Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
be3c0a43 | 3 | * Copyright (c) 1991-2002, Larry Wall |
79072805 LW |
4 | * |
5 | * You may distribute under the terms of either the GNU General Public | |
6 | * License or the Artistic License, as specified in the README file. | |
7 | * | |
a0d0e21e LW |
8 | */ |
9 | ||
10 | /* | |
11 | * "'So that was the job I felt I had to do when I started,' thought Sam." | |
79072805 LW |
12 | */ |
13 | ||
14 | #include "EXTERN.h" | |
864dbfa3 | 15 | #define PERL_IN_DOOP_C |
79072805 LW |
16 | #include "perl.h" |
17 | ||
64ca3a65 | 18 | #ifndef PERL_MICRO |
79072805 LW |
19 | #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) |
20 | #include <signal.h> | |
21 | #endif | |
64ca3a65 | 22 | #endif |
79072805 | 23 | |
942e002e | 24 | STATIC I32 |
b250498f | 25 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 | 26 | { |
4757a243 | 27 | U8 *s; |
b250498f | 28 | U8 *d; |
4757a243 | 29 | U8 *send; |
b250498f | 30 | U8 *dstart; |
4757a243 | 31 | I32 matches = 0; |
9b877dbb | 32 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 | 33 | STRLEN len; |
4757a243 LW |
34 | short *tbl; |
35 | I32 ch; | |
79072805 | 36 | |
4757a243 LW |
37 | tbl = (short*)cPVOP->op_pv; |
38 | if (!tbl) | |
7d85a32c | 39 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 40 | |
4757a243 LW |
41 | s = (U8*)SvPV(sv, len); |
42 | send = s + len; | |
43 | ||
b250498f | 44 | /* First, take care of non-UTF8 input strings, because they're easy */ |
1aa99e6b | 45 | if (!SvUTF8(sv)) { |
01ec43d0 | 46 | while (s < send) { |
036b4402 | 47 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 | 48 | matches++; |
eb160463 | 49 | *s++ = (U8)ch; |
01ec43d0 GS |
50 | } |
51 | else | |
52 | s++; | |
53 | } | |
54 | SvSETMAGIC(sv); | |
b250498f GS |
55 | return matches; |
56 | } | |
4757a243 | 57 | |
b250498f | 58 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb IH |
59 | if (grows) |
60 | New(0, d, len*2+1, U8); | |
61 | else | |
62 | d = s; | |
b250498f GS |
63 | dstart = d; |
64 | while (s < send) { | |
ba210ebe | 65 | STRLEN ulen; |
9b877dbb | 66 | UV c; |
b250498f | 67 | |
b250498f | 68 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 | 69 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb | 70 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f | 71 | matches++; |
9041c2e3 | 72 | d = uvchr_to_utf8(d, ch); |
b250498f | 73 | s += ulen; |
01ec43d0 GS |
74 | } |
75 | else { /* No match -> copy */ | |
9b877dbb IH |
76 | Copy(s, d, ulen, U8); |
77 | d += ulen; | |
78 | s += ulen; | |
b250498f GS |
79 | } |
80 | } | |
9b877dbb IH |
81 | if (grows) { |
82 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
83 | Safefree(dstart); | |
84 | } | |
85 | else { | |
86 | *d = '\0'; | |
87 | SvCUR_set(sv, d - dstart); | |
88 | } | |
b250498f | 89 | SvUTF8_on(sv); |
b250498f | 90 | SvSETMAGIC(sv); |
4757a243 LW |
91 | return matches; |
92 | } | |
93 | ||
942e002e | 94 | STATIC I32 |
21304a3b | 95 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 | 96 | { |
4757a243 LW |
97 | U8 *s; |
98 | U8 *send; | |
99 | I32 matches = 0; | |
100 | STRLEN len; | |
101 | short *tbl; | |
8973db79 | 102 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 LW |
103 | |
104 | tbl = (short*)cPVOP->op_pv; | |
105 | if (!tbl) | |
7d85a32c | 106 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 LW |
107 | |
108 | s = (U8*)SvPV(sv, len); | |
109 | send = s + len; | |
110 | ||
1aa99e6b IH |
111 | if (!SvUTF8(sv)) |
112 | while (s < send) { | |
113 | if (tbl[*s++] >= 0) | |
036b4402 | 114 | matches++; |
1aa99e6b IH |
115 | } |
116 | else | |
117 | while (s < send) { | |
118 | UV c; | |
119 | STRLEN ulen; | |
9041c2e3 | 120 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 JH |
121 | if (c < 0x100) { |
122 | if (tbl[c] >= 0) | |
123 | matches++; | |
124 | } else if (complement) | |
1aa99e6b IH |
125 | matches++; |
126 | s += ulen; | |
127 | } | |
4757a243 LW |
128 | |
129 | return matches; | |
130 | } | |
131 | ||
942e002e | 132 | STATIC I32 |
21304a3b | 133 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 | 134 | { |
4757a243 LW |
135 | U8 *s; |
136 | U8 *send; | |
137 | U8 *d; | |
5e44153e | 138 | U8 *dstart; |
1aa99e6b | 139 | I32 isutf8; |
4757a243 | 140 | I32 matches = 0; |
9b877dbb | 141 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 JH |
142 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
143 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
b7953727 | 144 | STRLEN len, rlen = 0; |
4757a243 LW |
145 | short *tbl; |
146 | I32 ch; | |
147 | ||
148 | tbl = (short*)cPVOP->op_pv; | |
149 | if (!tbl) | |
7d85a32c | 150 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 LW |
151 | |
152 | s = (U8*)SvPV(sv, len); | |
1aa99e6b | 153 | isutf8 = SvUTF8(sv); |
4757a243 LW |
154 | send = s + len; |
155 | ||
1aa99e6b IH |
156 | if (!isutf8) { |
157 | dstart = d = s; | |
158 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
159 | U8* p = send; | |
160 | while (s < send) { | |
161 | if ((ch = tbl[*s]) >= 0) { | |
eb160463 | 162 | *d = (U8)ch; |
036b4402 | 163 | matches++; |
1aa99e6b IH |
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 IH |
174 | else { |
175 | while (s < send) { | |
176 | if ((ch = tbl[*s]) >= 0) { | |
177 | matches++; | |
eb160463 | 178 | *d++ = (U8)ch; |
1aa99e6b IH |
179 | } |
180 | else if (ch == -1) /* -1 is unmapped character */ | |
181 | *d++ = *s; | |
182 | else if (ch == -2) /* -2 is delete character */ | |
183 | matches++; | |
184 | s++; | |
185 | } | |
186 | } | |
76ef7183 | 187 | *d = '\0'; |
1aa99e6b | 188 | SvCUR_set(sv, d - dstart); |
4757a243 | 189 | } |
