This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlunifaq: Note: \d, \D do not have Unicode bug
[perl5.git] / hv.c
CommitLineData
a0d0e21e 1/* hv.c
79072805 2 *
1129b882
NC
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
79072805
LW
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
a0d0e21e
LW
9 */
10
11/*
4ac71550
TC
12 * I sit beside the fire and think
13 * of all that I have seen.
14 * --Bilbo
15 *
16 * [p.278 of _The Lord of the Rings_, II/iii: "The Ring Goes South"]
79072805
LW
17 */
18
d5afce77
RB
19/*
20=head1 Hash Manipulation Functions
166f8a29
DM
21
22A HV structure represents a Perl hash. It consists mainly of an array
23of pointers, each of which points to a linked list of HE structures. The
24array is indexed by the hash function of the key, so each linked list
25represents all the hash entries with the same hash value. Each HE contains
26a pointer to the actual value, plus a pointer to a HEK structure which
27holds the key and hash value.
28
29=cut
30
d5afce77
RB
31*/
32
79072805 33#include "EXTERN.h"
864dbfa3 34#define PERL_IN_HV_C
3d78eb94 35#define PERL_HASH_INTERNAL_ACCESS
79072805
LW
36#include "perl.h"
37
d8012aaf 38#define HV_MAX_LENGTH_BEFORE_SPLIT 14
fdcd69b6 39
d75ce684 40static const char S_strtab_error[]
5d2b1485
NC
41 = "Cannot modify shared string table in hv_%s";
42
cac9b346
NC
43STATIC void
44S_more_he(pTHX)
45{
97aff369 46 dVAR;
d8fca402
NC
47 /* We could generate this at compile time via (another) auxiliary C
48 program? */
49 const size_t arena_size = Perl_malloc_good_size(PERL_ARENA_SIZE);
50 HE* he = (HE*) Perl_get_arena(aTHX_ arena_size, HE_SVSLOT);
51 HE * const heend = &he[arena_size / sizeof(HE) - 1];
cac9b346 52
d2a0f284 53 PL_body_roots[HE_SVSLOT] = he;
cac9b346
NC
54 while (he < heend) {
55 HeNEXT(he) = (HE*)(he + 1);
56 he++;
57 }
58 HeNEXT(he) = 0;
59}
60
c941fb51
NC
61#ifdef PURIFY
62
63#define new_HE() (HE*)safemalloc(sizeof(HE))
64#define del_HE(p) safefree((char*)p)
65
66#else
67
76e3520e 68STATIC HE*
cea2e8a9 69S_new_he(pTHX)
4633a7c4 70{
97aff369 71 dVAR;
4633a7c4 72 HE* he;
0bd48802 73 void ** const root = &PL_body_roots[HE_SVSLOT];
6a93a7e5 74
6a93a7e5 75 if (!*root)
cac9b346 76 S_more_he(aTHX);
10edeb5d 77 he = (HE*) *root;
ce3e5c45 78 assert(he);
6a93a7e5 79 *root = HeNEXT(he);
333f433b 80 return he;
4633a7c4
LW
81}
82
c941fb51
NC
83#define new_HE() new_he()
84#define del_HE(p) \
85 STMT_START { \
6a93a7e5
NC
86 HeNEXT(p) = (HE*)(PL_body_roots[HE_SVSLOT]); \
87 PL_body_roots[HE_SVSLOT] = p; \
c941fb51 88 } STMT_END
d33b2eba 89
d33b2eba 90
d33b2eba
GS
91
92#endif
93
76e3520e 94STATIC HEK *
5f66b61c 95S_save_hek_flags(const char *str, I32 len, U32 hash, int flags)
bbce6d69 96{
35a4481c 97 const int flags_masked = flags & HVhek_MASK;
bbce6d69 98 char *k;
99 register HEK *hek;
1c846c1f 100
7918f24d
NC
101 PERL_ARGS_ASSERT_SAVE_HEK_FLAGS;
102
a02a5408 103 Newx(k, HEK_BASESIZE + len + 2, char);
bbce6d69 104 hek = (HEK*)k;
ff68c719 105 Copy(str, HEK_KEY(hek), len, char);
e05949c7 106 HEK_KEY(hek)[len] = 0;
ff68c719 107 HEK_LEN(hek) = len;
108 HEK_HASH(hek) = hash;
45e34800 109 HEK_FLAGS(hek) = (unsigned char)flags_masked | HVhek_UNSHARED;
dcf933a4
NC
110
111 if (flags & HVhek_FREEKEY)
112 Safefree(str);
bbce6d69 113 return hek;
114}
115
4a31713e 116/* free the pool of temporary HE/HEK pairs returned by hv_fetch_ent
dd28f7bb
DM
117 * for tied hashes */
118
119void
120Perl_free_tied_hv_pool(pTHX)
121{
97aff369 122 dVAR;
dd28f7bb
DM
123 HE *he = PL_hv_fetch_ent_mh;
124 while (he) {
9d4ba2ae 125 HE * const ohe = he;
dd28f7bb 126 Safefree(HeKEY_hek(he));
dd28f7bb
DM
127 he = HeNEXT(he);
128 del_HE(ohe);
129 }
4608196e 130 PL_hv_fetch_ent_mh = NULL;
dd28f7bb
DM
131}
132
d18c6117 133#if defined(USE_ITHREADS)
0bff533c
NC
134HEK *
135Perl_hek_dup(pTHX_ HEK *source, CLONE_PARAMS* param)
136{
566771cc 137 HEK *shared;
9d4ba2ae 138
7918f24d 139 PERL_ARGS_ASSERT_HEK_DUP;
9d4ba2ae 140 PERL_UNUSED_ARG(param);
0bff533c 141
566771cc
NC
142 if (!source)
143 return NULL;
144
145 shared = (HEK*)ptr_table_fetch(PL_ptr_table, source);
0bff533c
NC
146 if (shared) {
147 /* We already shared this hash key. */
454f1e26 148 (void)share_hek_hek(shared);
0bff533c
NC
149 }
150 else {
658b4a4a 151 shared
6e838c70
NC
152 = share_hek_flags(HEK_KEY(source), HEK_LEN(source),
153 HEK_HASH(source), HEK_FLAGS(source));
658b4a4a 154 ptr_table_store(PL_ptr_table, source, shared);
0bff533c 155 }
658b4a4a 156 return shared;
0bff533c
NC
157}
158
d18c6117 159HE *
5c4138a0 160Perl_he_dup(pTHX_ const HE *e, bool shared, CLONE_PARAMS* param)
d18c6117
GS
161{
162 HE *ret;
163
7918f24d
NC
164 PERL_ARGS_ASSERT_HE_DUP;
165
d18c6117 166 if (!e)
4608196e 167 return NULL;
7766f137
GS
168 /* look for it in the table first */
169 ret = (HE*)ptr_table_fetch(PL_ptr_table, e);
170 if (ret)
171 return ret;
172
173 /* create anew and remember what it is */
d33b2eba 174 ret = new_HE();
7766f137
GS
175 ptr_table_store(PL_ptr_table, e, ret);
176
d2d73c3e 177 HeNEXT(ret) = he_dup(HeNEXT(e),shared, param);
dd28f7bb
DM
178 if (HeKLEN(e) == HEf_SVKEY) {
179 char *k;
ad64d0ec 180 Newx(k, HEK_BASESIZE + sizeof(const SV *), char);
dd28f7bb 181 HeKEY_hek(ret) = (HEK*)k;
a09252eb 182 HeKEY_sv(ret) = sv_dup_inc(HeKEY_sv(e), param);
dd28f7bb 183 }
c21d1a0f 184 else if (shared) {
0bff533c
NC
185 /* This is hek_dup inlined, which seems to be important for speed
186 reasons. */
1b6737cc 187 HEK * const source = HeKEY_hek(e);
658b4a4a 188 HEK *shared = (HEK*)ptr_table_fetch(PL_ptr_table, source);
c21d1a0f
NC
189
190 if (shared) {
191 /* We already shared this hash key. */
454f1e26 192 (void)share_hek_hek(shared);
c21d1a0f
NC
193 }
194 else {
658b4a4a 195 shared
6e838c70
NC
196 = share_hek_flags(HEK_KEY(source), HEK_LEN(source),
197 HEK_HASH(source), HEK_FLAGS(source));
658b4a4a 198 ptr_table_store(PL_ptr_table, source, shared);
c21d1a0f 199 }
658b4a4a 200 HeKEY_hek(ret) = shared;
c21d1a0f 201 }
d18c6117 202 else
19692e8d
NC
203 HeKEY_hek(ret) = save_hek_flags(HeKEY(e), HeKLEN(e), HeHASH(e),
204 HeKFLAGS(e));
a09252eb 205 HeVAL(ret) = sv_dup_inc(HeVAL(e), param);
d18c6117
GS
206 return ret;
207}
208#endif /* USE_ITHREADS */
209
1b1f1335 210static void
2393f1b9
JH
211S_hv_notallowed(pTHX_ int flags, const char *key, I32 klen,
212 const char *msg)
1b1f1335 213{
1b6737cc 214 SV * const sv = sv_newmortal();
7918f24d
NC
215
216 PERL_ARGS_ASSERT_HV_NOTALLOWED;
217
19692e8d 218 if (!(flags & HVhek_FREEKEY)) {
1b1f1335
NIS
219 sv_setpvn(sv, key, klen);
220 }
221 else {
222 /* Need to free saved eventually assign to mortal SV */
34c3c4e3 223 /* XXX is this line an error ???: SV *sv = sv_newmortal(); */
1b1f1335
NIS
224 sv_usepvn(sv, (char *) key, klen);
225 }
19692e8d 226 if (flags & HVhek_UTF8) {
1b1f1335
NIS
227 SvUTF8_on(sv);
228 }
be2597df 229 Perl_croak(aTHX_ msg, SVfARG(sv));
1b1f1335
NIS
230}
231
fde52b5c 232/* (klen == HEf_SVKEY) is special for MAGICAL hv entries, meaning key slot
233 * contains an SV* */
234
34a6f7b4
NC
235/*
236=for apidoc hv_store
237
238Stores an SV in a hash. The hash key is specified as C<key> and C<klen> is
239the length of the key. The C<hash> parameter is the precomputed hash
240value; if it is zero then Perl will compute it. The return value will be
241NULL if the operation failed or if the value did not need to be actually
242stored within the hash (as in the case of tied hashes). Otherwise it can
243be dereferenced to get the original C<SV*>. Note that the caller is
244responsible for suitably incrementing the reference count of C<val> before
245the call, and decrementing it if the function returned NULL. Effectively
246a successful hv_store takes ownership of one reference to C<val>. This is
247usually what you want; a newly created SV has a reference count of one, so
248if all your code does is create SVs then store them in a hash, hv_store
249will own the only reference to the new SV, and your code doesn't need to do
250anything further to tidy up. hv_store is not implemented as a call to
251hv_store_ent, and does not create a temporary SV for the key, so if your
252key data is not already in SV form then use hv_store in preference to
253hv_store_ent.
254
255See L<perlguts/"Understanding the Magic of Tied Hashes and Arrays"> for more
256information on how to use this function on tied hashes.
257
34a6f7b4
NC
258=for apidoc hv_store_ent
259
260Stores C<val> in a hash. The hash key is specified as C<key>. The C<hash>
261parameter is the precomputed hash value; if it is zero then Perl will
262compute it. The return value is the new hash entry so created. It will be
263NULL if the operation failed or if the value did not need to be actually
264stored within the hash (as in the case of tied hashes). Otherwise the
265contents of the return value can be accessed using the C<He?> macros
266described here. Note that the caller is responsible for suitably
267incrementing the reference count of C<val> before the call, and
268decrementing it if the function returned NULL. Effectively a successful
269hv_store_ent takes ownership of one reference to C<val>. This is
270usually what you want; a newly created SV has a reference count of one, so
271if all your code does is create SVs then store them in a hash, hv_store
272will own the only reference to the new SV, and your code doesn't need to do
273anything further to tidy up. Note that hv_store_ent only reads the C<key>;
274unlike C<val> it does not take ownership of it, so maintaining the correct
275reference count on C<key> is entirely the caller's responsibility. hv_store
276is not implemented as a call to hv_store_ent, and does not create a temporary
277SV for the key, so if your key data is not already in SV form then use
278hv_store in preference to hv_store_ent.
279
280See L<perlguts/"Understanding the Magic of Tied Hashes and Arrays"> for more
281information on how to use this function on tied hashes.
282
34a6f7b4
NC
283=for apidoc hv_exists
284
285Returns a boolean indicating whether the specified hash key exists. The
286C<klen> is the length of the key.
287
954c1994
GS
288=for apidoc hv_fetch
289
290Returns the SV which corresponds to the specified key in the hash. The
291C<klen> is the length of the key. If C<lval> is set then the fetch will be
292part of a store. Check that the return value is non-null before
d1be9408 293dereferencing it to an C<SV*>.
954c1994 294
96f1132b 295See L<perlguts/"Understanding the Magic of Tied Hashes and Arrays"> for more
954c1994
GS
296information on how to use this function on tied hashes.
297
34a6f7b4
NC
298=for apidoc hv_exists_ent
299
300Returns a boolean indicating whether the specified hash key exists. C<hash>
301can be a valid precomputed hash value, or 0 to ask for it to be
302computed.
303
304=cut
305*/
306
d1be9408 307/* returns an HE * structure with the all fields set */
fde52b5c 308/* note that hent_val will be a mortal sv for MAGICAL hashes */
954c1994
GS
309/*
310=for apidoc hv_fetch_ent
311
312Returns the hash entry which corresponds to the specified key in the hash.
313C<hash> must be a valid precomputed hash number for the given C<key>, or 0
314if you want the function to compute it. IF C<lval> is set then the fetch
315will be part of a store. Make sure the return value is non-null before
316accessing it. The return value when C<tb> is a tied hash is a pointer to a
317static location, so be sure to make a copy of the structure if you need to
1c846c1f 318store it somewhere.
954c1994 319
96f1132b 320See L<perlguts/"Understanding the Magic of Tied Hashes and Arrays"> for more
954c1994
GS
321information on how to use this function on tied hashes.
322
323=cut
324*/
325
a038e571
NC
326/* Common code for hv_delete()/hv_exists()/hv_fetch()/hv_store() */
327void *
328Perl_hv_common_key_len(pTHX_ HV *hv, const char *key, I32 klen_i32,
329 const int action, SV *val, const U32 hash)
330{
331 STRLEN klen;
332 int flags;
333
7918f24d
NC
334 PERL_ARGS_ASSERT_HV_COMMON_KEY_LEN;
335
a038e571
NC
336 if (klen_i32 < 0) {
337 klen = -klen_i32;
338 flags = HVhek_UTF8;
339 } else {
340 klen = klen_i32;
341 flags = 0;
342 }
343 return hv_common(hv, NULL, key, klen, flags, action, val, hash);
344}
345
63c89345 346void *
d3ba3f5c
NC
347Perl_hv_common(pTHX_ HV *hv, SV *keysv, const char *key, STRLEN klen,
348 int flags, int action, SV *val, register U32 hash)
113738bb 349{
27da23d5 350 dVAR;
b2c64049 351 XPVHV* xhv;
b2c64049
NC
352 HE *entry;
353 HE **oentry;
fde52b5c 354 SV *sv;
da58a35d 355 bool is_utf8;
113738bb 356 int masked_flags;
3c84c864 357 const int return_svp = action & HV_FETCH_JUST_SV;
fde52b5c 358
359 if (!hv)
a4fc7abc 360 return NULL;
8265e3d1
NC
361 if (SvTYPE(hv) == SVTYPEMASK)
362 return NULL;
363
364 assert(SvTYPE(hv) == SVt_PVHV);
fde52b5c 365
bdee33e4 366 if (SvSMAGICAL(hv) && SvGMAGICAL(hv) && !(action & HV_DISABLE_UVAR_XKEY)) {
fda2d18a 367 MAGIC* mg;
ad64d0ec 368 if ((mg = mg_find((const SV *)hv, PERL_MAGIC_uvar))) {
fda2d18a
NC
369 struct ufuncs * const uf = (struct ufuncs *)mg->mg_ptr;
370 if (uf->uf_set == NULL) {
371 SV* obj = mg->mg_obj;
372
373 if (!keysv) {
59cd0e26
NC
374 keysv = newSVpvn_flags(key, klen, SVs_TEMP |
375 ((flags & HVhek_UTF8)
376 ? SVf_UTF8 : 0));
fda2d18a
NC
377 }
378
379 mg->mg_obj = keysv; /* pass key */
380 uf->uf_index = action; /* pass action */
ad64d0ec 381 magic_getuvar(MUTABLE_SV(hv), mg);
fda2d18a
NC
382 keysv = mg->mg_obj; /* may have changed */
383 mg->mg_obj = obj;
384
385 /* If the key may have changed, then we need to invalidate
386 any passed-in computed hash value. */
387 hash = 0;
388 }
389 }
bdee33e4 390 }
113738bb 391 if (keysv) {
e593d2fe
AE
392 if (flags & HVhek_FREEKEY)
393 Safefree(key);
5c144d81 394 key = SvPV_const(keysv, klen);
113738bb 395 is_utf8 = (SvUTF8(keysv) != 0);
44b87b50
NC
396 if (SvIsCOW_shared_hash(keysv)) {
397 flags = HVhek_KEYCANONICAL | (is_utf8 ? HVhek_UTF8 : 0);
398 } else {
399 flags = 0;
400 }
113738bb 401 } else {
c1fe5510 402 is_utf8 = ((flags & HVhek_UTF8) ? TRUE : FALSE);
113738bb 403 }
113738bb 404
9dbc5603 405 if (action & HV_DELETE) {
3c84c864
NC
406 return (void *) hv_delete_common(hv, keysv, key, klen,
407 flags | (is_utf8 ? HVhek_UTF8 : 0),
408 action, hash);
9dbc5603
NC
409 }
410
b2c64049 411 xhv = (XPVHV*)SvANY(hv);
7f66fda2 412 if (SvMAGICAL(hv)) {
6136c704 413 if (SvRMAGICAL(hv) && !(action & (HV_FETCH_ISSTORE|HV_FETCH_ISEXISTS))) {
ad64d0ec
NC
414 if (mg_find((const SV *)hv, PERL_MAGIC_tied)
415 || SvGMAGICAL((const SV *)hv))
e62cc96a 416 {
3c84c864 417 /* FIXME should be able to skimp on the HE/HEK here when
7f66fda2 418 HV_FETCH_JUST_SV is true. */
7f66fda2 419 if (!keysv) {
740cce10
NC
420 keysv = newSVpvn_utf8(key, klen, is_utf8);
421 } else {
7f66fda2 422 keysv = newSVsv(keysv);
113738bb 423 }
44a2ac75 424 sv = sv_newmortal();
ad64d0ec 425 mg_copy(MUTABLE_SV(hv), sv, (char *)keysv, HEf_SVKEY);
7f66fda2
NC
426
427 /* grab a fake HE/HEK pair from the pool or make a new one */
428 entry = PL_hv_fetch_ent_mh;
429 if (entry)
430 PL_hv_fetch_ent_mh = HeNEXT(entry);
431 else {
432 char *k;
433 entry = new_HE();
ad64d0ec 434 Newx(k, HEK_BASESIZE + sizeof(const SV *), char);
7f66fda2
NC
435 HeKEY_hek(entry) = (HEK*)k;
436 }
4608196e 437 HeNEXT(entry) = NULL;
7f66fda2
NC
438 HeSVKEY_set(entry, keysv);
439 HeVAL(entry) = sv;
440 sv_upgrade(sv, SVt_PVLV);
441 LvTYPE(sv) = 'T';
442 /* so we can free entry when freeing sv */
ad64d0ec 443 LvTARG(sv) = MUTABLE_SV(entry);
7f66fda2
NC
444
445 /* XXX remove at some point? */
446 if (flags & HVhek_FREEKEY)
447 Safefree(key);
448
3c84c864
NC
449 if (return_svp) {
450 return entry ? (void *) &HeVAL(entry) : NULL;
451 }
452 return (void *) entry;
113738bb 453 }
7f66fda2 454#ifdef ENV_IS_CASELESS
ad64d0ec 455 else if (mg_find((const SV *)hv, PERL_MAGIC_env)) {
7f66fda2
NC
456 U32 i;
457 for (i = 0; i < klen; ++i)
458 if (isLOWER(key[i])) {
086cb327
NC
459 /* Would be nice if we had a routine to do the
460 copy and upercase in a single pass through. */
0bd48802 461 const char * const nkey = strupr(savepvn(key,klen));
086cb327
NC
462 /* Note that this fetch is for nkey (the uppercased
463 key) whereas the store is for key (the original) */
63c89345
NC
464 void *result = hv_common(hv, NULL, nkey, klen,
465 HVhek_FREEKEY, /* free nkey */
466 0 /* non-LVAL fetch */
3c84c864
NC
467 | HV_DISABLE_UVAR_XKEY
468 | return_svp,
63c89345
NC
469 NULL /* no value */,
470 0 /* compute hash */);
26488bcf 471 if (!result && (action & HV_FETCH_LVALUE)) {
086cb327
NC
472 /* This call will free key if necessary.
