Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
b5f8cc5c |
4 | * 2000, 2001, 2002, 2004, by Larry Wall and others |
79072805 |
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; |
4757a243 |
33 | U8 *send; |
b250498f |
34 | U8 *dstart; |
4757a243 |
35 | I32 matches = 0; |
9b877dbb |
36 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
37 | STRLEN len; |
4757a243 |
38 | short *tbl; |
39 | I32 ch; |
79072805 |
40 | |
4757a243 |
41 | tbl = (short*)cPVOP->op_pv; |
42 | if (!tbl) |
7d85a32c |
43 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
44 | |
4757a243 |
45 | s = (U8*)SvPV(sv, len); |
46 | send = s + len; |
47 | |
1e54db1a |
48 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
49 | if (!SvUTF8(sv)) { |
01ec43d0 |
50 | while (s < send) { |
036b4402 |
51 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 |
52 | matches++; |
eb160463 |
53 | *s++ = (U8)ch; |
01ec43d0 |
54 | } |
55 | else |
56 | s++; |
57 | } |
58 | SvSETMAGIC(sv); |
b250498f |
59 | return matches; |
60 | } |
4757a243 |
61 | |
b250498f |
62 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
63 | if (grows) |
64 | New(0, d, len*2+1, U8); |
65 | else |
66 | d = s; |
b250498f |
67 | dstart = d; |
68 | while (s < send) { |
ba210ebe |
69 | STRLEN ulen; |
9b877dbb |
70 | UV c; |
b250498f |
71 | |
b250498f |
72 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 |
73 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb |
74 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
75 | matches++; |
9041c2e3 |
76 | d = uvchr_to_utf8(d, ch); |
b250498f |
77 | s += ulen; |
01ec43d0 |
78 | } |
79 | else { /* No match -> copy */ |
0376ff32 |
80 | Move(s, d, ulen, U8); |
9b877dbb |
81 | d += ulen; |
82 | s += ulen; |
b250498f |
83 | } |
84 | } |
9b877dbb |
85 | if (grows) { |
86 | sv_setpvn(sv, (char*)dstart, d - dstart); |
87 | Safefree(dstart); |
88 | } |
89 | else { |
90 | *d = '\0'; |
91 | SvCUR_set(sv, d - dstart); |
92 | } |
b250498f |
93 | SvUTF8_on(sv); |
b250498f |
94 | SvSETMAGIC(sv); |
4757a243 |
95 | return matches; |
96 | } |
97 | |
942e002e |
98 | STATIC I32 |
21304a3b |
99 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 |
100 | { |
4757a243 |
101 | U8 *s; |
102 | U8 *send; |
103 | I32 matches = 0; |
104 | STRLEN len; |
105 | short *tbl; |
8973db79 |
106 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 |
107 | |
108 | tbl = (short*)cPVOP->op_pv; |
109 | if (!tbl) |
7d85a32c |
110 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
111 | |
112 | s = (U8*)SvPV(sv, len); |
113 | send = s + len; |
114 | |
1aa99e6b |
115 | if (!SvUTF8(sv)) |
116 | while (s < send) { |
117 | if (tbl[*s++] >= 0) |
036b4402 |
118 | matches++; |
1aa99e6b |
119 | } |
120 | else |
121 | while (s < send) { |
122 | UV c; |
123 | STRLEN ulen; |
9041c2e3 |
124 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 |
125 | if (c < 0x100) { |
126 | if (tbl[c] >= 0) |
127 | matches++; |
128 | } else if (complement) |
1aa99e6b |
129 | matches++; |
130 | s += ulen; |
131 | } |
4757a243 |
132 | |
133 | return matches; |
134 | } |
135 | |
942e002e |
136 | STATIC I32 |
21304a3b |
137 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 |
138 | { |
4757a243 |
139 | U8 *s; |
140 | U8 *send; |
141 | U8 *d; |
5e44153e |
142 | U8 *dstart; |
1aa99e6b |
143 | I32 isutf8; |
4757a243 |
144 | I32 matches = 0; |
9b877dbb |
145 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 |
146 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
147 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
b7953727 |
148 | STRLEN len, rlen = 0; |
4757a243 |
149 | short *tbl; |
150 | I32 ch; |
151 | |
152 | tbl = (short*)cPVOP->op_pv; |
153 | if (!tbl) |
7d85a32c |
154 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
155 | |
156 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
157 | isutf8 = SvUTF8(sv); |
4757a243 |
158 | send = s + len; |
159 | |
1aa99e6b |
160 | if (!isutf8) { |
161 | dstart = d = s; |
162 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
163 | U8* p = send; |
164 | while (s < send) { |
165 | if ((ch = tbl[*s]) >= 0) { |
eb160463 |
166 | *d = (U8)ch; |
036b4402 |
167 | matches++; |
1aa99e6b |
168 | if (p != d - 1 || *p != *d) |
169 | p = d++; |
170 | } |
171 | else if (ch == -1) /* -1 is unmapped character */ |
172 | *d++ = *s; |
173 | else if (ch == -2) /* -2 is delete character */ |
174 | matches++; |
175 | s++; |
176 | } |
a0ed51b3 |
177 | } |
1aa99e6b |
178 | else { |
179 | while (s < send) { |
180 | if ((ch = tbl[*s]) >= 0) { |
181 | matches++; |
eb160463 |
182 | *d++ = (U8)ch; |
1aa99e6b |
183 | } |
184 | else if (ch == -1) /* -1 is unmapped character */ |
185 | *d++ = *s; |
186 | else if (ch == -2) /* -2 is delete character */ |
187 | matches++; |
188 | s++; |
189 | } |
190 | } |
76ef7183 |
191 | *d = '\0'; |
1aa99e6b |
192 | SvCUR_set(sv, d - dstart); |
4757a243 |
193 | } |
1aa99e6b |
194 | else { /* isutf8 */ |
9b877dbb |
195 | if (grows) |
196 | New(0, d, len*2+1, U8); |
197 | else |
198 | d = s; |
1aa99e6b |
199 | dstart = d; |
8973db79 |
200 | if (complement && !del) |
201 | rlen = tbl[0x100]; |
1aa99e6b |
202 | |
4aecb5b5 |
203 | #ifdef MACOS_TRADITIONAL |
204 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ |
205 | #endif |
206 | |
1aa99e6b |
207 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b |
208 | UV pch = 0xfeedface; |
209 | while (s < send) { |
210 | STRLEN len; |
9041c2e3 |