1aa99e6b | 190 | else { /* isutf8 */ |
9b877dbb IH |
191 | if (grows) |
192 | New(0, d, len*2+1, U8); | |
193 | else | |
194 | d = s; | |
1aa99e6b | 195 | dstart = d; |
8973db79 JH |
196 | if (complement && !del) |
197 | rlen = tbl[0x100]; | |
1aa99e6b | 198 | |
4aecb5b5 JH |
199 | #ifdef MACOS_TRADITIONAL |
200 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
201 | #endif | |
202 | ||
1aa99e6b | 203 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
204 | UV pch = 0xfeedface; |
205 | while (s < send) { | |
206 | STRLEN len; | |
9041c2e3 | 207 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b | 208 | |
8973db79 JH |
209 | if (comp > 0xff) { |
210 | if (!complement) { | |
211 | Copy(s, d, len, U8); | |
212 | d += len; | |
213 | } | |
214 | else { | |
215 | matches++; | |
216 | if (!del) { | |
05d340b8 JH |
217 | ch = (rlen == 0) ? comp : |
218 | (comp - 0x100 < rlen) ? | |
8973db79 | 219 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 220 | if ((UV)ch != pch) { |
9041c2e3 | 221 | d = uvchr_to_utf8(d, ch); |
eb160463 | 222 | pch = (UV)ch; |
8973db79 JH |
223 | } |
224 | s += len; | |
225 | continue; | |
226 | } | |
227 | } | |
9b877dbb | 228 | } |
1aa99e6b IH |
229 | else if ((ch = tbl[comp]) >= 0) { |
230 | matches++; | |
eb160463 | 231 | if ((UV)ch != pch) { |
9041c2e3 | 232 | d = uvchr_to_utf8(d, ch); |
eb160463 | 233 | pch = (UV)ch; |
1aa99e6b IH |
234 | } |
235 | s += len; | |
236 | continue; | |
237 | } | |
9b877dbb IH |
238 | else if (ch == -1) { /* -1 is unmapped character */ |
239 | Copy(s, d, len, U8); | |
240 | d += len; | |
241 | } | |
1aa99e6b IH |
242 | else if (ch == -2) /* -2 is delete character */ |
243 | matches++; | |
244 | s += len; | |
245 | pch = 0xfeedface; | |
246 | } | |
5d06d08e | 247 | } |
1aa99e6b IH |
248 | else { |
249 | while (s < send) { | |
250 | STRLEN len; | |
9041c2e3 | 251 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 JH |
252 | if (comp > 0xff) { |
253 | if (!complement) { | |
254 | Copy(s, d, len, U8); | |
255 | d += len; | |
256 | } | |
257 | else { | |
258 | matches++; | |
259 | if (!del) { | |
9041c2e3 NIS |
260 | if (comp - 0x100 < rlen) |
261 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 262 | else |
9041c2e3 | 263 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
264 | } |
265 | } | |
9b877dbb | 266 | } |
1aa99e6b | 267 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 268 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
269 | matches++; |
270 | } | |
271 | else if (ch == -1) { /* -1 is unmapped character */ | |
9b877dbb IH |
272 | Copy(s, d, len, U8); |
273 | d += len; | |
1aa99e6b IH |
274 | } |
275 | else if (ch == -2) /* -2 is delete character */ | |
276 | matches++; | |
277 | s += len; | |
278 | } | |
279 | } | |
9b877dbb IH |
280 | if (grows) { |
281 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
282 | Safefree(dstart); | |
283 | } | |
284 | else { | |
285 | *d = '\0'; | |
286 | SvCUR_set(sv, d - dstart); | |
287 | } | |
1aa99e6b | 288 | SvUTF8_on(sv); |
4757a243 | 289 | } |
5e44153e | 290 | SvSETMAGIC(sv); |
4757a243 LW |
291 | return matches; |
292 | } | |
293 | ||
942e002e | 294 | STATIC I32 |
21304a3b | 295 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 | 296 | { |
4757a243 LW |
297 | U8 *s; |
298 | U8 *send; | |
299 | U8 *d; | |
036b4402 | 300 | U8 *start; |
1aa99e6b | 301 | U8 *dstart, *dend; |
4757a243 | 302 | I32 matches = 0; |
9b877dbb | 303 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
304 | STRLEN len; |
305 | ||
306 | SV* rv = (SV*)cSVOP->op_sv; | |
307 | HV* hv = (HV*)SvRV(rv); | |
308 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
309 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
310 | UV extra = none + 1; | |
b7953727 | 311 | UV final = 0; |
4757a243 | 312 | UV uv; |
1aa99e6b IH |
313 | I32 isutf8; |
314 | U8 hibit = 0; | |
4757a243 LW |
315 | |
316 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
317 | isutf8 = SvUTF8(sv); |
318 | if (!isutf8) { | |
319 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
320 | while (t < e) { |
321 | U8 ch = *t++; | |
322 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 323 | break; |
c4d5f83a | 324 | } |
1aa99e6b IH |
325 | if (hibit) |
326 | s = bytes_to_utf8(s, &len); | |
327 | } | |
4757a243 | 328 | send = s + len; |
036b4402 | 329 | start = s; |
4757a243 LW |
330 | |
331 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
332 | if (svp) | |
333 | final = SvUV(*svp); | |
334 | ||
9b877dbb IH |
335 | if (grows) { |
336 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
337 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
338 | dend = d + len * 3; | |
339 | dstart = d; | |
340 | } | |
341 | else { | |
342 | dstart = d = s; | |
343 | dend = d + len; | |
344 | } | |
1aa99e6b | 345 | |
4757a243 | 346 | while (s < send) { |
3568d838 | 347 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 LW |
348 | s += UTF8SKIP(s); |
349 | matches++; | |
7d85a32c | 350 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
351 | } |
352 | else if (uv == none) { | |
1aa99e6b | 353 | int i = UTF8SKIP(s); |
9b877dbb IH |
354 | Copy(s, d, i, U8); |
355 | d += i; | |
356 | s += i; | |
4757a243 LW |
357 | } |
358 | else if (uv == extra) { | |
1aa99e6b | 359 | int i = UTF8SKIP(s); |
036b4402 | 360 | s += i; |
4757a243 | 361 | matches++; |
7d85a32c | 362 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
363 | } |
364 | else | |
365 | s += UTF8SKIP(s); | |
1aa99e6b | 366 | |
9b877dbb | 367 | if (d > dend) { |
1aa99e6b IH |
368 | STRLEN clen = d - dstart; |
369 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
9b877dbb | 370 | if (!grows) |
