Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
4bb101f2 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
af3babe4 | 4 | * 2000, 2001, 2002, 2004, 2005, 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 | ||
166f8a29 DM |
15 | /* This file contains some common functions needed to carry out certain |
16 | * ops. For example both pp_schomp() and pp_chomp() - scalar and array | |
17 | * chomp operations - call the function do_chomp() found in this file. | |
18 | */ | |
19 | ||
79072805 | 20 | #include "EXTERN.h" |
864dbfa3 | 21 | #define PERL_IN_DOOP_C |
79072805 LW |
22 | #include "perl.h" |
23 | ||
64ca3a65 | 24 | #ifndef PERL_MICRO |
79072805 LW |
25 | #include <signal.h> |
26 | #endif | |
27 | ||
942e002e | 28 | STATIC I32 |
b250498f | 29 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 | 30 | { |
4757a243 | 31 | U8 *s; |
b250498f | 32 | U8 *d; |
4757a243 | 33 | U8 *send; |
b250498f | 34 | U8 *dstart; |
4757a243 | 35 | I32 matches = 0; |
9b877dbb | 36 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 | 37 | STRLEN len; |
4757a243 LW |
38 | short *tbl; |
39 | I32 ch; | |
79072805 | 40 | |
4757a243 LW |
41 | tbl = (short*)cPVOP->op_pv; |
42 | if (!tbl) | |
7d85a32c | 43 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 44 | |
4757a243 LW |
45 | s = (U8*)SvPV(sv, len); |
46 | send = s + len; | |
47 | ||
1e54db1a | 48 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 49 | if (!SvUTF8(sv)) { |
01ec43d0 | 50 | while (s < send) { |
036b4402 | 51 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 | 52 | matches++; |
eb160463 | 53 | *s++ = (U8)ch; |
01ec43d0 GS |
54 | } |
55 | else | |
56 | s++; | |
57 | } | |
58 | SvSETMAGIC(sv); | |
b250498f GS |
59 | return matches; |
60 | } | |
4757a243 | 61 | |
b250498f | 62 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb IH |
63 | if (grows) |
64 | New(0, d, len*2+1, U8); | |
65 | else | |
66 | d = s; | |
b250498f GS |
67 | dstart = d; |
68 | while (s < send) { | |
ba210ebe | 69 | STRLEN ulen; |
9b877dbb | 70 | UV c; |
b250498f | 71 | |
b250498f | 72 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 | 73 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb | 74 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f | 75 | matches++; |
9041c2e3 | 76 | d = uvchr_to_utf8(d, ch); |
b250498f | 77 | s += ulen; |
01ec43d0 GS |
78 | } |
79 | else { /* No match -> copy */ | |
0376ff32 | 80 | Move(s, d, ulen, U8); |
9b877dbb IH |
81 | d += ulen; |
82 | s += ulen; | |
b250498f GS |
83 | } |
84 | } | |
9b877dbb IH |
85 | if (grows) { |
86 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
87 | Safefree(dstart); | |
88 | } | |
89 | else { | |
90 | *d = '\0'; | |
91 | SvCUR_set(sv, d - dstart); | |
92 | } | |
b250498f | 93 | SvUTF8_on(sv); |
b250498f | 94 | SvSETMAGIC(sv); |
4757a243 LW |
95 | return matches; |
96 | } | |
97 | ||
942e002e | 98 | STATIC I32 |
21304a3b | 99 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 | 100 | { |
4757a243 LW |
101 | U8 *s; |
102 | U8 *send; | |
103 | I32 matches = 0; | |
104 | STRLEN len; | |
105 | short *tbl; | |
8973db79 | 106 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 LW |
107 | |
108 | tbl = (short*)cPVOP->op_pv; | |
109 | if (!tbl) | |
7d85a32c | 110 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 LW |
111 | |
112 | s = (U8*)SvPV(sv, len); | |
113 | send = s + len; | |
114 | ||
1aa99e6b IH |
115 | if (!SvUTF8(sv)) |
116 | while (s < send) { | |
117 | if (tbl[*s++] >= 0) | |
036b4402 | 118 | matches++; |
1aa99e6b IH |
119 | } |
120 | else | |
121 | while (s < send) { | |
122 | UV c; | |
123 | STRLEN ulen; | |
9041c2e3 | 124 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 JH |
125 | if (c < 0x100) { |
126 | if (tbl[c] >= 0) | |
127 | matches++; | |
128 | } else if (complement) | |
1aa99e6b IH |
129 | matches++; |
130 | s += ulen; | |
131 | } | |
4757a243 LW |
132 | |
133 | return matches; | |
134 | } | |
135 | ||
942e002e | 136 | STATIC I32 |
21304a3b | 137 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 | 138 | { |
4757a243 LW |
139 | U8 *s; |
140 | U8 *send; | |
141 | U8 *d; | |
5e44153e | 142 | U8 *dstart; |
1aa99e6b | 143 | I32 isutf8; |
4757a243 | 144 | I32 matches = 0; |
9b877dbb | 145 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 JH |
146 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
147 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
b7953727 | 148 | STRLEN len, rlen = 0; |
4757a243 LW |
149 | short *tbl; |
150 | I32 ch; | |
151 | ||
152 | tbl = (short*)cPVOP->op_pv; | |
153 | if (!tbl) | |
7d85a32c | 154 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 LW |
155 | |
156 | s = (U8*)SvPV(sv, len); | |
1aa99e6b | 157 | isutf8 = SvUTF8(sv); |
4757a243 LW |
158 | send = s + len; |
159 | ||
1aa99e6b IH |
160 | if (!isutf8) { |
161 | dstart = d = s; | |
162 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
163 | U8* p = send; | |
164 | while (s < send) { | |
165 | if ((ch = tbl[*s]) >= 0) { | |
eb160463 | 166 | *d = (U8)ch; |
036b4402 | 167 | matches++; |
1aa99e6b IH |
168 | if (p != d - 1 || *p != *d) |
169 | p = d++; | |
170 | } | |
171 | else if (ch == -1) /* -1 is unmapped character */ | |
172 | *d++ = *s; | |
173 | else if (ch == -2) /* -2 is delete character */ | |
174 | matches++; | |
175 | s++; | |
176 | } | |
a0ed51b3 | 177 | } |
1aa99e6b IH |
178 | else { |
179 | while (s < send) { | |
180 | if ((ch = tbl[*s]) >= 0) { | |
181 | matches++; | |
eb160463 | 182 | *d++ = (U8)ch; |
1aa99e6b IH |
183 | } |
184 | else if (ch == -1) /* -1 is unmapped character */ | |
185 | *d++ = *s; | |
186 | else if (ch == -2) /* -2 is delete character */ | |
187 | matches++; | |
188 | s++; | |
189 | } | |
190 | } | |
76ef7183 | 191 | *d = '\0'; |
1aa99e6b | 192 | SvCUR_set(sv, d - dstart); |
4757a243 | 193 | } |
1aa99e6b | 194 | else { /* isutf8 */ |
9b877dbb IH |
195 | if (grows) |
196 | New(0, d, len*2+1, U8); | |
197 | else | |
198 | d = s; | |
1aa99e6b | 199 | dstart = d; |