211 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b |
212 | |
8973db79 |
213 | if (comp > 0xff) { |
214 | if (!complement) { |
215 | Copy(s, d, len, U8); |
216 | d += len; |
217 | } |
218 | else { |
219 | matches++; |
220 | if (!del) { |
05d340b8 |
221 | ch = (rlen == 0) ? comp : |
222 | (comp - 0x100 < rlen) ? |
8973db79 |
223 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 |
224 | if ((UV)ch != pch) { |
9041c2e3 |
225 | d = uvchr_to_utf8(d, ch); |
eb160463 |
226 | pch = (UV)ch; |
8973db79 |
227 | } |
228 | s += len; |
229 | continue; |
230 | } |
231 | } |
9b877dbb |
232 | } |
1aa99e6b |
233 | else if ((ch = tbl[comp]) >= 0) { |
234 | matches++; |
eb160463 |
235 | if ((UV)ch != pch) { |
9041c2e3 |
236 | d = uvchr_to_utf8(d, ch); |
eb160463 |
237 | pch = (UV)ch; |
1aa99e6b |
238 | } |
239 | s += len; |
240 | continue; |
241 | } |
9b877dbb |
242 | else if (ch == -1) { /* -1 is unmapped character */ |
243 | Copy(s, d, len, U8); |
244 | d += len; |
245 | } |
1aa99e6b |
246 | else if (ch == -2) /* -2 is delete character */ |
247 | matches++; |
248 | s += len; |
249 | pch = 0xfeedface; |
250 | } |
5d06d08e |
251 | } |
1aa99e6b |
252 | else { |
253 | while (s < send) { |
254 | STRLEN len; |
9041c2e3 |
255 | UV comp = utf8_to_uvchr(s, &len); |
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; |
9b877dbb |
307 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
308 | STRLEN len; |
309 | |
310 | SV* rv = (SV*)cSVOP->op_sv; |
311 | HV* hv = (HV*)SvRV(rv); |
312 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
313 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
314 | UV extra = none + 1; |
b7953727 |
315 | UV final = 0; |
4757a243 |
316 | UV uv; |
1aa99e6b |
317 | I32 isutf8; |
318 | U8 hibit = 0; |
4757a243 |
319 | |
320 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
321 | isutf8 = SvUTF8(sv); |
322 | if (!isutf8) { |
323 | U8 *t = s, *e = s + len; |
c4d5f83a |
324 | while (t < e) { |
325 | U8 ch = *t++; |
326 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
327 | break; |
c4d5f83a |
328 | } |
1aa99e6b |
329 | if (hibit) |
330 | s = bytes_to_utf8(s, &len); |
331 | } |
4757a243 |
332 | send = s + len; |
036b4402 |
333 | start = s; |
4757a243 |
334 | |
335 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
336 | if (svp) |
337 | final = SvUV(*svp); |
338 | |
9b877dbb |
339 | if (grows) { |
340 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
341 | New(0, d, len*3+UTF8_MAXLEN, U8); |
342 | dend = d + len * 3; |
343 | dstart = d; |
344 | } |
345 | else { |
346 | dstart = d = s; |
347 | dend = d + len; |
348 | } |
1aa99e6b |
349 | |
4757a243 |
350 | while (s < send) { |
3568d838 |
351 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 |
352 | s += UTF8SKIP(s); |
353 | matches++; |
7d85a32c |
354 | d = uvuni_to_utf8(d, uv); |
4757a243 |
355 | } |
356 | else if (uv == none) { |
1aa99e6b |
357 | int i = UTF8SKIP(s); |
0376ff32 |
358 | Move(s, d, i, U8); |
9b877dbb |
359 | d += i; |
360 | s += i; |
4757a243 |
361 | } |
362 | else if (uv == extra) { |
1aa99e6b |
363 | int i = UTF8SKIP(s); |
036b4402 |
364 | s += i; |
4757a243 |
365 | matches++; |
7d85a32c |
366 | d = uvuni_to_utf8(d, final); |
4757a243 |
367 | } |
368 | else |
369 | s += UTF8SKIP(s); |
1aa99e6b |
370 | |
9b877dbb |
371 | if (d > dend) { |
1aa99e6b |
372 | STRLEN clen = d - dstart; |
373 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
9b877dbb |
374 | if (!grows) |
7d85a32c |
375 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b |
376 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
377 | d = dstart + clen; |
378 | dend = dstart + nlen; |
379 | } |
4757a243 |
380 | } |
16ec844d |
381 | if (grows || hibit) { |
9b877dbb |
382 | sv_setpvn(sv, (char*)dstart, d - dstart); |
383 | Safefree(dstart); |
16ec844d |
384 | if (grows && hibit) |
385 | Safefree(start); |
9b877dbb |
386 | } |
387 | else { |
388 | *d = '\0'; |
389 | SvCUR_set(sv, d - dstart); |
390 | } |
4757a243 |
391 | SvSETMAGIC(sv); |
1aa99e6b |
392 | SvUTF8_on(sv); |
4757a243 |
393 | |
394 | return matches; |
395 | } |
396 | |
942e002e |
397 | STATIC I32 |
21304a3b |
398 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 |
399 | { |
4757a243 |
400 | U8 *s; |
b7953727 |
401 | U8 *start = 0, *send; |
4757a243 |
402 | I32 matches = 0; |
403 | STRLEN len; |
404 | |
405 | SV* rv = (SV*)cSVOP->op_sv; |
406 | HV* hv = (HV*)SvRV(rv); |
407 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
408 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
45005bfb |
409 | UV extra = none + 1; |
4757a243 |
410 | UV uv; |
1aa99e6b |
411 | U8 hibit = 0; |
4757a243 |
412 | |
413 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
414 | if (!SvUTF8(sv)) { |
415 | U8 *t = s, *e = s + len; |
c4d5f83a |
416 | while (t < e) { |
417 | U8 ch = *t++; |
418 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
419 | break; |
c4d5f83a |
420 | } |
1aa99e6b |
421 | if (hibit) |
422 | start = s = bytes_to_utf8(s, &len); |
423 | } |
4757a243 |
424 | send = s + len; |
425 | |
426 | while (s < send) { |
3568d838 |
427 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 |
428 | matches++; |
834a4ddd |
429 | s += UTF8SKIP(s); |
4757a243 |
430 | } |
1aa99e6b |
431 | if (hibit) |
432 | Safefree(start); |
4757a243 |
433 | |
434 | return matches; |
435 | } |
436 | |
942e002e |
437 | STATIC I32 |
21304a3b |
438 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 |
439 | { |
4757a243 |
440 | U8 *s; |