7d85a32c | 371 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b IH |
372 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
373 | d = dstart + clen; | |
374 | dend = dstart + nlen; | |
375 | } | |
4757a243 | 376 | } |
16ec844d | 377 | if (grows || hibit) { |
9b877dbb IH |
378 | sv_setpvn(sv, (char*)dstart, d - dstart); |
379 | Safefree(dstart); | |
16ec844d IH |
380 | if (grows && hibit) |
381 | Safefree(start); | |
9b877dbb IH |
382 | } |
383 | else { | |
384 | *d = '\0'; | |
385 | SvCUR_set(sv, d - dstart); | |
386 | } | |
4757a243 | 387 | SvSETMAGIC(sv); |
1aa99e6b | 388 | SvUTF8_on(sv); |
4757a243 LW |
389 | |
390 | return matches; | |
391 | } | |
392 | ||
942e002e | 393 | STATIC I32 |
21304a3b | 394 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 | 395 | { |
4757a243 | 396 | U8 *s; |
b7953727 | 397 | U8 *start = 0, *send; |
4757a243 LW |
398 | I32 matches = 0; |
399 | STRLEN len; | |
400 | ||
401 | SV* rv = (SV*)cSVOP->op_sv; | |
402 | HV* hv = (HV*)SvRV(rv); | |
403 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
404 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
45005bfb | 405 | UV extra = none + 1; |
4757a243 | 406 | UV uv; |
1aa99e6b | 407 | U8 hibit = 0; |
4757a243 LW |
408 | |
409 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
410 | if (!SvUTF8(sv)) { |
411 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
412 | while (t < e) { |
413 | U8 ch = *t++; | |
414 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 415 | break; |
c4d5f83a | 416 | } |
1aa99e6b IH |
417 | if (hibit) |
418 | start = s = bytes_to_utf8(s, &len); | |
419 | } | |
4757a243 LW |
420 | send = s + len; |
421 | ||
422 | while (s < send) { | |
3568d838 | 423 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 | 424 | matches++; |
834a4ddd | 425 | s += UTF8SKIP(s); |
4757a243 | 426 | } |
1aa99e6b IH |
427 | if (hibit) |
428 | Safefree(start); | |
4757a243 LW |
429 | |
430 | return matches; | |
431 | } | |
432 | ||
942e002e | 433 | STATIC I32 |
21304a3b | 434 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 | 435 | { |
4757a243 | 436 | U8 *s; |
1aa99e6b | 437 | U8 *start, *send; |
4757a243 LW |
438 | U8 *d; |
439 | I32 matches = 0; | |
440 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
4757a243 | 441 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb | 442 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 LW |
443 | SV* rv = (SV*)cSVOP->op_sv; |
444 | HV* hv = (HV*)SvRV(rv); | |
445 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
446 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
447 | UV extra = none + 1; | |
b7953727 | 448 | UV final = 0; |
45005bfb | 449 | bool havefinal = FALSE; |
4757a243 LW |
450 | UV uv; |
451 | STRLEN len; | |
1aa99e6b IH |
452 | U8 *dstart, *dend; |
453 | I32 isutf8; | |
454 | U8 hibit = 0; | |
4757a243 LW |
455 | |
456 | s = (U8*)SvPV(sv, len); | |
1aa99e6b IH |
457 | isutf8 = SvUTF8(sv); |
458 | if (!isutf8) { | |
459 | U8 *t = s, *e = s + len; | |
c4d5f83a NIS |
460 | while (t < e) { |
461 | U8 ch = *t++; | |
462 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) | |
1aa99e6b | 463 | break; |
c4d5f83a | 464 | } |
1aa99e6b IH |
465 | if (hibit) |
466 | s = bytes_to_utf8(s, &len); | |
467 | } | |
4757a243 | 468 | send = s + len; |
1aa99e6b | 469 | start = s; |
4757a243 LW |
470 | |
471 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
45005bfb | 472 | if (svp) { |
4757a243 | 473 | final = SvUV(*svp); |
45005bfb JH |
474 | havefinal = TRUE; |
475 | } | |
4757a243 | 476 | |
9b877dbb IH |
477 | if (grows) { |
478 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
479 | New(0, d, len*3+UTF8_MAXLEN, U8); | |
480 | dend = d + len * 3; | |
481 | dstart = d; | |
482 | } | |
483 | else { | |
484 | dstart = d = s; | |
485 | dend = d + len; | |
486 | } | |
4757a243 LW |
487 | |
488 | if (squash) { | |
489 | UV puv = 0xfeedface; | |
490 | while (s < send) { | |
3568d838 | 491 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 492 | |
9b877dbb IH |
493 | if (d > dend) { |
494 | STRLEN clen = d - dstart; | |
495 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
496 | if (!grows) | |
7d85a32c | 497 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
498 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
499 | d = dstart + clen; | |
500 | dend = dstart + nlen; | |
4757a243 LW |
501 | } |
502 | if (uv < none) { | |
503 | matches++; | |
94472101 | 504 | s += UTF8SKIP(s); |
4757a243 | 505 | if (uv != puv) { |
7d85a32c | 506 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
507 | puv = uv; |
508 | } | |
4757a243 LW |
509 | continue; |
510 | } | |
511 | else if (uv == none) { /* "none" is unmapped character */ | |
1aa99e6b | 512 | int i = UTF8SKIP(s); |
9b877dbb IH |
513 | Copy(s, d, i, U8); |
514 | d += i; | |
515 | s += i; | |
4757a243 LW |
516 | puv = 0xfeedface; |
517 | continue; | |
a0ed51b3 | 518 | } |
4757a243 LW |
519 | else if (uv == extra && !del) { |
520 | matches++; | |
45005bfb JH |
521 | if (havefinal) { |
522 | s += UTF8SKIP(s); | |
523 | if (puv != final) { | |
7d85a32c | 524 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
525 | puv = final; |
526 | } | |
527 | } | |
528 | else { | |
529 | STRLEN len; | |
7d85a32c | 530 | uv = utf8_to_uvuni(s, &len); |
45005bfb JH |
531 | if (uv != puv) { |
532 | Copy(s, d, len, U8); | |
533 | d += len; | |
534 | puv = uv; | |
535 | } | |
536 | s += len; | |
4757a243 | 537 | } |
4757a243 LW |
538 | continue; |
539 | } | |
01ec43d0 GS |
540 | matches++; /* "none+1" is delete character */ |
541 | s += UTF8SKIP(s); | |
a0ed51b3 | 542 | } |
79072805 LW |
543 | } |
544 | else { | |
4757a243 | 545 | while (s < send) { |