8973db79 JH |
200 | if (complement && !del) |
201 | rlen = tbl[0x100]; | |
1aa99e6b | 202 | |
4aecb5b5 JH |
203 | #ifdef MACOS_TRADITIONAL |
204 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
205 | #endif | |
206 | ||
1aa99e6b | 207 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
208 | UV pch = 0xfeedface; |
209 | while (s < send) { | |
210 | STRLEN len; | |
9041c2e3 | 211 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b | 212 | |
8973db79 JH |
213 | if (comp > 0xff) { |
214 | if (!complement) { | |
215 | Copy(s, d, len, U8); | |
216 | d += len; | |
217 | } | |
218 | else { | |
219 | matches++; | |
220 | if (!del) { | |
05d340b8 JH |
221 | ch = (rlen == 0) ? comp : |
222 | (comp - 0x100 < rlen) ? | |
8973db79 | 223 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 224 | if ((UV)ch != pch) { |
9041c2e3 | 225 | d = uvchr_to_utf8(d, ch); |
eb160463 | 226 | pch = (UV)ch; |
8973db79 JH |
227 | } |
228 | s += len; | |
229 | continue; | |
230 | } | |
231 | } | |
9b877dbb | 232 | } |
1aa99e6b IH |
233 | else if ((ch = tbl[comp]) >= 0) { |
234 | matches++; | |
eb160463 | 235 | if ((UV)ch != pch) { |
9041c2e3 | 236 | d = uvchr_to_utf8(d, ch); |
eb160463 | 237 | pch = (UV)ch; |
1aa99e6b IH |
238 | } |
239 | s += len; | |
240 | continue; | |
241 | } | |
9b877dbb IH |
242 | else if (ch == -1) { /* -1 is unmapped character */ |
243 | Copy(s, d, len, U8); | |
244 | d += len; | |
245 | } | |
1aa99e6b IH |
246 | else if (ch == -2) /* -2 is delete character */ |
247 | matches++; | |
248 | s += len; | |
249 | pch = 0xfeedface; | |
250 | } | |
5d06d08e | 251 | } |
1aa99e6b IH |
252 | else { |
253 | while (s < send) { | |
254 | STRLEN len; | |
9041c2e3 | 255 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
256 | if (comp > 0xff) { |
257 | if (!complement) { | |
0376ff32 | 258 | Move(s, d, len, U8); |
8973db79 JH |
259 | d += len; |
260 | } | |
261 | else { | |
262 | matches++; | |
263 | if (!del) { | |
9041c2e3 NIS |
264 | if (comp - 0x100 < rlen) |
265 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 266 | else |
9041c2e3 | 267 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
268 | } |
269 | } | |
9b877dbb | 270 | } |
1aa99e6b | 271 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 272 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
273 | matches++; |
274 | } | |
275 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
276 | Copy(s, d, len, U8); |
277 | d += len; | |
1aa99e6b IH |
278 | } |
279 | else if (ch == -2) /* -2 is delete character */ | |
280 | matches++; | |
281 | s += len; | |
282 | } | |
283 | } | |
9b877dbb IH |
284 | if (grows) { |
285 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
286 | Safefree(dstart); | |
287 | } | |
288 | else { | |
289 | *d = '\0'; | |
290 | SvCUR_set(sv, d - dstart); | |
291 | } | |
1aa99e6b | 292 | SvUTF8_on(sv); |
4757a243 | 293 | } |
5e44153e | 294 | SvSETMAGIC(sv); |
4757a243 LW |
295 | return matches; |
296 | } | |
297 | ||
942e002e | 298 | STATIC I32 |
21304a3b | 299 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 | 300 | { |
4757a243 LW |
301 | U8 *s; |
302 | U8 *send; | |
303 | U8 *d; | |
036b4402 | 304 | U8 *start; |
1aa99e6b | 305 | U8 *dstart, *dend; |
4757a243 | 306 | I32 matches = 0; |
9b877dbb | 307 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
308 | STRLEN len; |
309 | ||
310 | SV* rv = (SV*)cSVOP->op_sv; | |
311 | HV* hv = (HV*)SvRV(rv); | |
312 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
313 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
314 | UV extra = none + 1; | |
b7953727 | 315 | UV final = 0; |
4757a243 | 316 | UV uv; |
1aa99e6b IH |
317 | I32 isutf8; |
318 | U8 hibit = 0; | |
4757a243 LW |
319 | |
320 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
321 | isutf8 = SvUTF8(sv); |
322 | if (!isutf8) { | |
323 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
324 | while (t < e) { |
325 | U8 ch = *t++; | |
326 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 327 | break; |
c4d5f83a | 328 | } |
1aa99e6b IH |
329 | if (hibit) |
330 | s = bytes_to_utf8(s, &len); | |
331 | } | |
4757a243 | 332 | send = s + len; |
036b4402 | 333 | start = s; |
4757a243 LW |
334 | |
335 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
336 | if (svp) | |
337 | final = SvUV(*svp); | |
338 | ||
9b877dbb IH |
339 | if (grows) { |
340 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
89ebb4a3 | 341 | New(0, d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
342 | dend = d + len * 3; |
343 | dstart = d; | |
344 | } | |
345 | else { | |
346 | dstart = d = s; | |
347 | dend = d + len; | |
348 | } | |
1aa99e6b | 349 | |
4757a243 | 350 | while (s < send) { |
3568d838 | 351 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 LW |
352 | s += UTF8SKIP(s); |
353 | matches++; | |
7d85a32c | 354 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
355 | } |
356 | else if (uv == none) { | |
1aa99e6b | 357 | int i = UTF8SKIP(s); |
0376ff32 | 358 | Move(s, d, i, U8); |
9b877dbb IH |
359 | d += i; |
360 | s += i; | |
4757a243 LW |
361 | } |
362 | else if (uv == extra) { | |
1aa99e6b | 363 | int i = UTF8SKIP(s); |
036b4402 | 364 | s += i; |
4757a243 | 365 | matches++; |
7d85a32c | 366 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
367 | } |
368 | else | |
369 | s += UTF8SKIP(s); | |
1aa99e6b | 370 | |
9b877dbb | 371 | if (d > dend) { |
1aa99e6b | 372 | STRLEN clen = d - dstart; |
89ebb4a3 | 373 | STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb | 374 | if (!grows) |
7d85a32c | 375 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 | 376 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
377 | d = dstart + clen; |
378 | dend = dstart + nlen; | |
379 | } | |
4757a243 | 380 | } |
16ec844d | 381 | if (grows || hibit) { |
9b877dbb IH |
382 | sv_setpvn(sv, (char*)dstart, d - dstart); |
383 | Safefree(dstart); | |
16ec844d IH |
384 | if (grows && hibit) |
385 | Safefree(start); | |