1aa99e6b |
441 | U8 *start, *send; |
4757a243 |
442 | U8 *d; |
443 | I32 matches = 0; |
444 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
4757a243 |
445 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb |
446 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
447 | SV* rv = (SV*)cSVOP->op_sv; |
448 | HV* hv = (HV*)SvRV(rv); |
449 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
450 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
451 | UV extra = none + 1; |
b7953727 |
452 | UV final = 0; |
45005bfb |
453 | bool havefinal = FALSE; |
4757a243 |
454 | UV uv; |
455 | STRLEN len; |
1aa99e6b |
456 | U8 *dstart, *dend; |
457 | I32 isutf8; |
458 | U8 hibit = 0; |
4757a243 |
459 | |
460 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
461 | isutf8 = SvUTF8(sv); |
462 | if (!isutf8) { |
463 | U8 *t = s, *e = s + len; |
c4d5f83a |
464 | while (t < e) { |
465 | U8 ch = *t++; |
466 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
467 | break; |
c4d5f83a |
468 | } |
1aa99e6b |
469 | if (hibit) |
470 | s = bytes_to_utf8(s, &len); |
471 | } |
4757a243 |
472 | send = s + len; |
1aa99e6b |
473 | start = s; |
4757a243 |
474 | |
475 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
45005bfb |
476 | if (svp) { |
4757a243 |
477 | final = SvUV(*svp); |
45005bfb |
478 | havefinal = TRUE; |
479 | } |
4757a243 |
480 | |
9b877dbb |
481 | if (grows) { |
482 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
483 | New(0, d, len*3+UTF8_MAXLEN, U8); |
484 | dend = d + len * 3; |
485 | dstart = d; |
486 | } |
487 | else { |
488 | dstart = d = s; |
489 | dend = d + len; |
490 | } |
4757a243 |
491 | |
492 | if (squash) { |
493 | UV puv = 0xfeedface; |
494 | while (s < send) { |
3568d838 |
495 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
496 | |
9b877dbb |
497 | if (d > dend) { |
498 | STRLEN clen = d - dstart; |
499 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
500 | if (!grows) |
7d85a32c |
501 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
502 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
503 | d = dstart + clen; |
504 | dend = dstart + nlen; |
4757a243 |
505 | } |
506 | if (uv < none) { |
507 | matches++; |
94472101 |
508 | s += UTF8SKIP(s); |
4757a243 |
509 | if (uv != puv) { |
7d85a32c |
510 | d = uvuni_to_utf8(d, uv); |
4757a243 |
511 | puv = uv; |
512 | } |
4757a243 |
513 | continue; |
514 | } |
515 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
516 | int i = UTF8SKIP(s); |
0376ff32 |
517 | Move(s, d, i, U8); |
9b877dbb |
518 | d += i; |
519 | s += i; |
4757a243 |
520 | puv = 0xfeedface; |
521 | continue; |
a0ed51b3 |
522 | } |
4757a243 |
523 | else if (uv == extra && !del) { |
524 | matches++; |
45005bfb |
525 | if (havefinal) { |
526 | s += UTF8SKIP(s); |
527 | if (puv != final) { |
7d85a32c |
528 | d = uvuni_to_utf8(d, final); |
45005bfb |
529 | puv = final; |
530 | } |
531 | } |
532 | else { |
533 | STRLEN len; |
7d85a32c |
534 | uv = utf8_to_uvuni(s, &len); |
45005bfb |
535 | if (uv != puv) { |
0376ff32 |
536 | Move(s, d, len, U8); |
45005bfb |
537 | d += len; |
538 | puv = uv; |
539 | } |
540 | s += len; |
4757a243 |
541 | } |
4757a243 |
542 | continue; |
543 | } |
01ec43d0 |
544 | matches++; /* "none+1" is delete character */ |
545 | s += UTF8SKIP(s); |
a0ed51b3 |
546 | } |
79072805 |
547 | } |
548 | else { |
4757a243 |
549 | while (s < send) { |
3568d838 |
550 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
551 | if (d > dend) { |
552 | STRLEN clen = d - dstart; |
553 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
554 | if (!grows) |
7d85a32c |
555 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
556 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
557 | d = dstart + clen; |
558 | dend = dstart + nlen; |
a0ed51b3 |
559 | } |
4757a243 |
560 | if (uv < none) { |
561 | matches++; |
01ec43d0 |
562 | s += UTF8SKIP(s); |
7d85a32c |
563 | d = uvuni_to_utf8(d, uv); |
4757a243 |
564 | continue; |
a0ed51b3 |
565 | } |
4757a243 |
566 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
567 | int i = UTF8SKIP(s); |
0376ff32 |
568 | Move(s, d, i, U8); |
9b877dbb |
569 | d += i; |
570 | s += i; |
4757a243 |
571 | continue; |
79072805 |
572 | } |
4757a243 |
573 | else if (uv == extra && !del) { |
574 | matches++; |
01ec43d0 |
575 | s += UTF8SKIP(s); |
7d85a32c |
576 | d = uvuni_to_utf8(d, final); |
4757a243 |
577 | continue; |
578 | } |
01ec43d0 |
579 | matches++; /* "none+1" is delete character */ |
580 | s += UTF8SKIP(s); |
79072805 |
581 | } |
4757a243 |
582 | } |
16ec844d |
583 | if (grows || hibit) { |
9b877dbb |
584 | sv_setpvn(sv, (char*)dstart, d - dstart); |
585 | Safefree(dstart); |
16ec844d |
586 | if (grows && hibit) |
587 | Safefree(start); |
9b877dbb |
588 | } |
589 | else { |
590 | *d = '\0'; |
591 | SvCUR_set(sv, d - dstart); |
592 | } |
1aa99e6b |
593 | SvUTF8_on(sv); |
4757a243 |
594 | SvSETMAGIC(sv); |
595 | |
596 | return matches; |
597 | } |
598 | |
599 | I32 |
864dbfa3 |
600 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
601 | { |
602 | STRLEN len; |
8e84507e |
603 | I32 hasutf = (PL_op->op_private & |
036b4402 |
604 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
605 | |
2233f375 |
606 | if (SvREADONLY(sv)) { |
765f542d |
607 | if (SvIsCOW(sv)) |
608 | sv_force_normal_flags(sv, 0); |
2233f375 |
609 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
610 | Perl_croak(aTHX_ PL_no_modify); |