473 Do it this way to encourage compiler to tail
474 call optimise. */
63c89345
NC
475 result = hv_common(hv, keysv, key, klen, flags,
476 HV_FETCH_ISSTORE
3c84c864
NC
477 | HV_DISABLE_UVAR_XKEY
478 | return_svp,
63c89345 479 newSV(0), hash);
086cb327
NC
480 } else {
481 if (flags & HVhek_FREEKEY)
482 Safefree(key);
483 }
63c89345 484 return result;
7f66fda2 485 }
902173a3 486 }
7f66fda2
NC
487#endif
488 } /* ISFETCH */
489 else if (SvRMAGICAL(hv) && (action & HV_FETCH_ISEXISTS)) {
ad64d0ec
NC
490 if (mg_find((const SV *)hv, PERL_MAGIC_tied)
491 || SvGMAGICAL((const SV *)hv)) {
b2c64049
NC
492 /* I don't understand why hv_exists_ent has svret and sv,
493 whereas hv_exists only had one. */
9d4ba2ae 494 SV * const svret = sv_newmortal();
b2c64049 495 sv = sv_newmortal();
7f66fda2
NC
496
497 if (keysv || is_utf8) {
498 if (!keysv) {
740cce10 499 keysv = newSVpvn_utf8(key, klen, TRUE);
7f66fda2
NC
500 } else {
501 keysv = newSVsv(keysv);
502 }
ad64d0ec 503 mg_copy(MUTABLE_SV(hv), sv, (char *)sv_2mortal(keysv), HEf_SVKEY);
b2c64049 504 } else {
ad64d0ec 505 mg_copy(MUTABLE_SV(hv), sv, key, klen);
7f66fda2 506 }
b2c64049
NC
507 if (flags & HVhek_FREEKEY)
508 Safefree(key);
7f66fda2
NC
509 magic_existspack(svret, mg_find(sv, PERL_MAGIC_tiedelem));
510 /* This cast somewhat evil, but I'm merely using NULL/
511 not NULL to return the boolean exists.
512 And I know hv is not NULL. */
3c84c864 513 return SvTRUE(svret) ? (void *)hv : NULL;
e7152ba2 514 }
7f66fda2 515#ifdef ENV_IS_CASELESS
ad64d0ec 516 else if (mg_find((const SV *)hv, PERL_MAGIC_env)) {
7f66fda2 517 /* XXX This code isn't UTF8 clean. */
a15d23f8 518 char * const keysave = (char * const)key;
b2c64049
NC
519 /* Will need to free this, so set FREEKEY flag. */
520 key = savepvn(key,klen);
521 key = (const char*)strupr((char*)key);
6136c704 522 is_utf8 = FALSE;
7f66fda2 523 hash = 0;
8b4f7dd5 524 keysv = 0;
b2c64049
NC
525
526 if (flags & HVhek_FREEKEY) {
527 Safefree(keysave);
528 }
529 flags |= HVhek_FREEKEY;
7f66fda2 530 }
902173a3 531#endif
7f66fda2 532 } /* ISEXISTS */
b2c64049
NC
533 else if (action & HV_FETCH_ISSTORE) {
534 bool needs_copy;
535 bool needs_store;
536 hv_magic_check (hv, &needs_copy, &needs_store);
537 if (needs_copy) {
a3b680e6 538 const bool save_taint = PL_tainted;
b2c64049
NC
539 if (keysv || is_utf8) {
540 if (!keysv) {
740cce10 541 keysv = newSVpvn_utf8(key, klen, TRUE);
b2c64049
NC
542 }
543 if (PL_tainting)
544 PL_tainted = SvTAINTED(keysv);
545 keysv = sv_2mortal(newSVsv(keysv));
ad64d0ec 546 mg_copy(MUTABLE_SV(hv), val, (char*)keysv, HEf_SVKEY);
b2c64049 547 } else {
ad64d0ec 548 mg_copy(MUTABLE_SV(hv), val, key, klen);
b2c64049
NC
549 }
550
551 TAINT_IF(save_taint);
1baaf5d7 552 if (!needs_store) {
b2c64049
NC
553 if (flags & HVhek_FREEKEY)
554 Safefree(key);
4608196e 555 return NULL;
b2c64049
NC
556 }
557#ifdef ENV_IS_CASELESS
ad64d0ec 558 else if (mg_find((const SV *)hv, PERL_MAGIC_env)) {
b2c64049
NC
559 /* XXX This code isn't UTF8 clean. */
560 const char *keysave = key;
561 /* Will need to free this, so set FREEKEY flag. */
562 key = savepvn(key,klen);
563 key = (const char*)strupr((char*)key);
6136c704 564 is_utf8 = FALSE;
b2c64049 565 hash = 0;
8b4f7dd5 566 keysv = 0;
b2c64049
NC
567
568 if (flags & HVhek_FREEKEY) {
569 Safefree(keysave);
570 }
571 flags |= HVhek_FREEKEY;
572 }
573#endif
574 }
575 } /* ISSTORE */
7f66fda2 576 } /* SvMAGICAL */
fde52b5c 577
7b2c381c 578 if (!HvARRAY(hv)) {
b2c64049 579 if ((action & (HV_FETCH_LVALUE | HV_FETCH_ISSTORE))
fde52b5c 580#ifdef DYNAMIC_ENV_FETCH /* if it's an %ENV lookup, we may get it on the fly */
ad64d0ec
NC
581 || (SvRMAGICAL((const SV *)hv)
582 && mg_find((const SV *)hv, PERL_MAGIC_env))
fde52b5c 583#endif
d58e6666
NC
584 ) {
585 char *array;
a02a5408 586 Newxz(array,
cbec9347 587 PERL_HV_ARRAY_ALLOC_BYTES(xhv->xhv_max+1 /* HvMAX(hv)+1 */),
d58e6666
NC
588 char);
589 HvARRAY(hv) = (HE**)array;
590 }
7f66fda2
NC
591#ifdef DYNAMIC_ENV_FETCH
592 else if (action & HV_FETCH_ISEXISTS) {
593 /* for an %ENV exists, if we do an insert it's by a recursive
594 store call, so avoid creating HvARRAY(hv) right now. */
595 }
596#endif
113738bb
NC
597 else {
598 /* XXX remove at some point? */
599 if (flags & HVhek_FREEKEY)
600 Safefree(key);
601
3c84c864 602 return NULL;
113738bb 603 }
fde52b5c 604 }
605
44b87b50 606 if (is_utf8 & !(flags & HVhek_KEYCANONICAL)) {
41d88b63 607 char * const keysave = (char *)key;
f9a63242 608 key = (char*)bytes_from_utf8((U8*)key, &klen, &is_utf8);
19692e8d 609 if (is_utf8)
c1fe5510
NC
610 flags |= HVhek_UTF8;
611 else
612 flags &= ~HVhek_UTF8;
7f66fda2
NC
613 if (key != keysave) {
614 if (flags & HVhek_FREEKEY)
615 Safefree(keysave);
19692e8d 616 flags |= HVhek_WASUTF8 | HVhek_FREEKEY;
527df579
NC
617 /* If the caller calculated a hash, it was on the sequence of
618 octets that are the UTF-8 form. We've now changed the sequence
619 of octets stored to that of the equivalent byte representation,
620 so the hash we need is different. */
621 hash = 0;
7f66fda2 622 }
19692e8d 623 }
f9a63242 624
4b5190b5
NC
625 if (HvREHASH(hv)) {
626 PERL_HASH_INTERNAL(hash, key, klen);
b2c64049
NC
627 /* We don't have a pointer to the hv, so we have to replicate the
628 flag into every HEK, so that hv_iterkeysv can see it. */
629 /* And yes, you do need this even though you are not "storing" because
fdcd69b6
NC
630 you can flip the flags below if doing an lval lookup. (And that
631 was put in to give the semantics Andreas was expecting.) */
632 flags |= HVhek_REHASH;
4b5190b5 633 } else if (!hash) {
113738bb 634 if (keysv && (SvIsCOW_shared_hash(keysv))) {
c158a4fd 635 hash = SvSHARED_HASH(keysv);
46187eeb
NC
636 } else {
637 PERL_HASH(hash, key, klen);
638 }
639 }
effa1e2d 640
113738bb
NC
641 masked_flags = (flags & HVhek_MASK);
642
7f66fda2 643#ifdef DYNAMIC_ENV_FETCH
4608196e 644 if (!HvARRAY(hv)) entry = NULL;
7f66fda2
NC
645 else
646#endif
b2c64049 647 {
7b2c381c 648 entry = (HvARRAY(hv))[hash & (I32) HvMAX(hv)];
b2c64049 649 }
0298d7b9 650 for (; entry; entry = HeNEXT(entry)) {
fde52b5c 651 if (HeHASH(entry) != hash) /* strings can't be equal */
652 continue;
eb160463 653 if (HeKLEN(entry) != (I32)klen)
fde52b5c 654 continue;
1c846c1f 655 if (HeKEY(entry) != key && memNE(HeKEY(entry),key,klen)) /* is this it? */
fde52b5c 656 continue;
113738bb 657 if ((HeKFLAGS(entry) ^ masked_flags) & HVhek_UTF8)
c3654f1a 658 continue;
b2c64049
NC
659
660 if (action & (HV_FETCH_LVALUE|HV_FETCH_ISSTORE)) {
661 if (HeKFLAGS(entry) != masked_flags) {
662 /* We match if HVhek_UTF8 bit in our flags and hash key's
663 match. But if entry was set previously with HVhek_WASUTF8
664 and key now doesn't (or vice versa) then we should change
665 the key's flag, as this is assignment. */
666 if (HvSHAREKEYS(hv)) {
667 /* Need to swap the key we have for a key with the flags we
668 need. As keys are shared we can't just write to the
669 flag, so we share the new one, unshare the old one. */
6136c704 670 HEK * const new_hek = share_hek_flags(key, klen, hash,
6e838c70 671 masked_flags);
b2c64049
NC
672 unshare_hek (HeKEY_hek(entry));
673 HeKEY_hek(entry) = new_hek;
674 }
5d2b1485
NC
675 else if (hv == PL_strtab) {
676 /* PL_strtab is usually the only hash without HvSHAREKEYS,
677 so putting this test here is cheap */
678 if (flags & HVhek_FREEKEY)
679 Safefree(key);
680 Perl_croak(aTHX_ S_strtab_error,
681 action & HV_FETCH_LVALUE ? "fetch" : "store");
682 }
b2c64049
NC
683 else
684 HeKFLAGS(entry) = masked_flags;
685 if (masked_flags & HVhek_ENABLEHVKFLAGS)
686 HvHASKFLAGS_on(hv);
687 }
688 if (HeVAL(entry) == &PL_sv_placeholder) {
689 /* yes, can store into placeholder slot */
690 if (action & HV_FETCH_LVALUE) {
691 if (SvMAGICAL(hv)) {
692 /* This preserves behaviour with the old hv_fetch
693 implementation which at this point would bail out
694 with a break; (at "if we find a placeholder, we
695 pretend we haven't found anything")
696
697 That break mean that if a placeholder were found, it
698 caused a call into hv_store, which in turn would
699 check magic, and if there is no magic end up pretty
700 much back at this point (in hv_store's code). */
701 break;
702 }
703 /* LVAL fetch which actaully needs a store. */
561b68a9 704 val = newSV(0);
ca732855 705 HvPLACEHOLDERS(hv)--;
b2c64049
NC
706 } else {
707 /* store */
708 if (val != &PL_sv_placeholder)
ca732855 709 HvPLACEHOLDERS(hv)--;
b2c64049
NC
710 }
711 HeVAL(entry) = val;
712 } else if (action & HV_FETCH_ISSTORE) {
713 SvREFCNT_dec(HeVAL(entry));
714 HeVAL(entry) = val;
715 }
27bcc0a7 716 } else if (HeVAL(entry) == &PL_sv_placeholder) {
b2c64049
NC
717 /* if we find a placeholder, we pretend we haven't found
718 anything */
8aacddc1 719 break;
b2c64049 720 }
113738bb
NC
721 if (flags & HVhek_FREEKEY)
722 Safefree(key);
3c84c864
NC
723 if (return_svp) {
724 return entry ? (void *) &HeVAL(entry) : NULL;
725 }
fde52b5c 726 return entry;
727 }
728#ifdef DYNAMIC_ENV_FETCH /* %ENV lookup? If so, try to fetch the value now */
0ed29950 729 if (!(action & HV_FETCH_ISSTORE)
ad64d0ec
NC
730 && SvRMAGICAL((const SV *)hv)
731 && mg_find((const SV *)hv, PERL_MAGIC_env)) {
a6c40364 732 unsigned long len;
9d4ba2ae 733 const char * const env = PerlEnv_ENVgetenv_len(key,&len);
a6c40364
GS
734 if (env) {
735 sv = newSVpvn(env,len);
736 SvTAINTED_on(sv);
d3ba3f5c 737 return hv_common(hv, keysv, key, klen, flags,
3c84c864
NC
738 HV_FETCH_ISSTORE|HV_DISABLE_UVAR_XKEY|return_svp,
739 sv, hash);
a6c40364 740 }
fde52b5c 741 }
742#endif
7f66fda2
NC
743
744 if (!entry && SvREADONLY(hv) && !(action & HV_FETCH_ISEXISTS)) {
c445ea15 745 hv_notallowed(flags, key, klen,
c8cd6465
NC
746 "Attempt to access disallowed key '%"SVf"' in"
747 " a restricted hash");
1b1f1335 748 }
b2c64049
NC
749 if (!(action & (HV_FETCH_LVALUE|HV_FETCH_ISSTORE))) {
750 /* Not doing some form of store, so return failure. */
751 if (flags & HVhek_FREEKEY)
752 Safefree(key);
3c84c864 753 return NULL;
b2c64049 754 }
113738bb 755 if (action & HV_FETCH_LVALUE) {
561b68a9 756 val = newSV(0);
b2c64049
NC
757 if (SvMAGICAL(hv)) {
758 /* At this point the old hv_fetch code would call to hv_store,
759 which in turn might do some tied magic. So we need to make that
760 magic check happen. */
761 /* gonna assign to this, so it better be there */
fda2d18a
NC
762 /* If a fetch-as-store fails on the fetch, then the action is to
763 recurse once into "hv_store". If we didn't do this, then that
764 recursive call would call the key conversion routine again.
765 However, as we replace the original key with the converted
766 key, this would result in a double conversion, which would show
767 up as a bug if the conversion routine is not idempotent. */
d3ba3f5c 768 return hv_common(hv, keysv, key, klen, flags,
3c84c864
NC
769 HV_FETCH_ISSTORE|HV_DISABLE_UVAR_XKEY|return_svp,
770 val, hash);
b2c64049
NC
771 /* XXX Surely that could leak if the fetch-was-store fails?
