Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
b94e2f88 |
4 | * 2000, 2001, 2002, 2004, 2005, 2006, by Larry Wall and others |
79072805 |
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 |
9 | */ |
10 | |
11 | /* |
12 | * "'So that was the job I felt I had to do when I started,' thought Sam." |
79072805 |
13 | */ |
14 | |
166f8a29 |
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 |
22 | #include "perl.h" |
23 | |
64ca3a65 |
24 | #ifndef PERL_MICRO |
79072805 |
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; |
f54cb97a |
33 | const U8 *send; |
b250498f |
34 | U8 *dstart; |
4757a243 |
35 | I32 matches = 0; |
f54cb97a |
36 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
37 | STRLEN len; |
79072805 |
38 | |
c4420975 |
39 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
40 | if (!tbl) |
7d85a32c |
41 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
42 | |
4757a243 |
43 | s = (U8*)SvPV(sv, len); |
44 | send = s + len; |
45 | |
1e54db1a |
46 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
47 | if (!SvUTF8(sv)) { |
01ec43d0 |
48 | while (s < send) { |
f54cb97a |
49 | const I32 ch = tbl[*s]; |
50 | if (ch >= 0) { |
01ec43d0 |
51 | matches++; |
c4420975 |
52 | *s = (U8)ch; |
01ec43d0 |
53 | } |
c4420975 |
54 | s++; |
01ec43d0 |
55 | } |
56 | SvSETMAGIC(sv); |
b250498f |
57 | return matches; |
58 | } |
4757a243 |
59 | |
b250498f |
60 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
61 | if (grows) |
a02a5408 |
62 | Newx(d, len*2+1, U8); |
9b877dbb |
63 | else |
64 | d = s; |
b250498f |
65 | dstart = d; |
66 | while (s < send) { |
ba210ebe |
67 | STRLEN ulen; |
f54cb97a |
68 | I32 ch; |
b250498f |
69 | |
b250498f |
70 | /* Need to check this, otherwise 128..255 won't match */ |
f54cb97a |
71 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb |
72 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
73 | matches++; |
9041c2e3 |
74 | d = uvchr_to_utf8(d, ch); |
b250498f |
75 | s += ulen; |
01ec43d0 |
76 | } |
77 | else { /* No match -> copy */ |
0376ff32 |
78 | Move(s, d, ulen, U8); |
9b877dbb |
79 | d += ulen; |
80 | s += ulen; |
b250498f |
81 | } |
82 | } |
9b877dbb |
83 | if (grows) { |
84 | sv_setpvn(sv, (char*)dstart, d - dstart); |
85 | Safefree(dstart); |
86 | } |
87 | else { |
88 | *d = '\0'; |
89 | SvCUR_set(sv, d - dstart); |
90 | } |
b250498f |
91 | SvUTF8_on(sv); |
b250498f |
92 | SvSETMAGIC(sv); |
4757a243 |
93 | return matches; |
94 | } |
95 | |
942e002e |
96 | STATIC I32 |
21304a3b |
97 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 |
98 | { |
83003860 |
99 | const U8 *s; |
100 | const U8 *send; |
4757a243 |
101 | I32 matches = 0; |
102 | STRLEN len; |
4757a243 |
103 | |
f54cb97a |
104 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
105 | if (!tbl) |
7d85a32c |
106 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
107 | |
83003860 |
108 | s = (const U8*)SvPV_const(sv, len); |
4757a243 |
109 | send = s + len; |
110 | |
1aa99e6b |
111 | if (!SvUTF8(sv)) |
112 | while (s < send) { |
113 | if (tbl[*s++] >= 0) |
036b4402 |
114 | matches++; |
1aa99e6b |
115 | } |
fabdb6c0 |
116 | else { |
117 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
1aa99e6b |
118 | while (s < send) { |
1aa99e6b |
119 | STRLEN ulen; |
f54cb97a |
120 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 |
121 | if (c < 0x100) { |
122 | if (tbl[c] >= 0) |
123 | matches++; |
124 | } else if (complement) |
1aa99e6b |
125 | matches++; |
126 | s += ulen; |
127 | } |
fabdb6c0 |
128 | } |
4757a243 |
129 | |
130 | return matches; |
131 | } |
132 | |
942e002e |
133 | STATIC I32 |
21304a3b |
134 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 |
135 | { |
4757a243 |
136 | U8 *s; |
137 | U8 *send; |
138 | U8 *d; |
5e44153e |
139 | U8 *dstart; |
1aa99e6b |
140 | I32 isutf8; |
4757a243 |
141 | I32 matches = 0; |
b7953727 |
142 | STRLEN len, rlen = 0; |
4757a243 |
143 | |
f54cb97a |
144 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
145 | if (!tbl) |
7d85a32c |
146 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
147 | |
148 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
149 | isutf8 = SvUTF8(sv); |
4757a243 |
150 | send = s + len; |
151 | |
1aa99e6b |
152 | if (!isutf8) { |
153 | dstart = d = s; |
154 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a |
155 | const U8* p = send; |
1aa99e6b |
156 | while (s < send) { |
f54cb97a |
157 | const I32 ch = tbl[*s]; |
158 | if (ch >= 0) { |
eb160463 |
159 | *d = (U8)ch; |
036b4402 |
160 | matches++; |
1aa99e6b |
161 | if (p != d - 1 || *p != *d) |
162 | p = d++; |
163 | } |
164 | else if (ch == -1) /* -1 is unmapped character */ |
165 | *d++ = *s; |
166 | else if (ch == -2) /* -2 is delete character */ |
167 | matches++; |
168 | s++; |
169 | } |
a0ed51b3 |
170 | } |
1aa99e6b |
171 | else { |
172 | while (s < send) { |
f54cb97a |
173 | const I32 ch = tbl[*s]; |
174 | if (ch >= 0) { |
1aa99e6b |
175 | matches++; |
eb160463 |
176 | *d++ = (U8)ch; |
1aa99e6b |
177 | } |
178 | else if (ch == -1) /* -1 is unmapped character */ |
179 | *d++ = *s; |
180 | else if (ch == -2) /* -2 is delete character */ |
181 | matches++; |
182 | s++; |
183 | } |
184 | } |
76ef7183 |
185 | *d = '\0'; |
1aa99e6b |
186 | SvCUR_set(sv, d - dstart); |
4757a243 |
187 | } |
1aa99e6b |
188 | else { /* isutf8 */ |
fabdb6c0 |
189 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
190 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
191 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
192 | |
9b877dbb |
193 | if (grows) |
a02a5408 |
194 | Newx(d, len*2+1, U8); |
9b877dbb |
195 | else |
196 | d = s; |
1aa99e6b |
197 | dstart = d; |
8973db79 |
198 | if (complement && !del) |
199 | rlen = tbl[0x100]; |
1aa99e6b |
200 | |
4aecb5b5 |
201 | #ifdef MACOS_TRADITIONAL |
202 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ |
203 | #endif |
204 | |
1aa99e6b |
205 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b |
206 | UV pch = 0xfeedface; |
207 | while (s < send) { |
208 | STRLEN len; |
f54cb97a |
209 | const UV comp = utf8_to_uvchr(s, &len); |
210 | I32 ch; |
1aa99e6b |
211 | |
8973db79 |
212 | if (comp > 0xff) { |
213 | if (!complement) { |
214 | Copy(s, d, len, U8); |
215 | d += len; |
216 | } |
217 | else { |
218 | matches++; |
219 | if (!del) { |
05d340b8 |
220 | ch = (rlen == 0) ? comp : |
221 | (comp - 0x100 < rlen) ? |
8973db79 |
222 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 |
223 | if ((UV)ch != pch) { |
9041c2e3 |
224 | d = uvchr_to_utf8(d, ch); |
eb160463 |
225 | pch = (UV)ch; |
8973db79 |
226 | } |
227 | s += len; |
228 | continue; |
229 | } |
230 | } |
9b877dbb |
231 | } |
1aa99e6b |
232 | else if ((ch = tbl[comp]) >= 0) { |
233 | matches++; |
eb160463 |
234 | if ((UV)ch != pch) { |
9041c2e3 |
235 | d = uvchr_to_utf8(d, ch); |
eb160463 |
236 | pch = (UV)ch; |
1aa99e6b |
237 | } |
238 | s += len; |
239 | continue; |
240 | } |
9b877dbb |
241 | else if (ch == -1) { /* -1 is unmapped character */ |
242 | Copy(s, d, len, U8); |
243 | d += len; |
244 | } |
1aa99e6b |
245 | else if (ch == -2) /* -2 is delete character */ |
246 | matches++; |
247 | s += len; |
248 | pch = 0xfeedface; |
249 | } |
5d06d08e |
250 | } |
1aa99e6b |
251 | else { |
252 | while (s < send) { |
253 | STRLEN len; |
4373e329 |
254 | const UV comp = utf8_to_uvchr(s, &len); |
f54cb97a |
255 | I32 ch; |
8973db79 |
256 | if (comp > 0xff) { |
257 | if (!complement) { |
0376ff32 |
258 | Move(s, d, len, U8); |
8973db79 |
259 | d += len; |
260 | } |
261 | else { |
262 | matches++; |
263 | if (!del) { |
9041c2e3 |
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 |
268 | } |
269 | } |
9b877dbb |
270 | } |
1aa99e6b |
271 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 |
272 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
273 | matches++; |
274 | } |
275 | else if (ch == -1) { /* -1 is unmapped character */ |
9b877dbb |
276 | Copy(s, d, len, U8); |
277 | d += len; |
1aa99e6b |
278 | } |
279 | else if (ch == -2) /* -2 is delete character */ |
280 | matches++; |
281 | s += len; |
282 | } |
283 | } |
9b877dbb |
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 |
295 | return matches; |
296 | } |
297 | |
942e002e |
298 | STATIC I32 |
21304a3b |
299 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 |
300 | { |
4757a243 |
301 | U8 *s; |
302 | U8 *send; |
303 | U8 *d; |
036b4402 |
304 | U8 *start; |
1aa99e6b |
305 | U8 *dstart, *dend; |
4757a243 |
306 | I32 matches = 0; |
f54cb97a |
307 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
308 | STRLEN len; |
309 | |
53c1dcc0 |
310 | SV* const rv = (SV*)cSVOP->op_sv; |
311 | HV* const hv = (HV*)SvRV(rv); |
c4420975 |
312 | SV* const * svp = hv_fetch(hv, "NONE", 4, FALSE); |
f54cb97a |
313 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
314 | const UV extra = none + 1; |
b7953727 |
315 | UV final = 0; |
1aa99e6b |
316 | I32 isutf8; |
317 | U8 hibit = 0; |
4757a243 |
318 | |
319 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
320 | isutf8 = SvUTF8(sv); |
321 | if (!isutf8) { |
c4420975 |
322 | const U8 *t = s; |
323 | const U8 * const e = s + len; |
c4d5f83a |
324 | while (t < e) { |
f54cb97a |
325 | const U8 ch = *t++; |
fabdb6c0 |
326 | hibit = !NATIVE_IS_INVARIANT(ch); |
327 | if (hibit) { |
328 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
329 | break; |
fabdb6c0 |
330 | } |
c4d5f83a |
331 | } |
1aa99e6b |
332 | } |
4757a243 |
333 | send = s + len; |
036b4402 |
334 | start = s; |
4757a243 |
335 | |
336 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
337 | if (svp) |
338 | final = SvUV(*svp); |
339 | |
9b877dbb |
340 | if (grows) { |
341 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
342 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
343 | dend = d + len * 3; |
344 | dstart = d; |
345 | } |
346 | else { |
347 | dstart = d = s; |
348 | dend = d + len; |
349 | } |
1aa99e6b |
350 | |
4757a243 |
351 | while (s < send) { |
0bcc34c2 |
352 | const UV uv = swash_fetch(rv, s, TRUE); |
353 | if (uv < none) { |
4757a243 |
354 | s += UTF8SKIP(s); |
355 | matches++; |
7d85a32c |
356 | d = uvuni_to_utf8(d, uv); |
4757a243 |
357 | } |
358 | else if (uv == none) { |
4373e329 |
359 | const int i = UTF8SKIP(s); |
0376ff32 |
360 | Move(s, d, i, U8); |
9b877dbb |
361 | d += i; |
362 | s += i; |
4757a243 |
363 | } |
364 | else if (uv == extra) { |
4373e329 |
365 | s += UTF8SKIP(s); |
4757a243 |
366 | matches++; |
7d85a32c |
367 | d = uvuni_to_utf8(d, final); |
4757a243 |
368 | } |
369 | else |
370 | s += UTF8SKIP(s); |
1aa99e6b |
371 | |
9b877dbb |
372 | if (d > dend) { |
4373e329 |
373 | const STRLEN clen = d - dstart; |
374 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
375 | if (!grows) |
7d85a32c |
376 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 |
377 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
378 | d = dstart + clen; |
379 | dend = dstart + nlen; |
380 | } |
4757a243 |
381 | } |
16ec844d |
382 | if (grows || hibit) { |
9b877dbb |
383 | sv_setpvn(sv, (char*)dstart, d - dstart); |
384 | Safefree(dstart); |
16ec844d |
385 | if (grows && hibit) |
386 | Safefree(start); |
9b877dbb |
387 | } |
388 | else { |
389 | *d = '\0'; |
390 | SvCUR_set(sv, d - dstart); |
391 | } |
4757a243 |
392 | SvSETMAGIC(sv); |
1aa99e6b |
393 | SvUTF8_on(sv); |
4757a243 |
394 | |
395 | return matches; |
396 | } |
397 | |
942e002e |
398 | STATIC I32 |
21304a3b |
399 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 |
400 | { |
93524f2b |
401 | const U8 *s; |
cbbf8932 |
402 | const U8 *start = NULL; |
403 | const U8 *send; |
4757a243 |
404 | I32 matches = 0; |
405 | STRLEN len; |
406 | |
53c1dcc0 |