3568d838 | 546 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb IH |
547 | if (d > dend) { |
548 | STRLEN clen = d - dstart; | |
549 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; | |
550 | if (!grows) | |
7d85a32c | 551 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b IH |
552 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
553 | d = dstart + clen; | |
554 | dend = dstart + nlen; | |
a0ed51b3 | 555 | } |
4757a243 LW |
556 | if (uv < none) { |
557 | matches++; | |
01ec43d0 | 558 | s += UTF8SKIP(s); |
7d85a32c | 559 | d = uvuni_to_utf8(d, uv); |
4757a243 | 560 | continue; |
a0ed51b3 | 561 | } |
4757a243 | 562 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b | 563 | int i = UTF8SKIP(s); |
9b877dbb IH |
564 | Copy(s, d, i, U8); |
565 | d += i; | |
566 | s += i; | |
4757a243 | 567 | continue; |
79072805 | 568 | } |
4757a243 LW |
569 | else if (uv == extra && !del) { |
570 | matches++; | |
01ec43d0 | 571 | s += UTF8SKIP(s); |
7d85a32c | 572 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
573 | continue; |
574 | } | |
01ec43d0 GS |
575 | matches++; /* "none+1" is delete character */ |
576 | s += UTF8SKIP(s); | |
79072805 | 577 | } |
4757a243 | 578 | } |
16ec844d | 579 | if (grows || hibit) { |
9b877dbb IH |
580 | sv_setpvn(sv, (char*)dstart, d - dstart); |
581 | Safefree(dstart); | |
16ec844d IH |
582 | if (grows && hibit) |
583 | Safefree(start); | |
9b877dbb IH |
584 | } |
585 | else { | |
586 | *d = '\0'; | |
587 | SvCUR_set(sv, d - dstart); | |
588 | } | |
1aa99e6b | 589 | SvUTF8_on(sv); |
4757a243 LW |
590 | SvSETMAGIC(sv); |
591 | ||
592 | return matches; | |
593 | } | |
594 | ||
595 | I32 | |
864dbfa3 | 596 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
597 | { |
598 | STRLEN len; | |
8e84507e | 599 | I32 hasutf = (PL_op->op_private & |
036b4402 | 600 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 601 | |
2233f375 NC |
602 | if (SvREADONLY(sv)) { |
603 | if (SvFAKE(sv)) | |
604 | sv_force_normal(sv); | |
605 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) | |
606 | Perl_croak(aTHX_ PL_no_modify); | |
607 | } | |
4757a243 LW |
608 | (void)SvPV(sv, len); |
609 | if (!len) | |
610 | return 0; | |
611 | if (!SvPOKp(sv)) | |
612 | (void)SvPV_force(sv, len); | |
2de7b02f GS |
613 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) |
614 | (void)SvPOK_only_UTF8(sv); | |
4757a243 | 615 | |
cea2e8a9 | 616 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 617 | |
036b4402 | 618 | switch (PL_op->op_private & ~hasutf & 63) { |
4757a243 | 619 | case 0: |
01ec43d0 GS |
620 | if (hasutf) |
621 | return do_trans_simple_utf8(sv); | |
622 | else | |
623 | return do_trans_simple(sv); | |
4757a243 LW |
624 | |
625 | case OPpTRANS_IDENTICAL: | |
05d340b8 | 626 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 GS |
627 | if (hasutf) |
628 | return do_trans_count_utf8(sv); | |
629 | else | |
630 | return do_trans_count(sv); | |
4757a243 LW |
631 | |
632 | default: | |
01ec43d0 | 633 | if (hasutf) |
036b4402 | 634 | return do_trans_complex_utf8(sv); |
4757a243 | 635 | else |
036b4402 | 636 | return do_trans_complex(sv); |
79072805 | 637 | } |
79072805 LW |
638 | } |
639 | ||
640 | void | |
864dbfa3 | 641 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
642 | { |
643 | SV **oldmark = mark; | |
644 | register I32 items = sp - mark; | |
79072805 | 645 | register STRLEN len; |
463ee0b2 | 646 | STRLEN delimlen; |
463ee0b2 | 647 | STRLEN tmplen; |
79072805 | 648 | |
5ba99574 NC |
649 | (void) SvPV(del, delimlen); /* stringify and get the delimlen */ |
650 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ | |
516a5887 | 651 | |
79072805 LW |
652 | mark++; |
653 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 654 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
655 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
656 | while (items-- > 0) { | |
1426bbf4 | 657 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 LW |
658 | SvPV(*mark, tmplen); |
659 | len += tmplen; | |
79072805 LW |
660 | } |
661 | mark++; | |
662 | } | |
663 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
664 | ||
665 | mark = oldmark; | |
db7c17d7 | 666 | items = sp - mark; |
79072805 LW |
667 | ++mark; |
668 | } | |
669 | ||
8d6d96c1 HS |
670 | sv_setpv(sv, ""); |
671 | if (PL_tainting && SvMAGICAL(sv)) | |
672 | SvTAINTED_off(sv); | |
673 | ||
463ee0b2 | 674 | if (items-- > 0) { |
92d29cee JH |
675 | if (*mark) |
676 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
677 | mark++; |
678 | } | |
8d6d96c1 | 679 | |
c512ce4f | 680 | if (delimlen) { |
79072805 | 681 | for (; items > 0; items--,mark++) { |
c512ce4f | 682 | sv_catsv(sv,del); |
79072805 LW |
683 | sv_catsv(sv,*mark); |
684 | } | |
685 | } | |
686 | else { | |
687 | for (; items > 0; items--,mark++) | |
688 | sv_catsv(sv,*mark); | |
689 | } | |
690 | SvSETMAGIC(sv); | |
691 | } | |
692 | ||
693 | void | |
864dbfa3 | 694 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 695 | { |
46fc3d4c | 696 | STRLEN patlen; |
697 | char *pat = SvPV(*sarg, patlen); | |
698 | bool do_taint = FALSE; | |
699 | ||
5b781b5b | 700 | SvUTF8_off(sv); |
2cf2cfc6 A |
701 | if (DO_UTF8(*sarg)) |
702 | SvUTF8_on(sv); | |
46fc3d4c | 703 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 | 704 | SvSETMAGIC(sv); |
46fc3d4c | 705 | if (do_taint) |
706 | SvTAINTED_on(sv); | |
79072805 LW |
707 | } |
708 | ||
33b45480 | 709 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
710 | UV |
711 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
712 | { | |
713 | STRLEN srclen, len; | |
714 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