611 | } |
4757a243 |
612 | (void)SvPV(sv, len); |
613 | if (!len) |
614 | return 0; |
d59e14db |
615 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
616 | if (!SvPOKp(sv)) |
617 | (void)SvPV_force(sv, len); |
2de7b02f |
618 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
619 | } |
4757a243 |
620 | |
cea2e8a9 |
621 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
622 | |
a77e643a |
623 | switch (PL_op->op_private & ~hasutf & ( |
624 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| |
625 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { |
4757a243 |
626 | case 0: |
01ec43d0 |
627 | if (hasutf) |
628 | return do_trans_simple_utf8(sv); |
629 | else |
630 | return do_trans_simple(sv); |
4757a243 |
631 | |
632 | case OPpTRANS_IDENTICAL: |
05d340b8 |
633 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
634 | if (hasutf) |
635 | return do_trans_count_utf8(sv); |
636 | else |
637 | return do_trans_count(sv); |
4757a243 |
638 | |
639 | default: |
01ec43d0 |
640 | if (hasutf) |
036b4402 |
641 | return do_trans_complex_utf8(sv); |
4757a243 |
642 | else |
036b4402 |
643 | return do_trans_complex(sv); |
79072805 |
644 | } |
79072805 |
645 | } |
646 | |
647 | void |
864dbfa3 |
648 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
649 | { |
650 | SV **oldmark = mark; |
651 | register I32 items = sp - mark; |
79072805 |
652 | register STRLEN len; |
463ee0b2 |
653 | STRLEN delimlen; |
463ee0b2 |
654 | STRLEN tmplen; |
79072805 |
655 | |
5ba99574 |
656 | (void) SvPV(del, delimlen); /* stringify and get the delimlen */ |
657 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
658 | |
79072805 |
659 | mark++; |
660 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
07f14f54 |
661 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 |
662 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
663 | while (items-- > 0) { |
1426bbf4 |
664 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 |
665 | SvPV(*mark, tmplen); |
666 | len += tmplen; |
79072805 |
667 | } |
668 | mark++; |
669 | } |
670 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
671 | |
672 | mark = oldmark; |
db7c17d7 |
673 | items = sp - mark; |
79072805 |
674 | ++mark; |
675 | } |
676 | |
fc0199cd |
677 | sv_setpvn(sv, "", 0); |
e4803c42 |
678 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
679 | if (SvUTF8(sv)) |
680 | SvUTF8_off(sv); |
681 | |
8d6d96c1 |
682 | if (PL_tainting && SvMAGICAL(sv)) |
683 | SvTAINTED_off(sv); |
684 | |
463ee0b2 |
685 | if (items-- > 0) { |
92d29cee |
686 | if (*mark) |
687 | sv_catsv(sv, *mark); |
463ee0b2 |
688 | mark++; |
689 | } |
8d6d96c1 |
690 | |
c512ce4f |
691 | if (delimlen) { |
79072805 |
692 | for (; items > 0; items--,mark++) { |
c512ce4f |
693 | sv_catsv(sv,del); |
79072805 |
694 | sv_catsv(sv,*mark); |
695 | } |
696 | } |
697 | else { |
698 | for (; items > 0; items--,mark++) |
699 | sv_catsv(sv,*mark); |
700 | } |
701 | SvSETMAGIC(sv); |
702 | } |
703 | |
704 | void |
864dbfa3 |
705 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
706 | { |
46fc3d4c |
707 | STRLEN patlen; |
708 | char *pat = SvPV(*sarg, patlen); |
709 | bool do_taint = FALSE; |
710 | |
5b781b5b |
711 | SvUTF8_off(sv); |
2cf2cfc6 |
712 | if (DO_UTF8(*sarg)) |
713 | SvUTF8_on(sv); |
46fc3d4c |
714 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
715 | SvSETMAGIC(sv); |
46fc3d4c |
716 | if (do_taint) |
717 | SvTAINTED_on(sv); |
79072805 |
718 | } |
719 | |
33b45480 |
720 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
721 | UV |
722 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
723 | { |
724 | STRLEN srclen, len; |
725 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); |
726 | UV retnum = 0; |
727 | |
a50d7633 |
728 | if (offset < 0) |
81e118e0 |
729 | return retnum; |
8e84507e |
730 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
731 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
732 | |
dcad2880 |
733 | if (SvUTF8(sv)) |
33b45480 |
734 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
735 | |
81e118e0 |
736 | offset *= size; /* turn into bit offset */ |
737 | len = (offset + size + 7) / 8; /* required number of bytes */ |
738 | if (len > srclen) { |
739 | if (size <= 8) |
740 | retnum = 0; |
741 | else { |
742 | offset >>= 3; /* turn into byte offset */ |
743 | if (size == 16) { |
eb160463 |
744 | if ((STRLEN)offset >= srclen) |
81e118e0 |
745 | retnum = 0; |
746 | else |
628e1a40 |
747 | retnum = (UV) s[offset] << 8; |
81e118e0 |
748 | } |
749 | else if (size == 32) { |
eb160463 |
750 | if ((STRLEN)offset >= srclen) |
81e118e0 |
751 | retnum = 0; |
eb160463 |
752 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 |
753 | retnum = |
754 | ((UV) s[offset ] << 24); |
eb160463 |
755 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 |
756 | retnum = |
757 | ((UV) s[offset ] << 24) + |
758 | ((UV) s[offset + 1] << 16); |
759 | else |
760 | retnum = |
761 | ((UV) s[offset ] << 24) + |
762 | ((UV) s[offset + 1] << 16) + |
763 | ( s[offset + 2] << 8); |
764 | } |
d7d93a81 |
765 | #ifdef UV_IS_QUAD |
c5a0f51a |
766 | else if (size == 64) { |
c5a0f51a |
767 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
768 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
769 | "Bit vector size > 32 non-portable"); |
770 | if (offset >= srclen) |
771 | retnum = 0; |
772 | else if (offset + 1 >= srclen) |
773 | retnum = |
774 | (UV) s[offset ] << 56; |
775 | else if (offset + 2 >= srclen) |