9b877dbb IH |
386 | } |
387 | else { | |
388 | *d = '\0'; | |
389 | SvCUR_set(sv, d - dstart); | |
390 | } | |
4757a243 | 391 | SvSETMAGIC(sv); |
1aa99e6b | 392 | SvUTF8_on(sv); |
4757a243 LW |
393 | |
394 | return matches; | |
395 | } | |
396 | ||
942e002e | 397 | STATIC I32 |
21304a3b | 398 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 | 399 | { |
4757a243 | 400 | U8 *s; |
b7953727 | 401 | U8 *start = 0, *send; |
4757a243 LW |
402 | I32 matches = 0; |
403 | STRLEN len; | |
404 | ||
405 | SV* rv = (SV*)cSVOP->op_sv; | |
406 | HV* hv = (HV*)SvRV(rv); | |
407 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
408 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
45005bfb | 409 | UV extra = none + 1; |
4757a243 | 410 | UV uv; |
1aa99e6b | 411 | U8 hibit = 0; |
4757a243 LW |
412 | |
413 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
414 | if (!SvUTF8(sv)) { |
415 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
416 | while (t < e) { |
417 | U8 ch = *t++; | |
418 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 419 | break; |
c4d5f83a | 420 | } |
1aa99e6b IH |
421 | if (hibit) |
422 | start = s = bytes_to_utf8(s, &len); | |
423 | } | |
4757a243 LW |
424 | send = s + len; |
425 | ||
426 | while (s < send) { | |
3568d838 | 427 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 | 428 | matches++; |
834a4ddd | 429 | s += UTF8SKIP(s); |
4757a243 | 430 | } |
1aa99e6b IH |
431 | if (hibit) |
432 | Safefree(start); | |
4757a243 LW |
433 | |
434 | return matches; | |
435 | } | |
436 | ||
942e002e | 437 | STATIC I32 |
21304a3b | 438 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 | 439 | { |
4757a243 | 440 | U8 *s; |
1aa99e6b | 441 | U8 *start, *send; |
4757a243 LW |
442 | U8 *d; |
443 | I32 matches = 0; | |
444 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 445 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 446 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
447 | SV* rv = (SV*)cSVOP->op_sv; |
448 | HV* hv = (HV*)SvRV(rv); | |
449 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
450 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
451 | UV extra = none + 1; | |
b7953727 | 452 | UV final = 0; |
45005bfb | 453 | bool havefinal = FALSE; |
4757a243 LW |
454 | UV uv; |
455 | STRLEN len; | |
1aa99e6b IH |
456 | U8 *dstart, *dend; |
457 | I32 isutf8; | |
458 | U8 hibit = 0; | |
4757a243 LW |
459 | |
460 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
461 | isutf8 = SvUTF8(sv); |
462 | if (!isutf8) { | |
463 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
464 | while (t < e) { |
465 | U8 ch = *t++; | |
466 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 467 | break; |
c4d5f83a | 468 | } |
1aa99e6b IH |
469 | if (hibit) |
470 | s = bytes_to_utf8(s, &len); | |
471 | } | |
4757a243 | 472 | send = s + len; |
1aa99e6b | 473 | start = s; |
4757a243 LW |
474 | |
475 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
45005bfb | 476 | if (svp) { |
4757a243 | 477 | final = SvUV(*svp); |
45005bfb JH |
478 | havefinal = TRUE; |
479 | } | |
4757a243 | 480 | |
9b877dbb IH |
481 | if (grows) { |
482 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
89ebb4a3 | 483 | New(0, d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
484 | dend = d + len * 3; |
485 | dstart = d; | |
486 | } | |
487 | else { | |
488 | dstart = d = s; | |
489 | dend = d + len; | |
490 | } | |
4757a243 LW |
491 | |
492 | if (squash) { | |
493 | UV puv = 0xfeedface; | |
494 | while (s < send) { | |
3568d838 | 495 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 496 | |
9b877dbb IH |
497 | if (d > dend) { |
498 | STRLEN clen = d - dstart; | |
89ebb4a3 | 499 | STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb | 500 | if (!grows) |
7d85a32c | 501 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 502 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
503 | d = dstart + clen; |
504 | dend = dstart + nlen; | |
4757a243 LW |
505 | } |
506 | if (uv < none) { | |
507 | matches++; | |
94472101 | 508 | s += UTF8SKIP(s); |
4757a243 | 509 | if (uv != puv) { |
7d85a32c | 510 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
511 | puv = uv; |
512 | } | |
4757a243 LW |
513 | continue; |
514 | } | |
515 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 516 | int i = UTF8SKIP(s); |
0376ff32 | 517 | Move(s, d, i, U8); |
9b877dbb IH |
518 | d += i; |
519 | s += i; | |
4757a243 LW |
520 | puv = 0xfeedface; |
521 | continue; | |
a0ed51b3 | 522 | } |
4757a243 LW |
523 | else if (uv == extra && !del) { |
524 | matches++; | |
45005bfb JH |
525 | if (havefinal) { |
526 | s += UTF8SKIP(s); | |
527 | if (puv != final) { | |
7d85a32c | 528 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
529 | puv = final; |
530 | } | |
531 | } | |
532 | else { | |
533 | STRLEN len; | |
7d85a32c | 534 | uv = utf8_to_uvuni(s, &len); |
45005bfb | 535 | if (uv != puv) { |
0376ff32 | 536 | Move(s, d, len, U8); |
45005bfb JH |
537 | d += len; |
538 | puv = uv; | |
539 | } | |
540 | s += len; | |
4757a243 | 541 | } |
4757a243 LW |
542 | continue; |
543 | } | |
01ec43d0 GS |
544 | matches++; /* "none+1" is delete character */ |
545 | s += UTF8SKIP(s); | |
a0ed51b3 | 546 | } |
79072805 LW |
547 | } |
548 | else { | |
4757a243 | 549 | while (s < send) { |
3568d838 | 550 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb IH |
551 | if (d > dend) { |
552 | STRLEN clen = d - dstart; | |
89ebb4a3 | 553 | STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb | 554 | if (!grows) |
7d85a32c | 555 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 556 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
557 | d = dstart + clen; |
558 | dend = dstart + nlen; | |
a0ed51b3 | 559 | } |
4757a243 LW |
560 | if (uv < none) { |
561 | matches++; | |
01ec43d0 | 562 | s += UTF8SKIP(s); |
7d85a32c | 563 | d = uvuni_to_utf8(d, uv); |
4757a243 | 564 | continue; |
a0ed51b3 | 565 | } |