715 | UV retnum = 0; | |
716 | ||
a50d7633 | 717 | if (offset < 0) |
81e118e0 | 718 | return retnum; |
8e84507e | 719 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 720 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 721 | |
dcad2880 | 722 | if (SvUTF8(sv)) |
33b45480 | 723 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 724 | |
81e118e0 JH |
725 | offset *= size; /* turn into bit offset */ |
726 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
727 | if (len > srclen) { | |
728 | if (size <= 8) | |
729 | retnum = 0; | |
730 | else { | |
731 | offset >>= 3; /* turn into byte offset */ | |
732 | if (size == 16) { | |
eb160463 | 733 | if ((STRLEN)offset >= srclen) |
81e118e0 JH |
734 | retnum = 0; |
735 | else | |
628e1a40 | 736 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
737 | } |
738 | else if (size == 32) { | |
eb160463 | 739 | if ((STRLEN)offset >= srclen) |
81e118e0 | 740 | retnum = 0; |
eb160463 | 741 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 JH |
742 | retnum = |
743 | ((UV) s[offset ] << 24); | |
eb160463 | 744 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 JH |
745 | retnum = |
746 | ((UV) s[offset ] << 24) + | |
747 | ((UV) s[offset + 1] << 16); | |
748 | else | |
749 | retnum = | |
750 | ((UV) s[offset ] << 24) + | |
751 | ((UV) s[offset + 1] << 16) + | |
752 | ( s[offset + 2] << 8); | |
753 | } | |
d7d93a81 | 754 | #ifdef UV_IS_QUAD |
c5a0f51a | 755 | else if (size == 64) { |
c5a0f51a | 756 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 757 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
758 | "Bit vector size > 32 non-portable"); |
759 | if (offset >= srclen) | |
760 | retnum = 0; | |
761 | else if (offset + 1 >= srclen) | |
762 | retnum = | |
763 | (UV) s[offset ] << 56; | |
764 | else if (offset + 2 >= srclen) | |
765 | retnum = | |
766 | ((UV) s[offset ] << 56) + | |
767 | ((UV) s[offset + 1] << 48); | |
768 | else if (offset + 3 >= srclen) | |
769 | retnum = | |
770 | ((UV) s[offset ] << 56) + | |
771 | ((UV) s[offset + 1] << 48) + | |
772 | ((UV) s[offset + 2] << 40); | |
773 | else if (offset + 4 >= srclen) | |
774 | retnum = | |
775 | ((UV) s[offset ] << 56) + | |
776 | ((UV) s[offset + 1] << 48) + | |
777 | ((UV) s[offset + 2] << 40) + | |
778 | ((UV) s[offset + 3] << 32); | |
779 | else if (offset + 5 >= srclen) | |
780 | retnum = | |
781 | ((UV) s[offset ] << 56) + | |
782 | ((UV) s[offset + 1] << 48) + | |
783 | ((UV) s[offset + 2] << 40) + | |
784 | ((UV) s[offset + 3] << 32) + | |
785 | ( s[offset + 4] << 24); | |
786 | else if (offset + 6 >= srclen) | |
787 | retnum = | |
788 | ((UV) s[offset ] << 56) + | |
789 | ((UV) s[offset + 1] << 48) + | |
790 | ((UV) s[offset + 2] << 40) + | |
791 | ((UV) s[offset + 3] << 32) + | |
792 | ((UV) s[offset + 4] << 24) + | |
793 | ((UV) s[offset + 5] << 16); | |
794 | else | |
8e84507e | 795 | retnum = |
c5a0f51a JH |
796 | ((UV) s[offset ] << 56) + |
797 | ((UV) s[offset + 1] << 48) + | |
798 | ((UV) s[offset + 2] << 40) + | |
799 | ((UV) s[offset + 3] << 32) + | |
800 | ((UV) s[offset + 4] << 24) + | |
801 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 802 | ( s[offset + 6] << 8); |
c5a0f51a JH |
803 | } |
804 | #endif | |
81e118e0 JH |
805 | } |
806 | } | |
807 | else if (size < 8) | |
808 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
809 | else { | |
810 | offset >>= 3; /* turn into byte offset */ | |
811 | if (size == 8) | |
812 | retnum = s[offset]; | |
813 | else if (size == 16) | |
814 | retnum = | |
628e1a40 | 815 | ((UV) s[offset] << 8) + |
81e118e0 JH |
816 | s[offset + 1]; |
817 | else if (size == 32) | |
818 | retnum = | |
819 | ((UV) s[offset ] << 24) + | |
820 | ((UV) s[offset + 1] << 16) + | |
821 | ( s[offset + 2] << 8) + | |
822 | s[offset + 3]; | |
d7d93a81 | 823 | #ifdef UV_IS_QUAD |
c5a0f51a | 824 | else if (size == 64) { |
c5a0f51a | 825 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 826 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
827 | "Bit vector size > 32 non-portable"); |
828 | retnum = | |
829 | ((UV) s[offset ] << 56) + | |
830 | ((UV) s[offset + 1] << 48) + | |
831 | ((UV) s[offset + 2] << 40) + | |
832 | ((UV) s[offset + 3] << 32) + | |
833 | ((UV) s[offset + 4] << 24) + | |
834 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 835 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
836 | s[offset + 7]; |
837 | } | |
838 | #endif | |
81e118e0 JH |
839 | } |
840 | ||
841 | return retnum; | |
842 | } | |
843 | ||
33b45480 SB |
844 | /* currently converts input to bytes if possible but doesn't sweat failures, |
845 | * although it does ensure that the string it clobbers is not marked as | |
846 | * utf8-valid any more | |
847 | */ | |
79072805 | 848 | void |
864dbfa3 | 849 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
850 | { |
851 | SV *targ = LvTARG(sv); | |
852 | register I32 offset; | |
853 | register I32 size; | |
8990e307 | 854 | register unsigned char *s; |
81e118e0 | 855 | register UV lval; |
79072805 | 856 | I32 mask; |
a0d0e21e LW |
857 | STRLEN targlen; |
858 | STRLEN len; | |
79072805 | 859 | |
8990e307 LW |
860 | if (!targ) |
861 | return; | |
a0d0e21e | 862 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 863 | if (SvUTF8(targ)) { |
33b45480 SB |
864 | /* This is handled by the SvPOK_only below... |
865 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
866 | SvUTF8_off(targ); | |
867 | */ | |
868 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
869 | } |
870 | ||
4ebbc975 | 871 | (void)SvPOK_only(targ); |
81e118e0 | 872 | lval = SvUV(sv); |
79072805 | 873 | offset = LvTARGOFF(sv); |
fe58ced6 | 874 | if (offset < 0) |