407 | SV* const rv = (SV*)cSVOP->op_sv; |
408 | HV* const hv = (HV*)SvRV(rv); |
c4420975 |
409 | SV* const * const svp = hv_fetch(hv, "NONE", 4, FALSE); |
f54cb97a |
410 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
411 | const UV extra = none + 1; |
1aa99e6b |
412 | U8 hibit = 0; |
4757a243 |
413 | |
93524f2b |
414 | s = (const U8*)SvPV_const(sv, len); |
1aa99e6b |
415 | if (!SvUTF8(sv)) { |
4373e329 |
416 | const U8 *t = s; |
c4420975 |
417 | const U8 * const e = s + len; |
c4d5f83a |
418 | while (t < e) { |
4373e329 |
419 | const U8 ch = *t++; |
fabdb6c0 |
420 | hibit = !NATIVE_IS_INVARIANT(ch); |
421 | if (hibit) { |
422 | start = s = bytes_to_utf8(s, &len); |
1aa99e6b |
423 | break; |
fabdb6c0 |
424 | } |
c4d5f83a |
425 | } |
1aa99e6b |
426 | } |
4757a243 |
427 | send = s + len; |
428 | |
429 | while (s < send) { |
c4420975 |
430 | const UV uv = swash_fetch(rv, s, TRUE); |
431 | if (uv < none || uv == extra) |
4757a243 |
432 | matches++; |
834a4ddd |
433 | s += UTF8SKIP(s); |
4757a243 |
434 | } |
1aa99e6b |
435 | if (hibit) |
436 | Safefree(start); |
4757a243 |
437 | |
438 | return matches; |
439 | } |
440 | |
942e002e |
441 | STATIC I32 |
21304a3b |
442 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 |
443 | { |
1aa99e6b |
444 | U8 *start, *send; |
4757a243 |
445 | U8 *d; |
446 | I32 matches = 0; |
4373e329 |
447 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
448 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
449 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
53c1dcc0 |
450 | SV * const rv = (SV*)cSVOP->op_sv; |
451 | HV * const hv = (HV*)SvRV(rv); |
c4420975 |
452 | SV * const *svp = hv_fetch(hv, "NONE", 4, FALSE); |
f54cb97a |
453 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
454 | const UV extra = none + 1; |
b7953727 |
455 | UV final = 0; |
45005bfb |
456 | bool havefinal = FALSE; |
4757a243 |
457 | STRLEN len; |
1aa99e6b |
458 | U8 *dstart, *dend; |
1aa99e6b |
459 | U8 hibit = 0; |
4757a243 |
460 | |
f54cb97a |
461 | U8 *s = (U8*)SvPV(sv, len); |
462 | const I32 isutf8 = SvUTF8(sv); |
1aa99e6b |
463 | if (!isutf8) { |
53c1dcc0 |
464 | const U8 *t = s; |
465 | const U8 * const e = s + len; |
c4d5f83a |
466 | while (t < e) { |
f54cb97a |
467 | const U8 ch = *t++; |
fabdb6c0 |
468 | hibit = !NATIVE_IS_INVARIANT(ch); |
469 | if (hibit) { |
470 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
471 | break; |
fabdb6c0 |
472 | } |
c4d5f83a |
473 | } |
1aa99e6b |
474 | } |
4757a243 |
475 | send = s + len; |
1aa99e6b |
476 | start = s; |
4757a243 |
477 | |
478 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
45005bfb |
479 | if (svp) { |
4757a243 |
480 | final = SvUV(*svp); |
45005bfb |
481 | havefinal = TRUE; |
482 | } |
4757a243 |
483 | |
9b877dbb |
484 | if (grows) { |
485 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
486 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
487 | dend = d + len * 3; |
488 | dstart = d; |
489 | } |
490 | else { |
491 | dstart = d = s; |
492 | dend = d + len; |
493 | } |
4757a243 |
494 | |
495 | if (squash) { |
496 | UV puv = 0xfeedface; |
497 | while (s < send) { |
f54cb97a |
498 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
499 | |
9b877dbb |
500 | if (d > dend) { |
4373e329 |
501 | const STRLEN clen = d - dstart; |
502 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
503 | if (!grows) |
7d85a32c |
504 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
505 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
506 | d = dstart + clen; |
507 | dend = dstart + nlen; |
4757a243 |
508 | } |
509 | if (uv < none) { |
510 | matches++; |
94472101 |
511 | s += UTF8SKIP(s); |
4757a243 |
512 | if (uv != puv) { |
7d85a32c |
513 | d = uvuni_to_utf8(d, uv); |
4757a243 |
514 | puv = uv; |
515 | } |
4757a243 |
516 | continue; |
517 | } |
518 | else if (uv == none) { /* "none" is unmapped character */ |
4373e329 |
519 | const int i = UTF8SKIP(s); |
0376ff32 |
520 | Move(s, d, i, U8); |
9b877dbb |
521 | d += i; |
522 | s += i; |
4757a243 |
523 | puv = 0xfeedface; |
524 | continue; |
a0ed51b3 |
525 | } |
4757a243 |
526 | else if (uv == extra && !del) { |
527 | matches++; |
45005bfb |
528 | if (havefinal) { |
529 | s += UTF8SKIP(s); |
530 | if (puv != final) { |
7d85a32c |
531 | d = uvuni_to_utf8(d, final); |
45005bfb |
532 | puv = final; |
533 | } |
534 | } |
535 | else { |
536 | STRLEN len; |
7d85a32c |
537 | uv = utf8_to_uvuni(s, &len); |
45005bfb |
538 | if (uv != puv) { |
0376ff32 |
539 | Move(s, d, len, U8); |
45005bfb |
540 | d += len; |
541 | puv = uv; |
542 | } |
543 | s += len; |
4757a243 |
544 | } |
4757a243 |
545 | continue; |
546 | } |
01ec43d0 |
547 | matches++; /* "none+1" is delete character */ |
548 | s += UTF8SKIP(s); |
a0ed51b3 |
549 | } |
79072805 |
550 | } |
551 | else { |
4757a243 |
552 | while (s < send) { |
f54cb97a |
553 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
554 | if (d > dend) { |
f54cb97a |
555 | const STRLEN clen = d - dstart; |
556 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
557 | if (!grows) |
7d85a32c |
558 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
559 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
560 | d = dstart + clen; |
561 | dend = dstart + nlen; |
a0ed51b3 |
562 | } |
4757a243 |
563 | if (uv < none) { |
564 | matches++; |
01ec43d0 |
565 | s += UTF8SKIP(s); |
7d85a32c |
566 | d = uvuni_to_utf8(d, uv); |
4757a243 |
567 | continue; |
a0ed51b3 |
568 | } |
4757a243 |
569 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a |
570 | const int i = UTF8SKIP(s); |
0376ff32 |
571 | Move(s, d, i, U8); |
9b877dbb |
572 | d += i; |
573 | s += i; |
4757a243 |
574 | continue; |
79072805 |
575 | } |
4757a243 |
576 | else if (uv == extra && !del) { |
577 | matches++; |
01ec43d0 |
578 | s += UTF8SKIP(s); |
7d85a32c |