4757a243 | 566 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 567 | int i = UTF8SKIP(s); |
0376ff32 | 568 | Move(s, d, i, U8); |
9b877dbb IH |
569 | d += i; |
570 | s += i; | |
4757a243 | 571 | continue; |
79072805 | 572 | } |
4757a243 LW |
573 | else if (uv == extra && !del) { |
574 | matches++; | |
01ec43d0 | 575 | s += UTF8SKIP(s); |
7d85a32c | 576 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
577 | continue; |
578 | } | |
01ec43d0 GS |
579 | matches++; /* "none+1" is delete character */ |
580 | s += UTF8SKIP(s); | |
79072805 | 581 | } |
4757a243 | 582 | } |
16ec844d | 583 | if (grows || hibit) { |
9b877dbb IH |
584 | sv_setpvn(sv, (char*)dstart, d - dstart); |
585 | Safefree(dstart); | |
16ec844d IH |
586 | if (grows && hibit) |
587 | Safefree(start); | |
9b877dbb IH |
588 | } |
589 | else { | |
590 | *d = '\0'; | |
591 | SvCUR_set(sv, d - dstart); | |
592 | } | |
1aa99e6b | 593 | SvUTF8_on(sv); |
4757a243 LW |
594 | SvSETMAGIC(sv); |
595 | ||
596 | return matches; | |
597 | } | |
598 | ||
599 | I32 | |
864dbfa3 | 600 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
601 | { |
602 | STRLEN len; | |
8e84507e | 603 | I32 hasutf = (PL_op->op_private & |
036b4402 | 604 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 605 | |
2233f375 | 606 | if (SvREADONLY(sv)) { |
765f542d NC |
607 | if (SvIsCOW(sv)) |
608 | sv_force_normal_flags(sv, 0); | |
2233f375 NC |
609 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
610 | Perl_croak(aTHX_ PL_no_modify); | |
611 | } | |
4757a243 LW |
612 | (void)SvPV(sv, len); |
613 | if (!len) | |
614 | return 0; | |
d59e14db RGS |
615 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
616 | if (!SvPOKp(sv)) | |
617 | (void)SvPV_force(sv, len); | |
2de7b02f | 618 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 619 | } |
4757a243 | 620 | |
cea2e8a9 | 621 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 622 | |
a77e643a RGS |
623 | switch (PL_op->op_private & ~hasutf & ( |
624 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| | |
625 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
4757a243 | 626 | case 0: |
01ec43d0 GS |
627 | if (hasutf) |
628 | return do_trans_simple_utf8(sv); | |
629 | else | |
630 | return do_trans_simple(sv); | |
4757a243 LW |
631 | |
632 | case OPpTRANS_IDENTICAL: | |
05d340b8 | 633 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 GS |
634 | if (hasutf) |
635 | return do_trans_count_utf8(sv); | |
636 | else | |
637 | return do_trans_count(sv); | |
4757a243 LW |
638 | |
639 | default: | |
01ec43d0 | 640 | if (hasutf) |
036b4402 | 641 | return do_trans_complex_utf8(sv); |
4757a243 | 642 | else |
036b4402 | 643 | return do_trans_complex(sv); |
79072805 | 644 | } |
79072805 LW |
645 | } |
646 | ||
647 | void | |
864dbfa3 | 648 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
649 | { |
650 | SV **oldmark = mark; | |
651 | register I32 items = sp - mark; | |
79072805 | 652 | register STRLEN len; |
463ee0b2 | 653 | STRLEN delimlen; |
463ee0b2 | 654 | STRLEN tmplen; |
79072805 | 655 | |
5ba99574 NC |
656 | (void) SvPV(del, delimlen); /* stringify and get the delimlen */ |
657 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ | |
516a5887 | 658 | |
79072805 LW |
659 | mark++; |
660 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 661 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
662 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
663 | while (items-- > 0) { | |
1426bbf4 | 664 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 LW |
665 | SvPV(*mark, tmplen); |
666 | len += tmplen; | |
79072805 LW |
667 | } |
668 | mark++; | |
669 | } | |
670 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
671 | ||
672 | mark = oldmark; | |
db7c17d7 | 673 | items = sp - mark; |
79072805 LW |
674 | ++mark; |
675 | } | |
676 | ||
fc0199cd | 677 | sv_setpvn(sv, "", 0); |
e4803c42 | 678 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 679 | SvUTF8_off(sv); |
e4803c42 | 680 | |
8d6d96c1 HS |
681 | if (PL_tainting && SvMAGICAL(sv)) |
682 | SvTAINTED_off(sv); | |
683 | ||
463ee0b2 | 684 | if (items-- > 0) { |
92d29cee JH |
685 | if (*mark) |
686 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
687 | mark++; |
688 | } | |
8d6d96c1 | 689 | |
c512ce4f | 690 | if (delimlen) { |
79072805 | 691 | for (; items > 0; items--,mark++) { |
c512ce4f | 692 | sv_catsv(sv,del); |
79072805 LW |
693 | sv_catsv(sv,*mark); |
694 | } | |
695 | } | |
696 | else { | |
697 | for (; items > 0; items--,mark++) | |
698 | sv_catsv(sv,*mark); | |
699 | } | |
700 | SvSETMAGIC(sv); | |
701 | } | |
702 | ||
703 | void | |
864dbfa3 | 704 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 705 | { |
46fc3d4c | 706 | STRLEN patlen; |
707 | char *pat = SvPV(*sarg, patlen); | |
708 | bool do_taint = FALSE; | |
709 | ||
5b781b5b | 710 | SvUTF8_off(sv); |
2cf2cfc6 A |
711 | if (DO_UTF8(*sarg)) |
712 | SvUTF8_on(sv); | |
46fc3d4c | 713 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 | 714 | SvSETMAGIC(sv); |
46fc3d4c | 715 | if (do_taint) |
716 | SvTAINTED_on(sv); | |
79072805 LW |
717 | } |
718 | ||
33b45480 | 719 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
720 | UV |
721 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
722 | { | |
723 | STRLEN srclen, len; | |
724 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
725 | UV retnum = 0; | |
726 | ||
a50d7633 | 727 | if (offset < 0) |
81e118e0 | 728 | return retnum; |
8e84507e | 729 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 730 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 731 | |
dcad2880 | 732 | if (SvUTF8(sv)) |
33b45480 | 733 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 734 | |
81e118e0 JH |
735 | offset *= size; /* turn into bit offset */ |
736 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
737 | if (len > srclen) { | |
738 | if (size <= 8) | |
739 | retnum = 0; | |
740 | else { | |
741 | offset >>= 3; /* turn into byte offset */ | |
742 | if (size == 16) { | |
