } |
984 | else | |
985 | sv_setpvn(astr, "", 0); | |
986 | } | |
7e2040f0 | 987 | else if (s && len) { |
a0d0e21e LW |
988 | s += --len; |
989 | sv_setpvn(astr, s, 1); | |
990 | *s = '\0'; | |
991 | SvCUR_set(sv, len); | |
2c19a612 | 992 | SvUTF8_off(sv); |
a0d0e21e | 993 | SvNIOK_off(sv); |
79072805 LW |
994 | } |
995 | else | |
a0d0e21e LW |
996 | sv_setpvn(astr, "", 0); |
997 | SvSETMAGIC(sv); | |
7e2040f0 | 998 | } |
a0d0e21e LW |
999 | |
1000 | I32 | |
864dbfa3 | 1001 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 1002 | { |
c07a80fd | 1003 | register I32 count; |
a0d0e21e | 1004 | STRLEN len; |
9b33ce3b | 1005 | STRLEN n_a; |
a0d0e21e | 1006 | char *s; |
c07a80fd | 1007 | |
3280af22 | 1008 | if (RsSNARF(PL_rs)) |
c07a80fd | 1009 | return 0; |
4c5a6083 GS |
1010 | if (RsRECORD(PL_rs)) |
1011 | return 0; | |
c07a80fd | 1012 | count = 0; |
a0d0e21e LW |
1013 | if (SvTYPE(sv) == SVt_PVAV) { |
1014 | register I32 i; | |
1015 | I32 max; | |
1016 | AV* av = (AV*)sv; | |
1017 | max = AvFILL(av); | |
1018 | for (i = 0; i <= max; i++) { | |
1019 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 1020 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1021 | count += do_chomp(sv); |
1022 | } | |
1023 | return count; | |
1024 | } | |
aa854799 | 1025 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
1026 | HV* hv = (HV*)sv; |
1027 | HE* entry; | |
1028 | (void)hv_iterinit(hv); | |
1029 | /*SUPPRESS 560*/ | |
155aba94 | 1030 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1031 | count += do_chomp(hv_iterval(hv,entry)); |
1032 | return count; | |
1033 | } | |
991e6d41 NC |
1034 | else if (SvREADONLY(sv)) { |
1035 | if (SvFAKE(sv)) { | |
1036 | /* SV is copy-on-write */ | |
1037 | sv_force_normal_flags(sv, 0); | |
1038 | } | |
1039 | if (SvREADONLY(sv)) | |
1040 | Perl_croak(aTHX_ PL_no_modify); | |
1041 | } | |
a0d0e21e | 1042 | s = SvPV(sv, len); |
a0d0e21e LW |
1043 | if (s && len) { |
1044 | s += --len; | |
3280af22 | 1045 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1046 | if (*s != '\n') |
1047 | goto nope; | |
1048 | ++count; | |
1049 | while (len && s[-1] == '\n') { | |
1050 | --len; | |
1051 | --s; | |
1052 | ++count; | |
1053 | } | |
1054 | } | |
a0d0e21e | 1055 | else { |
c07a80fd | 1056 | STRLEN rslen; |
3280af22 | 1057 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 1058 | if (rslen == 1) { |
1059 | if (*s != *rsptr) | |
1060 | goto nope; | |
1061 | ++count; | |
1062 | } | |
1063 | else { | |
8c2cee6f | 1064 | if (len < rslen - 1) |
c07a80fd | 1065 | goto nope; |
1066 | len -= rslen - 1; | |
1067 | s -= rslen - 1; | |
36477c24 | 1068 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1069 | goto nope; |
1070 | count += rslen; | |
1071 | } | |
a0d0e21e | 1072 | } |
9b33ce3b | 1073 | s = SvPV_force(sv, n_a); |
a0d0e21e | 1074 | SvCUR_set(sv, len); |
9b33ce3b | 1075 | *SvEND(sv) = '\0'; |
a0d0e21e | 1076 | SvNIOK_off(sv); |
9b33ce3b | 1077 | SvSETMAGIC(sv); |
a0d0e21e LW |
1078 | } |
1079 | nope: | |
a0d0e21e | 1080 | return count; |
8e84507e | 1081 | } |
79072805 LW |
1082 | |
1083 | void | |
864dbfa3 | 1084 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1085 | { |
1086 | #ifdef LIBERAL | |
1087 | register long *dl; | |
1088 | register long *ll; | |
1089 | register long *rl; | |
1090 | #endif | |
1091 | register char *dc; | |
463ee0b2 LW |
1092 | STRLEN leftlen; |
1093 | STRLEN rightlen; | |
7a4c00b4 | 1094 | register char *lc; |
1095 | register char *rc; | |
79072805 | 1096 | register I32 len; |
a0d0e21e | 1097 | I32 lensave; |
7a4c00b4 | 1098 | char *lsave; |
1099 | char *rsave; | |
0c57e439 GS |
1100 | bool left_utf = DO_UTF8(left); |
1101 | bool right_utf = DO_UTF8(right); | |
b7953727 | 1102 | I32 needlen = 0; |
0c57e439 GS |
1103 | |
1104 | if (left_utf && !right_utf) | |
1105 | sv_utf8_upgrade(right); | |
a1ca4561 | 1106 | else if (!left_utf && right_utf) |
0c57e439 | 1107 | sv_utf8_upgrade(left); |
79072805 | 1108 | |
1fbd88dc CS |
1109 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1110 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 1111 | lsave = lc = SvPV(left, leftlen); |