ed9aa3b7 | 875 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 876 | size = LvTARGLEN(sv); |
8e84507e | 877 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 878 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 879 | |
81e118e0 JH |
880 | offset *= size; /* turn into bit offset */ |
881 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
882 | if (len > targlen) { |
883 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 884 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
885 | SvCUR_set(targ, len); |
886 | } | |
8e84507e | 887 | |
79072805 LW |
888 | if (size < 8) { |
889 | mask = (1 << size) - 1; | |
890 | size = offset & 7; | |
891 | lval &= mask; | |
81e118e0 | 892 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
893 | s[offset] &= ~(mask << size); |
894 | s[offset] |= lval << size; | |
895 | } | |
896 | else { | |
81e118e0 | 897 | offset >>= 3; /* turn into byte offset */ |
79072805 | 898 | if (size == 8) |
eb160463 | 899 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 900 | else if (size == 16) { |
eb160463 GS |
901 | s[offset ] = (U8)((lval >> 8) & 0xff); |
902 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
903 | } |
904 | else if (size == 32) { | |
eb160463 GS |
905 | s[offset ] = (U8)((lval >> 24) & 0xff); |
906 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
907 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
908 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 909 | } |
d7d93a81 | 910 | #ifdef UV_IS_QUAD |
c5a0f51a | 911 | else if (size == 64) { |
c5a0f51a | 912 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 913 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 914 | "Bit vector size > 32 non-portable"); |
eb160463 GS |
915 | s[offset ] = (U8)((lval >> 56) & 0xff); |
916 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
917 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
918 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
919 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
920 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
921 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
922 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 923 | } |
dc1e3f56 | 924 | #endif |
79072805 | 925 | } |
7bb043c3 | 926 | SvSETMAGIC(targ); |
79072805 LW |
927 | } |
928 | ||
929 | void | |
864dbfa3 | 930 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 931 | { |
463ee0b2 | 932 | STRLEN len; |
a0d0e21e | 933 | char *s; |
8e84507e | 934 | |
79072805 | 935 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
936 | register I32 i; |
937 | I32 max; | |
938 | AV* av = (AV*)sv; | |
939 | max = AvFILL(av); | |
940 | for (i = 0; i <= max; i++) { | |
941 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 942 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
943 | do_chop(astr, sv); |
944 | } | |
945 | return; | |
79072805 | 946 | } |
aa854799 | 947 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
948 | HV* hv = (HV*)sv; |
949 | HE* entry; | |
950 | (void)hv_iterinit(hv); | |
951 | /*SUPPRESS 560*/ | |
155aba94 | 952 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
953 | do_chop(astr,hv_iterval(hv,entry)); |
954 | return; | |
79072805 | 955 | } |
991e6d41 NC |
956 | else if (SvREADONLY(sv)) { |
957 | if (SvFAKE(sv)) { | |
958 | /* SV is copy-on-write */ | |
959 | sv_force_normal_flags(sv, 0); | |
960 | } | |
961 | if (SvREADONLY(sv)) | |
962 | Perl_croak(aTHX_ PL_no_modify); | |
963 | } | |
a0d0e21e | 964 | s = SvPV(sv, len); |
748a9306 | 965 | if (len && !SvPOK(sv)) |
a0d0e21e | 966 | s = SvPV_force(sv, len); |
7e2040f0 | 967 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
968 | if (s && len) { |
969 | char *send = s + len; | |
970 | char *start = s; | |
971 | s = send - 1; | |
a0dbb045 JH |
972 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
973 | s--; | |
9041c2e3 | 974 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 JH |
975 | sv_setpvn(astr, s, send - s); |
976 | *s = '\0'; | |
977 | SvCUR_set(sv, s - start); | |
978 | SvNIOK_off(sv); | |
979 | SvUTF8_on(astr); | |
980 | } | |
a0ed51b3 LW |
981 | } |
982 | else | |
983 | sv_setpvn(astr, "", 0); | |
984 | } | |
7e2040f0 | 985 | else if (s && len) { |
a0d0e21e LW |
986 | s += --len; |
987 | sv_setpvn(astr, s, 1); | |
988 | *s = '\0'; | |
989 | SvCUR_set(sv, len); | |
2c19a612 | 990 | SvUTF8_off(sv); |
a0d0e21e | 991 | SvNIOK_off(sv); |
79072805 LW |
992 | } |
993 | else | |
a0d0e21e LW |
994 | sv_setpvn(astr, "", 0); |
995 | SvSETMAGIC(sv); | |
7e2040f0 | 996 | } |
a0d0e21e LW |
997 | |
998 | I32 | |
864dbfa3 | 999 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 1000 | { |
c07a80fd | 1001 | register I32 count; |
a0d0e21e | 1002 | STRLEN len; |
9b33ce3b | 1003 | STRLEN n_a; |
a0d0e21e | 1004 | char *s; |
c07a80fd | 1005 | |
3280af22 | 1006 | if (RsSNARF(PL_rs)) |
c07a80fd | 1007 | return 0; |
4c5a6083 GS |
1008 | if (RsRECORD(PL_rs)) |
1009 | return 0; | |
c07a80fd | 1010 | count = 0; |
a0d0e21e LW |
1011 | if (SvTYPE(sv) == SVt_PVAV) { |
1012 | register I32 i; | |
1013 | I32 max; | |
1014 | AV* av = (AV*)sv; | |
1015 | max = AvFILL(av); | |
1016 | for (i = 0; i <= max; i++) { | |
1017 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 1018 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1019 | count += do_chomp(sv); |
1020 | } | |
1021 | return count; | |
1022 | } | |
aa854799 | 1023 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
1024 | HV* hv = (HV*)sv; |
1025 | HE* entry; | |
1026 | (void)hv_iterinit(hv); | |
1027 | /*SUPPRESS 560*/ | |
155aba94 | 1028 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1029 | count += do_chomp(hv_iterval(hv,entry)); |