776 | retnum = |
777 | ((UV) s[offset ] << 56) + |
778 | ((UV) s[offset + 1] << 48); |
779 | else if (offset + 3 >= srclen) |
780 | retnum = |
781 | ((UV) s[offset ] << 56) + |
782 | ((UV) s[offset + 1] << 48) + |
783 | ((UV) s[offset + 2] << 40); |
784 | else if (offset + 4 >= srclen) |
785 | retnum = |
786 | ((UV) s[offset ] << 56) + |
787 | ((UV) s[offset + 1] << 48) + |
788 | ((UV) s[offset + 2] << 40) + |
789 | ((UV) s[offset + 3] << 32); |
790 | else if (offset + 5 >= srclen) |
791 | retnum = |
792 | ((UV) s[offset ] << 56) + |
793 | ((UV) s[offset + 1] << 48) + |
794 | ((UV) s[offset + 2] << 40) + |
795 | ((UV) s[offset + 3] << 32) + |
796 | ( s[offset + 4] << 24); |
797 | else if (offset + 6 >= srclen) |
798 | retnum = |
799 | ((UV) s[offset ] << 56) + |
800 | ((UV) s[offset + 1] << 48) + |
801 | ((UV) s[offset + 2] << 40) + |
802 | ((UV) s[offset + 3] << 32) + |
803 | ((UV) s[offset + 4] << 24) + |
804 | ((UV) s[offset + 5] << 16); |
805 | else |
8e84507e |
806 | retnum = |
c5a0f51a |
807 | ((UV) s[offset ] << 56) + |
808 | ((UV) s[offset + 1] << 48) + |
809 | ((UV) s[offset + 2] << 40) + |
810 | ((UV) s[offset + 3] << 32) + |
811 | ((UV) s[offset + 4] << 24) + |
812 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
813 | ( s[offset + 6] << 8); |
c5a0f51a |
814 | } |
815 | #endif |
81e118e0 |
816 | } |
817 | } |
818 | else if (size < 8) |
819 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
820 | else { |
821 | offset >>= 3; /* turn into byte offset */ |
822 | if (size == 8) |
823 | retnum = s[offset]; |
824 | else if (size == 16) |
825 | retnum = |
628e1a40 |
826 | ((UV) s[offset] << 8) + |
81e118e0 |
827 | s[offset + 1]; |
828 | else if (size == 32) |
829 | retnum = |
830 | ((UV) s[offset ] << 24) + |
831 | ((UV) s[offset + 1] << 16) + |
832 | ( s[offset + 2] << 8) + |
833 | s[offset + 3]; |
d7d93a81 |
834 | #ifdef UV_IS_QUAD |
c5a0f51a |
835 | else if (size == 64) { |
c5a0f51a |
836 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
837 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
838 | "Bit vector size > 32 non-portable"); |
839 | retnum = |
840 | ((UV) s[offset ] << 56) + |
841 | ((UV) s[offset + 1] << 48) + |
842 | ((UV) s[offset + 2] << 40) + |
843 | ((UV) s[offset + 3] << 32) + |
844 | ((UV) s[offset + 4] << 24) + |
845 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
846 | ( s[offset + 6] << 8) + |
c5a0f51a |
847 | s[offset + 7]; |
848 | } |
849 | #endif |
81e118e0 |
850 | } |
851 | |
852 | return retnum; |
853 | } |
854 | |
33b45480 |
855 | /* currently converts input to bytes if possible but doesn't sweat failures, |
856 | * although it does ensure that the string it clobbers is not marked as |
857 | * utf8-valid any more |
858 | */ |
79072805 |
859 | void |
864dbfa3 |
860 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
861 | { |
862 | SV *targ = LvTARG(sv); |
863 | register I32 offset; |
864 | register I32 size; |
8990e307 |
865 | register unsigned char *s; |
81e118e0 |
866 | register UV lval; |
79072805 |
867 | I32 mask; |
a0d0e21e |
868 | STRLEN targlen; |
869 | STRLEN len; |
79072805 |
870 | |
8990e307 |
871 | if (!targ) |
872 | return; |
a0d0e21e |
873 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
874 | if (SvUTF8(targ)) { |
33b45480 |
875 | /* This is handled by the SvPOK_only below... |
876 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
877 | SvUTF8_off(targ); |
878 | */ |
879 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
880 | } |
881 | |
4ebbc975 |
882 | (void)SvPOK_only(targ); |
81e118e0 |
883 | lval = SvUV(sv); |
79072805 |
884 | offset = LvTARGOFF(sv); |
fe58ced6 |
885 | if (offset < 0) |
ed9aa3b7 |
886 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
887 | size = LvTARGLEN(sv); |
8e84507e |
888 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
889 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
890 | |
81e118e0 |
891 | offset *= size; /* turn into bit offset */ |
892 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
893 | if (len > targlen) { |
894 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
895 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
896 | SvCUR_set(targ, len); |
897 | } |
8e84507e |
898 | |
79072805 |
899 | if (size < 8) { |
900 | mask = (1 << size) - 1; |
901 | size = offset & 7; |
902 | lval &= mask; |
81e118e0 |
903 | offset >>= 3; /* turn into byte offset */ |
79072805 |
904 | s[offset] &= ~(mask << size); |
905 | s[offset] |= lval << size; |
906 | } |
907 | else { |
81e118e0 |
908 | offset >>= 3; /* turn into byte offset */ |
79072805 |
909 | if (size == 8) |
eb160463 |
910 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
911 | else if (size == 16) { |
eb160463 |
912 | s[offset ] = (U8)((lval >> 8) & 0xff); |
913 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
914 | } |
915 | else if (size == 32) { |
eb160463 |
916 | s[offset ] = (U8)((lval >> 24) & 0xff); |
917 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
918 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
919 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
920 | } |
d7d93a81 |
921 | #ifdef UV_IS_QUAD |
c5a0f51a |
922 | else if (size == 64) { |
c5a0f51a |
923 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
924 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
925 | "Bit vector size > 32 non-portable"); |
eb160463 |
926 | s[offset ] = (U8)((lval >> 56) & 0xff); |