1112 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 1113 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1114 | lensave = len; |
c9b3c8d0 JH |
1115 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1116 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1117 | Newz(801, dc, needlen + 1, char); | |
1118 | } | |
1119 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d GS |
1120 | STRLEN n_a; |
1121 | dc = SvPV_force(sv, n_a); | |
eb160463 GS |
1122 | if (SvCUR(sv) < (STRLEN)len) { |
1123 | dc = SvGROW(sv, (STRLEN)(len + 1)); | |
ff68c719 | 1124 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1125 | } | |
c9b3c8d0 JH |
1126 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1127 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1128 | } |
1129 | else { | |
c9b3c8d0 JH |
1130 | needlen = ((optype == OP_BIT_AND) |
1131 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1132 | Newz(801, dc, needlen + 1, char); |
1133 | (void)sv_usepvn(sv, dc, needlen); | |
1134 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1135 | } |
a0d0e21e LW |
1136 | SvCUR_set(sv, len); |
1137 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1138 | if (left_utf || right_utf) { |
1139 | UV duc, luc, ruc; | |
c9b3c8d0 | 1140 | char *dcsave = dc; |
0c57e439 GS |
1141 | STRLEN lulen = leftlen; |
1142 | STRLEN rulen = rightlen; | |
ba210ebe | 1143 | STRLEN ulen; |
0c57e439 GS |
1144 | |
1145 | switch (optype) { | |
1146 | case OP_BIT_AND: | |
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 | 1156 | } |
c9b3c8d0 JH |
1157 | if (sv == left || sv == right) |
1158 | (void)sv_usepvn(sv, dcsave, needlen); | |
1159 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1160 | break; |
1161 | case OP_BIT_XOR: | |
1162 | while (lulen && rulen) { | |
9041c2e3 | 1163 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1164 | lc += ulen; |
1165 | lulen -= ulen; | |
9041c2e3 | 1166 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1167 | rc += ulen; |
1168 | rulen -= ulen; | |
1169 | duc = luc ^ ruc; | |
9041c2e3 | 1170 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1171 | } |
1172 | goto mop_up_utf; | |
1173 | case OP_BIT_OR: | |
1174 | while (lulen && rulen) { | |
9041c2e3 | 1175 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1176 | lc += ulen; |
1177 | lulen -= ulen; | |
9041c2e3 | 1178 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1179 | rc += ulen; |
1180 | rulen -= ulen; | |
1181 | duc = luc | ruc; | |
9041c2e3 | 1182 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1183 | } |
1184 | mop_up_utf: | |
c9b3c8d0 JH |
1185 | if (sv == left || sv == right) |
1186 | (void)sv_usepvn(sv, dcsave, needlen); | |
1187 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1188 | if (rulen) |
1189 | sv_catpvn(sv, rc, rulen); | |
1190 | else if (lulen) | |
1191 | sv_catpvn(sv, lc, lulen); | |
1192 | else | |
1193 | *SvEND(sv) = '\0'; | |
1194 | break; | |
1195 | } | |
1196 | SvUTF8_on(sv); | |
1197 | goto finish; | |
1198 | } | |
1199 | else | |
79072805 LW |
1200 | #ifdef LIBERAL |
1201 | if (len >= sizeof(long)*4 && | |
1202 | !((long)dc % sizeof(long)) && | |
1203 | !((long)lc % sizeof(long)) && | |
1204 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1205 | { | |
1206 | I32 remainder = len % (sizeof(long)*4); | |
1207 | len /= (sizeof(long)*4); | |
1208 | ||
1209 | dl = (long*)dc; | |
1210 | ll = (long*)lc; | |
1211 | rl = (long*)rc; | |
1212 | ||
1213 | switch (optype) { | |
1214 | case OP_BIT_AND: | |
1215 | while (len--) { | |
1216 | *dl++ = *ll++ & *rl++; | |
1217 | *dl++ = *ll++ & *rl++; | |
1218 | *dl++ = *ll++ & *rl++; | |
1219 | *dl++ = *ll++ & *rl++; | |
1220 | } | |
1221 | break; | |
a0d0e21e | 1222 | case OP_BIT_XOR: |
79072805 LW |
1223 | while (len--) { |
1224 | *dl++ = *ll++ ^ *rl++; | |
1225 | *dl++ = *ll++ ^ *rl++; | |
1226 | *dl++ = *ll++ ^ *rl++; | |
1227 | *dl++ = *ll++ ^ *rl++; | |
1228 | } | |
1229 | break; | |
1230 | case OP_BIT_OR: | |
1231 | while (len--) { | |
1232 | *dl++ = *ll++ | *rl++; | |
1233 | *dl++ = *ll++ | *rl++; | |
1234 | *dl++ = *ll++ | *rl++; | |