772 Just like the hv_fetch. */
113738bb
NC
773 }
774 }
775
b2c64049
NC
776 /* Welcome to hv_store... */
777
7b2c381c 778 if (!HvARRAY(hv)) {
b2c64049
NC
779 /* Not sure if we can get here. I think the only case of oentry being
780 NULL is for %ENV with dynamic env fetch. But that should disappear
781 with magic in the previous code. */
d58e6666 782 char *array;
a02a5408 783 Newxz(array,
b2c64049 784 PERL_HV_ARRAY_ALLOC_BYTES(xhv->xhv_max+1 /* HvMAX(hv)+1 */),
d58e6666
NC
785 char);
786 HvARRAY(hv) = (HE**)array;
b2c64049
NC
787 }
788
7b2c381c 789 oentry = &(HvARRAY(hv))[hash & (I32) xhv->xhv_max];
ab4af705 790
b2c64049
NC
791 entry = new_HE();
792 /* share_hek_flags will do the free for us. This might be considered
793 bad API design. */
794 if (HvSHAREKEYS(hv))
6e838c70 795 HeKEY_hek(entry) = share_hek_flags(key, klen, hash, flags);
5d2b1485
NC
796 else if (hv == PL_strtab) {
797 /* PL_strtab is usually the only hash without HvSHAREKEYS, so putting
798 this test here is cheap */
799 if (flags & HVhek_FREEKEY)
800 Safefree(key);
801 Perl_croak(aTHX_ S_strtab_error,
802 action & HV_FETCH_LVALUE ? "fetch" : "store");
803 }
b2c64049
NC
804 else /* gotta do the real thing */
805 HeKEY_hek(entry) = save_hek_flags(key, klen, hash, flags);
806 HeVAL(entry) = val;
807 HeNEXT(entry) = *oentry;
808 *oentry = entry;
809
810 if (val == &PL_sv_placeholder)
ca732855 811 HvPLACEHOLDERS(hv)++;
b2c64049
NC
812 if (masked_flags & HVhek_ENABLEHVKFLAGS)
813 HvHASKFLAGS_on(hv);
814
0298d7b9
NC
815 {
816 const HE *counter = HeNEXT(entry);
817
4c7185a0 818 xhv->xhv_keys++; /* HvTOTALKEYS(hv)++ */
0298d7b9 819 if (!counter) { /* initial entry? */
5ac36297 820 } else if (xhv->xhv_keys > xhv->xhv_max) {
0298d7b9
NC
821 hsplit(hv);
822 } else if(!HvREHASH(hv)) {
823 U32 n_links = 1;
824
825 while ((counter = HeNEXT(counter)))
826 n_links++;
827
828 if (n_links > HV_MAX_LENGTH_BEFORE_SPLIT) {
829 /* Use only the old HvKEYS(hv) > HvMAX(hv) condition to limit
830 bucket splits on a rehashed hash, as we're not going to
831 split it again, and if someone is lucky (evil) enough to
832 get all the keys in one list they could exhaust our memory
833 as we repeatedly double the number of buckets on every
834 entry. Linear search feels a less worse thing to do. */
835 hsplit(hv);
836 }
837 }
fde52b5c 838 }
b2c64049 839
3c84c864
NC
840 if (return_svp) {
841 return entry ? (void *) &HeVAL(entry) : NULL;
842 }
843 return (void *) entry;
fde52b5c 844}
845
864dbfa3 846STATIC void
b0e6ae5b 847S_hv_magic_check(HV *hv, bool *needs_copy, bool *needs_store)
d0066dc7 848{
a3b680e6 849 const MAGIC *mg = SvMAGIC(hv);
7918f24d
NC
850
851 PERL_ARGS_ASSERT_HV_MAGIC_CHECK;
852
d0066dc7
OT
853 *needs_copy = FALSE;
854 *needs_store = TRUE;
855 while (mg) {
856 if (isUPPER(mg->mg_type)) {
857 *needs_copy = TRUE;
d60c5a05 858 if (mg->mg_type == PERL_MAGIC_tied) {
d0066dc7 859 *needs_store = FALSE;
4ab2a30b 860 return; /* We've set all there is to set. */
d0066dc7
OT
861 }
862 }
863 mg = mg->mg_moremagic;
864 }
865}
866
954c1994 867/*
a3bcc51e
TP
868=for apidoc hv_scalar
869
870Evaluates the hash in scalar context and returns the result. Handles magic when the hash is tied.
871
872=cut
873*/
874
875SV *
876Perl_hv_scalar(pTHX_ HV *hv)
877{
a3bcc51e 878 SV *sv;
823a54a3 879
7918f24d
NC
880 PERL_ARGS_ASSERT_HV_SCALAR;
881
823a54a3 882 if (SvRMAGICAL(hv)) {
ad64d0ec 883 MAGIC * const mg = mg_find((const SV *)hv, PERL_MAGIC_tied);
823a54a3
AL
884 if (mg)
885 return magic_scalarpack(hv, mg);
886 }
a3bcc51e
TP
887
888 sv = sv_newmortal();
f4431c56 889 if (HvTOTALKEYS((const HV *)hv))
a3bcc51e
TP
890 Perl_sv_setpvf(aTHX_ sv, "%ld/%ld",
891 (long)HvFILL(hv), (long)HvMAX(hv) + 1);
892 else
893 sv_setiv(sv, 0);
894
895 return sv;
896}
897
898/*
954c1994
GS
899=for apidoc hv_delete
900
901Deletes a key/value pair in the hash. The value SV is removed from the
1c846c1f 902hash and returned to the caller. The C<klen> is the length of the key.
954c1994
GS
903The C<flags> value will normally be zero; if set to G_DISCARD then NULL
904will be returned.
905
954c1994
GS
906=for apidoc hv_delete_ent
907
908Deletes a key/value pair in the hash. The value SV is removed from the
909hash and returned to the caller. The C<flags> value will normally be zero;
910if set to G_DISCARD then NULL will be returned. C<hash> can be a valid
911precomputed hash value, or 0 to ask for it to be computed.
912
913=cut
914*/
915
8f8d40ab 916STATIC SV *
cd6d36ac
NC
917S_hv_delete_common(pTHX_ HV *hv, SV *keysv, const char *key, STRLEN klen,
918 int k_flags, I32 d_flags, U32 hash)
f1317c8d 919{
27da23d5 920 dVAR;
cbec9347 921 register XPVHV* xhv;
fde52b5c 922 register HE *entry;
923 register HE **oentry;
9e720f71 924 HE *const *first_entry;
9dbc5603 925 bool is_utf8 = (k_flags & HVhek_UTF8) ? TRUE : FALSE;
7a9669ca 926 int masked_flags;
1c846c1f 927
fde52b5c 928 if (SvRMAGICAL(hv)) {
0a0bb7c7
OT
929 bool needs_copy;
930 bool needs_store;
931 hv_magic_check (hv, &needs_copy, &needs_store);
932
f1317c8d 933 if (needs_copy) {
6136c704 934 SV *sv;
63c89345
NC
935 entry = (HE *) hv_common(hv, keysv, key, klen,
936 k_flags & ~HVhek_FREEKEY,
937 HV_FETCH_LVALUE|HV_DISABLE_UVAR_XKEY,
938 NULL, hash);
7a9669ca 939 sv = entry ? HeVAL(entry) : NULL;
f1317c8d
NC
940 if (sv) {
941 if (SvMAGICAL(sv)) {
942 mg_clear(sv);
943 }
944 if (!needs_store) {
945 if (mg_find(sv, PERL_MAGIC_tiedelem)) {
946 /* No longer an element */
947 sv_unmagic(sv, PERL_MAGIC_tiedelem);
948 return sv;
949 }
a0714e2c 950 return NULL; /* element cannot be deleted */
f1317c8d 951 }
902173a3 952#ifdef ENV_IS_CASELESS
ad64d0ec 953 else if (mg_find((const SV *)hv, PERL_MAGIC_env)) {
8167a60a 954 /* XXX This code isn't UTF8 clean. */
59cd0e26 955 keysv = newSVpvn_flags(key, klen, SVs_TEMP);
8167a60a
NC
956 if (k_flags & HVhek_FREEKEY) {
957 Safefree(key);
958 }
959 key = strupr(SvPVX(keysv));
960 is_utf8 = 0;
961 k_flags = 0;
962 hash = 0;
7f66fda2 963 }
510ac311 964#endif
2fd1c6b8 965 }
2fd1c6b8 966 }
fde52b5c 967 }
cbec9347 968 xhv = (XPVHV*)SvANY(hv);
7b2c381c 969 if (!HvARRAY(hv))
a0714e2c 970 return NULL;
fde52b5c 971
19692e8d 972 if (is_utf8) {
c445ea15 973 const char * const keysave = key;
b464bac0 974 key = (char*)bytes_from_utf8((U8*)key, &klen, &is_utf8);
cd6d36ac 975
19692e8d 976 if (is_utf8)
cd6d36ac
NC
977 k_flags |= HVhek_UTF8;
978 else
979 k_flags &= ~HVhek_UTF8;
7f66fda2
NC
980 if (key != keysave) {
981 if (k_flags & HVhek_FREEKEY) {
982 /* This shouldn't happen if our caller does what we expect,
983 but strictly the API allows it. */
984 Safefree(keysave);
985 }
986 k_flags |= HVhek_WASUTF8 | HVhek_FREEKEY;
987 }
ad64d0ec 988 HvHASKFLAGS_on(MUTABLE_SV(hv));
19692e8d 989 }
f9a63242 990
4b5190b5
NC
991 if (HvREHASH(hv)) {
992 PERL_HASH_INTERNAL(hash, key, klen);
993 } else if (!hash) {
7a9669ca 994 if (keysv && (SvIsCOW_shared_hash(keysv))) {
c158a4fd 995 hash = SvSHARED_HASH(keysv);
7a9669ca
NC
996 } else {
997 PERL_HASH(hash, key, klen);
998 }
4b5190b5 999 }
fde52b5c 1000
7a9669ca
NC
1001 masked_flags = (k_flags & HVhek_MASK);
1002
9e720f71 1003 first_entry = oentry = &(HvARRAY(hv))[hash & (I32) HvMAX(hv)];
fde52b5c 1004 entry = *oentry;
9e720f71 1005 for (; entry; oentry = &HeNEXT(entry), entry = *oentry) {
6136c704 1006 SV *sv;
fde52b5c 1007 if (HeHASH(entry) != hash) /* strings can't be equal */
1008 continue;
eb160463 1009 if (HeKLEN(entry) != (I32)klen)
fde52b5c 1010 continue;
1c846c1f 1011 if (HeKEY(entry) != key && memNE(HeKEY(entry),key,klen)) /* is this it? */
fde52b5c 1012 continue;
7a9669ca 1013 if ((HeKFLAGS(entry) ^ masked_flags) & HVhek_UTF8)
c3654f1a 1014 continue;
8aacddc1 1015
5d2b1485
NC
1016 if (hv == PL_strtab) {
1017 if (k_flags & HVhek_FREEKEY)
1018 Safefree(key);
1019 Perl_croak(aTHX_ S_strtab_error, "delete");
1020 }
1021
8aacddc1 1022 /* if placeholder is here, it's already been deleted.... */
6136c704
AL
1023 if (HeVAL(entry) == &PL_sv_placeholder) {
1024 if (k_flags & HVhek_FREEKEY)
1025 Safefree(key);
1026 return NULL;
8aacddc1 1027 }
6136c704 1028 if (SvREADONLY(hv) && HeVAL(entry) && SvREADONLY(HeVAL(entry))) {
d4c19fe8 1029 hv_notallowed(k_flags, key, klen,
c8cd6465
NC
1030 "Attempt to delete readonly key '%"SVf"' from"
1031 " a restricted hash");
8aacddc1 1032 }
b84d0860
NC
1033 if (k_flags & HVhek_FREEKEY)
1034 Safefree(key);
8aacddc1 1035
cd6d36ac 1036 if (d_flags & G_DISCARD)
a0714e2c 1037 sv = NULL;
94f7643d 1038 else {
79d01fbf 1039 sv = sv_2mortal(HeVAL(entry));
7996736c 1040 HeVAL(entry) = &PL_sv_placeholder;
94f7643d 1041 }
8aacddc1
NIS
1042
1043 /*
1044 * If a restricted hash, rather than really deleting the entry, put
1045 * a placeholder there. This marks the key as being "approved", so
1046 * we can still access via not-really-existing key without raising
1047 * an error.
1048 */
1049 if (SvREADONLY(hv)) {
754604c4 1050 SvREFCNT_dec(HeVAL(entry));
7996736c 1051 HeVAL(entry) = &PL_sv_placeholder;
8aacddc1
NIS
1052 /* We'll be saving this slot, so the number of allocated keys
1053 * doesn't go down, but the number placeholders goes up */
ca732855 1054 HvPLACEHOLDERS(hv)++;
8aacddc1 1055 } else {
a26e96df 1056 *oentry = HeNEXT(entry);
b79f7545 1057 if (SvOOK(hv) && entry == HvAUX(hv)->xhv_eiter /* HvEITER(hv) */)
8aacddc1
NIS
1058 HvLAZYDEL_on(hv);
1059 else
1060 hv_free_ent(hv, entry);
4c7185a0 1061 xhv->xhv_keys--; /* HvTOTALKEYS(hv)-- */
574c8022 1062 if (xhv->xhv_keys == 0)
19692e8d 1063 HvHASKFLAGS_off(hv);
8aacddc1 1064 }
79072805
LW
1065 return sv;
1066 }
8aacddc1 1067 if (SvREADONLY(hv)) {
d4c19fe8 1068 hv_notallowed(k_flags, key, klen,
c8cd6465
NC
1069 "Attempt to delete disallowed key '%"SVf"' from"
1070 " a restricted hash");
8aacddc1
NIS
1071 }
1072
19692e8d 1073 if (k_flags & HVhek_FREEKEY)
f9a63242 1074 Safefree(key);
a0714e2c 1075 return NULL;
79072805
LW
1076}
1077
76e3520e 1078STATIC void
cea2e8a9 1079S_hsplit(pTHX_ HV *hv)
79072805 1080{
97aff369 1081 dVAR;
1e05feb3 1082 register XPVHV* const xhv = (XPVHV*)SvANY(hv);
a3b680e6 1083 const I32 oldsize = (I32) xhv->xhv_max+1; /* HvMAX(hv)+1 (sick) */
79072805
LW
1084 register I32 newsize = oldsize * 2;
1085 register I32 i;
7b2c381c 1086 char *a = (char*) HvARRAY(hv);
72311751 1087 register HE **aep;
4b5190b5
NC
1088 int longest_chain = 0;
1089 int was_shared;
79072805 1090
7918f24d
NC
1091 PERL_ARGS_ASSERT_HSPLIT;
1092
18026298 1093 /*PerlIO_printf(PerlIO_stderr(), "hsplit called for %p which had %d\n",
6c9570dc 1094 (void*)hv, (int) oldsize);*/
18026298 1095
5d88ecd7 1096 if (HvPLACEHOLDERS_get(hv) && !SvREADONLY(hv)) {
18026298
NC
1097 /* Can make this clear any placeholders first for non-restricted hashes,
1098 even though Storable rebuilds restricted hashes by putting in all the
1099 placeholders (first) before turning on the readonly flag, because
1100 Storable always pre-splits the hash. */
1101 hv_clear_placeholders(hv);
1102 }
1103
3280af22 1104 PL_nomemok = TRUE;
8d6dde3e 1105#if defined(STRANGE_MALLOC) || defined(MYMALLOC)
b79f7545
NC
1106 Renew(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize)
1107 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0), char);
422a93e5 1108 if (!a) {
4a33f861 1109 PL_nomemok = FALSE;
422a93e5
GA
1110 return;
1111 }
b79f7545 1112 if (SvOOK(hv)) {
31f0e52e 1113 Move(&a[oldsize * sizeof(HE*)], &a[newsize * sizeof(HE*)], 1, struct xpvhv_aux);
b79f7545 1114 }
4633a7c4 1115#else
a02a5408 1116 Newx(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize)
b79f7545 1117 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0), char);
422a93e5 1118 if (!a) {
3280af22 1119 PL_nomemok = FALSE;
422a93e5
GA
1120 return;
1121 }
7b2c381c 1122 Copy(HvARRAY(hv), a, oldsize * sizeof(HE*), char);
b79f7545
NC
1123 if (SvOOK(hv)) {
1124 Copy(HvAUX(hv), &a[newsize * sizeof(HE*)], 1, struct xpvhv_aux);
1125 }
fba3b22e 1126 if (oldsize >= 64) {
7b2c381c 1127 offer_nice_chunk(HvARRAY(hv),
b79f7545
NC
1128 PERL_HV_ARRAY_ALLOC_BYTES(oldsize)
1129 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0));
4633a7c4
LW
1130 }
1131 else
7b2c381c 1132 Safefree(HvARRAY(hv));
4633a7c4
LW
1133#endif
1134
3280af22 1135 PL_nomemok = FALSE;
72311751 1136 Zero(&a[oldsize * sizeof(HE*)], (newsize-oldsize) * sizeof(HE*), char); /* zero 2nd half*/
cbec9347 1137 xhv->xhv_max = --newsize; /* HvMAX(hv) = --newsize */
7b2c381c 1138 HvARRAY(hv) = (HE**) a;
72311751 1139 aep = (HE**)a;
79072805 1140
72311751 1141 for (i=0; i<oldsize; i++,aep++) {
4b5190b5
NC
1142 int left_length = 0;
1143 int right_length = 0;
a50a3493
NC
1144 HE **oentry = aep;
1145 HE *entry = *aep;
a3b680e6 1146 register HE **bep;
4b5190b5 1147
a50a3493 1148 if (!entry) /* non-existent */
79072805 1149 continue;
72311751 1150 bep = aep+oldsize;
4c9d89c5 1151 do {
eb160463 1152 if ((HeHASH(entry) & newsize) != (U32)i) {
fde52b5c 1153 *oentry = HeNEXT(entry);
72311751 1154 HeNEXT(entry) = *bep;
72311751 1155 *bep = entry;
4b5190b5 1156 right_length++;
79072805 1157 }
4b5190b5 1158 else {
fde52b5c 1159 oentry = &HeNEXT(entry);
4b5190b5
NC
1160 left_length++;
1161 }
4c9d89c5
NC
1162 entry = *oentry;
1163 } while (entry);
4b5190b5
NC
1164 /* I think we don't actually need to keep track of the longest length,
1165 merely flag if anything is too long. But for the moment while
1166 developing this code I'll track it. */
1167 if (left_length > longest_chain)
1168 longest_chain = left_length;
1169 if (right_length > longest_chain)
1170 longest_chain = right_length;
1171 }
1172
1173
1174 /* Pick your policy for "hashing isn't working" here: */
fdcd69b6 1175 if (longest_chain <= HV_MAX_LENGTH_BEFORE_SPLIT /* split worked? */
4b5190b5
NC
1176 || HvREHASH(hv)) {
1177 return;
79072805 1178 }
4b5190b5
NC
1179
1180 if (hv == PL_strtab) {
1181 /* Urg. Someone is doing something nasty to the string table.
