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 | { |
97aff369 |
31 | dVAR; |
4757a243 |
32 | U8 *s; |
b250498f |
33 | U8 *d; |
f54cb97a |
34 | const U8 *send; |
b250498f |
35 | U8 *dstart; |
4757a243 |
36 | I32 matches = 0; |
f54cb97a |
37 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
38 | STRLEN len; |
79072805 |
39 | |
c4420975 |
40 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
41 | if (!tbl) |
7d85a32c |
42 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
43 | |
4757a243 |
44 | s = (U8*)SvPV(sv, len); |
45 | send = s + len; |
46 | |
1e54db1a |
47 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
48 | if (!SvUTF8(sv)) { |
01ec43d0 |
49 | while (s < send) { |
f54cb97a |
50 | const I32 ch = tbl[*s]; |
51 | if (ch >= 0) { |
01ec43d0 |
52 | matches++; |
c4420975 |
53 | *s = (U8)ch; |
01ec43d0 |
54 | } |
c4420975 |
55 | s++; |
01ec43d0 |
56 | } |
57 | SvSETMAGIC(sv); |
b250498f |
58 | return matches; |
59 | } |
4757a243 |
60 | |
b250498f |
61 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
62 | if (grows) |
a02a5408 |
63 | Newx(d, len*2+1, U8); |
9b877dbb |
64 | else |
65 | d = s; |
b250498f |
66 | dstart = d; |
67 | while (s < send) { |
ba210ebe |
68 | STRLEN ulen; |
f54cb97a |
69 | I32 ch; |
b250498f |
70 | |
b250498f |
71 | /* Need to check this, otherwise 128..255 won't match */ |
f54cb97a |
72 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb |
73 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
74 | matches++; |
9041c2e3 |
75 | d = uvchr_to_utf8(d, ch); |
b250498f |
76 | s += ulen; |
01ec43d0 |
77 | } |
78 | else { /* No match -> copy */ |
0376ff32 |
79 | Move(s, d, ulen, U8); |
9b877dbb |
80 | d += ulen; |
81 | s += ulen; |
b250498f |
82 | } |
83 | } |
9b877dbb |
84 | if (grows) { |
85 | sv_setpvn(sv, (char*)dstart, d - dstart); |
86 | Safefree(dstart); |
87 | } |
88 | else { |
89 | *d = '\0'; |
90 | SvCUR_set(sv, d - dstart); |
91 | } |
b250498f |
92 | SvUTF8_on(sv); |
b250498f |
93 | SvSETMAGIC(sv); |
4757a243 |
94 | return matches; |
95 | } |
96 | |
942e002e |
97 | STATIC I32 |
21304a3b |
98 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 |
99 | { |
97aff369 |
100 | dVAR; |
83003860 |
101 | const U8 *s; |
102 | const U8 *send; |
4757a243 |
103 | I32 matches = 0; |
104 | STRLEN len; |
4757a243 |
105 | |
f54cb97a |
106 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
107 | if (!tbl) |
7d85a32c |
108 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
109 | |
83003860 |
110 | s = (const U8*)SvPV_const(sv, len); |
4757a243 |
111 | send = s + len; |
112 | |
1aa99e6b |
113 | if (!SvUTF8(sv)) |
114 | while (s < send) { |
115 | if (tbl[*s++] >= 0) |
036b4402 |
116 | matches++; |
1aa99e6b |
117 | } |
fabdb6c0 |
118 | else { |
119 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
1aa99e6b |
120 | while (s < send) { |
1aa99e6b |
121 | STRLEN ulen; |
f54cb97a |
122 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 |
123 | if (c < 0x100) { |
124 | if (tbl[c] >= 0) |
125 | matches++; |
126 | } else if (complement) |
1aa99e6b |
127 | matches++; |
128 | s += ulen; |
129 | } |
fabdb6c0 |
130 | } |
4757a243 |
131 | |
132 | return matches; |
133 | } |
134 | |
942e002e |
135 | STATIC I32 |
21304a3b |
136 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 |
137 | { |
97aff369 |
138 | dVAR; |
4757a243 |
139 | U8 *s; |
140 | U8 *send; |
141 | U8 *d; |
5e44153e |
142 | U8 *dstart; |
1aa99e6b |
143 | I32 isutf8; |
4757a243 |
144 | I32 matches = 0; |
b7953727 |
145 | STRLEN len, rlen = 0; |
4757a243 |
146 | |
f54cb97a |
147 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
148 | if (!tbl) |
7d85a32c |
149 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
150 | |
151 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
152 | isutf8 = SvUTF8(sv); |
4757a243 |
153 | send = s + len; |
154 | |
1aa99e6b |
155 | if (!isutf8) { |
156 | dstart = d = s; |
157 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a |
158 | const U8* p = send; |
1aa99e6b |
159 | while (s < send) { |
f54cb97a |
160 | const I32 ch = tbl[*s]; |
161 | if (ch >= 0) { |
eb160463 |
162 | *d = (U8)ch; |
036b4402 |
163 | matches++; |
1aa99e6b |
164 | if (p != d - 1 || *p != *d) |
165 | p = d++; |
166 | } |
167 | else if (ch == -1) /* -1 is unmapped character */ |
168 | *d++ = *s; |
169 | else if (ch == -2) /* -2 is delete character */ |
170 | matches++; |
171 | s++; |
172 | } |
a0ed51b3 |
173 | } |
1aa99e6b |
174 | else { |
175 | while (s < send) { |
f54cb97a |
176 | const I32 ch = tbl[*s]; |
177 | if (ch >= 0) { |
1aa99e6b |
178 | matches++; |
eb160463 |
179 | *d++ = (U8)ch; |
1aa99e6b |
180 | } |
181 | else if (ch == -1) /* -1 is unmapped character */ |
182 | *d++ = *s; |
183 | else if (ch == -2) /* -2 is delete character */ |
184 | matches++; |
185 | s++; |
186 | } |
187 | } |
76ef7183 |
188 | *d = '\0'; |
1aa99e6b |
189 | SvCUR_set(sv, d - dstart); |
4757a243 |
190 | } |
1aa99e6b |
191 | else { /* isutf8 */ |
fabdb6c0 |
192 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
193 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
194 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
195 | |
9b877dbb |
196 | if (grows) |
a02a5408 |
197 | Newx(d, len*2+1, U8); |
9b877dbb |
198 | else |
199 | d = s; |
1aa99e6b |
200 | dstart = d; |
8973db79 |
201 | if (complement && !del) |
202 | rlen = tbl[0x100]; |
1aa99e6b |
203 | |
4aecb5b5 |
204 | #ifdef MACOS_TRADITIONAL |
205 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ |
206 | #endif |
207 | |
1aa99e6b |
208 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b |
209 | UV pch = 0xfeedface; |
210 | while (s < send) { |
211 | STRLEN len; |
f54cb97a |
212 | const UV comp = utf8_to_uvchr(s, &len); |
213 | I32 ch; |
1aa99e6b |
214 | |
8973db79 |
215 | if (comp > 0xff) { |
216 | if (!complement) { |
217 | Copy(s, d, len, U8); |