eb160463 | 743 | if ((STRLEN)offset >= srclen) |
81e118e0 JH |
744 | retnum = 0; |
745 | else | |
628e1a40 | 746 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
747 | } |
748 | else if (size == 32) { | |
eb160463 | 749 | if ((STRLEN)offset >= srclen) |
81e118e0 | 750 | retnum = 0; |
eb160463 | 751 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 JH |
752 | retnum = |
753 | ((UV) s[offset ] << 24); | |
eb160463 | 754 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 JH |
755 | retnum = |
756 | ((UV) s[offset ] << 24) + | |
757 | ((UV) s[offset + 1] << 16); | |
758 | else | |
759 | retnum = | |
760 | ((UV) s[offset ] << 24) + | |
761 | ((UV) s[offset + 1] << 16) + | |
762 | ( s[offset + 2] << 8); | |
763 | } | |
d7d93a81 | 764 | #ifdef UV_IS_QUAD |
c5a0f51a | 765 | else if (size == 64) { |
c5a0f51a | 766 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 767 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
768 | "Bit vector size > 32 non-portable"); |
769 | if (offset >= srclen) | |
770 | retnum = 0; | |
771 | else if (offset + 1 >= srclen) | |
772 | retnum = | |
773 | (UV) s[offset ] << 56; | |
774 | else if (offset + 2 >= srclen) | |
775 | retnum = | |
776 | ((UV) s[offset ] << 56) + | |
777 | ((UV) s[offset + 1] << 48); | |
778 | else if (offset + 3 >= srclen) | |
779 | retnum = | |
780 | ((UV) s[offset ] << 56) + | |
781 | ((UV) s[offset + 1] << 48) + | |
782 | ((UV) s[offset + 2] << 40); | |
783 | else if (offset + 4 >= srclen) | |
784 | retnum = | |
785 | ((UV) s[offset ] << 56) + | |
786 | ((UV) s[offset + 1] << 48) + | |
787 | ((UV) s[offset + 2] << 40) + | |
788 | ((UV) s[offset + 3] << 32); | |
789 | else if (offset + 5 >= srclen) | |
790 | retnum = | |
791 | ((UV) s[offset ] << 56) + | |
792 | ((UV) s[offset + 1] << 48) + | |
793 | ((UV) s[offset + 2] << 40) + | |
794 | ((UV) s[offset + 3] << 32) + | |
795 | ( s[offset + 4] << 24); | |
796 | else if (offset + 6 >= srclen) | |
797 | retnum = | |
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); | |
804 | else | |
8e84507e | 805 | retnum = |
c5a0f51a JH |
806 | ((UV) s[offset ] << 56) + |
807 | ((UV) s[offset + 1] << 48) + | |
808 | ((UV) s[offset + 2] << 40) + | |
809 | ((UV) s[offset + 3] << 32) + | |
810 | ((UV) s[offset + 4] << 24) + | |
811 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 812 | ( s[offset + 6] << 8); |
c5a0f51a JH |
813 | } |
814 | #endif | |
81e118e0 JH |
815 | } |
816 | } | |
817 | else if (size < 8) | |
818 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
819 | else { | |
820 | offset >>= 3; /* turn into byte offset */ | |
821 | if (size == 8) | |
822 | retnum = s[offset]; | |
823 | else if (size == 16) | |
824 | retnum = | |
628e1a40 | 825 | ((UV) s[offset] << 8) + |
81e118e0 JH |
826 | s[offset + 1]; |
827 | else if (size == 32) | |
828 | retnum = | |
829 | ((UV) s[offset ] << 24) + | |
830 | ((UV) s[offset + 1] << 16) + | |
831 | ( s[offset + 2] << 8) + | |
832 | s[offset + 3]; | |
d7d93a81 | 833 | #ifdef UV_IS_QUAD |
c5a0f51a | 834 | else if (size == 64) { |
c5a0f51a | 835 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 836 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
837 | "Bit vector size > 32 non-portable"); |
838 | retnum = | |
839 | ((UV) s[offset ] << 56) + | |
840 | ((UV) s[offset + 1] << 48) + | |
841 | ((UV) s[offset + 2] << 40) + | |
842 | ((UV) s[offset + 3] << 32) + | |
843 | ((UV) s[offset + 4] << 24) + | |
844 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 845 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
846 | s[offset + 7]; |
847 | } | |
848 | #endif | |
81e118e0 JH |
849 | } |
850 | ||
851 | return retnum; | |
852 | } | |
853 | ||
33b45480 SB |
854 | /* currently converts input to bytes if possible but doesn't sweat failures, |
855 | * although it does ensure that the string it clobbers is not marked as | |
856 | * utf8-valid any more | |
857 | */ | |
79072805 | 858 | void |
864dbfa3 | 859 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
860 | { |
861 | SV *targ = LvTARG(sv); | |
862 | register I32 offset; | |
863 | register I32 size; | |
8990e307 | 864 | register unsigned char *s; |
81e118e0 | 865 | register UV lval; |
79072805 | 866 | I32 mask; |
a0d0e21e LW |
867 | STRLEN targlen; |
868 | STRLEN len; | |
79072805 | 869 | |
8990e307 LW |
870 | if (!targ) |
871 | return; | |
a0d0e21e | 872 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 873 | if (SvUTF8(targ)) { |
33b45480 SB |
874 | /* This is handled by the SvPOK_only below... |
875 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
876 | SvUTF8_off(targ); | |
877 | */ | |
878 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
879 | } |
880 | ||
4ebbc975 | 881 | (void)SvPOK_only(targ); |
81e118e0 | 882 | lval = SvUV(sv); |
79072805 | 883 | offset = LvTARGOFF(sv); |
fe58ced6 | 884 | if (offset < 0) |
ed9aa3b7 | 885 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 886 | size = LvTARGLEN(sv); |
8e84507e | 887 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 888 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 889 | |
81e118e0 JH |
890 | offset *= size; /* turn into bit offset */ |
891 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
892 | if (len > targlen) { |
893 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 894 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
895 | SvCUR_set(targ, len); |
896 | } | |
8e84507e | 897 | |
79072805 LW |
898 | if (size < 8) { |
899 | mask = (1 << size) - 1; | |
900 | size = offset & 7; | |
901 | lval &= mask; | |
81e118e0 | 902 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
903 | s[offset] &= ~(mask << size); |
904 | s[offset] |= lval << size; | |
905 | } | |
906 | else { | |
81e118e0 | 907 | offset >>= 3; /* turn into byte offset */ |
79072805 | 908 | if (size == 8) |
eb160463 | 909 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 910 | else if (size == 16) { |
eb160463 GS |
911 | s[offset ] = (U8)((lval >> 8) & 0xff); |