579 | d = uvuni_to_utf8(d, final); |
4757a243 |
580 | continue; |
581 | } |
01ec43d0 |
582 | matches++; /* "none+1" is delete character */ |
583 | s += UTF8SKIP(s); |
79072805 |
584 | } |
4757a243 |
585 | } |
16ec844d |
586 | if (grows || hibit) { |
9b877dbb |
587 | sv_setpvn(sv, (char*)dstart, d - dstart); |
588 | Safefree(dstart); |
16ec844d |
589 | if (grows && hibit) |
590 | Safefree(start); |
9b877dbb |
591 | } |
592 | else { |
593 | *d = '\0'; |
594 | SvCUR_set(sv, d - dstart); |
595 | } |
1aa99e6b |
596 | SvUTF8_on(sv); |
4757a243 |
597 | SvSETMAGIC(sv); |
598 | |
599 | return matches; |
600 | } |
601 | |
602 | I32 |
864dbfa3 |
603 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
604 | { |
605 | STRLEN len; |
4373e329 |
606 | const I32 hasutf = (PL_op->op_private & |
036b4402 |
607 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
608 | |
2233f375 |
609 | if (SvREADONLY(sv)) { |
765f542d |
610 | if (SvIsCOW(sv)) |
611 | sv_force_normal_flags(sv, 0); |
2233f375 |
612 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
613 | Perl_croak(aTHX_ PL_no_modify); |
614 | } |
10516c54 |
615 | (void)SvPV_const(sv, len); |
4757a243 |
616 | if (!len) |
617 | return 0; |
d59e14db |
618 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
619 | if (!SvPOKp(sv)) |
620 | (void)SvPV_force(sv, len); |
2de7b02f |
621 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
622 | } |
4757a243 |
623 | |
cea2e8a9 |
624 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
625 | |
a77e643a |
626 | switch (PL_op->op_private & ~hasutf & ( |
627 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| |
628 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { |
4757a243 |
629 | case 0: |
01ec43d0 |
630 | if (hasutf) |
631 | return do_trans_simple_utf8(sv); |
632 | else |
633 | return do_trans_simple(sv); |
4757a243 |
634 | |
635 | case OPpTRANS_IDENTICAL: |
05d340b8 |
636 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
637 | if (hasutf) |
638 | return do_trans_count_utf8(sv); |
639 | else |
640 | return do_trans_count(sv); |
4757a243 |
641 | |
642 | default: |
01ec43d0 |
643 | if (hasutf) |
036b4402 |
644 | return do_trans_complex_utf8(sv); |
4757a243 |
645 | else |
036b4402 |
646 | return do_trans_complex(sv); |
79072805 |
647 | } |
79072805 |
648 | } |
649 | |
650 | void |
864dbfa3 |
651 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
652 | { |
53c1dcc0 |
653 | SV ** const oldmark = mark; |
79072805 |
654 | register I32 items = sp - mark; |
79072805 |
655 | register STRLEN len; |
463ee0b2 |
656 | STRLEN delimlen; |
79072805 |
657 | |
5c144d81 |
658 | (void) SvPV_const(del, delimlen); /* stringify and get the delimlen */ |
5ba99574 |
659 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
660 | |
79072805 |
661 | mark++; |
662 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
862a34c6 |
663 | SvUPGRADE(sv, SVt_PV); |
79072805 |
664 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
665 | while (items-- > 0) { |
1426bbf4 |
666 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a |
667 | STRLEN tmplen; |
5c144d81 |
668 | SvPV_const(*mark, tmplen); |
463ee0b2 |
669 | len += tmplen; |
79072805 |
670 | } |
671 | mark++; |
672 | } |
673 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
674 | |
675 | mark = oldmark; |
db7c17d7 |
676 | items = sp - mark; |
79072805 |
677 | ++mark; |
678 | } |
679 | |
fc0199cd |
680 | sv_setpvn(sv, "", 0); |
e4803c42 |
681 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 |
682 | SvUTF8_off(sv); |
e4803c42 |
683 | |
8d6d96c1 |
684 | if (PL_tainting && SvMAGICAL(sv)) |
685 | SvTAINTED_off(sv); |
686 | |
463ee0b2 |
687 | if (items-- > 0) { |
92d29cee |
688 | if (*mark) |
689 | sv_catsv(sv, *mark); |
463ee0b2 |
690 | mark++; |
691 | } |
8d6d96c1 |
692 | |
c512ce4f |
693 | if (delimlen) { |
79072805 |
694 | for (; items > 0; items--,mark++) { |
c512ce4f |
695 | sv_catsv(sv,del); |
79072805 |
696 | sv_catsv(sv,*mark); |
697 | } |
698 | } |
699 | else { |
700 | for (; items > 0; items--,mark++) |
701 | sv_catsv(sv,*mark); |
702 | } |
703 | SvSETMAGIC(sv); |
704 | } |
705 | |
706 | void |
864dbfa3 |
707 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
708 | { |
46fc3d4c |
709 | STRLEN patlen; |
53c1dcc0 |
710 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c |
711 | bool do_taint = FALSE; |
712 | |
5b781b5b |
713 | SvUTF8_off(sv); |
2cf2cfc6 |
714 | if (DO_UTF8(*sarg)) |
715 | SvUTF8_on(sv); |
46fc3d4c |
716 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
717 | SvSETMAGIC(sv); |
46fc3d4c |
718 | if (do_taint) |
719 | SvTAINTED_on(sv); |
79072805 |
720 | } |
721 | |
33b45480 |
722 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
723 | UV |
724 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
725 | { |
726 | STRLEN srclen, len; |
10516c54 |
727 | const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); |
81e118e0 |
728 | UV retnum = 0; |
729 | |
a50d7633 |
730 | if (offset < 0) |
81e118e0 |
731 | return retnum; |
8e84507e |
732 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
733 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
734 | |
dcad2880 |
735 | if (SvUTF8(sv)) |
33b45480 |
736 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
737 | |
81e118e0 |
738 | offset *= size; /* turn into bit offset */ |
739 | len = (offset + size + 7) / 8; /* required number of bytes */ |
740 | if (len > srclen) { |
741 | if (size <= 8) |
742 | retnum = 0; |
743 | else { |
744 | offset >>= 3; /* turn into byte offset */ |
745 | if (size == 16) { |
eb160463 |
746 | if ((STRLEN)offset >= srclen) |
81e118e0 |
747 | retnum = 0; |
748 | else |
628e1a40 |
749 | retnum = (UV) s[offset] << 8; |
81e118e0 |
750 | } |
751 | else if (size == 32) { |
eb160463 |
752 | if ((STRLEN)offset >= srclen) |
81e118e0 |
753 | retnum = 0; |
eb160463 |