218 | d += len; |
219 | } |
220 | else { |
221 | matches++; |
222 | if (!del) { |
05d340b8 |
223 | ch = (rlen == 0) ? comp : |
224 | (comp - 0x100 < rlen) ? |
8973db79 |
225 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 |
226 | if ((UV)ch != pch) { |
9041c2e3 |
227 | d = uvchr_to_utf8(d, ch); |
eb160463 |
228 | pch = (UV)ch; |
8973db79 |
229 | } |
230 | s += len; |
231 | continue; |
232 | } |
233 | } |
9b877dbb |
234 | } |
1aa99e6b |
235 | else if ((ch = tbl[comp]) >= 0) { |
236 | matches++; |
eb160463 |
237 | if ((UV)ch != pch) { |
9041c2e3 |
238 | d = uvchr_to_utf8(d, ch); |
eb160463 |
239 | pch = (UV)ch; |
1aa99e6b |
240 | } |
241 | s += len; |
242 | continue; |
243 | } |
9b877dbb |
244 | else if (ch == -1) { /* -1 is unmapped character */ |
245 | Copy(s, d, len, U8); |
246 | d += len; |
247 | } |
1aa99e6b |
248 | else if (ch == -2) /* -2 is delete character */ |
249 | matches++; |
250 | s += len; |
251 | pch = 0xfeedface; |
252 | } |
5d06d08e |
253 | } |
1aa99e6b |
254 | else { |
255 | while (s < send) { |
256 | STRLEN len; |
4373e329 |
257 | const UV comp = utf8_to_uvchr(s, &len); |
f54cb97a |
258 | I32 ch; |
8973db79 |
259 | if (comp > 0xff) { |
260 | if (!complement) { |
0376ff32 |
261 | Move(s, d, len, U8); |
8973db79 |
262 | d += len; |
263 | } |
264 | else { |
265 | matches++; |
266 | if (!del) { |
9041c2e3 |
267 | if (comp - 0x100 < rlen) |
268 | d = uvchr_to_utf8(d, tbl[comp+1]); |
8973db79 |
269 | else |
9041c2e3 |
270 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 |
271 | } |
272 | } |
9b877dbb |
273 | } |
1aa99e6b |
274 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 |
275 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
276 | matches++; |
277 | } |
278 | else if (ch == -1) { /* -1 is unmapped character */ |
9b877dbb |
279 | Copy(s, d, len, U8); |
280 | d += len; |
1aa99e6b |
281 | } |
282 | else if (ch == -2) /* -2 is delete character */ |
283 | matches++; |
284 | s += len; |
285 | } |
286 | } |
9b877dbb |
287 | if (grows) { |
288 | sv_setpvn(sv, (char*)dstart, d - dstart); |
289 | Safefree(dstart); |
290 | } |
291 | else { |
292 | *d = '\0'; |
293 | SvCUR_set(sv, d - dstart); |
294 | } |
1aa99e6b |
295 | SvUTF8_on(sv); |
4757a243 |
296 | } |
5e44153e |
297 | SvSETMAGIC(sv); |
4757a243 |
298 | return matches; |
299 | } |
300 | |
942e002e |
301 | STATIC I32 |
21304a3b |
302 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 |
303 | { |
97aff369 |
304 | dVAR; |
4757a243 |
305 | U8 *s; |
306 | U8 *send; |
307 | U8 *d; |
036b4402 |
308 | U8 *start; |
1aa99e6b |
309 | U8 *dstart, *dend; |
4757a243 |
310 | I32 matches = 0; |
f54cb97a |
311 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
312 | STRLEN len; |
313 | |
53c1dcc0 |
314 | SV* const rv = (SV*)cSVOP->op_sv; |
315 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
316 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
317 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
318 | const UV extra = none + 1; |
b7953727 |
319 | UV final = 0; |
1aa99e6b |
320 | I32 isutf8; |
321 | U8 hibit = 0; |
4757a243 |
322 | |
323 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
324 | isutf8 = SvUTF8(sv); |
325 | if (!isutf8) { |
c4420975 |
326 | const U8 *t = s; |
327 | const U8 * const e = s + len; |
c4d5f83a |
328 | while (t < e) { |
f54cb97a |
329 | const U8 ch = *t++; |
fabdb6c0 |
330 | hibit = !NATIVE_IS_INVARIANT(ch); |
331 | if (hibit) { |
332 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
333 | break; |
fabdb6c0 |
334 | } |
c4d5f83a |
335 | } |
1aa99e6b |
336 | } |
4757a243 |
337 | send = s + len; |
036b4402 |
338 | start = s; |
4757a243 |
339 | |
a4fc7abc |
340 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 |
341 | if (svp) |
342 | final = SvUV(*svp); |
343 | |
9b877dbb |
344 | if (grows) { |
345 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
346 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
347 | dend = d + len * 3; |
348 | dstart = d; |
349 | } |
350 | else { |
351 | dstart = d = s; |
352 | dend = d + len; |
353 | } |
1aa99e6b |
354 | |
4757a243 |
355 | while (s < send) { |
0bcc34c2 |
356 | const UV uv = swash_fetch(rv, s, TRUE); |
357 | if (uv < none) { |
4757a243 |
358 | s += UTF8SKIP(s); |
359 | matches++; |
7d85a32c |
360 | d = uvuni_to_utf8(d, uv); |
4757a243 |
361 | } |
362 | else if (uv == none) { |
4373e329 |
363 | const int i = UTF8SKIP(s); |
0376ff32 |
364 | Move(s, d, i, U8); |
9b877dbb |
365 | d += i; |
366 | s += i; |
4757a243 |
367 | } |
368 | else if (uv == extra) { |
4373e329 |
369 | s += UTF8SKIP(s); |
4757a243 |
370 | matches++; |
7d85a32c |
371 | d = uvuni_to_utf8(d, final); |
4757a243 |
372 | } |
373 | else |
374 | s += UTF8SKIP(s); |
1aa99e6b |
375 | |
9b877dbb |
376 | if (d > dend) { |
4373e329 |
377 | const STRLEN clen = d - dstart; |
378 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
379 | if (!grows) |
7d85a32c |
380 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 |
381 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
382 | d = dstart + clen; |
383 | dend = dstart + nlen; |
384 | } |
4757a243 |
385 | } |
16ec844d |
386 | if (grows || hibit) { |
9b877dbb |
387 | sv_setpvn(sv, (char*)dstart, d - dstart); |
388 | Safefree(dstart); |
16ec844d |
389 | if (grows && hibit) |
390 | Safefree(start); |
9b877dbb |
391 | } |
392 | else { |
393 | *d = '\0'; |
394 | SvCUR_set(sv, d - dstart); |
395 | } |
4757a243 |
396 | SvSETMAGIC(sv); |
1aa99e6b |
397 | SvUTF8_on(sv); |
4757a243 |
398 | |
399 | return matches; |
400 | } |
401 | |
942e002e |
402 | STATIC I32 |
21304a3b |
403 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 |
404 | { |
97aff369 |
405 | dVAR; |
93524f2b |
406 | const U8 *s; |
cbbf8932 |
407 | const U8 *start = NULL; |
408 | const U8 *send; |
4757a243 |
409 | I32 matches = 0; |
410 | STRLEN len; |
411 | |
53c1dcc0 |