1182 Can't win. */
1183 return;
1184 }
1185
1186 /* Awooga. Awooga. Pathological data. */
6c9570dc 1187 /*PerlIO_printf(PerlIO_stderr(), "%p %d of %d with %d/%d buckets\n", (void*)hv,
4b5190b5
NC
1188 longest_chain, HvTOTALKEYS(hv), HvFILL(hv), 1+HvMAX(hv));*/
1189
1190 ++newsize;
a02a5408 1191 Newxz(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize)
b79f7545
NC
1192 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0), char);
1193 if (SvOOK(hv)) {
1194 Copy(HvAUX(hv), &a[newsize * sizeof(HE*)], 1, struct xpvhv_aux);
1195 }
1196
4b5190b5
NC
1197 was_shared = HvSHAREKEYS(hv);
1198
4b5190b5
NC
1199 HvSHAREKEYS_off(hv);
1200 HvREHASH_on(hv);
1201
7b2c381c 1202 aep = HvARRAY(hv);
4b5190b5
NC
1203
1204 for (i=0; i<newsize; i++,aep++) {
a3b680e6 1205 register HE *entry = *aep;
4b5190b5
NC
1206 while (entry) {
1207 /* We're going to trash this HE's next pointer when we chain it
1208 into the new hash below, so store where we go next. */
9d4ba2ae 1209 HE * const next = HeNEXT(entry);
4b5190b5 1210 UV hash;
a3b680e6 1211 HE **bep;
4b5190b5
NC
1212
1213 /* Rehash it */
1214 PERL_HASH_INTERNAL(hash, HeKEY(entry), HeKLEN(entry));
1215
1216 if (was_shared) {
1217 /* Unshare it. */
aec46f14 1218 HEK * const new_hek
4b5190b5
NC
1219 = save_hek_flags(HeKEY(entry), HeKLEN(entry),
1220 hash, HeKFLAGS(entry));
1221 unshare_hek (HeKEY_hek(entry));
1222 HeKEY_hek(entry) = new_hek;
1223 } else {
1224 /* Not shared, so simply write the new hash in. */
1225 HeHASH(entry) = hash;
1226 }
1227 /*PerlIO_printf(PerlIO_stderr(), "%d ", HeKFLAGS(entry));*/
1228 HEK_REHASH_on(HeKEY_hek(entry));
1229 /*PerlIO_printf(PerlIO_stderr(), "%d\n", HeKFLAGS(entry));*/
1230
1231 /* Copy oentry to the correct new chain. */
1232 bep = ((HE**)a) + (hash & (I32) xhv->xhv_max);
4b5190b5
NC
1233 HeNEXT(entry) = *bep;
1234 *bep = entry;
1235
1236 entry = next;
1237 }
1238 }
7b2c381c
NC
1239 Safefree (HvARRAY(hv));
1240 HvARRAY(hv) = (HE **)a;
79072805
LW
1241}
1242
72940dca 1243void
864dbfa3 1244Perl_hv_ksplit(pTHX_ HV *hv, IV newmax)
72940dca 1245{
97aff369 1246 dVAR;
cbec9347 1247 register XPVHV* xhv = (XPVHV*)SvANY(hv);
a3b680e6 1248 const I32 oldsize = (I32) xhv->xhv_max+1; /* HvMAX(hv)+1 (sick) */
72940dca 1249 register I32 newsize;
1250 register I32 i;
72311751
GS
1251 register char *a;
1252 register HE **aep;
72940dca 1253
7918f24d
NC
1254 PERL_ARGS_ASSERT_HV_KSPLIT;
1255
72940dca 1256 newsize = (I32) newmax; /* possible truncation here */
1257 if (newsize != newmax || newmax <= oldsize)
1258 return;
1259 while ((newsize & (1 + ~newsize)) != newsize) {
1260 newsize &= ~(newsize & (1 + ~newsize)); /* get proper power of 2 */
1261 }
1262 if (newsize < newmax)
1263 newsize *= 2;
1264 if (newsize < newmax)
1265 return; /* overflow detection */
1266
7b2c381c 1267 a = (char *) HvARRAY(hv);
72940dca 1268 if (a) {
3280af22 1269 PL_nomemok = TRUE;
8d6dde3e 1270#if defined(STRANGE_MALLOC) || defined(MYMALLOC)
b79f7545
NC
1271 Renew(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize)
1272 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0), char);
8aacddc1 1273 if (!a) {
4a33f861 1274 PL_nomemok = FALSE;
422a93e5
GA
1275 return;
1276 }
b79f7545 1277 if (SvOOK(hv)) {
7a9b70e9 1278 Copy(&a[oldsize * sizeof(HE*)], &a[newsize * sizeof(HE*)], 1, struct xpvhv_aux);
b79f7545 1279 }
72940dca 1280#else
a02a5408 1281 Newx(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize)
b79f7545 1282 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0), char);
8aacddc1 1283 if (!a) {
3280af22 1284 PL_nomemok = FALSE;
422a93e5
GA
1285 return;
1286 }
7b2c381c 1287 Copy(HvARRAY(hv), a, oldsize * sizeof(HE*), char);
b79f7545
NC
1288 if (SvOOK(hv)) {
1289 Copy(HvAUX(hv), &a[newsize * sizeof(HE*)], 1, struct xpvhv_aux);
1290 }
fba3b22e 1291 if (oldsize >= 64) {
7b2c381c 1292 offer_nice_chunk(HvARRAY(hv),
b79f7545
NC
1293 PERL_HV_ARRAY_ALLOC_BYTES(oldsize)
1294 + (SvOOK(hv) ? sizeof(struct xpvhv_aux) : 0));
72940dca 1295 }
1296 else
7b2c381c 1297 Safefree(HvARRAY(hv));
72940dca 1298#endif
3280af22 1299 PL_nomemok = FALSE;
72311751 1300 Zero(&a[oldsize * sizeof(HE*)], (newsize-oldsize) * sizeof(HE*), char); /* zero 2nd half*/
72940dca 1301 }
1302 else {
a02a5408 1303 Newxz(a, PERL_HV_ARRAY_ALLOC_BYTES(newsize), char);
72940dca 1304 }
cbec9347 1305 xhv->xhv_max = --newsize; /* HvMAX(hv) = --newsize */
7b2c381c 1306 HvARRAY(hv) = (HE **) a;
f4431c56 1307 if (!xhv->xhv_keys /* !HvTOTALKEYS(hv) */) /* skip rest if no entries */
72940dca 1308 return;
1309
72311751
GS
1310 aep = (HE**)a;
1311 for (i=0; i<oldsize; i++,aep++) {
a50a3493
NC
1312 HE **oentry = aep;
1313 HE *entry = *aep;
1314
1315 if (!entry) /* non-existent */
72940dca 1316 continue;
4c9d89c5 1317 do {
6136c704
AL
1318 register I32 j = (HeHASH(entry) & newsize);
1319
1320 if (j != i) {
72940dca 1321 j -= i;
1322 *oentry = HeNEXT(entry);
4d0fbddd 1323 HeNEXT(entry) = aep[j];
72311751 1324 aep[j] = entry;
72940dca 1325 }
1326 else
1327 oentry = &HeNEXT(entry);
4c9d89c5
NC
1328 entry = *oentry;
1329 } while (entry);
72940dca 1330 }
1331}
1332
b3ac6de7 1333HV *
864dbfa3 1334Perl_newHVhv(pTHX_ HV *ohv)
b3ac6de7 1335{
749123ff 1336 dVAR;
9d4ba2ae 1337 HV * const hv = newHV();
f4431c56 1338 STRLEN hv_max;
4beac62f 1339
f4431c56 1340 if (!ohv || !HvTOTALKEYS(ohv))
4beac62f 1341 return hv;
4beac62f 1342 hv_max = HvMAX(ohv);
b3ac6de7 1343
ad64d0ec 1344 if (!SvMAGICAL((const SV *)ohv)) {
b56ba0bf 1345 /* It's an ordinary hash, so copy it fast. AMS 20010804 */
eb160463 1346 STRLEN i;
a3b680e6 1347 const bool shared = !!HvSHAREKEYS(ohv);
aec46f14 1348 HE **ents, ** const oents = (HE **)HvARRAY(ohv);
ff875642 1349 char *a;
a02a5408 1350 Newx(a, PERL_HV_ARRAY_ALLOC_BYTES(hv_max+1), char);
ff875642 1351 ents = (HE**)a;
b56ba0bf
AMS
1352
1353 /* In each bucket... */
1354 for (i = 0; i <= hv_max; i++) {
6136c704 1355 HE *prev = NULL;
aec46f14 1356 HE *oent = oents[i];
b56ba0bf
AMS
1357
1358 if (!oent) {
1359 ents[i] = NULL;
1360 continue;
1361 }
1362
1363 /* Copy the linked list of entries. */
aec46f14 1364 for (; oent; oent = HeNEXT(oent)) {
a3b680e6
AL
1365 const U32 hash = HeHASH(oent);
1366 const char * const key = HeKEY(oent);
1367 const STRLEN len = HeKLEN(oent);
1368 const int flags = HeKFLAGS(oent);
6136c704 1369 HE * const ent = new_HE();
c3acb9e0 1370 SV *const val = HeVAL(oent);
b56ba0bf 1371
c3acb9e0 1372 HeVAL(ent) = SvIMMORTAL(val) ? val : newSVsv(val);
19692e8d 1373 HeKEY_hek(ent)
6e838c70 1374 = shared ? share_hek_flags(key, len, hash, flags)
19692e8d 1375 : save_hek_flags(key, len, hash, flags);
b56ba0bf
AMS
1376 if (prev)
1377 HeNEXT(prev) = ent;
1378 else
1379 ents[i] = ent;
1380 prev = ent;
1381 HeNEXT(ent) = NULL;
1382 }
1383 }
1384
1385 HvMAX(hv) = hv_max;
8aacddc1 1386 HvTOTALKEYS(hv) = HvTOTALKEYS(ohv);
b56ba0bf 1387 HvARRAY(hv) = ents;
aec46f14 1388 } /* not magical */
b56ba0bf
AMS
1389 else {
1390 /* Iterate over ohv, copying keys and values one at a time. */
b3ac6de7 1391 HE *entry;
bfcb3514
NC
1392 const I32 riter = HvRITER_get(ohv);
1393 HE * const eiter = HvEITER_get(ohv);
f4431c56 1394 STRLEN hv_fill = HvFILL(ohv);
b56ba0bf
AMS
1395
1396 /* Can we use fewer buckets? (hv_max is always 2^n-1) */
1397 while (hv_max && hv_max + 1 >= hv_fill * 2)
1398 hv_max = hv_max / 2;
1399 HvMAX(hv) = hv_max;
1400
4a76a316 1401 hv_iterinit(ohv);
e16e2ff8 1402 while ((entry = hv_iternext_flags(ohv, 0))) {
c3acb9e0 1403 SV *const val = HeVAL(entry);
04fe65b0 1404 (void)hv_store_flags(hv, HeKEY(entry), HeKLEN(entry),
c3acb9e0
NC
1405 SvIMMORTAL(val) ? val : newSVsv(val),
1406 HeHASH(entry), HeKFLAGS(entry));
b3ac6de7 1407 }
bfcb3514
NC
1408 HvRITER_set(ohv, riter);
1409 HvEITER_set(ohv, eiter);
b3ac6de7 1410 }
1c846c1f 1411
b3ac6de7
IZ
1412 return hv;
1413}
1414
5b9c0671
NC
1415/* A rather specialised version of newHVhv for copying %^H, ensuring all the
1416 magic stays on it. */
1417HV *
1418Perl_hv_copy_hints_hv(pTHX_ HV *const ohv)
1419{
1420 HV * const hv = newHV();
5b9c0671 1421
f4431c56 1422 if (ohv && HvTOTALKEYS(ohv)) {
5b9c0671 1423 STRLEN hv_max = HvMAX(ohv);
f4431c56 1424 STRLEN hv_fill = HvFILL(ohv);
5b9c0671
NC
1425 HE *entry;
1426 const I32 riter = HvRITER_get(ohv);
1427 HE * const eiter = HvEITER_get(ohv);
1428
1429 while (hv_max && hv_max + 1 >= hv_fill * 2)
1430 hv_max = hv_max / 2;
1431 HvMAX(hv) = hv_max;
1432
1433 hv_iterinit(ohv);
1434 while ((entry = hv_iternext_flags(ohv, 0))) {
1435 SV *const sv = newSVsv(HeVAL(entry));
e3b1b6b1 1436 SV *heksv = newSVhek(HeKEY_hek(entry));
5b9c0671 1437 sv_magic(sv, NULL, PERL_MAGIC_hintselem,
e3b1b6b1
RGS
1438 (char *)heksv, HEf_SVKEY);
1439 SvREFCNT_dec(heksv);
04fe65b0
RGS
1440 (void)hv_store_flags(hv, HeKEY(entry), HeKLEN(entry),
1441 sv, HeHASH(entry), HeKFLAGS(entry));
5b9c0671
NC
1442 }
1443 HvRITER_set(ohv, riter);
1444 HvEITER_set(ohv, eiter);
1445 }
1446 hv_magic(hv, NULL, PERL_MAGIC_hints);
1447 return hv;
1448}
1449
79072805 1450void
864dbfa3 1451Perl_hv_free_ent(pTHX_ HV *hv, register HE *entry)
79072805 1452{
97aff369 1453 dVAR;
16bdeea2
GS
1454 SV *val;
1455
7918f24d
NC
1456 PERL_ARGS_ASSERT_HV_FREE_ENT;
1457
68dc0745 1458 if (!entry)
79072805 1459 return;
16bdeea2 1460 val = HeVAL(entry);
803f2748
DM
1461 if (val && isGV(val) && isGV_with_GP(val) && GvCVu(val) && HvNAME_get(hv))
1462 mro_method_changed_in(hv); /* deletion of method from stash */
16bdeea2 1463 SvREFCNT_dec(val);
68dc0745 1464 if (HeKLEN(entry) == HEf_SVKEY) {
1465 SvREFCNT_dec(HeKEY_sv(entry));
8aacddc1 1466 Safefree(HeKEY_hek(entry));
44a8e56a 1467 }
1468 else if (HvSHAREKEYS(hv))
68dc0745 1469 unshare_hek(HeKEY_hek(entry));
fde52b5c 1470 else
68dc0745 1471 Safefree(HeKEY_hek(entry));
d33b2eba 1472 del_HE(entry);
79072805
LW
1473}
1474
f1c32fec 1475
79072805 1476void
864dbfa3 1477Perl_hv_delayfree_ent(pTHX_ HV *hv, register HE *entry)
79072805 1478{
97aff369 1479 dVAR;
7918f24d
NC
1480
1481 PERL_ARGS_ASSERT_HV_DELAYFREE_ENT;
1482
68dc0745 1483 if (!entry)
79072805 1484 return;
bc4947fc
NC
1485 /* SvREFCNT_inc to counter the SvREFCNT_dec in hv_free_ent */
1486 sv_2mortal(SvREFCNT_inc(HeVAL(entry))); /* free between statements */
68dc0745 1487 if (HeKLEN(entry) == HEf_SVKEY) {
bc4947fc 1488 sv_2mortal(SvREFCNT_inc(HeKEY_sv(entry)));
44a8e56a 1489 }
bc4947fc 1490 hv_free_ent(hv, entry);
79072805
LW
1491}
1492
954c1994
GS
1493/*
1494=for apidoc hv_clear
1495
1496Clears a hash, making it empty.