754 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 |
755 | retnum = |
756 | ((UV) s[offset ] << 24); |
eb160463 |
757 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 |
758 | retnum = |
759 | ((UV) s[offset ] << 24) + |
760 | ((UV) s[offset + 1] << 16); |
761 | else |
762 | retnum = |
763 | ((UV) s[offset ] << 24) + |
764 | ((UV) s[offset + 1] << 16) + |
765 | ( s[offset + 2] << 8); |
766 | } |
d7d93a81 |
767 | #ifdef UV_IS_QUAD |
c5a0f51a |
768 | else if (size == 64) { |
c5a0f51a |
769 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
770 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
771 | "Bit vector size > 32 non-portable"); |
772 | if (offset >= srclen) |
773 | retnum = 0; |
774 | else if (offset + 1 >= srclen) |
775 | retnum = |
776 | (UV) s[offset ] << 56; |
777 | else if (offset + 2 >= srclen) |
778 | retnum = |
779 | ((UV) s[offset ] << 56) + |
780 | ((UV) s[offset + 1] << 48); |
781 | else if (offset + 3 >= srclen) |
782 | retnum = |
783 | ((UV) s[offset ] << 56) + |
784 | ((UV) s[offset + 1] << 48) + |
785 | ((UV) s[offset + 2] << 40); |
786 | else if (offset + 4 >= srclen) |
787 | retnum = |
788 | ((UV) s[offset ] << 56) + |
789 | ((UV) s[offset + 1] << 48) + |
790 | ((UV) s[offset + 2] << 40) + |
791 | ((UV) s[offset + 3] << 32); |
792 | else if (offset + 5 >= srclen) |
793 | retnum = |
794 | ((UV) s[offset ] << 56) + |
795 | ((UV) s[offset + 1] << 48) + |
796 | ((UV) s[offset + 2] << 40) + |
797 | ((UV) s[offset + 3] << 32) + |
798 | ( s[offset + 4] << 24); |
799 | else if (offset + 6 >= srclen) |
800 | retnum = |
801 | ((UV) s[offset ] << 56) + |
802 | ((UV) s[offset + 1] << 48) + |
803 | ((UV) s[offset + 2] << 40) + |
804 | ((UV) s[offset + 3] << 32) + |
805 | ((UV) s[offset + 4] << 24) + |
806 | ((UV) s[offset + 5] << 16); |
807 | else |
8e84507e |
808 | retnum = |
c5a0f51a |
809 | ((UV) s[offset ] << 56) + |
810 | ((UV) s[offset + 1] << 48) + |
811 | ((UV) s[offset + 2] << 40) + |
812 | ((UV) s[offset + 3] << 32) + |
813 | ((UV) s[offset + 4] << 24) + |
814 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
815 | ( s[offset + 6] << 8); |
c5a0f51a |
816 | } |
817 | #endif |
81e118e0 |
818 | } |
819 | } |
820 | else if (size < 8) |
821 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
822 | else { |
823 | offset >>= 3; /* turn into byte offset */ |
824 | if (size == 8) |
825 | retnum = s[offset]; |
826 | else if (size == 16) |
827 | retnum = |
628e1a40 |
828 | ((UV) s[offset] << 8) + |
81e118e0 |
829 | s[offset + 1]; |
830 | else if (size == 32) |
831 | retnum = |
832 | ((UV) s[offset ] << 24) + |
833 | ((UV) s[offset + 1] << 16) + |
834 | ( s[offset + 2] << 8) + |
835 | s[offset + 3]; |
d7d93a81 |
836 | #ifdef UV_IS_QUAD |
c5a0f51a |
837 | else if (size == 64) { |
c5a0f51a |
838 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
839 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
840 | "Bit vector size > 32 non-portable"); |
841 | retnum = |
842 | ((UV) s[offset ] << 56) + |
843 | ((UV) s[offset + 1] << 48) + |
844 | ((UV) s[offset + 2] << 40) + |
845 | ((UV) s[offset + 3] << 32) + |
846 | ((UV) s[offset + 4] << 24) + |
847 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
848 | ( s[offset + 6] << 8) + |
c5a0f51a |
849 | s[offset + 7]; |
850 | } |
851 | #endif |
81e118e0 |
852 | } |
853 | |
854 | return retnum; |
855 | } |
856 | |
33b45480 |
857 | /* currently converts input to bytes if possible but doesn't sweat failures, |
858 | * although it does ensure that the string it clobbers is not marked as |
859 | * utf8-valid any more |
860 | */ |
79072805 |
861 | void |
864dbfa3 |
862 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
863 | { |
79072805 |
864 | register I32 offset; |
865 | register I32 size; |
8990e307 |
866 | register unsigned char *s; |
81e118e0 |
867 | register UV lval; |
79072805 |
868 | I32 mask; |
a0d0e21e |
869 | STRLEN targlen; |
870 | STRLEN len; |
c4420975 |
871 | SV * const targ = LvTARG(sv); |
79072805 |
872 | |
8990e307 |
873 | if (!targ) |
874 | return; |
a0d0e21e |
875 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
876 | if (SvUTF8(targ)) { |
33b45480 |
877 | /* This is handled by the SvPOK_only below... |
878 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
879 | SvUTF8_off(targ); |
880 | */ |
881 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
882 | } |
883 | |
4ebbc975 |
884 | (void)SvPOK_only(targ); |
81e118e0 |
885 | lval = SvUV(sv); |
79072805 |
886 | offset = LvTARGOFF(sv); |
fe58ced6 |
887 | if (offset < 0) |
ed9aa3b7 |
888 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
889 | size = LvTARGLEN(sv); |
8e84507e |
890 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
891 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
892 | |
81e118e0 |
893 | offset *= size; /* turn into bit offset */ |
894 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
895 | if (len > targlen) { |
896 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
897 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
898 | SvCUR_set(targ, len); |
899 | } |
8e84507e |
900 | |
79072805 |
901 | if (size < 8) { |
902 | mask = (1 << size) - 1; |
903 | size = offset & 7; |
904 | lval &= mask; |
81e118e0 |
905 | offset >>= 3; /* turn into byte offset */ |
79072805 |
906 | s[offset] &= ~(mask << size); |
907 | s[offset] |= lval << size; |
908 | } |
909 | else { |
81e118e0 |
910 | offset >>= 3; /* turn into byte offset */ |
79072805 |
911 | if (size == 8) |
eb160463 |
912 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
913 | else if (size == 16) { |
eb160463 |
914 | s[offset ] = (U8)((lval >> 8) & 0xff); |
915 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
916 | } |
917 | else if (size == 32) { |
eb160463 |
918 | s[offset ] = (U8)((lval >> 24) & 0xff); |