1030 | return count; | |
1031 | } | |
991e6d41 NC |
1032 | else if (SvREADONLY(sv)) { |
1033 | if (SvFAKE(sv)) { | |
1034 | /* SV is copy-on-write */ | |
1035 | sv_force_normal_flags(sv, 0); | |
1036 | } | |
1037 | if (SvREADONLY(sv)) | |
1038 | Perl_croak(aTHX_ PL_no_modify); | |
1039 | } | |
a0d0e21e | 1040 | s = SvPV(sv, len); |
a0d0e21e LW |
1041 | if (s && len) { |
1042 | s += --len; | |
3280af22 | 1043 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1044 | if (*s != '\n') |
1045 | goto nope; | |
1046 | ++count; | |
1047 | while (len && s[-1] == '\n') { | |
1048 | --len; | |
1049 | --s; | |
1050 | ++count; | |
1051 | } | |
1052 | } | |
a0d0e21e | 1053 | else { |
c07a80fd | 1054 | STRLEN rslen; |
3280af22 | 1055 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 1056 | if (rslen == 1) { |
1057 | if (*s != *rsptr) | |
1058 | goto nope; | |
1059 | ++count; | |
1060 | } | |
1061 | else { | |
8c2cee6f | 1062 | if (len < rslen - 1) |
c07a80fd | 1063 | goto nope; |
1064 | len -= rslen - 1; | |
1065 | s -= rslen - 1; | |
36477c24 | 1066 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1067 | goto nope; |
1068 | count += rslen; | |
1069 | } | |
a0d0e21e | 1070 | } |
9b33ce3b | 1071 | s = SvPV_force(sv, n_a); |
a0d0e21e | 1072 | SvCUR_set(sv, len); |
9b33ce3b | 1073 | *SvEND(sv) = '\0'; |
a0d0e21e | 1074 | SvNIOK_off(sv); |
9b33ce3b | 1075 | SvSETMAGIC(sv); |
a0d0e21e LW |
1076 | } |
1077 | nope: | |
a0d0e21e | 1078 | return count; |
8e84507e | 1079 | } |
79072805 LW |
1080 | |
1081 | void | |
864dbfa3 | 1082 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1083 | { |
1084 | #ifdef LIBERAL | |
1085 | register long *dl; | |
1086 | register long *ll; | |
1087 | register long *rl; | |
1088 | #endif | |
1089 | register char *dc; | |
463ee0b2 LW |
1090 | STRLEN leftlen; |
1091 | STRLEN rightlen; | |
7a4c00b4 | 1092 | register char *lc; |
1093 | register char *rc; | |
79072805 | 1094 | register I32 len; |
a0d0e21e | 1095 | I32 lensave; |
7a4c00b4 | 1096 | char *lsave; |
1097 | char *rsave; | |
0c57e439 GS |
1098 | bool left_utf = DO_UTF8(left); |
1099 | bool right_utf = DO_UTF8(right); | |
b7953727 | 1100 | I32 needlen = 0; |
0c57e439 GS |
1101 | |
1102 | if (left_utf && !right_utf) | |
1103 | sv_utf8_upgrade(right); | |
a1ca4561 | 1104 | else if (!left_utf && right_utf) |
0c57e439 | 1105 | sv_utf8_upgrade(left); |
79072805 | 1106 | |
1fbd88dc CS |
1107 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1108 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 1109 | lsave = lc = SvPV(left, leftlen); |
1110 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 1111 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1112 | lensave = len; |
c9b3c8d0 JH |
1113 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1114 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
1115 | Newz(801, dc, needlen + 1, char); | |
1116 | } | |
1117 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2d8e6c8d GS |
1118 | STRLEN n_a; |
1119 | dc = SvPV_force(sv, n_a); | |
eb160463 GS |
1120 | if (SvCUR(sv) < (STRLEN)len) { |
1121 | dc = SvGROW(sv, (STRLEN)(len + 1)); | |
ff68c719 | 1122 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1123 | } | |
c9b3c8d0 JH |
1124 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1125 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1126 | } |
1127 | else { | |
c9b3c8d0 JH |
1128 | needlen = ((optype == OP_BIT_AND) |
1129 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
ff68c719 | 1130 | Newz(801, dc, needlen + 1, char); |
1131 | (void)sv_usepvn(sv, dc, needlen); | |
1132 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1133 | } |
a0d0e21e LW |
1134 | SvCUR_set(sv, len); |
1135 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1136 | if (left_utf || right_utf) { |
1137 | UV duc, luc, ruc; | |
c9b3c8d0 | 1138 | char *dcsave = dc; |
0c57e439 GS |
1139 | STRLEN lulen = leftlen; |
1140 | STRLEN rulen = rightlen; | |
ba210ebe | 1141 | STRLEN ulen; |
0c57e439 GS |
1142 | |
1143 | switch (optype) { | |
1144 | case OP_BIT_AND: | |
1145 | while (lulen && rulen) { | |
9041c2e3 | 1146 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1147 | lc += ulen; |
1148 | lulen -= ulen; | |
9041c2e3 | 1149 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1150 | rc += ulen; |
1151 | rulen -= ulen; | |
1152 | duc = luc & ruc; | |
9041c2e3 | 1153 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1154 | } |
c9b3c8d0 JH |
1155 | if (sv == left || sv == right) |
1156 | (void)sv_usepvn(sv, dcsave, needlen); | |
1157 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1158 | break; |
1159 | case OP_BIT_XOR: | |
1160 | while (lulen && rulen) { | |
9041c2e3 | 1161 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1162 | lc += ulen; |
1163 | lulen -= ulen; | |
9041c2e3 | 1164 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1165 | rc += ulen; |
1166 | rulen -= ulen; | |
1167 | duc = luc ^ ruc; | |
9041c2e3 | 1168 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1169 | } |
1170 | goto mop_up_utf; | |
1171 | case OP_BIT_OR: | |
1172 | while (lulen && rulen) { | |
9041c2e3 | 1173 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1174 | lc += ulen; |
1175 | lulen -= ulen; | |
9041c2e3 | 1176 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1177 | rc += ulen; |
1178 | rulen -= ulen; | |
1179 | duc = luc | ruc; | |
9041c2e3 | 1180 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1181 | } |
1182 | mop_up_utf: | |
c9b3c8d0 JH |
1183 | if (sv == left || sv == right) |
1184 | (void)sv_usepvn(sv, dcsave, needlen); | |
1185 | SvCUR_set(sv, dc - dcsave); | |