1497
1498=cut
1499*/
1500
79072805 1501void
864dbfa3 1502Perl_hv_clear(pTHX_ HV *hv)
79072805 1503{
27da23d5 1504 dVAR;
cbec9347 1505 register XPVHV* xhv;
79072805
LW
1506 if (!hv)
1507 return;
49293501 1508
ecae49c0
NC
1509 DEBUG_A(Perl_hv_assert(aTHX_ hv));
1510
34c3c4e3
DM
1511 xhv = (XPVHV*)SvANY(hv);
1512
7b2c381c 1513 if (SvREADONLY(hv) && HvARRAY(hv) != NULL) {
34c3c4e3 1514 /* restricted hash: convert all keys to placeholders */
b464bac0
AL
1515 STRLEN i;
1516 for (i = 0; i <= xhv->xhv_max; i++) {
7b2c381c 1517 HE *entry = (HvARRAY(hv))[i];
3a676441
JH
1518 for (; entry; entry = HeNEXT(entry)) {
1519 /* not already placeholder */
7996736c 1520 if (HeVAL(entry) != &PL_sv_placeholder) {
3a676441 1521 if (HeVAL(entry) && SvREADONLY(HeVAL(entry))) {
6136c704 1522 SV* const keysv = hv_iterkeysv(entry);
3a676441 1523 Perl_croak(aTHX_
95b63a38
JH
1524 "Attempt to delete readonly key '%"SVf"' from a restricted hash",
1525 (void*)keysv);
3a676441
JH
1526 }
1527 SvREFCNT_dec(HeVAL(entry));
7996736c 1528 HeVAL(entry) = &PL_sv_placeholder;
ca732855 1529 HvPLACEHOLDERS(hv)++;
3a676441 1530 }
34c3c4e3
DM
1531 }
1532 }
df8c6964 1533 goto reset;
49293501
MS
1534 }
1535
463ee0b2 1536 hfreeentries(hv);
ca732855 1537 HvPLACEHOLDERS_set(hv, 0);
7b2c381c 1538 if (HvARRAY(hv))
41f62432 1539 Zero(HvARRAY(hv), xhv->xhv_max+1 /* HvMAX(hv)+1 */, HE*);
a0d0e21e
LW
1540
1541 if (SvRMAGICAL(hv))
ad64d0ec 1542 mg_clear(MUTABLE_SV(hv));
574c8022 1543
19692e8d 1544 HvHASKFLAGS_off(hv);
bb443f97 1545 HvREHASH_off(hv);
df8c6964 1546 reset:
b79f7545 1547 if (SvOOK(hv)) {
dd69841b
BB
1548 if(HvNAME_get(hv))
1549 mro_isa_changed_in(hv);
bfcb3514
NC
1550 HvEITER_set(hv, NULL);
1551 }
79072805
LW
1552}
1553
3540d4ce
AB
1554/*
1555=for apidoc hv_clear_placeholders
1556
1557Clears any placeholders from a hash. If a restricted hash has any of its keys
1558marked as readonly and the key is subsequently deleted, the key is not actually
1559deleted but is marked by assigning it a value of &PL_sv_placeholder. This tags
1560it so it will be ignored by future operations such as iterating over the hash,
4cdaeff7 1561but will still allow the hash to have a value reassigned to the key at some
3540d4ce
AB
1562future point. This function clears any such placeholder keys from the hash.
1563See Hash::Util::lock_keys() for an example of its use.
1564
1565=cut
1566*/
1567
1568void
1569Perl_hv_clear_placeholders(pTHX_ HV *hv)
1570{
27da23d5 1571 dVAR;
b3ca2e83
NC
1572 const U32 items = (U32)HvPLACEHOLDERS_get(hv);
1573
7918f24d
NC
1574 PERL_ARGS_ASSERT_HV_CLEAR_PLACEHOLDERS;
1575
b3ca2e83
NC
1576 if (items)
1577 clear_placeholders(hv, items);
1578}
1579
1580static void
1581S_clear_placeholders(pTHX_ HV *hv, U32 items)
1582{
1583 dVAR;
b464bac0 1584 I32 i;
d3677389 1585
7918f24d
NC
1586 PERL_ARGS_ASSERT_CLEAR_PLACEHOLDERS;
1587
d3677389
NC
1588 if (items == 0)
1589 return;
1590
b464bac0 1591 i = HvMAX(hv);
d3677389
NC
1592 do {
1593 /* Loop down the linked list heads */
6136c704 1594 bool first = TRUE;
d3677389 1595 HE **oentry = &(HvARRAY(hv))[i];
cf6db12b 1596 HE *entry;
d3677389 1597
cf6db12b 1598 while ((entry = *oentry)) {
d3677389
NC
1599 if (HeVAL(entry) == &PL_sv_placeholder) {
1600 *oentry = HeNEXT(entry);
2e58978b 1601 if (entry == HvEITER_get(hv))
d3677389
NC
1602 HvLAZYDEL_on(hv);
1603 else
1604 hv_free_ent(hv, entry);
1605
1606 if (--items == 0) {
1607 /* Finished. */
5d88ecd7 1608 HvTOTALKEYS(hv) -= (IV)HvPLACEHOLDERS_get(hv);
d3677389
NC
1609 if (HvKEYS(hv) == 0)
1610 HvHASKFLAGS_off(hv);
5d88ecd7 1611 HvPLACEHOLDERS_set(hv, 0);
d3677389
NC
1612 return;
1613 }
213ce8b3
NC
1614 } else {
1615 oentry = &HeNEXT(entry);
6136c704 1616 first = FALSE;
d3677389
NC
1617 }
1618 }
1619 } while (--i >= 0);
1620 /* You can't get here, hence assertion should always fail. */
1621 assert (items == 0);
1622 assert (0);
3540d4ce
AB
1623}
1624
76e3520e 1625STATIC void
cea2e8a9 1626S_hfreeentries(pTHX_ HV *hv)
79072805 1627{
23976bdd 1628 /* This is the array that we're going to restore */
fd7de8a8 1629 HE **const orig_array = HvARRAY(hv);
23976bdd
NC
1630 HEK *name;
1631 int attempts = 100;
3abe233e 1632
7918f24d
NC
1633 PERL_ARGS_ASSERT_HFREEENTRIES;
1634
fd7de8a8 1635 if (!orig_array)
79072805 1636 return;
a0d0e21e 1637
23976bdd
NC
1638 if (SvOOK(hv)) {
1639 /* If the hash is actually a symbol table with a name, look after the
1640 name. */
1641 struct xpvhv_aux *iter = HvAUX(hv);
1642
1643 name = iter->xhv_name;
1644 iter->xhv_name = NULL;
1645 } else {
1646 name = NULL;
1647 }
1648
23976bdd
NC
1649 /* orig_array remains unchanged throughout the loop. If after freeing all
1650 the entries it turns out that one of the little blighters has triggered
1651 an action that has caused HvARRAY to be re-allocated, then we set
1652 array to the new HvARRAY, and try again. */
1653
1654 while (1) {
1655 /* This is the one we're going to try to empty. First time round
1656 it's the original array. (Hopefully there will only be 1 time
1657 round) */
6136c704 1658 HE ** const array = HvARRAY(hv);
7440661e 1659 I32 i = HvMAX(hv);
23976bdd
NC
1660
1661 /* Because we have taken xhv_name out, the only allocated pointer
1662 in the aux structure that might exist is the backreference array.
1663 */
1664
1665 if (SvOOK(hv)) {
7440661e 1666 HE *entry;
e1a479c5 1667 struct mro_meta *meta;
09aad8f0 1668 struct xpvhv_aux *iter = HvAUX(hv);
cdb996f4
DM
1669 /* weak references: if called from sv_clear(), the backrefs
1670 * should already have been killed; if there are any left, its
1671 * because we're doing hv_clear() or hv_undef(), and the HV
1672 * will continue to live.
1673 * Because while freeing the entries we fake up a NULL HvARRAY
1674 * (and hence HvAUX), we need to store the backref array
1675 * somewhere else; but it still needs to be visible in case
1676 * any the things we free happen to call sv_del_backref().
1677 * We do this by storing it in magic instead.
1678 * If, during the entry freeing, a destructor happens to add
1679 * a new weak backref, then sv_add_backref will look in both
1680 * places (magic in HvAUX) for the AV, but will create a new
8ac9a371
DM
1681 * AV in HvAUX if it can't find one (if it finds it in magic,
1682 * it moves it back into HvAUX. So at the end of the iteration
1683 * we have to allow for this. */
1684
09aad8f0
DM
1685
1686 if (iter->xhv_backreferences) {
5648c0ae
DM
1687 if (SvTYPE(iter->xhv_backreferences) == SVt_PVAV) {
1688 /* The sv_magic will increase the reference count of the AV,
1689 so we need to drop it first. */
09aad8f0 1690 SvREFCNT_dec(iter->xhv_backreferences);
5648c0ae
DM
1691 if (AvFILLp(iter->xhv_backreferences) == -1) {
1692 /* Turns out that the array is empty. Just free it. */
1693 SvREFCNT_dec(iter->xhv_backreferences);
1b8791d1 1694
5648c0ae
DM
1695 } else {
1696 sv_magic(MUTABLE_SV(hv),
1697 MUTABLE_SV(iter->xhv_backreferences),
1698 PERL_MAGIC_backref, NULL, 0);
1699 }
1700 }
1701 else {
1702 MAGIC *mg;
1703 sv_magic(MUTABLE_SV(hv), NULL, PERL_MAGIC_backref, NULL, 0);
1704 mg = mg_find(MUTABLE_SV(hv), PERL_MAGIC_backref);
1705 mg->mg_obj = (SV*)iter->xhv_backreferences;
09aad8f0
DM
1706 }
1707 iter->xhv_backreferences = NULL;
5b285ea4 1708 }
86f55936 1709
23976bdd
NC
1710 entry = iter->xhv_eiter; /* HvEITER(hv) */
1711 if (entry && HvLAZYDEL(hv)) { /* was deleted earlier? */
1712 HvLAZYDEL_off(hv);
1713 hv_free_ent(hv, entry);
1714 }
1715 iter->xhv_riter = -1; /* HvRITER(hv) = -1 */
4608196e 1716 iter->xhv_eiter = NULL; /* HvEITER(hv) = NULL */
b79f7545 1717
e1a479c5 1718 if((meta = iter->xhv_mro_meta)) {
9953ff72
NC
1719 if (meta->mro_linear_all) {
1720 SvREFCNT_dec(MUTABLE_SV(meta->mro_linear_all));
1721 meta->mro_linear_all = NULL;
553e831a 1722 /* This is just acting as a shortcut pointer. */
3a6fa573
NC
1723 meta->mro_linear_current = NULL;
1724 } else if (meta->mro_linear_current) {
553e831a
NC
1725 /* Only the current MRO is stored, so this owns the data.
1726 */
3a6fa573
NC
1727 SvREFCNT_dec(meta->mro_linear_current);
1728 meta->mro_linear_current = NULL;
553e831a 1729 }
e1a479c5 1730 if(meta->mro_nextmethod) SvREFCNT_dec(meta->mro_nextmethod);
a49ba3fc 1731 SvREFCNT_dec(meta->isa);
e1a479c5
BB
1732 Safefree(meta);
1733 iter->xhv_mro_meta = NULL;
1734 }
1735
23976bdd 1736 /* There are now no allocated pointers in the aux structure. */
2f86008e 1737
23976bdd
NC
1738 SvFLAGS(hv) &= ~SVf_OOK; /* Goodbye, aux structure. */
1739 /* What aux structure? */
a0d0e21e 1740 }
bfcb3514 1741
23976bdd 1742 /* make everyone else think the array is empty, so that the destructors
cef0c2ea 1743 * called for freed entries can't recursively mess with us */
23976bdd 1744 HvARRAY(hv) = NULL;
23976bdd
NC
1745 ((XPVHV*) SvANY(hv))->xhv_keys = 0;
1746
7440661e
NC
1747
1748 do {
1749 /* Loop down the linked list heads */
1750 HE *entry = array[i];
1751
1752 while (entry) {
23976bdd
NC
1753 register HE * const oentry = entry;
1754 entry = HeNEXT(entry);
1755 hv_free_ent(hv, oentry);
1756 }
7440661e 1757 } while (--i >= 0);
b79f7545 1758
23976bdd
NC
1759 /* As there are no allocated pointers in the aux structure, it's now
1760 safe to free the array we just cleaned up, if it's not the one we're
1761 going to put back. */
1762 if (array != orig_array) {
1763 Safefree(array);
1764 }
b79f7545 1765
23976bdd
NC
1766 if (!HvARRAY(hv)) {
1767 /* Good. No-one added anything this time round. */
1768 break;
bfcb3514 1769 }
b79f7545 1770
23976bdd
NC
1771 if (SvOOK(hv)) {
1772 /* Someone attempted to iterate or set the hash name while we had
1773 the array set to 0. We'll catch backferences on the next time
1774 round the while loop. */
1775 assert(HvARRAY(hv));
1b8791d1 1776
23976bdd
NC
1777 if (HvAUX(hv)->xhv_name) {
1778 unshare_hek_or_pvn(HvAUX(hv)->xhv_name, 0, 0, 0);
1779 }
1780 }
1781
1782 if (--attempts == 0) {
1783 Perl_die(aTHX_ "panic: hfreeentries failed to free hash - something is repeatedly re-creating entries");
1784 }
6136c704 1785 }
23976bdd
NC
1786
1787 HvARRAY(hv) = orig_array;
1788
1789 /* If the hash was actually a symbol table, put the name back. */
1790 if (name) {
1791 /* We have restored the original array. If name is non-NULL, then
1792 the original array had an aux structure at the end. So this is
1793 valid: */
1794 SvFLAGS(hv) |= SVf_OOK;
1795 HvAUX(hv)->xhv_name = name;
1b8791d1 1796 }
79072805
LW
1797}
1798
954c1994
GS
1799/*
1800=for apidoc hv_undef
1801
1802Undefines the hash.
1803
1804=cut
1805*/
1806
79072805 1807void
864dbfa3 1808Perl_hv_undef(pTHX_ HV *hv)
79072805 1809{
97aff369 1810 dVAR;
cbec9347 1811 register XPVHV* xhv;
bfcb3514 1812 const char *name;
86f55936 1813
79072805
LW
1814 if (!hv)
1815 return;
ecae49c0 1816 DEBUG_A(Perl_hv_assert(aTHX_ hv));
cbec9347 1817 xhv = (XPVHV*)SvANY(hv);
dd69841b 1818
0fa56319 1819 if ((name = HvNAME_get(hv)) && !PL_dirty)
dd69841b
BB
1820 mro_isa_changed_in(hv);
1821
463ee0b2 1822 hfreeentries(hv);
dd69841b 1823 if (name) {
04fe65b0
RGS
1824 if (PL_stashcache)
1825 (void)hv_delete(PL_stashcache, name, HvNAMELEN_get(hv), G_DISCARD);
bd61b366 1826 hv_name_set(hv, NULL, 0, 0);
85e6fe83 1827 }
b79f7545
NC
1828 SvFLAGS(hv) &= ~SVf_OOK;
1829 Safefree(HvARRAY(hv));
cbec9347 1830 xhv->xhv_max = 7; /* HvMAX(hv) = 7 (it's a normal hash) */
7b2c381c 1831 HvARRAY(hv) = 0;
ca732855 1832 HvPLACEHOLDERS_set(hv, 0);
a0d0e21e
LW
1833
1834 if (SvRMAGICAL(hv))
ad64d0ec 1835 mg_clear(MUTABLE_SV(hv));
79072805
LW
1836}
1837
4d0fbddd
NC
1838/*
1839=for apidoc hv_fill
1840
1841Returns the number of hash buckets that happen to be in use. This function is
1842wrapped by the macro C<HvFILL>.
1843
1844Previously this value was stored in the HV structure, rather than being
1845calculated on demand.
1846
1847=cut
1848*/
1849
1850STRLEN
1851Perl_hv_fill(pTHX_ HV const *const hv)
1852{
1853 STRLEN count = 0;
1854 HE **ents = HvARRAY(hv);
1855
1856 PERL_ARGS_ASSERT_HV_FILL;
1857
1858 if (ents) {
fcd24582
NC
1859 HE *const *const last = ents + HvMAX(hv);
1860 count = last + 1 - ents;
4d0fbddd
NC
1861
1862 do {
fcd24582
NC
1863 if (!*ents)
1864 --count;
1865 } while (++ents <= last);
4d0fbddd
NC
1866 }
1867 return count;
1868}
1869
b464bac0 1870static struct xpvhv_aux*
5f66b61c 1871S_hv_auxinit(HV *hv) {
bfcb3514 1872 struct xpvhv_aux *iter;
b79f7545 1873 char *array;
bfcb3514 1874
7918f24d
NC
1875 PERL_ARGS_ASSERT_HV_AUXINIT;
1876
b79f7545 1877 if (!HvARRAY(hv)) {
a02a5408 1878 Newxz(array, PERL_HV_ARRAY_ALLOC_BYTES(HvMAX(hv) + 1)
b79f7545
NC
1879 + sizeof(struct xpvhv_aux), char);
1880 } else {
1881 array = (char *) HvARRAY(hv);
1882 Renew(array, PERL_HV_ARRAY_ALLOC_BYTES(HvMAX(hv) + 1)
1883 + sizeof(struct xpvhv_aux), char);
1884 }
1885 HvARRAY(hv) = (HE**) array;
1886 /* SvOOK_on(hv) attacks the IV flags. */
1887 SvFLAGS(hv) |= SVf_OOK;
1888 iter = HvAUX(hv);
bfcb3514
NC
1889
1890 iter->xhv_riter = -1; /* HvRITER(hv) = -1 */
4608196e 1891 iter->xhv_eiter = NULL; /* HvEITER(hv) = NULL */
bfcb3514 1892 iter->xhv_name = 0;
86f55936 1893 iter->xhv_backreferences = 0;
e1a479c5 1894 iter->xhv_mro_meta = NULL;
bfcb3514
NC
1895 return iter;
1896}
1897
954c1994
GS
1898/*
1899=for apidoc hv_iterinit
1900
1901Prepares a starting point to traverse a hash table. Returns the number of
1902keys in the hash (i.e. the same as C<HvKEYS(tb)>). The return value is
1c846c1f 1903currently only meaningful for hashes without tie magic.
954c1994
GS
1904
1905NOTE: Before version 5.004_65, C<hv_iterinit> used to return the number of
1906hash buckets that happen to be in use. If you still need that esoteric
1907value, you can get it through the macro C<HvFILL(tb)>.