919 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
920 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
921 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
922 | } |
d7d93a81 |
923 | #ifdef UV_IS_QUAD |
c5a0f51a |
924 | else if (size == 64) { |
c5a0f51a |
925 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
926 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
927 | "Bit vector size > 32 non-portable"); |
eb160463 |
928 | s[offset ] = (U8)((lval >> 56) & 0xff); |
929 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
930 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
931 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
932 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
933 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
934 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
935 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
936 | } |
dc1e3f56 |
937 | #endif |
79072805 |
938 | } |
7bb043c3 |
939 | SvSETMAGIC(targ); |
79072805 |
940 | } |
941 | |
942 | void |
864dbfa3 |
943 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
944 | { |
463ee0b2 |
945 | STRLEN len; |
a0d0e21e |
946 | char *s; |
8e84507e |
947 | |
79072805 |
948 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
949 | register I32 i; |
c4420975 |
950 | AV* const av = (AV*)sv; |
f54cb97a |
951 | const I32 max = AvFILL(av); |
952 | |
953 | for (i = 0; i <= max; i++) { |
a0d0e21e |
954 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
955 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
956 | do_chop(astr, sv); |
957 | } |
958 | return; |
79072805 |
959 | } |
aa854799 |
960 | else if (SvTYPE(sv) == SVt_PVHV) { |
c4420975 |
961 | HV* const hv = (HV*)sv; |
a0d0e21e |
962 | HE* entry; |
963 | (void)hv_iterinit(hv); |
155aba94 |
964 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
965 | do_chop(astr,hv_iterval(hv,entry)); |
966 | return; |
79072805 |
967 | } |
991e6d41 |
968 | else if (SvREADONLY(sv)) { |
969 | if (SvFAKE(sv)) { |
970 | /* SV is copy-on-write */ |
971 | sv_force_normal_flags(sv, 0); |
972 | } |
973 | if (SvREADONLY(sv)) |
974 | Perl_croak(aTHX_ PL_no_modify); |
975 | } |
565a3db3 |
976 | |
977 | if (PL_encoding && !SvUTF8(sv)) { |
978 | /* like in do_chomp(), utf8-ize the sv as a side-effect |
979 | * if we're using encoding. */ |
980 | sv_recode_to_utf8(sv, PL_encoding); |
981 | } |
982 | |
a0d0e21e |
983 | s = SvPV(sv, len); |
748a9306 |
984 | if (len && !SvPOK(sv)) |
a0d0e21e |
985 | s = SvPV_force(sv, len); |
7e2040f0 |
986 | if (DO_UTF8(sv)) { |
a0ed51b3 |
987 | if (s && len) { |
c4420975 |
988 | char * const send = s + len; |
0bcc34c2 |
989 | char * const start = s; |
a0ed51b3 |
990 | s = send - 1; |
a0dbb045 |
991 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
992 | s--; |
ace7757b |
993 | if (is_utf8_string((U8*)s, send - s)) { |
a0dbb045 |
994 | sv_setpvn(astr, s, send - s); |
995 | *s = '\0'; |
996 | SvCUR_set(sv, s - start); |
997 | SvNIOK_off(sv); |
998 | SvUTF8_on(astr); |
999 | } |
a0ed51b3 |
1000 | } |
1001 | else |
1002 | sv_setpvn(astr, "", 0); |
1003 | } |
7e2040f0 |
1004 | else if (s && len) { |
a0d0e21e |
1005 | s += --len; |
1006 | sv_setpvn(astr, s, 1); |
1007 | *s = '\0'; |
1008 | SvCUR_set(sv, len); |
2c19a612 |
1009 | SvUTF8_off(sv); |
a0d0e21e |
1010 | SvNIOK_off(sv); |
79072805 |
1011 | } |
1012 | else |
a0d0e21e |
1013 | sv_setpvn(astr, "", 0); |
1014 | SvSETMAGIC(sv); |
7e2040f0 |
1015 | } |
a0d0e21e |
1016 | |
1017 | I32 |
864dbfa3 |
1018 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1019 | { |
c07a80fd |
1020 | register I32 count; |
a0d0e21e |
1021 | STRLEN len; |
1022 | char *s; |
c4c87a06 |
1023 | char *temp_buffer = NULL; |
a6aa349d |
1024 | SV* svrecode = Nullsv; |
c07a80fd |
1025 | |
3280af22 |
1026 | if (RsSNARF(PL_rs)) |
c07a80fd |
1027 | return 0; |
4c5a6083 |
1028 | if (RsRECORD(PL_rs)) |
1029 | return 0; |
c07a80fd |
1030 | count = 0; |
a0d0e21e |
1031 | if (SvTYPE(sv) == SVt_PVAV) { |
1032 | register I32 i; |
890ce7af |
1033 | AV* const av = (AV*)sv; |
f54cb97a |
1034 | const I32 max = AvFILL(av); |
1035 | |
1036 | for (i = 0; i <= max; i++) { |
a0d0e21e |
1037 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1038 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1039 | count += do_chomp(sv); |
1040 | } |
1041 | return count; |
1042 | } |
aa854799 |
1043 | else if (SvTYPE(sv) == SVt_PVHV) { |
890ce7af |
1044 | HV* const hv = (HV*)sv; |
a0d0e21e |
1045 | HE* entry; |
1046 | (void)hv_iterinit(hv); |
155aba94 |
1047 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1048 | count += do_chomp(hv_iterval(hv,entry)); |
1049 | return count; |
1050 | } |
991e6d41 |
1051 | else if (SvREADONLY(sv)) { |
1052 | if (SvFAKE(sv)) { |
1053 | /* SV is copy-on-write */ |
1054 | sv_force_normal_flags(sv, 0); |
1055 | } |
1056 | if (SvREADONLY(sv)) |
1057 | Perl_croak(aTHX_ PL_no_modify); |
1058 | } |
a6aa349d |
1059 | |
1060 | if (PL_encoding) { |
1061 | if (!SvUTF8(sv)) { |
1062 | /* XXX, here sv is utf8-ized as a side-effect! |
1063 | If encoding.pm is used properly, almost string-generating |
1064 | operations, including literal strings, chr(), input data, etc. |
1065 | should have been utf8-ized already, right? |
1066 | */ |
1067 | sv_recode_to_utf8(sv, PL_encoding); |
1068 | } |
1069 | } |
1070 | |
a0d0e21e |
1071 | s = SvPV(sv, len); |
a0d0e21e |
1072 | if (s && len) { |
1073 | s += --len; |
3280af22 |
1074 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1075 | if (*s != '\n') |
1076 | goto nope; |
1077 | ++count; |
1078 | while (len && s[-1] == '\n') { |
1079 | --len; |
1080 | --s; |
1081 | ++count; |
1082 | } |
1083 | } |
a0d0e21e |
1084 | else { |
a6aa349d |
1085 | STRLEN rslen, rs_charlen; |
93524f2b |
1086 | const char *rsptr = SvPV_const(PL_rs, rslen); |
a6aa349d |
1087 | |