412 | SV* const rv = (SV*)cSVOP->op_sv; |
413 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
414 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
415 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
416 | const UV extra = none + 1; |
1aa99e6b |
417 | U8 hibit = 0; |
4757a243 |
418 | |
93524f2b |
419 | s = (const U8*)SvPV_const(sv, len); |
1aa99e6b |
420 | if (!SvUTF8(sv)) { |
4373e329 |
421 | const U8 *t = s; |
c4420975 |
422 | const U8 * const e = s + len; |
c4d5f83a |
423 | while (t < e) { |
4373e329 |
424 | const U8 ch = *t++; |
fabdb6c0 |
425 | hibit = !NATIVE_IS_INVARIANT(ch); |
426 | if (hibit) { |
427 | start = s = bytes_to_utf8(s, &len); |
1aa99e6b |
428 | break; |
fabdb6c0 |
429 | } |
c4d5f83a |
430 | } |
1aa99e6b |
431 | } |
4757a243 |
432 | send = s + len; |
433 | |
434 | while (s < send) { |
c4420975 |
435 | const UV uv = swash_fetch(rv, s, TRUE); |
436 | if (uv < none || uv == extra) |
4757a243 |
437 | matches++; |
834a4ddd |
438 | s += UTF8SKIP(s); |
4757a243 |
439 | } |
1aa99e6b |
440 | if (hibit) |
441 | Safefree(start); |
4757a243 |
442 | |
443 | return matches; |
444 | } |
445 | |
942e002e |
446 | STATIC I32 |
21304a3b |
447 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 |
448 | { |
97aff369 |
449 | dVAR; |
1aa99e6b |
450 | U8 *start, *send; |
4757a243 |
451 | U8 *d; |
452 | I32 matches = 0; |
4373e329 |
453 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
454 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
455 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
53c1dcc0 |
456 | SV * const rv = (SV*)cSVOP->op_sv; |
457 | HV * const hv = (HV*)SvRV(rv); |
a4fc7abc |
458 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
459 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
460 | const UV extra = none + 1; |
b7953727 |
461 | UV final = 0; |
45005bfb |
462 | bool havefinal = FALSE; |
4757a243 |
463 | STRLEN len; |
1aa99e6b |
464 | U8 *dstart, *dend; |
1aa99e6b |
465 | U8 hibit = 0; |
4757a243 |
466 | |
f54cb97a |
467 | U8 *s = (U8*)SvPV(sv, len); |
468 | const I32 isutf8 = SvUTF8(sv); |
1aa99e6b |
469 | if (!isutf8) { |
53c1dcc0 |
470 | const U8 *t = s; |
471 | const U8 * const e = s + len; |
c4d5f83a |
472 | while (t < e) { |
f54cb97a |
473 | const U8 ch = *t++; |
fabdb6c0 |
474 | hibit = !NATIVE_IS_INVARIANT(ch); |
475 | if (hibit) { |
476 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
477 | break; |
fabdb6c0 |
478 | } |
c4d5f83a |
479 | } |
1aa99e6b |
480 | } |
4757a243 |
481 | send = s + len; |
1aa99e6b |
482 | start = s; |
4757a243 |
483 | |
a4fc7abc |
484 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb |
485 | if (svp) { |
4757a243 |
486 | final = SvUV(*svp); |
45005bfb |
487 | havefinal = TRUE; |
488 | } |
4757a243 |
489 | |
9b877dbb |
490 | if (grows) { |
491 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
492 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
493 | dend = d + len * 3; |
494 | dstart = d; |
495 | } |
496 | else { |
497 | dstart = d = s; |
498 | dend = d + len; |
499 | } |
4757a243 |
500 | |
501 | if (squash) { |
502 | UV puv = 0xfeedface; |
503 | while (s < send) { |
f54cb97a |
504 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
505 | |
9b877dbb |
506 | if (d > dend) { |
4373e329 |
507 | const STRLEN clen = d - dstart; |
508 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
509 | if (!grows) |
7d85a32c |
510 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
511 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
512 | d = dstart + clen; |
513 | dend = dstart + nlen; |
4757a243 |
514 | } |
515 | if (uv < none) { |
516 | matches++; |
94472101 |
517 | s += UTF8SKIP(s); |
4757a243 |
518 | if (uv != puv) { |
7d85a32c |
519 | d = uvuni_to_utf8(d, uv); |
4757a243 |
520 | puv = uv; |
521 | } |
4757a243 |
522 | continue; |
523 | } |
524 | else if (uv == none) { /* "none" is unmapped character */ |
4373e329 |
525 | const int i = UTF8SKIP(s); |
0376ff32 |
526 | Move(s, d, i, U8); |
9b877dbb |
527 | d += i; |
528 | s += i; |
4757a243 |
529 | puv = 0xfeedface; |
530 | continue; |
a0ed51b3 |
531 | } |
4757a243 |
532 | else if (uv == extra && !del) { |
533 | matches++; |
45005bfb |
534 | if (havefinal) { |
535 | s += UTF8SKIP(s); |
536 | if (puv != final) { |
7d85a32c |
537 | d = uvuni_to_utf8(d, final); |
45005bfb |
538 | puv = final; |
539 | } |
540 | } |
541 | else { |
542 | STRLEN len; |
7d85a32c |
543 | uv = utf8_to_uvuni(s, &len); |
45005bfb |
544 | if (uv != puv) { |
0376ff32 |
545 | Move(s, d, len, U8); |
45005bfb |
546 | d += len; |
547 | puv = uv; |
548 | } |
549 | s += len; |
4757a243 |
550 | } |
4757a243 |
551 | continue; |
552 | } |
01ec43d0 |
553 | matches++; /* "none+1" is delete character */ |
554 | s += UTF8SKIP(s); |
a0ed51b3 |
555 | } |
79072805 |
556 | } |
557 | else { |
4757a243 |
558 | while (s < send) { |
f54cb97a |
559 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
560 | if (d > dend) { |
f54cb97a |
561 | const STRLEN clen = d - dstart; |
562 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
563 | if (!grows) |
7d85a32c |
564 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
565 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
566 | d = dstart + clen; |
567 | dend = dstart + nlen; |
a0ed51b3 |
568 | } |
4757a243 |
569 | if (uv < none) { |
570 | matches++; |
01ec43d0 |
571 | s += UTF8SKIP(s); |
7d85a32c |
572 | d = uvuni_to_utf8(d, uv); |
4757a243 |
573 | continue; |
a0ed51b3 |
574 | } |
4757a243 |
575 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a |
576 | const int i = UTF8SKIP(s); |
0376ff32 |
577 | Move(s, d, i, U8); |
9b877dbb |
578 | d += i; |
579 | s += i; |
4757a243 |
580 | continue; |
79072805 |
581 | } |
4757a243 |
582 | else if (uv == extra && !del) { |
583 | matches++; |
01ec43d0 |
584 | s += UTF8SKIP(s); |
7d85a32c |
585 | d = uvuni_to_utf8(d, final); |