912 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
913 | } |
914 | else if (size == 32) { | |
eb160463 GS |
915 | s[offset ] = (U8)((lval >> 24) & 0xff); |
916 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
917 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
918 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 919 | } |
d7d93a81 | 920 | #ifdef UV_IS_QUAD |
c5a0f51a | 921 | else if (size == 64) { |
c5a0f51a | 922 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 923 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 924 | "Bit vector size > 32 non-portable"); |
eb160463 GS |
925 | s[offset ] = (U8)((lval >> 56) & 0xff); |
926 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
927 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
928 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
929 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
930 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
931 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
932 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 933 | } |
dc1e3f56 | 934 | #endif |
79072805 | 935 | } |
7bb043c3 | 936 | SvSETMAGIC(targ); |
79072805 LW |
937 | } |
938 | ||
939 | void | |
864dbfa3 | 940 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 941 | { |
463ee0b2 | 942 | STRLEN len; |
a0d0e21e | 943 | char *s; |
8e84507e | 944 | |
79072805 | 945 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
946 | register I32 i; |
947 | I32 max; | |
948 | AV* av = (AV*)sv; | |
949 | max = AvFILL(av); | |
950 | for (i = 0; i <= max; i++) { | |
951 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 952 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
953 | do_chop(astr, sv); |
954 | } | |
955 | return; | |
79072805 | 956 | } |
aa854799 | 957 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
958 | HV* hv = (HV*)sv; |
959 | HE* entry; | |
960 | (void)hv_iterinit(hv); | |
961 | /*SUPPRESS 560*/ | |
155aba94 | 962 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
963 | do_chop(astr,hv_iterval(hv,entry)); |
964 | return; | |
79072805 | 965 | } |
991e6d41 NC |
966 | else if (SvREADONLY(sv)) { |
967 | if (SvFAKE(sv)) { | |
968 | /* SV is copy-on-write */ | |
969 | sv_force_normal_flags(sv, 0); | |
970 | } | |
971 | if (SvREADONLY(sv)) | |
972 | Perl_croak(aTHX_ PL_no_modify); | |
973 | } | |
a0d0e21e | 974 | s = SvPV(sv, len); |
748a9306 | 975 | if (len && !SvPOK(sv)) |
a0d0e21e | 976 | s = SvPV_force(sv, len); |
7e2040f0 | 977 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
978 | if (s && len) { |
979 | char *send = s + len; | |
980 | char *start = s; | |
981 | s = send - 1; | |
a0dbb045 JH |
982 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
983 | s--; | |
9041c2e3 | 984 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
985 | sv_setpvn(astr, s, send - s); |
986 | *s = '\0'; | |
987 | SvCUR_set(sv, s - start); | |
988 | SvNIOK_off(sv); | |
989 | SvUTF8_on(astr); | |
990 | } | |
a0ed51b3 LW |
991 | } |
992 | else | |
993 | sv_setpvn(astr, "", 0); | |
994 | } | |
7e2040f0 | 995 | else if (s && len) { |
a0d0e21e LW |
996 | s += --len; |
997 | sv_setpvn(astr, s, 1); | |
998 | *s = '\0'; | |
999 | SvCUR_set(sv, len); | |
2c19a612 | 1000 | SvUTF8_off(sv); |
a0d0e21e | 1001 | SvNIOK_off(sv); |
79072805 LW |
1002 | } |
1003 | else | |
a0d0e21e LW |
1004 | sv_setpvn(astr, "", 0); |
1005 | SvSETMAGIC(sv); | |
7e2040f0 | 1006 | } |
a0d0e21e LW |
1007 | |
1008 | I32 | |
864dbfa3 | 1009 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 1010 | { |
c07a80fd | 1011 | register I32 count; |
a0d0e21e | 1012 | STRLEN len; |
9b33ce3b | 1013 | STRLEN n_a; |
a0d0e21e | 1014 | char *s; |
c4c87a06 | 1015 | char *temp_buffer = NULL; |
a6aa349d | 1016 | SV* svrecode = Nullsv; |
c07a80fd | 1017 | |
3280af22 | 1018 | if (RsSNARF(PL_rs)) |
c07a80fd | 1019 | return 0; |
4c5a6083 GS |
1020 | if (RsRECORD(PL_rs)) |
1021 | return 0; | |
c07a80fd | 1022 | count = 0; |
a0d0e21e LW |
1023 | if (SvTYPE(sv) == SVt_PVAV) { |
1024 | register I32 i; | |
1025 | I32 max; | |
1026 | AV* av = (AV*)sv; | |
1027 | max = AvFILL(av); | |
1028 | for (i = 0; i <= max; i++) { | |
1029 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 1030 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1031 | count += do_chomp(sv); |
1032 | } | |
1033 | return count; | |
1034 | } | |
aa854799 | 1035 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
1036 | HV* hv = (HV*)sv; |
1037 | HE* entry; | |
1038 | (void)hv_iterinit(hv); | |
1039 | /*SUPPRESS 560*/ | |
155aba94 | 1040 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1041 | count += do_chomp(hv_iterval(hv,entry)); |
1042 | return count; | |
1043 | } | |
991e6d41 NC |
1044 | else if (SvREADONLY(sv)) { |
1045 | if (SvFAKE(sv)) { | |
1046 | /* SV is copy-on-write */ | |
1047 | sv_force_normal_flags(sv, 0); | |
1048 | } | |
1049 | if (SvREADONLY(sv)) | |
1050 | Perl_croak(aTHX_ PL_no_modify); | |
1051 | } | |
a6aa349d TS |
1052 | |
1053 | if (PL_encoding) { | |
1054 | if (!SvUTF8(sv)) { | |
1055 | /* XXX, here sv is utf8-ized as a side-effect! | |
1056 | If encoding.pm is used properly, almost string-generating | |
1057 | operations, including literal strings, chr(), input data, etc. | |
1058 | should have been utf8-ized already, right? | |
1059 | */ | |
1060 | sv_recode_to_utf8(sv, PL_encoding); | |
1061 | } | |
1062 | } | |
1063 | ||
a0d0e21e | 1064 | s = SvPV(sv, len); |
a0d0e21e LW |
1065 | if (s && len) { |
1066 | s += --len; | |
3280af22 | 1067 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1068 | if (*s != '\n') |
1069 | goto nope; | |
1070 | ++count; | |
1071 | while (len && s[-1] == '\n') { | |
1072 | --len; | |
1073 | --s; | |
1074 | ++count; | |
1075 | } | |
1076 | } | |
a0d0e21e | 1077 | else { |
a6aa349d | 1078 | STRLEN rslen, rs_charlen; |
3280af22 | 1079 | char *rsptr = SvPV(PL_rs, rslen); |
a6aa349d TS |
1080 | |
1081 | rs_charlen = SvUTF8(PL_rs) | |
1082 | ? sv_len_utf8(PL_rs) | |