1908
e16e2ff8 1909
954c1994
GS
1910=cut
1911*/
1912
79072805 1913I32
864dbfa3 1914Perl_hv_iterinit(pTHX_ HV *hv)
79072805 1915{
7918f24d
NC
1916 PERL_ARGS_ASSERT_HV_ITERINIT;
1917
1918 /* FIXME: Are we not NULL, or do we croak? Place bets now! */
1919
aa689395 1920 if (!hv)
cea2e8a9 1921 Perl_croak(aTHX_ "Bad hash");
bfcb3514 1922
b79f7545 1923 if (SvOOK(hv)) {
6136c704 1924 struct xpvhv_aux * const iter = HvAUX(hv);
0bd48802 1925 HE * const entry = iter->xhv_eiter; /* HvEITER(hv) */
bfcb3514
NC
1926 if (entry && HvLAZYDEL(hv)) { /* was deleted earlier? */
1927 HvLAZYDEL_off(hv);
1928 hv_free_ent(hv, entry);
1929 }
1930 iter->xhv_riter = -1; /* HvRITER(hv) = -1 */
4608196e 1931 iter->xhv_eiter = NULL; /* HvEITER(hv) = NULL */
bfcb3514 1932 } else {
6136c704 1933 hv_auxinit(hv);
72940dca 1934 }
44a2ac75 1935
cbec9347 1936 /* used to be xhv->xhv_fill before 5.004_65 */
5d88ecd7 1937 return HvTOTALKEYS(hv);
79072805 1938}
bfcb3514
NC
1939
1940I32 *
1941Perl_hv_riter_p(pTHX_ HV *hv) {
1942 struct xpvhv_aux *iter;
1943
7918f24d
NC
1944 PERL_ARGS_ASSERT_HV_RITER_P;
1945
bfcb3514
NC
1946 if (!hv)
1947 Perl_croak(aTHX_ "Bad hash");
1948
6136c704 1949 iter = SvOOK(hv) ? HvAUX(hv) : hv_auxinit(hv);
bfcb3514
NC
1950 return &(iter->xhv_riter);
1951}
1952
1953HE **
1954Perl_hv_eiter_p(pTHX_ HV *hv) {
1955 struct xpvhv_aux *iter;
1956
7918f24d
NC
1957 PERL_ARGS_ASSERT_HV_EITER_P;
1958
bfcb3514
NC
1959 if (!hv)
1960 Perl_croak(aTHX_ "Bad hash");
1961
6136c704 1962 iter = SvOOK(hv) ? HvAUX(hv) : hv_auxinit(hv);
bfcb3514
NC
1963 return &(iter->xhv_eiter);
1964}
1965
1966void
1967Perl_hv_riter_set(pTHX_ HV *hv, I32 riter) {
1968 struct xpvhv_aux *iter;
1969
7918f24d
NC
1970 PERL_ARGS_ASSERT_HV_RITER_SET;
1971
bfcb3514
NC
1972 if (!hv)
1973 Perl_croak(aTHX_ "Bad hash");
1974
b79f7545
NC
1975 if (SvOOK(hv)) {
1976 iter = HvAUX(hv);
1977 } else {
bfcb3514
NC
1978 if (riter == -1)
1979 return;
1980
6136c704 1981 iter = hv_auxinit(hv);
bfcb3514
NC
1982 }
1983 iter->xhv_riter = riter;
1984}
1985
1986void
1987Perl_hv_eiter_set(pTHX_ HV *hv, HE *eiter) {
1988 struct xpvhv_aux *iter;
1989
7918f24d
NC
1990 PERL_ARGS_ASSERT_HV_EITER_SET;
1991
bfcb3514
NC
1992 if (!hv)
1993 Perl_croak(aTHX_ "Bad hash");
1994
b79f7545
NC
1995 if (SvOOK(hv)) {
1996 iter = HvAUX(hv);
1997 } else {
bfcb3514
NC
1998 /* 0 is the default so don't go malloc()ing a new structure just to
1999 hold 0. */
2000 if (!eiter)
2001 return;
2002
6136c704 2003 iter = hv_auxinit(hv);
bfcb3514
NC
2004 }
2005 iter->xhv_eiter = eiter;
2006}
2007
bfcb3514 2008void
4164be69 2009Perl_hv_name_set(pTHX_ HV *hv, const char *name, U32 len, U32 flags)
bfcb3514 2010{
97aff369 2011 dVAR;
b79f7545 2012 struct xpvhv_aux *iter;
7423f6db 2013 U32 hash;
46c461b5 2014
7918f24d 2015 PERL_ARGS_ASSERT_HV_NAME_SET;
46c461b5 2016 PERL_UNUSED_ARG(flags);
bfcb3514 2017
4164be69
NC
2018 if (len > I32_MAX)
2019 Perl_croak(aTHX_ "panic: hv name too long (%"UVuf")", (UV) len);
2020
b79f7545
NC
2021 if (SvOOK(hv)) {
2022 iter = HvAUX(hv);
7423f6db
NC
2023 if (iter->xhv_name) {
2024 unshare_hek_or_pvn(iter->xhv_name, 0, 0, 0);
2025 }
16580ff5 2026 } else {
bfcb3514
NC
2027 if (name == 0)
2028 return;
2029
6136c704 2030 iter = hv_auxinit(hv);
bfcb3514 2031 }
7423f6db 2032 PERL_HASH(hash, name, len);
adf4e37a 2033 iter->xhv_name = name ? share_hek(name, len, hash) : NULL;
bfcb3514
NC
2034}
2035
86f55936
NC
2036AV **
2037Perl_hv_backreferences_p(pTHX_ HV *hv) {
6136c704 2038 struct xpvhv_aux * const iter = SvOOK(hv) ? HvAUX(hv) : hv_auxinit(hv);
7918f24d
NC
2039
2040 PERL_ARGS_ASSERT_HV_BACKREFERENCES_P;
96a5add6 2041 PERL_UNUSED_CONTEXT;
7918f24d 2042
86f55936
NC
2043 return &(iter->xhv_backreferences);
2044}
2045
09aad8f0
DM
2046void
2047Perl_hv_kill_backrefs(pTHX_ HV *hv) {
2048 AV *av;
2049
2050 PERL_ARGS_ASSERT_HV_KILL_BACKREFS;
2051
2052 if (!SvOOK(hv))
2053 return;
2054
2055 av = HvAUX(hv)->xhv_backreferences;
2056
2057 if (av) {
2058 HvAUX(hv)->xhv_backreferences = 0;
2059 Perl_sv_kill_backrefs(aTHX_ MUTABLE_SV(hv), av);
5648c0ae
DM
2060 if (SvTYPE(av) == SVt_PVAV)
2061 SvREFCNT_dec(av);
09aad8f0
DM
2062 }
2063}
2064
954c1994 2065/*
7a7b9979
NC
2066hv_iternext is implemented as a macro in hv.h
2067
954c1994
GS
2068=for apidoc hv_iternext
2069
2070Returns entries from a hash iterator. See C<hv_iterinit>.
2071
fe7bca90
NC
2072You may call C<hv_delete> or C<hv_delete_ent> on the hash entry that the
2073iterator currently points to, without losing your place or invalidating your
2074iterator. Note that in this case the current entry is deleted from the hash
2075with your iterator holding the last reference to it. Your iterator is flagged
2076to free the entry on the next call to C<hv_iternext>, so you must not discard
2077your iterator immediately else the entry will leak - call C<hv_iternext> to
2078trigger the resource deallocation.
2079
fe7bca90
NC
2080=for apidoc hv_iternext_flags
2081
2082Returns entries from a hash iterator. See C<hv_iterinit> and C<hv_iternext>.
2083The C<flags> value will normally be zero; if HV_ITERNEXT_WANTPLACEHOLDERS is
2084set the placeholders keys (for restricted hashes) will be returned in addition
2085to normal keys. By default placeholders are automatically skipped over.
7996736c
MHM
2086Currently a placeholder is implemented with a value that is
2087C<&Perl_sv_placeholder>. Note that the implementation of placeholders and
fe7bca90
NC
2088restricted hashes may change, and the implementation currently is
2089insufficiently abstracted for any change to be tidy.
e16e2ff8 2090
fe7bca90 2091=cut
e16e2ff8
NC
2092*/
2093
2094HE *
2095Perl_hv_iternext_flags(pTHX_ HV *hv, I32 flags)
2096{
27da23d5 2097 dVAR;
cbec9347 2098 register XPVHV* xhv;
79072805 2099 register HE *entry;
a0d0e21e 2100 HE *oldentry;
463ee0b2 2101 MAGIC* mg;
bfcb3514 2102 struct xpvhv_aux *iter;
79072805 2103
7918f24d
NC
2104 PERL_ARGS_ASSERT_HV_ITERNEXT_FLAGS;
2105
79072805 2106 if (!hv)
cea2e8a9 2107 Perl_croak(aTHX_ "Bad hash");
81714fb9 2108
cbec9347 2109 xhv = (XPVHV*)SvANY(hv);
bfcb3514 2110
b79f7545 2111 if (!SvOOK(hv)) {
bfcb3514
NC
2112 /* Too many things (well, pp_each at least) merrily assume that you can
2113 call iv_iternext without calling hv_iterinit, so we'll have to deal
2114 with it. */
2115 hv_iterinit(hv);
bfcb3514 2116 }
b79f7545 2117 iter = HvAUX(hv);
bfcb3514
NC
2118
2119 oldentry = entry = iter->xhv_eiter; /* HvEITER(hv) */
e62cc96a 2120 if (SvMAGICAL(hv) && SvRMAGICAL(hv)) {
ad64d0ec 2121 if ( ( mg = mg_find((const SV *)hv, PERL_MAGIC_tied) ) ) {
e62cc96a
YO
2122 SV * const key = sv_newmortal();
2123 if (entry) {
2124 sv_setsv(key, HeSVKEY_force(entry));
2125 SvREFCNT_dec(HeSVKEY(entry)); /* get rid of previous key */
2126 }
2127 else {
2128 char *k;
2129 HEK *hek;
2130
2131 /* one HE per MAGICAL hash */
2132 iter->xhv_eiter = entry = new_HE(); /* HvEITER(hv) = new_HE() */
2133 Zero(entry, 1, HE);
ad64d0ec 2134 Newxz(k, HEK_BASESIZE + sizeof(const SV *), char);
e62cc96a
YO
2135 hek = (HEK*)k;
2136 HeKEY_hek(entry) = hek;
2137 HeKLEN(entry) = HEf_SVKEY;
2138 }
ad64d0ec 2139 magic_nextpack(MUTABLE_SV(hv),mg,key);
e62cc96a
YO
2140 if (SvOK(key)) {
2141 /* force key to stay around until next time */
2142 HeSVKEY_set(entry, SvREFCNT_inc_simple_NN(key));
2143 return entry; /* beware, hent_val is not set */
2144 }
ef8d46e8 2145 SvREFCNT_dec(HeVAL(entry));
e62cc96a
YO
2146 Safefree(HeKEY_hek(entry));
2147 del_HE(entry);
2148 iter->xhv_eiter = NULL; /* HvEITER(hv) = NULL */
2149 return NULL;
81714fb9 2150 }
79072805 2151 }
7ee146b1 2152#if defined(DYNAMIC_ENV_FETCH) && !defined(__riscos__) /* set up %ENV for iteration */
ad64d0ec
NC
2153 if (!entry && SvRMAGICAL((const SV *)hv)
2154 && mg_find((const SV *)hv, PERL_MAGIC_env)) {
f675dbe5 2155 prime_env_iter();
03026e68
JM
2156#ifdef VMS
2157 /* The prime_env_iter() on VMS just loaded up new hash values
2158 * so the iteration count needs to be reset back to the beginning
2159 */
2160 hv_iterinit(hv);
2161 iter = HvAUX(hv);
2162 oldentry = entry = iter->xhv_eiter; /* HvEITER(hv) */
2163#endif
2164 }
f675dbe5 2165#endif
463ee0b2 2166
b79f7545
NC
2167 /* hv_iterint now ensures this. */
2168 assert (HvARRAY(hv));
2169
015a5f36 2170 /* At start of hash, entry is NULL. */
fde52b5c 2171 if (entry)
8aacddc1 2172 {
fde52b5c 2173 entry = HeNEXT(entry);
e16e2ff8
NC
2174 if (!(flags & HV_ITERNEXT_WANTPLACEHOLDERS)) {
2175 /*
2176 * Skip past any placeholders -- don't want to include them in
2177 * any iteration.
2178 */
7996736c 2179 while (entry && HeVAL(entry) == &PL_sv_placeholder) {
e16e2ff8
NC
2180 entry = HeNEXT(entry);
2181 }
8aacddc1
NIS
2182 }
2183 }
015a5f36 2184
9eb4ebd1
NC
2185 /* Skip the entire loop if the hash is empty. */
2186 if ((flags & HV_ITERNEXT_WANTPLACEHOLDERS)
2187 ? HvTOTALKEYS(hv) : HvUSEDKEYS(hv)) {
900ac051
MM
2188 while (!entry) {
2189 /* OK. Come to the end of the current list. Grab the next one. */
2190
2191 iter->xhv_riter++; /* HvRITER(hv)++ */
2192 if (iter->xhv_riter > (I32)xhv->xhv_max /* HvRITER(hv) > HvMAX(hv) */) {
2193 /* There is no next one. End of the hash. */
2194 iter->xhv_riter = -1; /* HvRITER(hv) = -1 */
2195 break;
2196 }
2197 entry = (HvARRAY(hv))[iter->xhv_riter];
8aacddc1 2198
900ac051
MM
2199 if (!(flags & HV_ITERNEXT_WANTPLACEHOLDERS)) {
2200 /* If we have an entry, but it's a placeholder, don't count it.
2201 Try the next. */
2202 while (entry && HeVAL(entry) == &PL_sv_placeholder)
2203 entry = HeNEXT(entry);
2204 }
2205 /* Will loop again if this linked list starts NULL
2206 (for HV_ITERNEXT_WANTPLACEHOLDERS)
2207 or if we run through it and find only placeholders. */
015a5f36 2208 }
fde52b5c 2209 }
79072805 2210
72940dca 2211 if (oldentry && HvLAZYDEL(hv)) { /* was deleted earlier? */
2212 HvLAZYDEL_off(hv);
68dc0745 2213 hv_free_ent(hv, oldentry);
72940dca 2214 }
a0d0e21e 2215
fdcd69b6 2216 /*if (HvREHASH(hv) && entry && !HeKREHASH(entry))
6c9570dc 2217 PerlIO_printf(PerlIO_stderr(), "Awooga %p %p\n", (void*)hv, (void*)entry);*/
fdcd69b6 2218
bfcb3514 2219 iter->xhv_eiter = entry; /* HvEITER(hv) = entry */
79072805
LW
2220 return entry;
2221}
2222
954c1994
GS
2223/*
2224=for apidoc hv_iterkey
2225
2226Returns the key from the current position of the hash iterator. See
2227C<hv_iterinit>.
2228
2229=cut
2230*/
2231
79072805 2232char *
864dbfa3 2233Perl_hv_iterkey(pTHX_ register HE *entry, I32 *retlen)
79072805 2234{
7918f24d
NC
2235 PERL_ARGS_ASSERT_HV_ITERKEY;
2236
fde52b5c 2237 if (HeKLEN(entry) == HEf_SVKEY) {
fb73857a 2238 STRLEN len;
0bd48802 2239 char * const p = SvPV(HeKEY_sv(entry), len);
fb73857a 2240 *retlen = len;
2241 return p;
fde52b5c 2242 }
2243 else {
2244 *retlen = HeKLEN(entry);
2245 return HeKEY(entry);
2246 }
2247}
2248
2249/* unlike hv_iterval(), this always returns a mortal copy of the key */
954c1994
GS
2250/*
2251=for apidoc hv_iterkeysv
2252
2253Returns the key as an C<SV*> from the current position of the hash
2254iterator. The return value will always be a mortal copy of the key. Also
2255see C<hv_iterinit>.
2256
2257=cut
2258*/
2259
fde52b5c 2260SV *
864dbfa3 2261Perl_hv_iterkeysv(pTHX_ register HE *entry)
fde52b5c 2262{
7918f24d
NC
2263 PERL_ARGS_ASSERT_HV_ITERKEYSV;
2264
c1b02ed8 2265 return sv_2mortal(newSVhek(HeKEY_hek(entry)));
79072805
LW
2266}
2267
954c1994
GS
2268/*
2269=for apidoc hv_iterval
2270
2271Returns the value from the current position of the hash iterator. See
2272C<hv_iterkey>.
2273
2274=cut
2275*/
2276
79072805 2277SV *
864dbfa3 2278Perl_hv_iterval(pTHX_ HV *hv, register HE *entry)
79072805 2279{
7918f24d
NC
2280 PERL_ARGS_ASSERT_HV_ITERVAL;
2281
8990e307 2282 if (SvRMAGICAL(hv)) {
ad64d0ec 2283 if (mg_find((const SV *)hv, PERL_MAGIC_tied)) {
c4420975 2284 SV* const sv = sv_newmortal();
bbce6d69 2285 if (HeKLEN(entry) == HEf_SVKEY)
ad64d0ec 2286 mg_copy(MUTABLE_SV(hv), sv, (char*)HeKEY_sv(entry), HEf_SVKEY);
a3b680e6 2287 else
ad64d0ec 2288 mg_copy(MUTABLE_SV(hv), sv, HeKEY(entry), HeKLEN(entry));
463ee0b2
LW
2289 return sv;
2290 }
79072805 2291 }
fde52b5c 2292 return HeVAL(entry);
79072805
LW
2293}
2294
954c1994
GS
2295/*
2296=for apidoc hv_iternextsv
2297
2298Performs an C<hv_iternext>, C<hv_iterkey>, and C<hv_iterval> in one
2299operation.