4757a243 |
586 | continue; |
587 | } |
01ec43d0 |
588 | matches++; /* "none+1" is delete character */ |
589 | s += UTF8SKIP(s); |
79072805 |
590 | } |
4757a243 |
591 | } |
16ec844d |
592 | if (grows || hibit) { |
9b877dbb |
593 | sv_setpvn(sv, (char*)dstart, d - dstart); |
594 | Safefree(dstart); |
16ec844d |
595 | if (grows && hibit) |
596 | Safefree(start); |
9b877dbb |
597 | } |
598 | else { |
599 | *d = '\0'; |
600 | SvCUR_set(sv, d - dstart); |
601 | } |
1aa99e6b |
602 | SvUTF8_on(sv); |
4757a243 |
603 | SvSETMAGIC(sv); |
604 | |
605 | return matches; |
606 | } |
607 | |
608 | I32 |
864dbfa3 |
609 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
610 | { |
97aff369 |
611 | dVAR; |
4757a243 |
612 | STRLEN len; |
4373e329 |
613 | const I32 hasutf = (PL_op->op_private & |
036b4402 |
614 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
615 | |
2233f375 |
616 | if (SvREADONLY(sv)) { |
765f542d |
617 | if (SvIsCOW(sv)) |
618 | sv_force_normal_flags(sv, 0); |
2233f375 |
619 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
620 | Perl_croak(aTHX_ PL_no_modify); |
621 | } |
10516c54 |
622 | (void)SvPV_const(sv, len); |
4757a243 |
623 | if (!len) |
624 | return 0; |
d59e14db |
625 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
626 | if (!SvPOKp(sv)) |
627 | (void)SvPV_force(sv, len); |
2de7b02f |
628 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
629 | } |
4757a243 |
630 | |
cea2e8a9 |
631 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
632 | |
a77e643a |
633 | switch (PL_op->op_private & ~hasutf & ( |
634 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| |
635 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { |
4757a243 |
636 | case 0: |
01ec43d0 |
637 | if (hasutf) |
638 | return do_trans_simple_utf8(sv); |
639 | else |
640 | return do_trans_simple(sv); |
4757a243 |
641 | |
642 | case OPpTRANS_IDENTICAL: |
05d340b8 |
643 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
644 | if (hasutf) |
645 | return do_trans_count_utf8(sv); |
646 | else |
647 | return do_trans_count(sv); |
4757a243 |
648 | |
649 | default: |
01ec43d0 |
650 | if (hasutf) |
036b4402 |
651 | return do_trans_complex_utf8(sv); |
4757a243 |
652 | else |
036b4402 |
653 | return do_trans_complex(sv); |
79072805 |
654 | } |
79072805 |
655 | } |
656 | |
657 | void |
864dbfa3 |
658 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
659 | { |
97aff369 |
660 | dVAR; |
53c1dcc0 |
661 | SV ** const oldmark = mark; |
79072805 |
662 | register I32 items = sp - mark; |
79072805 |
663 | register STRLEN len; |
463ee0b2 |
664 | STRLEN delimlen; |
79072805 |
665 | |
5c144d81 |
666 | (void) SvPV_const(del, delimlen); /* stringify and get the delimlen */ |
5ba99574 |
667 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
668 | |
79072805 |
669 | mark++; |
670 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
862a34c6 |
671 | SvUPGRADE(sv, SVt_PV); |
79072805 |
672 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
673 | while (items-- > 0) { |
1426bbf4 |
674 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a |
675 | STRLEN tmplen; |
5c144d81 |
676 | SvPV_const(*mark, tmplen); |
463ee0b2 |
677 | len += tmplen; |
79072805 |
678 | } |
679 | mark++; |
680 | } |
681 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
682 | |
683 | mark = oldmark; |
db7c17d7 |
684 | items = sp - mark; |
79072805 |
685 | ++mark; |
686 | } |
687 | |
fc0199cd |
688 | sv_setpvn(sv, "", 0); |
e4803c42 |
689 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 |
690 | SvUTF8_off(sv); |
e4803c42 |
691 | |
8d6d96c1 |
692 | if (PL_tainting && SvMAGICAL(sv)) |
693 | SvTAINTED_off(sv); |
694 | |
463ee0b2 |
695 | if (items-- > 0) { |
92d29cee |
696 | if (*mark) |
697 | sv_catsv(sv, *mark); |
463ee0b2 |
698 | mark++; |
699 | } |
8d6d96c1 |
700 | |
c512ce4f |
701 | if (delimlen) { |
79072805 |
702 | for (; items > 0; items--,mark++) { |
c512ce4f |
703 | sv_catsv(sv,del); |
79072805 |
704 | sv_catsv(sv,*mark); |
705 | } |
706 | } |
707 | else { |
708 | for (; items > 0; items--,mark++) |
709 | sv_catsv(sv,*mark); |
710 | } |
711 | SvSETMAGIC(sv); |
712 | } |
713 | |
714 | void |
864dbfa3 |
715 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
716 | { |
97aff369 |
717 | dVAR; |
46fc3d4c |
718 | STRLEN patlen; |
53c1dcc0 |
719 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c |
720 | bool do_taint = FALSE; |
721 | |
5b781b5b |
722 | SvUTF8_off(sv); |
2cf2cfc6 |
723 | if (DO_UTF8(*sarg)) |
724 | SvUTF8_on(sv); |
46fc3d4c |
725 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
726 | SvSETMAGIC(sv); |
46fc3d4c |
727 | if (do_taint) |
728 | SvTAINTED_on(sv); |
79072805 |
729 | } |
730 | |
33b45480 |
731 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
732 | UV |
733 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
734 | { |
97aff369 |
735 | dVAR; |
81e118e0 |
736 | STRLEN srclen, len; |
10516c54 |
737 | const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); |
81e118e0 |
738 | UV retnum = 0; |
739 | |
a50d7633 |
740 | if (offset < 0) |
81e118e0 |
741 | return retnum; |
8e84507e |
742 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
743 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
744 | |
dcad2880 |
745 | if (SvUTF8(sv)) |
33b45480 |
746 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
747 | |
81e118e0 |
748 | offset *= size; /* turn into bit offset */ |
749 | len = (offset + size + 7) / 8; /* required number of bytes */ |
750 | if (len > srclen) { |
751 | if (size <= 8) |
752 | retnum = 0; |
753 | else { |
754 | offset >>= 3; /* turn into byte offset */ |
755 | if (size == 16) { |
eb160463 |
756 | if ((STRLEN)offset >= srclen) |
81e118e0 |
757 | retnum = 0; |
758 | else |
628e1a40 |
759 | retnum = (UV) s[offset] << 8; |
81e118e0 |
760 | } |
761 | else if (size == 32) { |
eb160463 |