1235 | *dl++ = *ll++ | *rl++; | |
1236 | } | |
1237 | } | |
1238 | ||
1239 | dc = (char*)dl; | |
1240 | lc = (char*)ll; | |
1241 | rc = (char*)rl; | |
1242 | ||
1243 | len = remainder; | |
1244 | } | |
1245 | #endif | |
a0d0e21e | 1246 | { |
a0d0e21e LW |
1247 | switch (optype) { |
1248 | case OP_BIT_AND: | |
1249 | while (len--) | |
1250 | *dc++ = *lc++ & *rc++; | |
1251 | break; | |
1252 | case OP_BIT_XOR: | |
1253 | while (len--) | |
1254 | *dc++ = *lc++ ^ *rc++; | |
1255 | goto mop_up; | |
1256 | case OP_BIT_OR: | |
1257 | while (len--) | |
1258 | *dc++ = *lc++ | *rc++; | |
1259 | mop_up: | |
1260 | len = lensave; | |
eb160463 | 1261 | if (rightlen > (STRLEN)len) |
a0d0e21e | 1262 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 | 1263 | else if (leftlen > (STRLEN)len) |
a0d0e21e | 1264 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1265 | else |
1266 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1267 | break; |
1268 | } | |
79072805 | 1269 | } |
0c57e439 | 1270 | finish: |
fb73857a | 1271 | SvTAINT(sv); |
79072805 | 1272 | } |
463ee0b2 LW |
1273 | |
1274 | OP * | |
cea2e8a9 | 1275 | Perl_do_kv(pTHX) |
463ee0b2 | 1276 | { |
39644a26 | 1277 | dSP; |
463ee0b2 | 1278 | HV *hv = (HV*)POPs; |
800e9ae0 | 1279 | HV *keys; |
463ee0b2 | 1280 | register HE *entry; |
463ee0b2 | 1281 | SV *tmpstr; |
54310121 | 1282 | I32 gimme = GIMME_V; |
533c011a NIS |
1283 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1284 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
8e84507e NIS |
1285 | |
1286 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1287 | dokeys = dovalues = TRUE; |
463ee0b2 | 1288 | |
85581909 | 1289 | if (!hv) { |
78f9721b | 1290 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1291 | dTARGET; /* make sure to clear its target here */ |
1292 | if (SvTYPE(TARG) == SVt_PVLV) | |
1293 | LvTARG(TARG) = Nullsv; | |
1294 | PUSHs(TARG); | |
1295 | } | |
463ee0b2 | 1296 | RETURN; |
85581909 | 1297 | } |
748a9306 | 1298 | |
6d822dc4 | 1299 | keys = hv; |
800e9ae0 | 1300 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1301 | |
54310121 | 1302 | if (gimme == G_VOID) |
aa689395 | 1303 | RETURN; |
1304 | ||
54310121 | 1305 | if (gimme == G_SCALAR) { |
6ee623d5 | 1306 | IV i; |
463ee0b2 LW |
1307 | dTARGET; |
1308 | ||
78f9721b | 1309 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1310 | if (SvTYPE(TARG) < SVt_PVLV) { |
1311 | sv_upgrade(TARG, SVt_PVLV); | |
14befaf4 | 1312 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 SB |
1313 | } |
1314 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1315 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1316 | if (LvTARG(TARG)) |
1317 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1318 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1319 | } |
85581909 SB |
1320 | PUSHs(TARG); |
1321 | RETURN; | |
1322 | } | |
1323 | ||
14befaf4 | 1324 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 | 1325 | i = HvKEYS(keys); |
463ee0b2 LW |
1326 | else { |
1327 | i = 0; | |
463ee0b2 | 1328 | /*SUPPRESS 560*/ |
800e9ae0 | 1329 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1330 | } |
1331 | PUSHi( i ); | |
1332 | RETURN; | |
1333 | } | |
1334 | ||
8ed4b672 | 1335 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1336 | |
463ee0b2 | 1337 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1338 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1339 | SPAGAIN; |
574c8022 JH |
1340 | if (dokeys) { |
1341 | SV* sv = hv_iterkeysv(entry); | |
574c8022 JH |
1342 | XPUSHs(sv); /* won't clobber stack_sp */ |
1343 | } | |
463ee0b2 | 1344 | if (dovalues) { |
463ee0b2 | 1345 | PUTBACK; |
6d822dc4 | 1346 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 | 1347 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1348 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1349 | HvMAX(keys)+1, |
1350 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1351 | SPAGAIN; |
46fc3d4c | 1352 | XPUSHs(tmpstr); |
463ee0b2 LW |
1353 | } |
1354 | PUTBACK; | |
1355 | } | |
1356 | return NORMAL; | |
1357 | } | |
4e35701f | 1358 |