2300
2301=cut
2302*/
2303
a0d0e21e 2304SV *
864dbfa3 2305Perl_hv_iternextsv(pTHX_ HV *hv, char **key, I32 *retlen)
a0d0e21e 2306{
0bd48802
AL
2307 HE * const he = hv_iternext_flags(hv, 0);
2308
7918f24d
NC
2309 PERL_ARGS_ASSERT_HV_ITERNEXTSV;
2310
0bd48802 2311 if (!he)
a0d0e21e
LW
2312 return NULL;
2313 *key = hv_iterkey(he, retlen);
2314 return hv_iterval(hv, he);
2315}
2316
954c1994 2317/*
bc5cdc23
NC
2318
2319Now a macro in hv.h
2320
954c1994
GS
2321=for apidoc hv_magic
2322
2323Adds magic to a hash. See C<sv_magic>.
2324
2325=cut
2326*/
2327
bbce6d69 2328/* possibly free a shared string if no one has access to it
fde52b5c 2329 * len and hash must both be valid for str.
2330 */
bbce6d69 2331void
864dbfa3 2332Perl_unsharepvn(pTHX_ const char *str, I32 len, U32 hash)
fde52b5c 2333{
19692e8d
NC
2334 unshare_hek_or_pvn (NULL, str, len, hash);
2335}
2336
2337
2338void
2339Perl_unshare_hek(pTHX_ HEK *hek)
2340{
bf11fd37 2341 assert(hek);
19692e8d
NC
2342 unshare_hek_or_pvn(hek, NULL, 0, 0);
2343}
2344
2345/* possibly free a shared string if no one has access to it
2346 hek if non-NULL takes priority over the other 3, else str, len and hash
2347 are used. If so, len and hash must both be valid for str.
2348 */
df132699 2349STATIC void
97ddebaf 2350S_unshare_hek_or_pvn(pTHX_ const HEK *hek, const char *str, I32 len, U32 hash)
19692e8d 2351{
97aff369 2352 dVAR;
cbec9347 2353 register XPVHV* xhv;
20454177 2354 HE *entry;
fde52b5c 2355 register HE **oentry;
45d1cc86 2356 HE **first;
c3654f1a 2357 bool is_utf8 = FALSE;
19692e8d 2358 int k_flags = 0;
aec46f14 2359 const char * const save = str;
cbbf8932 2360 struct shared_he *he = NULL;
c3654f1a 2361
19692e8d 2362 if (hek) {
cbae3960
NC
2363 /* Find the shared he which is just before us in memory. */
2364 he = (struct shared_he *)(((char *)hek)
2365 - STRUCT_OFFSET(struct shared_he,
2366 shared_he_hek));
2367
2368 /* Assert that the caller passed us a genuine (or at least consistent)
2369 shared hek */
2370 assert (he->shared_he_he.hent_hek == hek);
29404ae0 2371
de616631
NC
2372 if (he->shared_he_he.he_valu.hent_refcount - 1) {
2373 --he->shared_he_he.he_valu.hent_refcount;
29404ae0
NC
2374 return;
2375 }
29404ae0 2376
19692e8d
NC
2377 hash = HEK_HASH(hek);
2378 } else if (len < 0) {
2379 STRLEN tmplen = -len;
2380 is_utf8 = TRUE;
2381 /* See the note in hv_fetch(). --jhi */
2382 str = (char*)bytes_from_utf8((U8*)str, &tmplen, &is_utf8);
2383 len = tmplen;
2384 if (is_utf8)
2385 k_flags = HVhek_UTF8;
2386 if (str != save)
2387 k_flags |= HVhek_WASUTF8 | HVhek_FREEKEY;
c3654f1a 2388 }
1c846c1f 2389
de616631 2390 /* what follows was the moral equivalent of:
6b88bc9c 2391 if ((Svp = hv_fetch(PL_strtab, tmpsv, FALSE, hash))) {
a0714e2c 2392 if (--*Svp == NULL)
6b88bc9c 2393 hv_delete(PL_strtab, str, len, G_DISCARD, hash);
bbce6d69 2394 } */
cbec9347 2395 xhv = (XPVHV*)SvANY(PL_strtab);
fde52b5c 2396 /* assert(xhv_array != 0) */
45d1cc86 2397 first = oentry = &(HvARRAY(PL_strtab))[hash & (I32) HvMAX(PL_strtab)];
6c1b96a1
NC
2398 if (he) {
2399 const HE *const he_he = &(he->shared_he_he);
45d1cc86 2400 for (entry = *oentry; entry; oentry = &HeNEXT(entry), entry = *oentry) {
35ab5632
NC
2401 if (entry == he_he)
2402 break;
19692e8d
NC
2403 }
2404 } else {
35a4481c 2405 const int flags_masked = k_flags & HVhek_MASK;
45d1cc86 2406 for (entry = *oentry; entry; oentry = &HeNEXT(entry), entry = *oentry) {
19692e8d
NC
2407 if (HeHASH(entry) != hash) /* strings can't be equal */
2408 continue;
2409 if (HeKLEN(entry) != len)
2410 continue;
2411 if (HeKEY(entry) != str && memNE(HeKEY(entry),str,len)) /* is this it? */
2412 continue;
2413 if (HeKFLAGS(entry) != flags_masked)
2414 continue;
19692e8d
NC
2415 break;
2416 }
2417 }
2418
35ab5632
NC
2419 if (entry) {
2420 if (--entry->he_valu.hent_refcount == 0) {
19692e8d 2421 *oentry = HeNEXT(entry);
cbae3960 2422 Safefree(entry);
4c7185a0 2423 xhv->xhv_keys--; /* HvTOTALKEYS(hv)-- */
19692e8d 2424 }
fde52b5c 2425 }
19692e8d 2426
9b387841
NC
2427 if (!entry)
2428 Perl_ck_warner_d(aTHX_ packWARN(WARN_INTERNAL),
2429 "Attempt to free non-existent shared string '%s'%s"
2430 pTHX__FORMAT,
2431 hek ? HEK_KEY(hek) : str,
2432 ((k_flags & HVhek_UTF8) ? " (utf8)" : "") pTHX__VALUE);
19692e8d
NC
2433 if (k_flags & HVhek_FREEKEY)
2434 Safefree(str);
fde52b5c 2435}
2436
bbce6d69 2437/* get a (constant) string ptr from the global string table
2438 * string will get added if it is not already there.
fde52b5c 2439 * len and hash must both be valid for str.
2440 */
bbce6d69 2441HEK *
864dbfa3 2442Perl_share_hek(pTHX_ const char *str, I32 len, register U32 hash)
fde52b5c 2443{
da58a35d 2444 bool is_utf8 = FALSE;
19692e8d 2445 int flags = 0;
aec46f14 2446 const char * const save = str;
da58a35d 2447
7918f24d
NC
2448 PERL_ARGS_ASSERT_SHARE_HEK;
2449
da58a35d 2450 if (len < 0) {
77caf834 2451 STRLEN tmplen = -len;
da58a35d 2452 is_utf8 = TRUE;
77caf834
JH
2453 /* See the note in hv_fetch(). --jhi */
2454 str = (char*)bytes_from_utf8((U8*)str, &tmplen, &is_utf8);
2455 len = tmplen;
19692e8d
NC
2456 /* If we were able to downgrade here, then than means that we were passed
2457 in a key which only had chars 0-255, but was utf8 encoded. */
2458 if (is_utf8)
2459 flags = HVhek_UTF8;
2460 /* If we found we were able to downgrade the string to bytes, then
2461 we should flag that it needs upgrading on keys or each. Also flag
2462 that we need share_hek_flags to free the string. */
2463 if (str != save)
2464 flags |= HVhek_WASUTF8 | HVhek_FREEKEY;
2465 }
2466
6e838c70 2467 return share_hek_flags (str, len, hash, flags);
19692e8d
NC
2468}
2469
6e838c70 2470STATIC HEK *
19692e8d
NC
2471S_share_hek_flags(pTHX_ const char *str, I32 len, register U32 hash, int flags)
2472{
97aff369 2473 dVAR;
19692e8d 2474 register HE *entry;
35a4481c 2475 const int flags_masked = flags & HVhek_MASK;
263cb4a6 2476 const U32 hindex = hash & (I32) HvMAX(PL_strtab);
7918f24d
NC
2477 register XPVHV * const xhv = (XPVHV*)SvANY(PL_strtab);
2478
2479 PERL_ARGS_ASSERT_SHARE_HEK_FLAGS;
bbce6d69 2480
fde52b5c 2481 /* what follows is the moral equivalent of:
1c846c1f 2482
6b88bc9c 2483 if (!(Svp = hv_fetch(PL_strtab, str, len, FALSE)))
a0714e2c 2484 hv_store(PL_strtab, str, len, NULL, hash);
fdcd69b6
NC
2485
2486 Can't rehash the shared string table, so not sure if it's worth
2487 counting the number of entries in the linked list
bbce6d69 2488 */
7918f24d 2489
fde52b5c 2490 /* assert(xhv_array != 0) */
263cb4a6
NC
2491 entry = (HvARRAY(PL_strtab))[hindex];
2492 for (;entry; entry = HeNEXT(entry)) {
fde52b5c 2493 if (HeHASH(entry) != hash) /* strings can't be equal */
2494 continue;
2495 if (HeKLEN(entry) != len)
2496 continue;
1c846c1f 2497 if (HeKEY(entry) != str && memNE(HeKEY(entry),str,len)) /* is this it? */
fde52b5c 2498 continue;
19692e8d 2499 if (HeKFLAGS(entry) != flags_masked)
c3654f1a 2500 continue;
fde52b5c 2501 break;
2502 }
263cb4a6
NC
2503
2504 if (!entry) {
45d1cc86
NC
2505 /* What used to be head of the list.
2506 If this is NULL, then we're the first entry for this slot, which
2507 means we need to increate fill. */
cbae3960
NC
2508 struct shared_he *new_entry;
2509 HEK *hek;
2510 char *k;
263cb4a6
NC
2511 HE **const head = &HvARRAY(PL_strtab)[hindex];
2512 HE *const next = *head;
cbae3960
NC
2513
2514 /* We don't actually store a HE from the arena and a regular HEK.
2515 Instead we allocate one chunk of memory big enough for both,
2516 and put the HEK straight after the HE. This way we can find the
2517 HEK directly from the HE.
2518 */
2519
a02a5408 2520 Newx(k, STRUCT_OFFSET(struct shared_he,
cbae3960
NC
2521 shared_he_hek.hek_key[0]) + len + 2, char);
2522 new_entry = (struct shared_he *)k;
2523 entry = &(new_entry->shared_he_he);
2524 hek = &(new_entry->shared_he_hek);
2525
2526 Copy(str, HEK_KEY(hek), len, char);
2527 HEK_KEY(hek)[len] = 0;
2528 HEK_LEN(hek) = len;
2529 HEK_HASH(hek) = hash;
2530 HEK_FLAGS(hek) = (unsigned char)flags_masked;
2531
2532 /* Still "point" to the HEK, so that other code need not know what
2533 we're up to. */
2534 HeKEY_hek(entry) = hek;
de616631 2535 entry->he_valu.hent_refcount = 0;
263cb4a6
NC
2536 HeNEXT(entry) = next;
2537 *head = entry;
cbae3960 2538
4c7185a0 2539 xhv->xhv_keys++; /* HvTOTALKEYS(hv)++ */
263cb4a6 2540 if (!next) { /* initial entry? */
5ac36297 2541 } else if (xhv->xhv_keys > xhv->xhv_max /* HvKEYS(hv) > HvMAX(hv) */) {
cbec9347 2542 hsplit(PL_strtab);
bbce6d69 2543 }
2544 }
2545
de616631 2546 ++entry->he_valu.hent_refcount;
19692e8d
NC
2547
2548 if (flags & HVhek_FREEKEY)
f9a63242 2549 Safefree(str);
19692e8d 2550
6e838c70 2551 return HeKEY_hek(entry);
fde52b5c 2552}
ecae49c0 2553
ca732855
NC
2554I32 *
2555Perl_hv_placeholders_p(pTHX_ HV *hv)
2556{
2557 dVAR;
ad64d0ec 2558 MAGIC *mg = mg_find((const SV *)hv, PERL_MAGIC_rhash);
ca732855 2559
7918f24d
NC
2560 PERL_ARGS_ASSERT_HV_PLACEHOLDERS_P;
2561
ca732855 2562 if (!mg) {
ad64d0ec 2563 mg = sv_magicext(MUTABLE_SV(hv), 0, PERL_MAGIC_rhash, 0, 0, 0);
ca732855
NC
2564
2565 if (!mg) {
2566 Perl_die(aTHX_ "panic: hv_placeholders_p");
2567 }
2568 }
2569 return &(mg->mg_len);
2570}
2571
2572
2573I32
0c289d13 2574Perl_hv_placeholders_get(pTHX_ const HV *hv)
ca732855
NC
2575{
2576 dVAR;
0c289d13 2577 MAGIC * const mg = mg_find((const SV *)hv, PERL_MAGIC_rhash);
ca732855 2578
7918f24d
NC
2579 PERL_ARGS_ASSERT_HV_PLACEHOLDERS_GET;
2580
ca732855
NC
2581 return mg ? mg->mg_len : 0;
2582}
2583
2584void
ac1e784a 2585Perl_hv_placeholders_set(pTHX_ HV *hv, I32 ph)
ca732855
NC
2586{
2587 dVAR;
ad64d0ec 2588 MAGIC * const mg = mg_find((const SV *)hv, PERL_MAGIC_rhash);
ca732855 2589
7918f24d
NC
2590 PERL_ARGS_ASSERT_HV_PLACEHOLDERS_SET;
2591
ca732855
NC
2592 if (mg) {
2593 mg->mg_len = ph;
2594 } else if (ph) {
ad64d0ec 2595 if (!sv_magicext(MUTABLE_SV(hv), 0, PERL_MAGIC_rhash, 0, 0, ph))
ca732855
NC
2596 Perl_die(aTHX_ "panic: hv_placeholders_set");
2597 }
2598 /* else we don't need to add magic to record 0 placeholders. */
2599}
ecae49c0 2600
2a49f0f5 2601STATIC SV *
7b0bddfa
NC
2602S_refcounted_he_value(pTHX_ const struct refcounted_he *he)
2603{
0b2d3faa 2604 dVAR;
7b0bddfa 2605 SV *value;
7918f24d
NC
2606
2607 PERL_ARGS_ASSERT_REFCOUNTED_HE_VALUE;
2608
7b0bddfa
NC
2609 switch(he->refcounted_he_data[0] & HVrhek_typemask) {
2610 case HVrhek_undef:
2611 value = newSV(0);
2612 break;
2613 case HVrhek_delete:
2614 value = &PL_sv_placeholder;
2615 break;
2616 case HVrhek_IV:
44ebaf21
NC
2617 value = newSViv(he->refcounted_he_val.refcounted_he_u_iv);
2618 break;
2619 case HVrhek_UV:
2620 value = newSVuv(he->refcounted_he_val.refcounted_he_u_uv);
7b0bddfa
NC
2621 break;
2622 case HVrhek_PV:
44ebaf21 2623 case HVrhek_PV_UTF8:
7b0bddfa
NC
2624 /* Create a string SV that directly points to the bytes in our
2625 structure. */
b9f83d2f 2626 value = newSV_type(SVt_PV);
7b0bddfa
NC
2627 SvPV_set(value, (char *) he->refcounted_he_data + 1);
2628 SvCUR_set(value, he->refcounted_he_val.refcounted_he_u_len);
2629 /* This stops anything trying to free it */
2630 SvLEN_set(value, 0);
2631 SvPOK_on(value);
2632 SvREADONLY_on(value);
44ebaf21 2633 if ((he->refcounted_he_data[0] & HVrhek_typemask) == HVrhek_PV_UTF8)
7b0bddfa
NC
2634 SvUTF8_on(value);
2635 break;
2636 default:
2637 Perl_croak(aTHX_ "panic: refcounted_he_value bad flags %x",
2638 he->refcounted_he_data[0]);
2639 }
2640 return value;
2641}
2642
ecae49c0 2643/*
b3ca2e83
NC
2644=for apidoc refcounted_he_chain_2hv
2645
abc25d8c 2646Generates and returns a C<HV *> by walking up the tree starting at the passed
b3ca2e83
NC
2647in C<struct refcounted_he *>.