762 | if ((STRLEN)offset >= srclen) |
81e118e0 |
763 | retnum = 0; |
eb160463 |
764 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 |
765 | retnum = |
766 | ((UV) s[offset ] << 24); |
eb160463 |
767 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 |
768 | retnum = |
769 | ((UV) s[offset ] << 24) + |
770 | ((UV) s[offset + 1] << 16); |
771 | else |
772 | retnum = |
773 | ((UV) s[offset ] << 24) + |
774 | ((UV) s[offset + 1] << 16) + |
775 | ( s[offset + 2] << 8); |
776 | } |
d7d93a81 |
777 | #ifdef UV_IS_QUAD |
c5a0f51a |
778 | else if (size == 64) { |
c5a0f51a |
779 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
780 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
781 | "Bit vector size > 32 non-portable"); |
782 | if (offset >= srclen) |
783 | retnum = 0; |
784 | else if (offset + 1 >= srclen) |
785 | retnum = |
786 | (UV) s[offset ] << 56; |
787 | else if (offset + 2 >= srclen) |
788 | retnum = |
789 | ((UV) s[offset ] << 56) + |
790 | ((UV) s[offset + 1] << 48); |
791 | else if (offset + 3 >= srclen) |
792 | retnum = |
793 | ((UV) s[offset ] << 56) + |
794 | ((UV) s[offset + 1] << 48) + |
795 | ((UV) s[offset + 2] << 40); |
796 | else if (offset + 4 >= 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 | else if (offset + 5 >= srclen) |
803 | retnum = |
804 | ((UV) s[offset ] << 56) + |
805 | ((UV) s[offset + 1] << 48) + |
806 | ((UV) s[offset + 2] << 40) + |
807 | ((UV) s[offset + 3] << 32) + |
808 | ( s[offset + 4] << 24); |
809 | else if (offset + 6 >= srclen) |
810 | retnum = |
811 | ((UV) s[offset ] << 56) + |
812 | ((UV) s[offset + 1] << 48) + |
813 | ((UV) s[offset + 2] << 40) + |
814 | ((UV) s[offset + 3] << 32) + |
815 | ((UV) s[offset + 4] << 24) + |
816 | ((UV) s[offset + 5] << 16); |
817 | else |
8e84507e |
818 | retnum = |
c5a0f51a |
819 | ((UV) s[offset ] << 56) + |
820 | ((UV) s[offset + 1] << 48) + |
821 | ((UV) s[offset + 2] << 40) + |
822 | ((UV) s[offset + 3] << 32) + |
823 | ((UV) s[offset + 4] << 24) + |
824 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
825 | ( s[offset + 6] << 8); |
c5a0f51a |
826 | } |
827 | #endif |
81e118e0 |
828 | } |
829 | } |
830 | else if (size < 8) |
831 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
832 | else { |
833 | offset >>= 3; /* turn into byte offset */ |
834 | if (size == 8) |
835 | retnum = s[offset]; |
836 | else if (size == 16) |
837 | retnum = |
628e1a40 |
838 | ((UV) s[offset] << 8) + |
81e118e0 |
839 | s[offset + 1]; |
840 | else if (size == 32) |
841 | retnum = |
842 | ((UV) s[offset ] << 24) + |
843 | ((UV) s[offset + 1] << 16) + |
844 | ( s[offset + 2] << 8) + |
845 | s[offset + 3]; |
d7d93a81 |
846 | #ifdef UV_IS_QUAD |
c5a0f51a |
847 | else if (size == 64) { |
c5a0f51a |
848 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
849 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
850 | "Bit vector size > 32 non-portable"); |
851 | retnum = |
852 | ((UV) s[offset ] << 56) + |
853 | ((UV) s[offset + 1] << 48) + |
854 | ((UV) s[offset + 2] << 40) + |
855 | ((UV) s[offset + 3] << 32) + |
856 | ((UV) s[offset + 4] << 24) + |
857 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
858 | ( s[offset + 6] << 8) + |
c5a0f51a |
859 | s[offset + 7]; |
860 | } |
861 | #endif |
81e118e0 |
862 | } |
863 | |
864 | return retnum; |
865 | } |
866 | |
33b45480 |
867 | /* currently converts input to bytes if possible but doesn't sweat failures, |
868 | * although it does ensure that the string it clobbers is not marked as |
869 | * utf8-valid any more |
870 | */ |
79072805 |
871 | void |
864dbfa3 |
872 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
873 | { |
97aff369 |
874 | dVAR; |
79072805 |
875 | register I32 offset; |
876 | register I32 size; |
8990e307 |
877 | register unsigned char *s; |
81e118e0 |
878 | register UV lval; |
79072805 |
879 | I32 mask; |
a0d0e21e |
880 | STRLEN targlen; |
881 | STRLEN len; |
c4420975 |
882 | SV * const targ = LvTARG(sv); |
79072805 |
883 | |
8990e307 |
884 | if (!targ) |
885 | return; |
a0d0e21e |
886 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
887 | if (SvUTF8(targ)) { |
33b45480 |
888 | /* This is handled by the SvPOK_only below... |
889 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
890 | SvUTF8_off(targ); |
891 | */ |
892 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
893 | } |
894 | |
4ebbc975 |
895 | (void)SvPOK_only(targ); |
81e118e0 |
896 | lval = SvUV(sv); |
79072805 |
897 | offset = LvTARGOFF(sv); |
fe58ced6 |
898 | if (offset < 0) |
ed9aa3b7 |
899 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
900 | size = LvTARGLEN(sv); |
8e84507e |
901 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
902 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
903 | |
81e118e0 |
904 | offset *= size; /* turn into bit offset */ |
905 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
906 | if (len > targlen) { |
907 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
908 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
909 | SvCUR_set(targ, len); |
910 | } |
8e84507e |
911 | |
79072805 |
912 | if (size < 8) { |
913 | mask = (1 << size) - 1; |
914 | size = offset & 7; |
915 | lval &= mask; |
81e118e0 |
916 | offset >>= 3; /* turn into byte offset */ |
79072805 |
917 | s[offset] &= ~(mask << size); |
918 | s[offset] |= lval << size; |
919 | } |
920 | else { |
81e118e0 |
921 | offset >>= 3; /* turn into byte offset */ |
79072805 |
922 | if (size == 8) |
eb160463 |
923 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
924 | else if (size == 16) { |
eb160463 |
925 | s[offset ] = (U8)((lval >> 8) & 0xff); |
926 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
927 | } |
928 | else if (size == 32) { |
eb160463 |
929 | s[offset ] = (U8)((lval >> 24) & 0xff); |
930 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
931 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