927 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
928 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
929 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
930 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
931 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
932 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
933 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
934 | } |
dc1e3f56 |
935 | #endif |
79072805 |
936 | } |
7bb043c3 |
937 | SvSETMAGIC(targ); |
79072805 |
938 | } |
939 | |
940 | void |
864dbfa3 |
941 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
942 | { |
463ee0b2 |
943 | STRLEN len; |
a0d0e21e |
944 | char *s; |
8e84507e |
945 | |
79072805 |
946 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
947 | register I32 i; |
948 | I32 max; |
949 | AV* av = (AV*)sv; |
950 | max = AvFILL(av); |
951 | for (i = 0; i <= max; i++) { |
952 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
953 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
954 | do_chop(astr, sv); |
955 | } |
956 | return; |
79072805 |
957 | } |
aa854799 |
958 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
959 | HV* hv = (HV*)sv; |
960 | HE* entry; |
961 | (void)hv_iterinit(hv); |
962 | /*SUPPRESS 560*/ |
155aba94 |
963 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
964 | do_chop(astr,hv_iterval(hv,entry)); |
965 | return; |
79072805 |
966 | } |
991e6d41 |
967 | else if (SvREADONLY(sv)) { |
968 | if (SvFAKE(sv)) { |
969 | /* SV is copy-on-write */ |
970 | sv_force_normal_flags(sv, 0); |
971 | } |
972 | if (SvREADONLY(sv)) |
973 | Perl_croak(aTHX_ PL_no_modify); |
974 | } |
a0d0e21e |
975 | s = SvPV(sv, len); |
748a9306 |
976 | if (len && !SvPOK(sv)) |
a0d0e21e |
977 | s = SvPV_force(sv, len); |
7e2040f0 |
978 | if (DO_UTF8(sv)) { |
a0ed51b3 |
979 | if (s && len) { |
980 | char *send = s + len; |
981 | char *start = s; |
982 | s = send - 1; |
a0dbb045 |
983 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
984 | s--; |
9041c2e3 |
985 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 |
986 | sv_setpvn(astr, s, send - s); |
987 | *s = '\0'; |
988 | SvCUR_set(sv, s - start); |
989 | SvNIOK_off(sv); |
990 | SvUTF8_on(astr); |
991 | } |
a0ed51b3 |
992 | } |
993 | else |
994 | sv_setpvn(astr, "", 0); |
995 | } |
7e2040f0 |
996 | else if (s && len) { |
a0d0e21e |
997 | s += --len; |
998 | sv_setpvn(astr, s, 1); |
999 | *s = '\0'; |
1000 | SvCUR_set(sv, len); |
2c19a612 |
1001 | SvUTF8_off(sv); |
a0d0e21e |
1002 | SvNIOK_off(sv); |
79072805 |
1003 | } |
1004 | else |
a0d0e21e |
1005 | sv_setpvn(astr, "", 0); |
1006 | SvSETMAGIC(sv); |
7e2040f0 |
1007 | } |
a0d0e21e |
1008 | |
1009 | I32 |
864dbfa3 |
1010 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1011 | { |
c07a80fd |
1012 | register I32 count; |
a0d0e21e |
1013 | STRLEN len; |
9b33ce3b |
1014 | STRLEN n_a; |
a0d0e21e |
1015 | char *s; |
c4c87a06 |
1016 | char *temp_buffer = NULL; |
a6aa349d |
1017 | SV* svrecode = Nullsv; |
c07a80fd |
1018 | |
3280af22 |
1019 | if (RsSNARF(PL_rs)) |
c07a80fd |
1020 | return 0; |
4c5a6083 |
1021 | if (RsRECORD(PL_rs)) |
1022 | return 0; |
c07a80fd |
1023 | count = 0; |
a0d0e21e |
1024 | if (SvTYPE(sv) == SVt_PVAV) { |
1025 | register I32 i; |
1026 | I32 max; |
1027 | AV* av = (AV*)sv; |
1028 | max = AvFILL(av); |
1029 | for (i = 0; i <= max; i++) { |
1030 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1031 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1032 | count += do_chomp(sv); |
1033 | } |
1034 | return count; |
1035 | } |
aa854799 |
1036 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
1037 | HV* hv = (HV*)sv; |
1038 | HE* entry; |
1039 | (void)hv_iterinit(hv); |
1040 | /*SUPPRESS 560*/ |
155aba94 |
1041 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1042 | count += do_chomp(hv_iterval(hv,entry)); |
1043 | return count; |
1044 | } |
991e6d41 |
1045 | else if (SvREADONLY(sv)) { |
1046 | if (SvFAKE(sv)) { |
1047 | /* SV is copy-on-write */ |
1048 | sv_force_normal_flags(sv, 0); |
1049 | } |
1050 | if (SvREADONLY(sv)) |
1051 | Perl_croak(aTHX_ PL_no_modify); |
1052 | } |
a6aa349d |
1053 | |
1054 | if (PL_encoding) { |
1055 | if (!SvUTF8(sv)) { |
1056 | /* XXX, here sv is utf8-ized as a side-effect! |
1057 | If encoding.pm is used properly, almost string-generating |
1058 | operations, including literal strings, chr(), input data, etc. |
1059 | should have been utf8-ized already, right? |
1060 | */ |
1061 | sv_recode_to_utf8(sv, PL_encoding); |
1062 | } |
1063 | } |
1064 | |
a0d0e21e |
1065 | s = SvPV(sv, len); |
a0d0e21e |
1066 | if (s && len) { |
1067 | s += --len; |
3280af22 |
1068 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1069 | if (*s != '\n') |
1070 | goto nope; |
1071 | ++count; |
1072 | while (len && s[-1] == '\n') { |
1073 | --len; |
1074 | --s; |
1075 | ++count; |
1076 | } |
1077 | } |
a0d0e21e |
1078 | else { |
a6aa349d |
1079 | STRLEN rslen, rs_charlen; |
3280af22 |
1080 | char *rsptr = SvPV(PL_rs, rslen); |
a6aa349d |
1081 | |
1082 | rs_charlen = SvUTF8(PL_rs) |
1083 | ? sv_len_utf8(PL_rs) |
1084 | : rslen; |
1085 | |
c4c87a06 |
1086 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1087 | /* Assumption is that rs is shorter than the scalar. */ |
1088 | if (SvUTF8(PL_rs)) { |
1089 | /* RS is utf8, scalar is 8 bit. */ |
1090 | bool is_utf8 = TRUE; |
1091 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, |
1092 | &rslen, &is_utf8); |