2648
2649=cut
2650*/
2651HV *
2652Perl_refcounted_he_chain_2hv(pTHX_ const struct refcounted_he *chain)
2653{
7a89be66 2654 dVAR;
b3ca2e83
NC
2655 HV *hv = newHV();
2656 U32 placeholders = 0;
2657 /* We could chase the chain once to get an idea of the number of keys,
2658 and call ksplit. But for now we'll make a potentially inefficient
2659 hash with only 8 entries in its array. */
2660 const U32 max = HvMAX(hv);
2661
2662 if (!HvARRAY(hv)) {
2663 char *array;
2664 Newxz(array, PERL_HV_ARRAY_ALLOC_BYTES(max + 1), char);
2665 HvARRAY(hv) = (HE**)array;
2666 }
2667
2668 while (chain) {
cbb1fbea 2669#ifdef USE_ITHREADS
b6bbf3fa 2670 U32 hash = chain->refcounted_he_hash;
cbb1fbea
NC
2671#else
2672 U32 hash = HEK_HASH(chain->refcounted_he_hek);
2673#endif
b3ca2e83
NC
2674 HE **oentry = &((HvARRAY(hv))[hash & max]);
2675 HE *entry = *oentry;
b6bbf3fa 2676 SV *value;
cbb1fbea 2677
b3ca2e83
NC
2678 for (; entry; entry = HeNEXT(entry)) {
2679 if (HeHASH(entry) == hash) {
9f769845
NC
2680 /* We might have a duplicate key here. If so, entry is older
2681 than the key we've already put in the hash, so if they are
2682 the same, skip adding entry. */
2683#ifdef USE_ITHREADS
2684 const STRLEN klen = HeKLEN(entry);
2685 const char *const key = HeKEY(entry);
2686 if (klen == chain->refcounted_he_keylen
2687 && (!!HeKUTF8(entry)
2688 == !!(chain->refcounted_he_data[0] & HVhek_UTF8))
2689 && memEQ(key, REF_HE_KEY(chain), klen))
2690 goto next_please;
2691#else
2692 if (HeKEY_hek(entry) == chain->refcounted_he_hek)
2693 goto next_please;
2694 if (HeKLEN(entry) == HEK_LEN(chain->refcounted_he_hek)
2695 && HeKUTF8(entry) == HEK_UTF8(chain->refcounted_he_hek)
2696 && memEQ(HeKEY(entry), HEK_KEY(chain->refcounted_he_hek),
2697 HeKLEN(entry)))
2698 goto next_please;
2699#endif
b3ca2e83
NC
2700 }
2701 }
2702 assert (!entry);
2703 entry = new_HE();
2704
cbb1fbea
NC
2705#ifdef USE_ITHREADS
2706 HeKEY_hek(entry)
7b0bddfa 2707 = share_hek_flags(REF_HE_KEY(chain),
b6bbf3fa
NC
2708 chain->refcounted_he_keylen,
2709 chain->refcounted_he_hash,
2710 (chain->refcounted_he_data[0]
2711 & (HVhek_UTF8|HVhek_WASUTF8)));
cbb1fbea 2712#else
71ad1b0c 2713 HeKEY_hek(entry) = share_hek_hek(chain->refcounted_he_hek);
cbb1fbea 2714#endif
7b0bddfa
NC
2715 value = refcounted_he_value(chain);
2716 if (value == &PL_sv_placeholder)
b3ca2e83 2717 placeholders++;
b6bbf3fa 2718 HeVAL(entry) = value;
b3ca2e83
NC
2719
2720 /* Link it into the chain. */
2721 HeNEXT(entry) = *oentry;
b3ca2e83
NC
2722 *oentry = entry;
2723
2724 HvTOTALKEYS(hv)++;
2725
2726 next_please:
71ad1b0c 2727 chain = chain->refcounted_he_next;
b3ca2e83
NC
2728 }
2729
2730 if (placeholders) {
2731 clear_placeholders(hv, placeholders);
2732 HvTOTALKEYS(hv) -= placeholders;
2733 }
2734
2735 /* We could check in the loop to see if we encounter any keys with key
2736 flags, but it's probably not worth it, as this per-hash flag is only
2737 really meant as an optimisation for things like Storable. */
2738 HvHASKFLAGS_on(hv);
def9038f 2739 DEBUG_A(Perl_hv_assert(aTHX_ hv));
b3ca2e83
NC
2740
2741 return hv;
2742}
2743
7b0bddfa
NC
2744SV *
2745Perl_refcounted_he_fetch(pTHX_ const struct refcounted_he *chain, SV *keysv,
2746 const char *key, STRLEN klen, int flags, U32 hash)
2747{
0b2d3faa 2748 dVAR;
7b0bddfa
NC
2749 /* Just to be awkward, if you're using this interface the UTF-8-or-not-ness
2750 of your key has to exactly match that which is stored. */
2751 SV *value = &PL_sv_placeholder;
7b0bddfa 2752
cd1d2f8a
NC
2753 if (chain) {
2754 /* No point in doing any of this if there's nothing to find. */
2755 bool is_utf8;
7b0bddfa 2756
cd1d2f8a
NC
2757 if (keysv) {
2758 if (flags & HVhek_FREEKEY)
2759 Safefree(key);
2760 key = SvPV_const(keysv, klen);
2761 flags = 0;
2762 is_utf8 = (SvUTF8(keysv) != 0);
2763 } else {
2764 is_utf8 = ((flags & HVhek_UTF8) ? TRUE : FALSE);
2765 }
2766
2767 if (!hash) {
2768 if (keysv && (SvIsCOW_shared_hash(keysv))) {
2769 hash = SvSHARED_HASH(keysv);
2770 } else {
2771 PERL_HASH(hash, key, klen);
2772 }
2773 }
7b0bddfa 2774
cd1d2f8a 2775 for (; chain; chain = chain->refcounted_he_next) {
7b0bddfa 2776#ifdef USE_ITHREADS
cd1d2f8a
NC
2777 if (hash != chain->refcounted_he_hash)
2778 continue;
2779 if (klen != chain->refcounted_he_keylen)
2780 continue;
2781 if (memNE(REF_HE_KEY(chain),key,klen))
2782 continue;
2783 if (!!is_utf8 != !!(chain->refcounted_he_data[0] & HVhek_UTF8))
2784 continue;
7b0bddfa 2785#else
cd1d2f8a
NC
2786 if (hash != HEK_HASH(chain->refcounted_he_hek))
2787 continue;
2788 if (klen != (STRLEN)HEK_LEN(chain->refcounted_he_hek))
2789 continue;
2790 if (memNE(HEK_KEY(chain->refcounted_he_hek),key,klen))
2791 continue;
2792 if (!!is_utf8 != !!HEK_UTF8(chain->refcounted_he_hek))
2793 continue;
7b0bddfa
NC
2794#endif
2795
cd1d2f8a
NC
2796 value = sv_2mortal(refcounted_he_value(chain));
2797 break;
2798 }
7b0bddfa
NC
2799 }
2800
2801 if (flags & HVhek_FREEKEY)
2802 Safefree(key);
2803
2804 return value;
2805}
2806
b3ca2e83
NC
2807/*
2808=for apidoc refcounted_he_new
2809
ec2a1de7
NC
2810Creates a new C<struct refcounted_he>. As S<key> is copied, and value is
2811stored in a compact form, all references remain the property of the caller.
2812The C<struct refcounted_he> is returned with a reference count of 1.
b3ca2e83
NC
2813
2814=cut
2815*/
2816
2817struct refcounted_he *
2818Perl_refcounted_he_new(pTHX_ struct refcounted_he *const parent,
2819 SV *const key, SV *const value) {
7a89be66 2820 dVAR;
b6bbf3fa
NC
2821 STRLEN key_len;
2822 const char *key_p = SvPV_const(key, key_len);
2823 STRLEN value_len = 0;
95b63a38 2824 const char *value_p = NULL;
b6bbf3fa
NC
2825 char value_type;
2826 char flags;
d8c5b3c5 2827 bool is_utf8 = SvUTF8(key) ? TRUE : FALSE;
b6bbf3fa
NC
2828
2829 if (SvPOK(value)) {
2830 value_type = HVrhek_PV;
2831 } else if (SvIOK(value)) {
ad64d0ec 2832 value_type = SvUOK((const SV *)value) ? HVrhek_UV : HVrhek_IV;
b6bbf3fa
NC
2833 } else if (value == &PL_sv_placeholder) {
2834 value_type = HVrhek_delete;
2835 } else if (!SvOK(value)) {
2836 value_type = HVrhek_undef;
2837 } else {
2838 value_type = HVrhek_PV;
2839 }
b3ca2e83 2840
b6bbf3fa 2841 if (value_type == HVrhek_PV) {
012da8e5
NC
2842 /* Do it this way so that the SvUTF8() test is after the SvPV, in case
2843 the value is overloaded, and doesn't yet have the UTF-8flag set. */
b6bbf3fa 2844 value_p = SvPV_const(value, value_len);
012da8e5
NC
2845 if (SvUTF8(value))
2846 value_type = HVrhek_PV_UTF8;
b6bbf3fa 2847 }
012da8e5
NC
2848 flags = value_type;
2849
2850 if (is_utf8) {
2851 /* Hash keys are always stored normalised to (yes) ISO-8859-1.
2852 As we're going to be building hash keys from this value in future,
2853 normalise it now. */
2854 key_p = (char*)bytes_from_utf8((const U8*)key_p, &key_len, &is_utf8);
2855 flags |= is_utf8 ? HVhek_UTF8 : HVhek_WASUTF8;
2856 }
2857
2858 return refcounted_he_new_common(parent, key_p, key_len, flags, value_type,
2859 ((value_type == HVrhek_PV
2860 || value_type == HVrhek_PV_UTF8) ?
2861 (void *)value_p : (void *)value),
2862 value_len);
2863}
2864
1da83c39 2865static struct refcounted_he *
012da8e5
NC
2866S_refcounted_he_new_common(pTHX_ struct refcounted_he *const parent,
2867 const char *const key_p, const STRLEN key_len,
2868 const char flags, char value_type,
2869 const void *value, const STRLEN value_len) {
2870 dVAR;
2871 struct refcounted_he *he;
2872 U32 hash;
2873 const bool is_pv = value_type == HVrhek_PV || value_type == HVrhek_PV_UTF8;
2874 STRLEN key_offset = is_pv ? value_len + 2 : 1;
2875
2876 PERL_ARGS_ASSERT_REFCOUNTED_HE_NEW_COMMON;
b6bbf3fa 2877
b6bbf3fa 2878#ifdef USE_ITHREADS
10edeb5d
JH
2879 he = (struct refcounted_he*)
2880 PerlMemShared_malloc(sizeof(struct refcounted_he) - 1
2881 + key_len
2882 + key_offset);
6cef672b 2883#else
10edeb5d
JH
2884 he = (struct refcounted_he*)
2885 PerlMemShared_malloc(sizeof(struct refcounted_he) - 1
2886 + key_offset);
6cef672b 2887#endif
b3ca2e83 2888
71ad1b0c 2889 he->refcounted_he_next = parent;
b6bbf3fa 2890
012da8e5
NC
2891 if (is_pv) {
2892 Copy((char *)value, he->refcounted_he_data + 1, value_len + 1, char);
b6bbf3fa 2893 he->refcounted_he_val.refcounted_he_u_len = value_len;
b6bbf3fa 2894 } else if (value_type == HVrhek_IV) {
ad64d0ec 2895 he->refcounted_he_val.refcounted_he_u_iv = SvIVX((const SV *)value);
012da8e5 2896 } else if (value_type == HVrhek_UV) {
ad64d0ec 2897 he->refcounted_he_val.refcounted_he_u_uv = SvUVX((const SV *)value);
b6bbf3fa
NC
2898 }
2899
b6bbf3fa
NC
2900 PERL_HASH(hash, key_p, key_len);
2901
cbb1fbea 2902#ifdef USE_ITHREADS
b6bbf3fa
NC
2903 he->refcounted_he_hash = hash;
2904 he->refcounted_he_keylen = key_len;
2905 Copy(key_p, he->refcounted_he_data + key_offset, key_len, char);
cbb1fbea 2906#else
b6bbf3fa 2907 he->refcounted_he_hek = share_hek_flags(key_p, key_len, hash, flags);
cbb1fbea 2908#endif
b6bbf3fa
NC
2909
2910 if (flags & HVhek_WASUTF8) {
2911 /* If it was downgraded from UTF-8, then the pointer returned from
2912 bytes_from_utf8 is an allocated pointer that we must free. */
2913 Safefree(key_p);
2914 }
2915
2916 he->refcounted_he_data[0] = flags;
b3ca2e83
NC
2917 he->refcounted_he_refcnt = 1;
2918
2919 return he;
2920}
2921
2922/*
2923=for apidoc refcounted_he_free
2924
2925Decrements the reference count of the passed in C<struct refcounted_he *>
2926by one. If the reference count reaches zero the structure's memory is freed,
2927and C<refcounted_he_free> iterates onto the parent node.
2928
2929=cut
2930*/
2931
2932void
2933Perl_refcounted_he_free(pTHX_ struct refcounted_he *he) {
53d44271 2934 dVAR;
57ca3b03
AL
2935 PERL_UNUSED_CONTEXT;
2936
b3ca2e83
NC
2937 while (he) {
2938 struct refcounted_he *copy;
cbb1fbea 2939 U32 new_count;
b3ca2e83 2940
cbb1fbea
NC
2941 HINTS_REFCNT_LOCK;
2942 new_count = --he->refcounted_he_refcnt;
2943 HINTS_REFCNT_UNLOCK;
2944
2945 if (new_count) {
b3ca2e83 2946 return;
cbb1fbea 2947 }
b3ca2e83 2948
b6bbf3fa 2949#ifndef USE_ITHREADS
71ad1b0c 2950 unshare_hek_or_pvn (he->refcounted_he_hek, 0, 0, 0);
cbb1fbea 2951#endif
b3ca2e83 2952 copy = he;
71ad1b0c 2953 he = he->refcounted_he_next;
b6bbf3fa 2954 PerlMemShared_free(copy);
b3ca2e83
NC
2955 }
2956}
2957
47550813
NC
2958/* pp_entereval is aware that labels are stored with a key ':' at the top of
2959 the linked list. */
dca6062a
NC
2960const char *
2961Perl_fetch_cop_label(pTHX_ struct refcounted_he *const chain, STRLEN *len,
2962 U32 *flags) {
2963 if (!chain)
2964 return NULL;
2965#ifdef USE_ITHREADS
2966 if (chain->refcounted_he_keylen != 1)
2967 return NULL;
2968 if (*REF_HE_KEY(chain) != ':')
2969 return NULL;
2970#else
2971 if ((STRLEN)HEK_LEN(chain->refcounted_he_hek) != 1)
2972 return NULL;
2973 if (*HEK_KEY(chain->refcounted_he_hek) != ':')
2974 return NULL;
2975#endif
012da8e5
NC
2976 /* Stop anyone trying to really mess us up by adding their own value for
2977 ':' into %^H */
2978 if ((chain->refcounted_he_data[0] & HVrhek_typemask) != HVrhek_PV
2979 && (chain->refcounted_he_data[0] & HVrhek_typemask) != HVrhek_PV_UTF8)
2980 return NULL;
2981
dca6062a
NC
2982 if (len)
2983 *len = chain->refcounted_he_val.refcounted_he_u_len;
2984 if (flags) {
2985 *flags = ((chain->refcounted_he_data[0] & HVrhek_typemask)
2986 == HVrhek_PV_UTF8) ? SVf_UTF8 : 0;
2987 }
2988 return chain->refcounted_he_data + 1;
2989}
2990
012da8e5
NC
2991/* As newSTATEOP currently gets passed plain char* labels, we will only provide
2992 that interface. Once it works out how to pass in length and UTF-8 ness, this
2993 function will need superseding. */
2994struct refcounted_he *
2995Perl_store_cop_label(pTHX_ struct refcounted_he *const chain, const char *label)
2996{
547bb267
NC
2997 PERL_ARGS_ASSERT_STORE_COP_LABEL;
2998
012da8e5
NC
2999 return refcounted_he_new_common(chain, ":", 1, HVrhek_PV, HVrhek_PV,
3000 label, strlen(label));
3001}
3002
b3ca2e83 3003/*
ecae49c0
NC
3004=for apidoc hv_assert
3005
3006Check that a hash is in an internally consistent state.
3007
3008=cut
3009*/
3010
943795c2
NC
3011#ifdef DEBUGGING
3012
ecae49c0
NC
3013void
3014Perl_hv_assert(pTHX_ HV *hv)
3015{
57ca3b03
AL
3016 dVAR;
3017 HE* entry;
3018 int withflags = 0;
3019 int placeholders = 0;
3020 int real = 0;
3021 int bad = 0;
3022 const I32 riter = HvRITER_get(hv);
3023 HE *eiter = HvEITER_get(hv);
3024
7918f24d
NC
3025 PERL_ARGS_ASSERT_HV_ASSERT;
3026
57ca3b03
AL
3027 (void)hv_iterinit(hv);
3028
3029 while ((entry = hv_iternext_flags(hv, HV_ITERNEXT_WANTPLACEHOLDERS))) {
3030 /* sanity check the values */
3031 if (HeVAL(entry) == &PL_sv_placeholder)
3032 placeholders++;
3033 else
3034 real++;
3035 /* sanity check the keys */
3036 if (HeSVKEY(entry)) {
6f207bd3 3037 NOOP; /* Don't know what to check on SV keys. */
57ca3b03
AL
3038 } else if (HeKUTF8(entry)) {
3039 withflags++;
3040 if (HeKWASUTF8(entry)) {
3041 PerlIO_printf(Perl_debug_log,
d2a455e7 3042 "hash key has both WASUTF8 and UTF8: '%.*s'\n",
57ca3b03
AL
3043 (int) HeKLEN(entry), HeKEY(entry));
3044 bad = 1;
3045 }
3046 } else if (HeKWASUTF8(entry))
3047 withflags++;
3048 }
ad64d0ec 3049 if (!SvTIED_mg((const SV *)hv, PERL_MAGIC_tied)) {
57ca3b03
AL
3050 static const char bad_count[] = "Count %d %s(s), but hash reports %d\n";
3051 const int nhashkeys = HvUSEDKEYS(hv);
3052 const int nhashplaceholders = HvPLACEHOLDERS_get(hv);
3053
3054 if (nhashkeys != real) {
3055 PerlIO_printf(Perl_debug_log, bad_count, real, "keys", nhashkeys );
3056 bad = 1;
3057 }
3058 if (nhashplaceholders != placeholders) {
3059 PerlIO_printf(Perl_debug_log, bad_count, placeholders, "placeholder", nhashplaceholders );
3060 bad = 1;
3061 }
3062 }
3063 if (withflags && ! HvHASKFLAGS(hv)) {
3064 PerlIO_printf(Perl_debug_log,
3065 "Hash has HASKFLAGS off but I count %d key(s) with flags\n",
3066 withflags);
3067 bad = 1;
3068 }
3069 if (bad) {
ad64d0ec 3070 sv_dump(MUTABLE_SV(hv));
57ca3b03
AL
3071 }
3072 HvRITER_set(hv, riter); /* Restore hash iterator state */
3073 HvEITER_set(hv, eiter);
ecae49c0 3074}
af3babe4 3075
943795c2
NC
3076#endif
3077
af3babe4
NC
3078/*
3079 * Local variables:
3080 * c-indentation-style: bsd
3081 * c-basic-offset: 4
3082 * indent-tabs-mode: t
3083 * End:
3084 *
37442d52
RGS
3085 * ex: set ts=8 sts=4 sw=4 noet:
3086 */