1083 | : rslen; | |
1084 | ||
c4c87a06 NC |
1085 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1086 | /* Assumption is that rs is shorter than the scalar. */ | |
1087 | if (SvUTF8(PL_rs)) { | |
1088 | /* RS is utf8, scalar is 8 bit. */ | |
1089 | bool is_utf8 = TRUE; | |
1090 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, | |
1091 | &rslen, &is_utf8); | |
1092 | if (is_utf8) { | |
1093 | /* Cannot downgrade, therefore cannot possibly match | |
1094 | */ | |
1095 | assert (temp_buffer == rsptr); | |
1096 | temp_buffer = NULL; | |
1097 | goto nope; | |
1098 | } | |
1099 | rsptr = temp_buffer; | |
a6aa349d TS |
1100 | } |
1101 | else if (PL_encoding) { | |
1102 | /* RS is 8 bit, encoding.pm is used. | |
1103 | * Do not recode PL_rs as a side-effect. */ | |
1104 | svrecode = newSVpvn(rsptr, rslen); | |
1105 | sv_recode_to_utf8(svrecode, PL_encoding); | |
1106 | rsptr = SvPV(svrecode, rslen); | |
1107 | rs_charlen = sv_len_utf8(svrecode); | |
1108 | } | |
1109 | else { | |
c4c87a06 NC |
1110 | /* RS is 8 bit, scalar is utf8. */ |
1111 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); | |
1112 | rsptr = temp_buffer; | |
1113 | } | |
1114 | } | |
c07a80fd | 1115 | if (rslen == 1) { |
1116 | if (*s != *rsptr) | |
1117 | goto nope; | |
1118 | ++count; | |
1119 | } | |
1120 | else { | |
8c2cee6f | 1121 | if (len < rslen - 1) |
c07a80fd | 1122 | goto nope; |
1123 | len -= rslen - 1; | |
1124 | s -= rslen - 1; | |
36477c24 | 1125 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1126 | goto nope; |
a6aa349d | 1127 | count += rs_charlen; |
c07a80fd | 1128 | } |
a0d0e21e | 1129 | } |
9b33ce3b | 1130 | s = SvPV_force(sv, n_a); |
a0d0e21e | 1131 | SvCUR_set(sv, len); |
9b33ce3b | 1132 | *SvEND(sv) = '\0'; |
a0d0e21e | 1133 | SvNIOK_off(sv); |
9b33ce3b | 1134 | SvSETMAGIC(sv); |
a0d0e21e LW |
1135 | } |
1136 | nope: | |
a6aa349d TS |
1137 | |
1138 | if (svrecode) | |
1139 | SvREFCNT_dec(svrecode); | |
1140 | ||
c4c87a06 | 1141 | Safefree(temp_buffer); |
a0d0e21e | 1142 | return count; |
8e84507e | 1143 | } |
79072805 LW |
1144 | |
1145 | void | |
864dbfa3 | 1146 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1147 | { |
1148 | #ifdef LIBERAL | |
1149 | register long *dl; | |
1150 | register long *ll; | |
1151 | register long *rl; | |
1152 | #endif | |
1153 | register char *dc; | |
463ee0b2 LW |
1154 | STRLEN leftlen; |
1155 | STRLEN rightlen; | |
7a4c00b4 | 1156 | register char *lc; |
1157 | register char *rc; | |
79072805 | 1158 | register I32 len; |
a0d0e21e | 1159 | I32 lensave; |
7a4c00b4 | 1160 | char *lsave; |
1161 | char *rsave; | |
0c57e439 GS |
1162 | bool left_utf = DO_UTF8(left); |
1163 | bool right_utf = DO_UTF8(right); | |
b7953727 | 1164 | I32 needlen = 0; |
0c57e439 GS |
1165 | |
1166 | if (left_utf && !right_utf) | |
1167 | sv_utf8_upgrade(right); | |
a1ca4561 | 1168 | else if (!left_utf && right_utf) |
0c57e439 | 1169 | sv_utf8_upgrade(left); |
79072805 | 1170 | |
1fbd88dc CS |
1171 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1172 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
891f9566 YST |
1173 | lsave = lc = SvPV_nomg(left, leftlen); |
1174 | rsave = rc = SvPV_nomg(right, rightlen); | |
93a17b20 | 1175 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1176 | lensave = len; |
c9b3c8d0 JH |
1177 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1178 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1179 | Newz(801, dc, needlen + 1, char); | |
1180 | } | |
1181 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d | 1182 | STRLEN n_a; |
891f9566 | 1183 | dc = SvPV_force_nomg(sv, n_a); |
eb160463 GS |
1184 | if (SvCUR(sv) < (STRLEN)len) { |
1185 | dc = SvGROW(sv, (STRLEN)(len + 1)); | |
ff68c719 | 1186 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1187 | } | |
c9b3c8d0 JH |
1188 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1189 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1190 | } |
1191 | else { | |
c9b3c8d0 JH |
1192 | needlen = ((optype == OP_BIT_AND) |
1193 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1194 | Newz(801, dc, needlen + 1, char); |
1195 | (void)sv_usepvn(sv, dc, needlen); | |
1196 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1197 | } |
a0d0e21e LW |
1198 | SvCUR_set(sv, len); |
1199 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1200 | if (left_utf || right_utf) { |
1201 | UV duc, luc, ruc; | |
c9b3c8d0 | 1202 | char *dcsave = dc; |
0c57e439 GS |
1203 | STRLEN lulen = leftlen; |
1204 | STRLEN rulen = rightlen; | |
ba210ebe | 1205 | STRLEN ulen; |
0c57e439 GS |
1206 | |
1207 | switch (optype) { | |
1208 | case OP_BIT_AND: | |
1209 | while (lulen && rulen) { | |
9041c2e3 | 1210 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1211 | lc += ulen; |
1212 | lulen -= ulen; | |
9041c2e3 | 1213 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1214 | rc += ulen; |
1215 | rulen -= ulen; | |
1216 | duc = luc & ruc; | |
9041c2e3 | 1217 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1218 | } |
c9b3c8d0 JH |
1219 | if (sv == left || sv == right) |
1220 | (void)sv_usepvn(sv, dcsave, needlen); | |
1221 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1222 | break; |
1223 | case OP_BIT_XOR: | |
1224 | while (lulen && rulen) { | |
9041c2e3 | 1225 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1226 | lc += ulen; |
1227 | lulen -= ulen; | |
9041c2e3 | 1228 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1229 | rc += ulen; |
1230 | rulen -= ulen; | |
1231 | duc = luc ^ ruc; | |
9041c2e3 | 1232 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1233 | } |
1234 | goto mop_up_utf; | |
1235 | case OP_BIT_OR: | |
1236 | while (lulen && rulen) { | |
9041c2e3 | 1237 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1238 | lc += ulen; |
1239 | lulen -= ulen; | |
9041c2e3 | 1240 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1241 | rc += ulen; |
1242 | rulen -= ulen; | |
1243 | duc = luc | ruc; | |