932 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
933 | } |
d7d93a81 |
934 | #ifdef UV_IS_QUAD |
c5a0f51a |
935 | else if (size == 64) { |
c5a0f51a |
936 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
937 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
938 | "Bit vector size > 32 non-portable"); |
eb160463 |
939 | s[offset ] = (U8)((lval >> 56) & 0xff); |
940 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
941 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
942 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
943 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
944 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
945 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
946 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
947 | } |
dc1e3f56 |
948 | #endif |
79072805 |
949 | } |
7bb043c3 |
950 | SvSETMAGIC(targ); |
79072805 |
951 | } |
952 | |
953 | void |
864dbfa3 |
954 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
955 | { |
97aff369 |
956 | dVAR; |
463ee0b2 |
957 | STRLEN len; |
a0d0e21e |
958 | char *s; |
8e84507e |
959 | |
79072805 |
960 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
961 | register I32 i; |
c4420975 |
962 | AV* const av = (AV*)sv; |
f54cb97a |
963 | const I32 max = AvFILL(av); |
964 | |
965 | for (i = 0; i <= max; i++) { |
a0d0e21e |
966 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
967 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
968 | do_chop(astr, sv); |
969 | } |
970 | return; |
79072805 |
971 | } |
aa854799 |
972 | else if (SvTYPE(sv) == SVt_PVHV) { |
c4420975 |
973 | HV* const hv = (HV*)sv; |
a0d0e21e |
974 | HE* entry; |
975 | (void)hv_iterinit(hv); |
155aba94 |
976 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
977 | do_chop(astr,hv_iterval(hv,entry)); |
978 | return; |
79072805 |
979 | } |
991e6d41 |
980 | else if (SvREADONLY(sv)) { |
981 | if (SvFAKE(sv)) { |
982 | /* SV is copy-on-write */ |
983 | sv_force_normal_flags(sv, 0); |
984 | } |
985 | if (SvREADONLY(sv)) |
986 | Perl_croak(aTHX_ PL_no_modify); |
987 | } |
565a3db3 |
988 | |
989 | if (PL_encoding && !SvUTF8(sv)) { |
990 | /* like in do_chomp(), utf8-ize the sv as a side-effect |
991 | * if we're using encoding. */ |
992 | sv_recode_to_utf8(sv, PL_encoding); |
993 | } |
994 | |
a0d0e21e |
995 | s = SvPV(sv, len); |
748a9306 |
996 | if (len && !SvPOK(sv)) |
a0d0e21e |
997 | s = SvPV_force(sv, len); |
7e2040f0 |
998 | if (DO_UTF8(sv)) { |
a0ed51b3 |
999 | if (s && len) { |
c4420975 |
1000 | char * const send = s + len; |
0bcc34c2 |
1001 | char * const start = s; |
a0ed51b3 |
1002 | s = send - 1; |
a0dbb045 |
1003 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
1004 | s--; |
ace7757b |
1005 | if (is_utf8_string((U8*)s, send - s)) { |
a0dbb045 |
1006 | sv_setpvn(astr, s, send - s); |
1007 | *s = '\0'; |
1008 | SvCUR_set(sv, s - start); |
1009 | SvNIOK_off(sv); |
1010 | SvUTF8_on(astr); |
1011 | } |
a0ed51b3 |
1012 | } |
1013 | else |
1014 | sv_setpvn(astr, "", 0); |
1015 | } |
7e2040f0 |
1016 | else if (s && len) { |
a0d0e21e |
1017 | s += --len; |
1018 | sv_setpvn(astr, s, 1); |
1019 | *s = '\0'; |
1020 | SvCUR_set(sv, len); |
2c19a612 |
1021 | SvUTF8_off(sv); |
a0d0e21e |
1022 | SvNIOK_off(sv); |
79072805 |
1023 | } |
1024 | else |
a0d0e21e |
1025 | sv_setpvn(astr, "", 0); |
1026 | SvSETMAGIC(sv); |
7e2040f0 |
1027 | } |
a0d0e21e |
1028 | |
1029 | I32 |
864dbfa3 |
1030 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1031 | { |
97aff369 |
1032 | dVAR; |
c07a80fd |
1033 | register I32 count; |
a0d0e21e |
1034 | STRLEN len; |
1035 | char *s; |
c4c87a06 |
1036 | char *temp_buffer = NULL; |
a0714e2c |
1037 | SV* svrecode = NULL; |
c07a80fd |
1038 | |
3280af22 |
1039 | if (RsSNARF(PL_rs)) |
c07a80fd |
1040 | return 0; |
4c5a6083 |
1041 | if (RsRECORD(PL_rs)) |
1042 | return 0; |
c07a80fd |
1043 | count = 0; |
a0d0e21e |
1044 | if (SvTYPE(sv) == SVt_PVAV) { |
1045 | register I32 i; |
890ce7af |
1046 | AV* const av = (AV*)sv; |
f54cb97a |
1047 | const I32 max = AvFILL(av); |
1048 | |
1049 | for (i = 0; i <= max; i++) { |
a0d0e21e |
1050 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1051 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1052 | count += do_chomp(sv); |
1053 | } |
1054 | return count; |
1055 | } |
aa854799 |
1056 | else if (SvTYPE(sv) == SVt_PVHV) { |
890ce7af |
1057 | HV* const hv = (HV*)sv; |
a0d0e21e |
1058 | HE* entry; |
1059 | (void)hv_iterinit(hv); |
155aba94 |
1060 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1061 | count += do_chomp(hv_iterval(hv,entry)); |
1062 | return count; |
1063 | } |
991e6d41 |
1064 | else if (SvREADONLY(sv)) { |
1065 | if (SvFAKE(sv)) { |
1066 | /* SV is copy-on-write */ |
1067 | sv_force_normal_flags(sv, 0); |
1068 | } |
1069 | if (SvREADONLY(sv)) |
1070 | Perl_croak(aTHX_ PL_no_modify); |
1071 | } |
a6aa349d |
1072 | |
1073 | if (PL_encoding) { |
1074 | if (!SvUTF8(sv)) { |
1075 | /* XXX, here sv is utf8-ized as a side-effect! |
1076 | If encoding.pm is used properly, almost string-generating |
1077 | operations, including literal strings, chr(), input data, etc. |
1078 | should have been utf8-ized already, right? |
1079 | */ |
1080 | sv_recode_to_utf8(sv, PL_encoding); |
1081 | } |
1082 | } |
1083 | |
a0d0e21e |
1084 | s = SvPV(sv, len); |
a0d0e21e |
1085 | if (s && len) { |
1086 | s += --len; |
3280af22 |
1087 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1088 | if (*s != '\n') |
1089 | goto nope; |
1090 | ++count; |
1091 | while (len && s[-1] == '\n') { |
1092 | --len; |
1093 | --s; |
1094 | ++count; |
1095 | } |
1096 | } |
a0d0e21e |
1097 | else { |
a6aa349d |
1098 | STRLEN rslen, rs_charlen; |
93524f2b |
1099 | const char *rsptr = SvPV_const(PL_rs, rslen); |
a6aa349d |
1100 | |
1101 | rs_charlen = SvUTF8(PL_rs) |