1093 | if (is_utf8) { |
1094 | /* Cannot downgrade, therefore cannot possibly match |
1095 | */ |
1096 | assert (temp_buffer == rsptr); |
1097 | temp_buffer = NULL; |
1098 | goto nope; |
1099 | } |
1100 | rsptr = temp_buffer; |
a6aa349d |
1101 | } |
1102 | else if (PL_encoding) { |
1103 | /* RS is 8 bit, encoding.pm is used. |
1104 | * Do not recode PL_rs as a side-effect. */ |
1105 | svrecode = newSVpvn(rsptr, rslen); |
1106 | sv_recode_to_utf8(svrecode, PL_encoding); |
1107 | rsptr = SvPV(svrecode, rslen); |
1108 | rs_charlen = sv_len_utf8(svrecode); |
1109 | } |
1110 | else { |
c4c87a06 |
1111 | /* RS is 8 bit, scalar is utf8. */ |
1112 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); |
1113 | rsptr = temp_buffer; |
1114 | } |
1115 | } |
c07a80fd |
1116 | if (rslen == 1) { |
1117 | if (*s != *rsptr) |
1118 | goto nope; |
1119 | ++count; |
1120 | } |
1121 | else { |
8c2cee6f |
1122 | if (len < rslen - 1) |
c07a80fd |
1123 | goto nope; |
1124 | len -= rslen - 1; |
1125 | s -= rslen - 1; |
36477c24 |
1126 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1127 | goto nope; |
a6aa349d |
1128 | count += rs_charlen; |
c07a80fd |
1129 | } |
a0d0e21e |
1130 | } |
9b33ce3b |
1131 | s = SvPV_force(sv, n_a); |
a0d0e21e |
1132 | SvCUR_set(sv, len); |
9b33ce3b |
1133 | *SvEND(sv) = '\0'; |
a0d0e21e |
1134 | SvNIOK_off(sv); |
9b33ce3b |
1135 | SvSETMAGIC(sv); |
a0d0e21e |
1136 | } |
1137 | nope: |
a6aa349d |
1138 | |
1139 | if (svrecode) |
1140 | SvREFCNT_dec(svrecode); |
1141 | |
c4c87a06 |
1142 | Safefree(temp_buffer); |
a0d0e21e |
1143 | return count; |
8e84507e |
1144 | } |
79072805 |
1145 | |
1146 | void |
864dbfa3 |
1147 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1148 | { |
1149 | #ifdef LIBERAL |
1150 | register long *dl; |
1151 | register long *ll; |
1152 | register long *rl; |
1153 | #endif |
1154 | register char *dc; |
463ee0b2 |
1155 | STRLEN leftlen; |
1156 | STRLEN rightlen; |
7a4c00b4 |
1157 | register char *lc; |
1158 | register char *rc; |
79072805 |
1159 | register I32 len; |
a0d0e21e |
1160 | I32 lensave; |
7a4c00b4 |
1161 | char *lsave; |
1162 | char *rsave; |
0c57e439 |
1163 | bool left_utf = DO_UTF8(left); |
1164 | bool right_utf = DO_UTF8(right); |
b7953727 |
1165 | I32 needlen = 0; |
0c57e439 |
1166 | |
1167 | if (left_utf && !right_utf) |
1168 | sv_utf8_upgrade(right); |
a1ca4561 |
1169 | else if (!left_utf && right_utf) |
0c57e439 |
1170 | sv_utf8_upgrade(left); |
79072805 |
1171 | |
1fbd88dc |
1172 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1173 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
891f9566 |
1174 | lsave = lc = SvPV_nomg(left, leftlen); |
1175 | rsave = rc = SvPV_nomg(right, rightlen); |
93a17b20 |
1176 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1177 | lensave = len; |
c9b3c8d0 |
1178 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1179 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
1180 | Newz(801, dc, needlen + 1, char); |
1181 | } |
1182 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2d8e6c8d |
1183 | STRLEN n_a; |
891f9566 |
1184 | dc = SvPV_force_nomg(sv, n_a); |
eb160463 |
1185 | if (SvCUR(sv) < (STRLEN)len) { |
1186 | dc = SvGROW(sv, (STRLEN)(len + 1)); |
ff68c719 |
1187 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1188 | } |
c9b3c8d0 |
1189 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1190 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1191 | } |
1192 | else { |
c9b3c8d0 |
1193 | needlen = ((optype == OP_BIT_AND) |
1194 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
ff68c719 |
1195 | Newz(801, dc, needlen + 1, char); |
1196 | (void)sv_usepvn(sv, dc, needlen); |
1197 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1198 | } |
a0d0e21e |
1199 | SvCUR_set(sv, len); |
1200 | (void)SvPOK_only(sv); |
0c57e439 |
1201 | if (left_utf || right_utf) { |
1202 | UV duc, luc, ruc; |
c9b3c8d0 |
1203 | char *dcsave = dc; |
0c57e439 |
1204 | STRLEN lulen = leftlen; |
1205 | STRLEN rulen = rightlen; |
ba210ebe |
1206 | STRLEN ulen; |
0c57e439 |
1207 | |
1208 | switch (optype) { |
1209 | case OP_BIT_AND: |
1210 | while (lulen && rulen) { |
9041c2e3 |
1211 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1212 | lc += ulen; |
1213 | lulen -= ulen; |
9041c2e3 |
1214 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1215 | rc += ulen; |
1216 | rulen -= ulen; |
1217 | duc = luc & ruc; |
9041c2e3 |
1218 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1219 | } |
c9b3c8d0 |
1220 | if (sv == left || sv == right) |
1221 | (void)sv_usepvn(sv, dcsave, needlen); |
1222 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1223 | break; |
1224 | case OP_BIT_XOR: |
1225 | while (lulen && rulen) { |
9041c2e3 |
1226 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1227 | lc += ulen; |
1228 | lulen -= ulen; |
9041c2e3 |
1229 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1230 | rc += ulen; |
1231 | rulen -= ulen; |
1232 | duc = luc ^ ruc; |
9041c2e3 |
1233 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1234 | } |
1235 | goto mop_up_utf; |
1236 | case OP_BIT_OR: |
1237 | while (lulen && rulen) { |
9041c2e3 |
1238 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1239 | lc += ulen; |
1240 | lulen -= ulen; |
9041c2e3 |
1241 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1242 | rc += ulen; |
1243 | rulen -= ulen; |
1244 | duc = luc | ruc; |
9041c2e3 |