0c57e439 GS |
1186 | if (rulen) |
1187 | sv_catpvn(sv, rc, rulen); | |
1188 | else if (lulen) | |
1189 | sv_catpvn(sv, lc, lulen); | |
1190 | else | |
1191 | *SvEND(sv) = '\0'; | |
1192 | break; | |
1193 | } | |
1194 | SvUTF8_on(sv); | |
1195 | goto finish; | |
1196 | } | |
1197 | else | |
79072805 LW |
1198 | #ifdef LIBERAL |
1199 | if (len >= sizeof(long)*4 && | |
1200 | !((long)dc % sizeof(long)) && | |
1201 | !((long)lc % sizeof(long)) && | |
1202 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1203 | { | |
1204 | I32 remainder = len % (sizeof(long)*4); | |
1205 | len /= (sizeof(long)*4); | |
1206 | ||
1207 | dl = (long*)dc; | |
1208 | ll = (long*)lc; | |
1209 | rl = (long*)rc; | |
1210 | ||
1211 | switch (optype) { | |
1212 | case OP_BIT_AND: | |
1213 | while (len--) { | |
1214 | *dl++ = *ll++ & *rl++; | |
1215 | *dl++ = *ll++ & *rl++; | |
1216 | *dl++ = *ll++ & *rl++; | |
1217 | *dl++ = *ll++ & *rl++; | |
1218 | } | |
1219 | break; | |
a0d0e21e | 1220 | case OP_BIT_XOR: |
79072805 LW |
1221 | while (len--) { |
1222 | *dl++ = *ll++ ^ *rl++; | |
1223 | *dl++ = *ll++ ^ *rl++; | |
1224 | *dl++ = *ll++ ^ *rl++; | |
1225 | *dl++ = *ll++ ^ *rl++; | |
1226 | } | |
1227 | break; | |
1228 | case OP_BIT_OR: | |
1229 | while (len--) { | |
1230 | *dl++ = *ll++ | *rl++; | |
1231 | *dl++ = *ll++ | *rl++; | |
1232 | *dl++ = *ll++ | *rl++; | |
1233 | *dl++ = *ll++ | *rl++; | |
1234 | } | |
1235 | } | |
1236 | ||
1237 | dc = (char*)dl; | |
1238 | lc = (char*)ll; | |
1239 | rc = (char*)rl; | |
1240 | ||
1241 | len = remainder; | |
1242 | } | |
1243 | #endif | |
a0d0e21e | 1244 | { |
a0d0e21e LW |
1245 | switch (optype) { |
1246 | case OP_BIT_AND: | |
1247 | while (len--) | |
1248 | *dc++ = *lc++ & *rc++; | |
1249 | break; | |
1250 | case OP_BIT_XOR: | |
1251 | while (len--) | |
1252 | *dc++ = *lc++ ^ *rc++; | |
1253 | goto mop_up; | |
1254 | case OP_BIT_OR: | |
1255 | while (len--) | |
1256 | *dc++ = *lc++ | *rc++; | |
1257 | mop_up: | |
1258 | len = lensave; | |
eb160463 | 1259 | if (rightlen > (STRLEN)len) |
a0d0e21e | 1260 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 | 1261 | else if (leftlen > (STRLEN)len) |
a0d0e21e | 1262 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1263 | else |
1264 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1265 | break; |
1266 | } | |
79072805 | 1267 | } |
0c57e439 | 1268 | finish: |
fb73857a | 1269 | SvTAINT(sv); |
79072805 | 1270 | } |
463ee0b2 LW |
1271 | |
1272 | OP * | |
cea2e8a9 | 1273 | Perl_do_kv(pTHX) |
463ee0b2 | 1274 | { |
39644a26 | 1275 | dSP; |
463ee0b2 | 1276 | HV *hv = (HV*)POPs; |
800e9ae0 | 1277 | HV *keys; |
463ee0b2 | 1278 | register HE *entry; |
463ee0b2 | 1279 | SV *tmpstr; |
54310121 | 1280 | I32 gimme = GIMME_V; |
533c011a NIS |
1281 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1282 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
c750a3ec | 1283 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
8e84507e NIS |
1284 | |
1285 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) | |
a0d0e21e | 1286 | dokeys = dovalues = TRUE; |
463ee0b2 | 1287 | |
85581909 | 1288 | if (!hv) { |
78f9721b | 1289 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1290 | dTARGET; /* make sure to clear its target here */ |
1291 | if (SvTYPE(TARG) == SVt_PVLV) | |
1292 | LvTARG(TARG) = Nullsv; | |
1293 | PUSHs(TARG); | |
1294 | } | |
463ee0b2 | 1295 | RETURN; |
85581909 | 1296 | } |
748a9306 | 1297 | |
800e9ae0 JP |
1298 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1299 | (void)hv_iterinit(keys); /* always reset iterator regardless */ | |
748a9306 | 1300 | |
54310121 | 1301 | if (gimme == G_VOID) |
aa689395 | 1302 | RETURN; |
1303 | ||
54310121 | 1304 | if (gimme == G_SCALAR) { |
6ee623d5 | 1305 | IV i; |
463ee0b2 LW |
1306 | dTARGET; |
1307 | ||
78f9721b | 1308 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1309 | if (SvTYPE(TARG) < SVt_PVLV) { |
1310 | sv_upgrade(TARG, SVt_PVLV); | |
14befaf4 | 1311 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 SB |
1312 | } |
1313 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1314 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1315 | if (LvTARG(TARG)) |
1316 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1317 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1318 | } |
85581909 SB |
1319 | PUSHs(TARG); |
1320 | RETURN; | |
1321 | } | |
1322 | ||
14befaf4 | 1323 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 | 1324 | i = HvKEYS(keys); |
463ee0b2 LW |
1325 | else { |
1326 | i = 0; | |
463ee0b2 | 1327 | /*SUPPRESS 560*/ |
800e9ae0 | 1328 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1329 | } |
1330 | PUSHi( i ); | |
1331 | RETURN; | |
1332 | } | |
1333 | ||
8ed4b672 | 1334 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1335 | |
463ee0b2 | 1336 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1337 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1338 | SPAGAIN; |
574c8022 JH |
1339 | if (dokeys) { |
1340 | SV* sv = hv_iterkeysv(entry); | |
574c8022 JH |
1341 | XPUSHs(sv); /* won't clobber stack_sp */ |
1342 | } | |
463ee0b2 | 1343 | if (dovalues) { |
463ee0b2 | 1344 | PUTBACK; |
b6429b1b GS |
1345 | tmpstr = realhv ? |
1346 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); | |
cea2e8a9 | 1347 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1348 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1349 | HvMAX(keys)+1, |
1350 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1351 | SPAGAIN; |
46fc3d4c | 1352 | XPUSHs(tmpstr); |
463ee0b2 LW |
1353 | } |
1354 | PUTBACK; | |
1355 | } | |
1356 | return NORMAL; | |
1357 | } | |
4e35701f | 1358 |