1102 | ? sv_len_utf8(PL_rs) |
1103 | : rslen; |
1104 | |
c4c87a06 |
1105 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1106 | /* Assumption is that rs is shorter than the scalar. */ |
1107 | if (SvUTF8(PL_rs)) { |
1108 | /* RS is utf8, scalar is 8 bit. */ |
1109 | bool is_utf8 = TRUE; |
1110 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, |
1111 | &rslen, &is_utf8); |
1112 | if (is_utf8) { |
1113 | /* Cannot downgrade, therefore cannot possibly match |
1114 | */ |
1115 | assert (temp_buffer == rsptr); |
1116 | temp_buffer = NULL; |
1117 | goto nope; |
1118 | } |
1119 | rsptr = temp_buffer; |
a6aa349d |
1120 | } |
1121 | else if (PL_encoding) { |
1122 | /* RS is 8 bit, encoding.pm is used. |
1123 | * Do not recode PL_rs as a side-effect. */ |
1124 | svrecode = newSVpvn(rsptr, rslen); |
1125 | sv_recode_to_utf8(svrecode, PL_encoding); |
93524f2b |
1126 | rsptr = SvPV_const(svrecode, rslen); |
a6aa349d |
1127 | rs_charlen = sv_len_utf8(svrecode); |
1128 | } |
1129 | else { |
c4c87a06 |
1130 | /* RS is 8 bit, scalar is utf8. */ |
1131 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); |
1132 | rsptr = temp_buffer; |
1133 | } |
1134 | } |
c07a80fd |
1135 | if (rslen == 1) { |
1136 | if (*s != *rsptr) |
1137 | goto nope; |
1138 | ++count; |
1139 | } |
1140 | else { |
8c2cee6f |
1141 | if (len < rslen - 1) |
c07a80fd |
1142 | goto nope; |
1143 | len -= rslen - 1; |
1144 | s -= rslen - 1; |
36477c24 |
1145 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1146 | goto nope; |
a6aa349d |
1147 | count += rs_charlen; |
c07a80fd |
1148 | } |
a0d0e21e |
1149 | } |
8b6b16e7 |
1150 | s = SvPV_force_nolen(sv); |
a0d0e21e |
1151 | SvCUR_set(sv, len); |
9b33ce3b |
1152 | *SvEND(sv) = '\0'; |
a0d0e21e |
1153 | SvNIOK_off(sv); |
9b33ce3b |
1154 | SvSETMAGIC(sv); |
a0d0e21e |
1155 | } |
1156 | nope: |
a6aa349d |
1157 | |
1158 | if (svrecode) |
1159 | SvREFCNT_dec(svrecode); |
1160 | |
c4c87a06 |
1161 | Safefree(temp_buffer); |
a0d0e21e |
1162 | return count; |
8e84507e |
1163 | } |
79072805 |
1164 | |
1165 | void |
864dbfa3 |
1166 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1167 | { |
97aff369 |
1168 | dVAR; |
79072805 |
1169 | #ifdef LIBERAL |
1170 | register long *dl; |
1171 | register long *ll; |
1172 | register long *rl; |
1173 | #endif |
1174 | register char *dc; |
463ee0b2 |
1175 | STRLEN leftlen; |
1176 | STRLEN rightlen; |
e62f0680 |
1177 | register const char *lc; |
1178 | register const char *rc; |
79072805 |
1179 | register I32 len; |
a0d0e21e |
1180 | I32 lensave; |
e62f0680 |
1181 | const char *lsave; |
1182 | const char *rsave; |
f54cb97a |
1183 | const bool left_utf = DO_UTF8(left); |
1184 | const bool right_utf = DO_UTF8(right); |
b7953727 |
1185 | I32 needlen = 0; |
0c57e439 |
1186 | |
1187 | if (left_utf && !right_utf) |
1188 | sv_utf8_upgrade(right); |
a1ca4561 |
1189 | else if (!left_utf && right_utf) |
0c57e439 |
1190 | sv_utf8_upgrade(left); |
79072805 |
1191 | |
1fbd88dc |
1192 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1193 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
e62f0680 |
1194 | lsave = lc = SvPV_nomg_const(left, leftlen); |
1195 | rsave = rc = SvPV_nomg_const(right, rightlen); |
93a17b20 |
1196 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1197 | lensave = len; |
c9b3c8d0 |
1198 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1199 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
a02a5408 |
1200 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 |
1201 | } |
1202 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe |
1203 | dc = SvPV_force_nomg_nolen(sv); |
1a787b95 |
1204 | if (SvLEN(sv) < (STRLEN)(len + 1)) { |
eb160463 |
1205 | dc = SvGROW(sv, (STRLEN)(len + 1)); |
ff68c719 |
1206 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1207 | } |
c9b3c8d0 |
1208 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1209 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1210 | } |
1211 | else { |
c9b3c8d0 |
1212 | needlen = ((optype == OP_BIT_AND) |
1213 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
a02a5408 |
1214 | Newxz(dc, needlen + 1, char); |
ff68c719 |
1215 | (void)sv_usepvn(sv, dc, needlen); |
1216 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1217 | } |
a0d0e21e |
1218 | SvCUR_set(sv, len); |
1219 | (void)SvPOK_only(sv); |
0c57e439 |
1220 | if (left_utf || right_utf) { |
1221 | UV duc, luc, ruc; |
c4420975 |
1222 | char * const dcsave = dc; |
0c57e439 |
1223 | STRLEN lulen = leftlen; |
1224 | STRLEN rulen = rightlen; |
ba210ebe |
1225 | STRLEN ulen; |
0c57e439 |
1226 | |
1227 | switch (optype) { |
1228 | case OP_BIT_AND: |
1229 | while (lulen && rulen) { |
9041c2e3 |
1230 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1231 | lc += ulen; |
1232 | lulen -= ulen; |
9041c2e3 |
1233 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1234 | rc += ulen; |
1235 | rulen -= ulen; |
1236 | duc = luc & ruc; |
9041c2e3 |
1237 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1238 | } |
c9b3c8d0 |
1239 | if (sv == left || sv == right) |
1240 | (void)sv_usepvn(sv, dcsave, needlen); |
1241 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1242 | break; |
1243 | case OP_BIT_XOR: |
1244 | while (lulen && rulen) { |
9041c2e3 |
1245 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1246 | lc += ulen; |
1247 | lulen -= ulen; |
9041c2e3 |
1248 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1249 | rc += ulen; |
1250 | rulen -= ulen; |
1251 | duc = luc ^ ruc; |
9041c2e3 |
1252 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1253 | } |
1254 | goto mop_up_utf; |
1255 | case OP_BIT_OR: |
1256 | while (lulen && rulen) { |
9041c2e3 |
1257 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1258 | lc += ulen; |
1259 | lulen -= ulen; |
9041c2e3 |
1260 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1261 | rc += ulen; |