1088 | rs_charlen = SvUTF8(PL_rs) |
1089 | ? sv_len_utf8(PL_rs) |
1090 | : rslen; |
1091 | |
c4c87a06 |
1092 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1093 | /* Assumption is that rs is shorter than the scalar. */ |
1094 | if (SvUTF8(PL_rs)) { |
1095 | /* RS is utf8, scalar is 8 bit. */ |
1096 | bool is_utf8 = TRUE; |
1097 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, |
1098 | &rslen, &is_utf8); |
1099 | if (is_utf8) { |
1100 | /* Cannot downgrade, therefore cannot possibly match |
1101 | */ |
1102 | assert (temp_buffer == rsptr); |
1103 | temp_buffer = NULL; |
1104 | goto nope; |
1105 | } |
1106 | rsptr = temp_buffer; |
a6aa349d |
1107 | } |
1108 | else if (PL_encoding) { |
1109 | /* RS is 8 bit, encoding.pm is used. |
1110 | * Do not recode PL_rs as a side-effect. */ |
1111 | svrecode = newSVpvn(rsptr, rslen); |
1112 | sv_recode_to_utf8(svrecode, PL_encoding); |
93524f2b |
1113 | rsptr = SvPV_const(svrecode, rslen); |
a6aa349d |
1114 | rs_charlen = sv_len_utf8(svrecode); |
1115 | } |
1116 | else { |
c4c87a06 |
1117 | /* RS is 8 bit, scalar is utf8. */ |
1118 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); |
1119 | rsptr = temp_buffer; |
1120 | } |
1121 | } |
c07a80fd |
1122 | if (rslen == 1) { |
1123 | if (*s != *rsptr) |
1124 | goto nope; |
1125 | ++count; |
1126 | } |
1127 | else { |
8c2cee6f |
1128 | if (len < rslen - 1) |
c07a80fd |
1129 | goto nope; |
1130 | len -= rslen - 1; |
1131 | s -= rslen - 1; |
36477c24 |
1132 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1133 | goto nope; |
a6aa349d |
1134 | count += rs_charlen; |
c07a80fd |
1135 | } |
a0d0e21e |
1136 | } |
8b6b16e7 |
1137 | s = SvPV_force_nolen(sv); |
a0d0e21e |
1138 | SvCUR_set(sv, len); |
9b33ce3b |
1139 | *SvEND(sv) = '\0'; |
a0d0e21e |
1140 | SvNIOK_off(sv); |
9b33ce3b |
1141 | SvSETMAGIC(sv); |
a0d0e21e |
1142 | } |
1143 | nope: |
a6aa349d |
1144 | |
1145 | if (svrecode) |
1146 | SvREFCNT_dec(svrecode); |
1147 | |
c4c87a06 |
1148 | Safefree(temp_buffer); |
a0d0e21e |
1149 | return count; |
8e84507e |
1150 | } |
79072805 |
1151 | |
1152 | void |
864dbfa3 |
1153 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1154 | { |
1155 | #ifdef LIBERAL |
1156 | register long *dl; |
1157 | register long *ll; |
1158 | register long *rl; |
1159 | #endif |
1160 | register char *dc; |
463ee0b2 |
1161 | STRLEN leftlen; |
1162 | STRLEN rightlen; |
e62f0680 |
1163 | register const char *lc; |
1164 | register const char *rc; |
79072805 |
1165 | register I32 len; |
a0d0e21e |
1166 | I32 lensave; |
e62f0680 |
1167 | const char *lsave; |
1168 | const char *rsave; |
f54cb97a |
1169 | const bool left_utf = DO_UTF8(left); |
1170 | const bool right_utf = DO_UTF8(right); |
b7953727 |
1171 | I32 needlen = 0; |
0c57e439 |
1172 | |
1173 | if (left_utf && !right_utf) |
1174 | sv_utf8_upgrade(right); |
a1ca4561 |
1175 | else if (!left_utf && right_utf) |
0c57e439 |
1176 | sv_utf8_upgrade(left); |
79072805 |
1177 | |
1fbd88dc |
1178 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1179 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
e62f0680 |
1180 | lsave = lc = SvPV_nomg_const(left, leftlen); |
1181 | rsave = rc = SvPV_nomg_const(right, rightlen); |
93a17b20 |
1182 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1183 | lensave = len; |
c9b3c8d0 |
1184 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1185 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
a02a5408 |
1186 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 |
1187 | } |
1188 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe |
1189 | dc = SvPV_force_nomg_nolen(sv); |
1a787b95 |
1190 | if (SvLEN(sv) < (STRLEN)(len + 1)) { |
eb160463 |
1191 | dc = SvGROW(sv, (STRLEN)(len + 1)); |
ff68c719 |
1192 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1193 | } |
c9b3c8d0 |
1194 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1195 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1196 | } |
1197 | else { |
c9b3c8d0 |
1198 | needlen = ((optype == OP_BIT_AND) |
1199 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
a02a5408 |
1200 | Newxz(dc, needlen + 1, char); |
ff68c719 |
1201 | (void)sv_usepvn(sv, dc, needlen); |
1202 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1203 | } |
a0d0e21e |
1204 | SvCUR_set(sv, len); |
1205 | (void)SvPOK_only(sv); |
0c57e439 |
1206 | if (left_utf || right_utf) { |
1207 | UV duc, luc, ruc; |
c4420975 |
1208 | char * const dcsave = dc; |
0c57e439 |
1209 | STRLEN lulen = leftlen; |
1210 | STRLEN rulen = rightlen; |
ba210ebe |
1211 | STRLEN ulen; |
0c57e439 |
1212 | |
1213 | switch (optype) { |
1214 | case OP_BIT_AND: |
1215 | while (lulen && rulen) { |
9041c2e3 |
1216 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1217 | lc += ulen; |
1218 | lulen -= ulen; |
9041c2e3 |
1219 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1220 | rc += ulen; |
1221 | rulen -= ulen; |
1222 | duc = luc & ruc; |
9041c2e3 |
1223 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1224 | } |
c9b3c8d0 |
1225 | if (sv == left || sv == right) |
1226 | (void)sv_usepvn(sv, dcsave, needlen); |
1227 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1228 | break; |
1229 | case OP_BIT_XOR: |
1230 | while (lulen && rulen) { |
9041c2e3 |
1231 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1232 | lc += ulen; |
1233 | lulen -= ulen; |
9041c2e3 |
1234 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1235 | rc += ulen; |
1236 | rulen -= ulen; |
1237 | duc = luc ^ ruc; |
9041c2e3 |
1238 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1239 | } |
1240 | goto mop_up_utf; |
1241 | case OP_BIT_OR: |
1242 | while (lulen && rulen) { |
9041c2e3 |
1243 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1244 | lc += ulen; |
1245 | lulen -= ulen; |
9041c2e3 |
1246 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1247 | rc += ulen; |