1245 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1246 | } |
1247 | mop_up_utf: |
c9b3c8d0 |
1248 | if (sv == left || sv == right) |
1249 | (void)sv_usepvn(sv, dcsave, needlen); |
1250 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1251 | if (rulen) |
1252 | sv_catpvn(sv, rc, rulen); |
1253 | else if (lulen) |
1254 | sv_catpvn(sv, lc, lulen); |
1255 | else |
1256 | *SvEND(sv) = '\0'; |
1257 | break; |
1258 | } |
1259 | SvUTF8_on(sv); |
1260 | goto finish; |
1261 | } |
1262 | else |
79072805 |
1263 | #ifdef LIBERAL |
1264 | if (len >= sizeof(long)*4 && |
1265 | !((long)dc % sizeof(long)) && |
1266 | !((long)lc % sizeof(long)) && |
1267 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1268 | { |
1269 | I32 remainder = len % (sizeof(long)*4); |
1270 | len /= (sizeof(long)*4); |
1271 | |
1272 | dl = (long*)dc; |
1273 | ll = (long*)lc; |
1274 | rl = (long*)rc; |
1275 | |
1276 | switch (optype) { |
1277 | case OP_BIT_AND: |
1278 | while (len--) { |
1279 | *dl++ = *ll++ & *rl++; |
1280 | *dl++ = *ll++ & *rl++; |
1281 | *dl++ = *ll++ & *rl++; |
1282 | *dl++ = *ll++ & *rl++; |
1283 | } |
1284 | break; |
a0d0e21e |
1285 | case OP_BIT_XOR: |
79072805 |
1286 | while (len--) { |
1287 | *dl++ = *ll++ ^ *rl++; |
1288 | *dl++ = *ll++ ^ *rl++; |
1289 | *dl++ = *ll++ ^ *rl++; |
1290 | *dl++ = *ll++ ^ *rl++; |
1291 | } |
1292 | break; |
1293 | case OP_BIT_OR: |
1294 | while (len--) { |
1295 | *dl++ = *ll++ | *rl++; |
1296 | *dl++ = *ll++ | *rl++; |
1297 | *dl++ = *ll++ | *rl++; |
1298 | *dl++ = *ll++ | *rl++; |
1299 | } |
1300 | } |
1301 | |
1302 | dc = (char*)dl; |
1303 | lc = (char*)ll; |
1304 | rc = (char*)rl; |
1305 | |
1306 | len = remainder; |
1307 | } |
1308 | #endif |
a0d0e21e |
1309 | { |
a0d0e21e |
1310 | switch (optype) { |
1311 | case OP_BIT_AND: |
1312 | while (len--) |
1313 | *dc++ = *lc++ & *rc++; |
1314 | break; |
1315 | case OP_BIT_XOR: |
1316 | while (len--) |
1317 | *dc++ = *lc++ ^ *rc++; |
1318 | goto mop_up; |
1319 | case OP_BIT_OR: |
1320 | while (len--) |
1321 | *dc++ = *lc++ | *rc++; |
1322 | mop_up: |
1323 | len = lensave; |
eb160463 |
1324 | if (rightlen > (STRLEN)len) |
a0d0e21e |
1325 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1326 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1327 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1328 | else |
1329 | *SvEND(sv) = '\0'; |
a0d0e21e |
1330 | break; |
1331 | } |
79072805 |
1332 | } |
0c57e439 |
1333 | finish: |
fb73857a |
1334 | SvTAINT(sv); |
79072805 |
1335 | } |
463ee0b2 |
1336 | |
1337 | OP * |
cea2e8a9 |
1338 | Perl_do_kv(pTHX) |
463ee0b2 |
1339 | { |
39644a26 |
1340 | dSP; |
463ee0b2 |
1341 | HV *hv = (HV*)POPs; |
800e9ae0 |
1342 | HV *keys; |
463ee0b2 |
1343 | register HE *entry; |
463ee0b2 |
1344 | SV *tmpstr; |
54310121 |
1345 | I32 gimme = GIMME_V; |
533c011a |
1346 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1347 | I32 dovalues = (PL_op->op_type == OP_VALUES); |
8e84507e |
1348 | |
1349 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e |
1350 | dokeys = dovalues = TRUE; |
463ee0b2 |
1351 | |
85581909 |
1352 | if (!hv) { |
78f9721b |
1353 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1354 | dTARGET; /* make sure to clear its target here */ |
1355 | if (SvTYPE(TARG) == SVt_PVLV) |
1356 | LvTARG(TARG) = Nullsv; |
1357 | PUSHs(TARG); |
1358 | } |
463ee0b2 |
1359 | RETURN; |
85581909 |
1360 | } |
748a9306 |
1361 | |
6d822dc4 |
1362 | keys = hv; |
800e9ae0 |
1363 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1364 | |
54310121 |
1365 | if (gimme == G_VOID) |
aa689395 |
1366 | RETURN; |
1367 | |
54310121 |
1368 | if (gimme == G_SCALAR) { |
6ee623d5 |
1369 | IV i; |
463ee0b2 |
1370 | dTARGET; |
1371 | |
78f9721b |
1372 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1373 | if (SvTYPE(TARG) < SVt_PVLV) { |
1374 | sv_upgrade(TARG, SVt_PVLV); |
14befaf4 |
1375 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 |
1376 | } |
1377 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1378 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1379 | if (LvTARG(TARG)) |
1380 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1381 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1382 | } |
85581909 |
1383 | PUSHs(TARG); |
1384 | RETURN; |
1385 | } |
1386 | |
14befaf4 |
1387 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 |
1388 | i = HvKEYS(keys); |
463ee0b2 |
1389 | else { |
1390 | i = 0; |
463ee0b2 |
1391 | /*SUPPRESS 560*/ |
800e9ae0 |
1392 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1393 | } |
1394 | PUSHi( i ); |
1395 | RETURN; |
1396 | } |
1397 | |
8ed4b672 |
1398 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1399 | |
463ee0b2 |
1400 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1401 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1402 | SPAGAIN; |
574c8022 |
1403 | if (dokeys) { |
1404 | SV* sv = hv_iterkeysv(entry); |
574c8022 |
1405 | XPUSHs(sv); /* won't clobber stack_sp */ |
1406 | } |
463ee0b2 |
1407 | if (dovalues) { |
463ee0b2 |
1408 | PUTBACK; |
6d822dc4 |
1409 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1410 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1411 | (unsigned long)HeHASH(entry), |
800e9ae0 |
1412 | HvMAX(keys)+1, |
1413 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1414 | SPAGAIN; |
46fc3d4c |
1415 | XPUSHs(tmpstr); |
463ee0b2 |
1416 | } |
1417 | PUTBACK; |
1418 | } |
1419 | return NORMAL; |
1420 | } |
4e35701f |
1421 | |