1262 | rulen -= ulen; |
1263 | duc = luc | ruc; |
9041c2e3 |
1264 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1265 | } |
1266 | mop_up_utf: |
c9b3c8d0 |
1267 | if (sv == left || sv == right) |
1268 | (void)sv_usepvn(sv, dcsave, needlen); |
1269 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1270 | if (rulen) |
1271 | sv_catpvn(sv, rc, rulen); |
1272 | else if (lulen) |
1273 | sv_catpvn(sv, lc, lulen); |
1274 | else |
1275 | *SvEND(sv) = '\0'; |
1276 | break; |
1277 | } |
1278 | SvUTF8_on(sv); |
1279 | goto finish; |
1280 | } |
1281 | else |
79072805 |
1282 | #ifdef LIBERAL |
1283 | if (len >= sizeof(long)*4 && |
1284 | !((long)dc % sizeof(long)) && |
1285 | !((long)lc % sizeof(long)) && |
1286 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1287 | { |
f54cb97a |
1288 | const I32 remainder = len % (sizeof(long)*4); |
79072805 |
1289 | len /= (sizeof(long)*4); |
1290 | |
1291 | dl = (long*)dc; |
1292 | ll = (long*)lc; |
1293 | rl = (long*)rc; |
1294 | |
1295 | switch (optype) { |
1296 | case OP_BIT_AND: |
1297 | while (len--) { |
1298 | *dl++ = *ll++ & *rl++; |
1299 | *dl++ = *ll++ & *rl++; |
1300 | *dl++ = *ll++ & *rl++; |
1301 | *dl++ = *ll++ & *rl++; |
1302 | } |
1303 | break; |
a0d0e21e |
1304 | case OP_BIT_XOR: |
79072805 |
1305 | while (len--) { |
1306 | *dl++ = *ll++ ^ *rl++; |
1307 | *dl++ = *ll++ ^ *rl++; |
1308 | *dl++ = *ll++ ^ *rl++; |
1309 | *dl++ = *ll++ ^ *rl++; |
1310 | } |
1311 | break; |
1312 | case OP_BIT_OR: |
1313 | while (len--) { |
1314 | *dl++ = *ll++ | *rl++; |
1315 | *dl++ = *ll++ | *rl++; |
1316 | *dl++ = *ll++ | *rl++; |
1317 | *dl++ = *ll++ | *rl++; |
1318 | } |
1319 | } |
1320 | |
1321 | dc = (char*)dl; |
1322 | lc = (char*)ll; |
1323 | rc = (char*)rl; |
1324 | |
1325 | len = remainder; |
1326 | } |
1327 | #endif |
a0d0e21e |
1328 | { |
a0d0e21e |
1329 | switch (optype) { |
1330 | case OP_BIT_AND: |
1331 | while (len--) |
1332 | *dc++ = *lc++ & *rc++; |
1a787b95 |
1333 | *dc = '\0'; |
a0d0e21e |
1334 | break; |
1335 | case OP_BIT_XOR: |
1336 | while (len--) |
1337 | *dc++ = *lc++ ^ *rc++; |
1338 | goto mop_up; |
1339 | case OP_BIT_OR: |
1340 | while (len--) |
1341 | *dc++ = *lc++ | *rc++; |
1342 | mop_up: |
1343 | len = lensave; |
eb160463 |
1344 | if (rightlen > (STRLEN)len) |
a0d0e21e |
1345 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1346 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1347 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1348 | else |
1349 | *SvEND(sv) = '\0'; |
a0d0e21e |
1350 | break; |
1351 | } |
79072805 |
1352 | } |
0c57e439 |
1353 | finish: |
fb73857a |
1354 | SvTAINT(sv); |
79072805 |
1355 | } |
463ee0b2 |
1356 | |
1357 | OP * |
cea2e8a9 |
1358 | Perl_do_kv(pTHX) |
463ee0b2 |
1359 | { |
97aff369 |
1360 | dVAR; |
39644a26 |
1361 | dSP; |
c4420975 |
1362 | HV * const hv = (HV*)POPs; |
800e9ae0 |
1363 | HV *keys; |
463ee0b2 |
1364 | register HE *entry; |
f54cb97a |
1365 | const I32 gimme = GIMME_V; |
1366 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); |
1367 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); |
1368 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); |
463ee0b2 |
1369 | |
85581909 |
1370 | if (!hv) { |
78f9721b |
1371 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1372 | dTARGET; /* make sure to clear its target here */ |
1373 | if (SvTYPE(TARG) == SVt_PVLV) |
a0714e2c |
1374 | LvTARG(TARG) = NULL; |
85581909 |
1375 | PUSHs(TARG); |
1376 | } |
463ee0b2 |
1377 | RETURN; |
85581909 |
1378 | } |
748a9306 |
1379 | |
6d822dc4 |
1380 | keys = hv; |
800e9ae0 |
1381 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1382 | |
54310121 |
1383 | if (gimme == G_VOID) |
aa689395 |
1384 | RETURN; |
1385 | |
54310121 |
1386 | if (gimme == G_SCALAR) { |
6ee623d5 |
1387 | IV i; |
463ee0b2 |
1388 | dTARGET; |
1389 | |
78f9721b |
1390 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1391 | if (SvTYPE(TARG) < SVt_PVLV) { |
1392 | sv_upgrade(TARG, SVt_PVLV); |
a0714e2c |
1393 | sv_magic(TARG, NULL, PERL_MAGIC_nkeys, NULL, 0); |
85581909 |
1394 | } |
1395 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1396 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1397 | if (LvTARG(TARG)) |
1398 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1399 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1400 | } |
85581909 |
1401 | PUSHs(TARG); |
1402 | RETURN; |
1403 | } |
1404 | |
14befaf4 |
1405 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 |
1406 | i = HvKEYS(keys); |
463ee0b2 |
1407 | else { |
1408 | i = 0; |
800e9ae0 |
1409 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1410 | } |
1411 | PUSHi( i ); |
1412 | RETURN; |
1413 | } |
1414 | |
8ed4b672 |
1415 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1416 | |
463ee0b2 |
1417 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1418 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1419 | SPAGAIN; |
574c8022 |
1420 | if (dokeys) { |
890ce7af |
1421 | SV* const sv = hv_iterkeysv(entry); |
574c8022 |
1422 | XPUSHs(sv); /* won't clobber stack_sp */ |
1423 | } |
463ee0b2 |
1424 | if (dovalues) { |
f54cb97a |
1425 | SV *tmpstr; |
463ee0b2 |
1426 | PUTBACK; |
6d822dc4 |
1427 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1428 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1429 | (unsigned long)HeHASH(entry), |
cdd3ba14 |
1430 | (int)HvMAX(keys)+1, |
800e9ae0 |
1431 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1432 | SPAGAIN; |
46fc3d4c |
1433 | XPUSHs(tmpstr); |
463ee0b2 |
1434 | } |
1435 | PUTBACK; |
1436 | } |
1437 | return NORMAL; |
1438 | } |
4e35701f |
1439 | |
af3babe4 |
1440 | /* |
1441 | * Local variables: |
1442 | * c-indentation-style: bsd |
1443 | * c-basic-offset: 4 |
1444 | * indent-tabs-mode: t |
1445 | * End: |
1446 | * |
37442d52 |
1447 | * ex: set ts=8 sts=4 sw=4 noet: |
1448 | */ |