1248 | rulen -= ulen; |
1249 | duc = luc | ruc; |
9041c2e3 |
1250 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1251 | } |
1252 | mop_up_utf: |
c9b3c8d0 |
1253 | if (sv == left || sv == right) |
1254 | (void)sv_usepvn(sv, dcsave, needlen); |
1255 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1256 | if (rulen) |
1257 | sv_catpvn(sv, rc, rulen); |
1258 | else if (lulen) |
1259 | sv_catpvn(sv, lc, lulen); |
1260 | else |
1261 | *SvEND(sv) = '\0'; |
1262 | break; |
1263 | } |
1264 | SvUTF8_on(sv); |
1265 | goto finish; |
1266 | } |
1267 | else |
79072805 |
1268 | #ifdef LIBERAL |
1269 | if (len >= sizeof(long)*4 && |
1270 | !((long)dc % sizeof(long)) && |
1271 | !((long)lc % sizeof(long)) && |
1272 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1273 | { |
f54cb97a |
1274 | const I32 remainder = len % (sizeof(long)*4); |
79072805 |
1275 | len /= (sizeof(long)*4); |
1276 | |
1277 | dl = (long*)dc; |
1278 | ll = (long*)lc; |
1279 | rl = (long*)rc; |
1280 | |
1281 | switch (optype) { |
1282 | case OP_BIT_AND: |
1283 | while (len--) { |
1284 | *dl++ = *ll++ & *rl++; |
1285 | *dl++ = *ll++ & *rl++; |
1286 | *dl++ = *ll++ & *rl++; |
1287 | *dl++ = *ll++ & *rl++; |
1288 | } |
1289 | break; |
a0d0e21e |
1290 | case OP_BIT_XOR: |
79072805 |
1291 | while (len--) { |
1292 | *dl++ = *ll++ ^ *rl++; |
1293 | *dl++ = *ll++ ^ *rl++; |
1294 | *dl++ = *ll++ ^ *rl++; |
1295 | *dl++ = *ll++ ^ *rl++; |
1296 | } |
1297 | break; |
1298 | case OP_BIT_OR: |
1299 | while (len--) { |
1300 | *dl++ = *ll++ | *rl++; |
1301 | *dl++ = *ll++ | *rl++; |
1302 | *dl++ = *ll++ | *rl++; |
1303 | *dl++ = *ll++ | *rl++; |
1304 | } |
1305 | } |
1306 | |
1307 | dc = (char*)dl; |
1308 | lc = (char*)ll; |
1309 | rc = (char*)rl; |
1310 | |
1311 | len = remainder; |
1312 | } |
1313 | #endif |
a0d0e21e |
1314 | { |
a0d0e21e |
1315 | switch (optype) { |
1316 | case OP_BIT_AND: |
1317 | while (len--) |
1318 | *dc++ = *lc++ & *rc++; |
1a787b95 |
1319 | *dc = '\0'; |
a0d0e21e |
1320 | break; |
1321 | case OP_BIT_XOR: |
1322 | while (len--) |
1323 | *dc++ = *lc++ ^ *rc++; |
1324 | goto mop_up; |
1325 | case OP_BIT_OR: |
1326 | while (len--) |
1327 | *dc++ = *lc++ | *rc++; |
1328 | mop_up: |
1329 | len = lensave; |
eb160463 |
1330 | if (rightlen > (STRLEN)len) |
a0d0e21e |
1331 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1332 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1333 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1334 | else |
1335 | *SvEND(sv) = '\0'; |
a0d0e21e |
1336 | break; |
1337 | } |
79072805 |
1338 | } |
0c57e439 |
1339 | finish: |
fb73857a |
1340 | SvTAINT(sv); |
79072805 |
1341 | } |
463ee0b2 |
1342 | |
1343 | OP * |
cea2e8a9 |
1344 | Perl_do_kv(pTHX) |
463ee0b2 |
1345 | { |
39644a26 |
1346 | dSP; |
c4420975 |
1347 | HV * const hv = (HV*)POPs; |
800e9ae0 |
1348 | HV *keys; |
463ee0b2 |
1349 | register HE *entry; |
f54cb97a |
1350 | const I32 gimme = GIMME_V; |
1351 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); |
1352 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); |
1353 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); |
463ee0b2 |
1354 | |
85581909 |
1355 | if (!hv) { |
78f9721b |
1356 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1357 | dTARGET; /* make sure to clear its target here */ |
1358 | if (SvTYPE(TARG) == SVt_PVLV) |
1359 | LvTARG(TARG) = Nullsv; |
1360 | PUSHs(TARG); |
1361 | } |
463ee0b2 |
1362 | RETURN; |
85581909 |
1363 | } |
748a9306 |
1364 | |
6d822dc4 |
1365 | keys = hv; |
800e9ae0 |
1366 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1367 | |
54310121 |
1368 | if (gimme == G_VOID) |
aa689395 |
1369 | RETURN; |
1370 | |
54310121 |
1371 | if (gimme == G_SCALAR) { |
6ee623d5 |
1372 | IV i; |
463ee0b2 |
1373 | dTARGET; |
1374 | |
78f9721b |
1375 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1376 | if (SvTYPE(TARG) < SVt_PVLV) { |
1377 | sv_upgrade(TARG, SVt_PVLV); |
14befaf4 |
1378 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 |
1379 | } |
1380 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1381 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1382 | if (LvTARG(TARG)) |
1383 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1384 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1385 | } |
85581909 |
1386 | PUSHs(TARG); |
1387 | RETURN; |
1388 | } |
1389 | |
14befaf4 |
1390 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 |
1391 | i = HvKEYS(keys); |
463ee0b2 |
1392 | else { |
1393 | i = 0; |
800e9ae0 |
1394 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1395 | } |
1396 | PUSHi( i ); |
1397 | RETURN; |
1398 | } |
1399 | |
8ed4b672 |
1400 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1401 | |
463ee0b2 |
1402 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1403 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1404 | SPAGAIN; |
574c8022 |
1405 | if (dokeys) { |
890ce7af |
1406 | SV* const sv = hv_iterkeysv(entry); |
574c8022 |
1407 | XPUSHs(sv); /* won't clobber stack_sp */ |
1408 | } |
463ee0b2 |
1409 | if (dovalues) { |
f54cb97a |
1410 | SV *tmpstr; |
463ee0b2 |
1411 | PUTBACK; |
6d822dc4 |
1412 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1413 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1414 | (unsigned long)HeHASH(entry), |
cdd3ba14 |
1415 | (int)HvMAX(keys)+1, |
800e9ae0 |
1416 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1417 | SPAGAIN; |
46fc3d4c |
1418 | XPUSHs(tmpstr); |
463ee0b2 |
1419 | } |
1420 | PUTBACK; |
1421 | } |
1422 | return NORMAL; |
1423 | } |
4e35701f |
1424 | |
af3babe4 |
1425 | /* |
1426 | * Local variables: |
1427 | * c-indentation-style: bsd |
1428 | * c-basic-offset: 4 |
1429 | * indent-tabs-mode: t |
1430 | * End: |
1431 | * |
37442d52 |
1432 | * ex: set ts=8 sts=4 sw=4 noet: |
1433 | */ |