9041c2e3 | 1244 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1245 | } |
1246 | mop_up_utf: | |
c9b3c8d0 JH |
1247 | if (sv == left || sv == right) |
1248 | (void)sv_usepvn(sv, dcsave, needlen); | |
1249 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1250 | if (rulen) |
1251 | sv_catpvn(sv, rc, rulen); | |
1252 | else if (lulen) | |
1253 | sv_catpvn(sv, lc, lulen); | |
1254 | else | |
1255 | *SvEND(sv) = '\0'; | |
1256 | break; | |
1257 | } | |
1258 | SvUTF8_on(sv); | |
1259 | goto finish; | |
1260 | } | |
1261 | else | |
79072805 LW |
1262 | #ifdef LIBERAL |
1263 | if (len >= sizeof(long)*4 && | |
1264 | !((long)dc % sizeof(long)) && | |
1265 | !((long)lc % sizeof(long)) && | |
1266 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1267 | { | |
1268 | I32 remainder = len % (sizeof(long)*4); | |
1269 | len /= (sizeof(long)*4); | |
1270 | ||
1271 | dl = (long*)dc; | |
1272 | ll = (long*)lc; | |
1273 | rl = (long*)rc; | |
1274 | ||
1275 | switch (optype) { | |
1276 | case OP_BIT_AND: | |
1277 | while (len--) { | |
1278 | *dl++ = *ll++ & *rl++; | |
1279 | *dl++ = *ll++ & *rl++; | |
1280 | *dl++ = *ll++ & *rl++; | |
1281 | *dl++ = *ll++ & *rl++; | |
1282 | } | |
1283 | break; | |
a0d0e21e | 1284 | case OP_BIT_XOR: |
79072805 LW |
1285 | while (len--) { |
1286 | *dl++ = *ll++ ^ *rl++; | |
1287 | *dl++ = *ll++ ^ *rl++; | |
1288 | *dl++ = *ll++ ^ *rl++; | |
1289 | *dl++ = *ll++ ^ *rl++; | |
1290 | } | |
1291 | break; | |
1292 | case OP_BIT_OR: | |
1293 | while (len--) { | |
1294 | *dl++ = *ll++ | *rl++; | |
1295 | *dl++ = *ll++ | *rl++; | |
1296 | *dl++ = *ll++ | *rl++; | |
1297 | *dl++ = *ll++ | *rl++; | |
1298 | } | |
1299 | } | |
1300 | ||
1301 | dc = (char*)dl; | |
1302 | lc = (char*)ll; | |
1303 | rc = (char*)rl; | |
1304 | ||
1305 | len = remainder; | |
1306 | } | |
1307 | #endif | |
a0d0e21e | 1308 | { |
a0d0e21e LW |
1309 | switch (optype) { |
1310 | case OP_BIT_AND: | |
1311 | while (len--) | |
1312 | *dc++ = *lc++ & *rc++; | |
1313 | break; | |
1314 | case OP_BIT_XOR: | |
1315 | while (len--) | |
1316 | *dc++ = *lc++ ^ *rc++; | |
1317 | goto mop_up; | |
1318 | case OP_BIT_OR: | |
1319 | while (len--) | |
1320 | *dc++ = *lc++ | *rc++; | |
1321 | mop_up: | |
1322 | len = lensave; | |
eb160463 | 1323 | if (rightlen > (STRLEN)len) |
a0d0e21e | 1324 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 | 1325 | else if (leftlen > (STRLEN)len) |
a0d0e21e | 1326 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1327 | else |
1328 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1329 | break; |
1330 | } | |
79072805 | 1331 | } |
0c57e439 | 1332 | finish: |
fb73857a | 1333 | SvTAINT(sv); |
79072805 | 1334 | } |
463ee0b2 LW |
1335 | |
1336 | OP * | |
cea2e8a9 | 1337 | Perl_do_kv(pTHX) |
463ee0b2 | 1338 | { |
39644a26 | 1339 | dSP; |
463ee0b2 | 1340 | HV *hv = (HV*)POPs; |
800e9ae0 | 1341 | HV *keys; |
463ee0b2 | 1342 | register HE *entry; |
463ee0b2 | 1343 | SV *tmpstr; |
54310121 | 1344 | I32 gimme = GIMME_V; |
533c011a NIS |
1345 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1346 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
8e84507e NIS |
1347 | |
1348 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1349 | dokeys = dovalues = TRUE; |
463ee0b2 | 1350 | |
85581909 | 1351 | if (!hv) { |
78f9721b | 1352 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1353 | dTARGET; /* make sure to clear its target here */ |
1354 | if (SvTYPE(TARG) == SVt_PVLV) | |
1355 | LvTARG(TARG) = Nullsv; | |
1356 | PUSHs(TARG); | |
1357 | } | |
463ee0b2 | 1358 | RETURN; |
85581909 | 1359 | } |
748a9306 | 1360 | |
6d822dc4 | 1361 | keys = hv; |
800e9ae0 | 1362 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1363 | |
54310121 | 1364 | if (gimme == G_VOID) |
aa689395 | 1365 | RETURN; |
1366 | ||
54310121 | 1367 | if (gimme == G_SCALAR) { |
6ee623d5 | 1368 | IV i; |
463ee0b2 LW |
1369 | dTARGET; |
1370 | ||
78f9721b | 1371 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1372 | if (SvTYPE(TARG) < SVt_PVLV) { |
1373 | sv_upgrade(TARG, SVt_PVLV); | |
14befaf4 | 1374 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 SB |
1375 | } |
1376 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1377 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1378 | if (LvTARG(TARG)) |
1379 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1380 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1381 | } |
85581909 SB |
1382 | PUSHs(TARG); |
1383 | RETURN; | |
1384 | } | |
1385 | ||
14befaf4 | 1386 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 | 1387 | i = HvKEYS(keys); |
463ee0b2 LW |
1388 | else { |
1389 | i = 0; | |
463ee0b2 | 1390 | /*SUPPRESS 560*/ |
800e9ae0 | 1391 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1392 | } |
1393 | PUSHi( i ); | |
1394 | RETURN; | |
1395 | } | |
1396 | ||
8ed4b672 | 1397 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1398 | |
463ee0b2 | 1399 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1400 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1401 | SPAGAIN; |
574c8022 JH |
1402 | if (dokeys) { |
1403 | SV* sv = hv_iterkeysv(entry); | |
574c8022 JH |
1404 | XPUSHs(sv); /* won't clobber stack_sp */ |
1405 | } | |
463ee0b2 | 1406 | if (dovalues) { |
463ee0b2 | 1407 | PUTBACK; |
6d822dc4 | 1408 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 | 1409 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1410 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1411 | HvMAX(keys)+1, |
1412 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1413 | SPAGAIN; |
46fc3d4c | 1414 | XPUSHs(tmpstr); |
463ee0b2 LW |
1415 | } |
1416 | PUTBACK; | |
1417 | } | |
1418 | return NORMAL; | |
1419 | } | |
4e35701f | 1420 | |
af3babe4 NC |
1421 | /* |
1422 | * Local variables: | |
1423 | * c-indentation-style: bsd | |
1424 | * c-basic-offset: 4 | |
1425 | * indent-tabs-mode: t | |
1426 | * End: | |
1427 | * | |
1428 | * vim: shiftwidth=4: | |
1429 | */ |