This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Assert that IVs and NVs can never be tainted.
[perl5.git] / hv.h
CommitLineData
a0d0e21e 1/* hv.h
79072805 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1996, 1997, 1998, 1999,
1d325971 4 * 2000, 2001, 2002, 2005, 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 *
79072805
LW
9 */
10
5cbe4eec 11/* entry in hash value chain */
79072805 12struct he {
6f51351d
NC
13 /* Keep hent_next first in this structure, because sv_free_arenas take
14 advantage of this to share code between the he arenas and the SV
15 body arenas */
5cbe4eec
MLF
16 HE *hent_next; /* next entry in chain */
17 HEK *hent_hek; /* hash key */
18 SV *hent_val; /* scalar value that was hashed */
bbce6d69 19};
20
5cbe4eec 21/* hash key -- defined separately for use as shared pointer */
ff68c719 22struct hek {
5cbe4eec
MLF
23 U32 hek_hash; /* hash of key */
24 I32 hek_len; /* length of hash key */
25 char hek_key[1]; /* variable-length hash key */
e05949c7 26 /* the hash-key is \0-terminated */
1e54db1a
JH
27 /* after the \0 there is a byte for flags, such as whether the key
28 is UTF-8 */
79072805
LW
29};
30
cbae3960
NC
31struct shared_he {
32 struct he shared_he_he;
33 struct hek shared_he_hek;
34};
bfcb3514
NC
35
36/* Subject to change.
37 Don't access this directly.
38*/
39struct xpvhv_aux {
7423f6db 40 HEK *xhv_name; /* name, if a symbol table */
86f55936 41 AV *xhv_backreferences; /* back references for weak references */
bfcb3514
NC
42 HE *xhv_eiter; /* current entry of iterator */
43 I32 xhv_riter; /* current root of iterator */
44};
45
5cbe4eec 46/* hash structure: */
6ee623d5 47/* This structure must match the beginning of struct xpvmg in sv.h. */
79072805 48struct xpvhv {
311a25d9 49 NV xnv_nv; /* numeric value, if any */
463ee0b2
LW
50 STRLEN xhv_fill; /* how full xhv_array currently is */
51 STRLEN xhv_max; /* subscript of last element of xhv_array */
e4305a63 52 union {
311a25d9
NC
53 IV xivu_iv; /* integer value or pv offset */
54 UV xivu_uv;
55 void * xivu_p1;
56 } xiv_u;
79072805
LW
57 MAGIC* xmg_magic; /* magic for scalar array */
58 HV* xmg_stash; /* class package */
79072805
LW
59};
60
311a25d9 61#define xhv_keys xiv_u.xivu_iv
e4305a63 62
c5f87f51 63#if 0
59813432 64typedef struct xpvhv xpvhv_allocated;
9f1501b2
NC
65#else
66typedef struct {
67 STRLEN xhv_fill; /* how full xhv_array currently is */
68 STRLEN xhv_max; /* subscript of last element of xhv_array */
69 union {
311a25d9
NC
70 IV xivu_iv; /* integer value or pv offset */
71 UV xivu_uv;
72 void * xivu_p1;
73 } xiv_u;
9f1501b2
NC
74 MAGIC* xmg_magic; /* magic for scalar array */
75 HV* xmg_stash; /* class package */
76} xpvhv_allocated;
77#endif
e4305a63 78
5cbe4eec 79/* hash a key */
5afd6d42
JH
80/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
81 * from requirements by Colin Plumb.
82 * (http://burtleburtle.net/bob/hash/doobs.html) */
83/* The use of a temporary pointer and the casting games
84 * is needed to serve the dual purposes of
85 * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
df805c3e 86 * a "char" can be either signed or unsigned, depending on the compiler)
5afd6d42 87 * (b) catering for old code that uses a "char"
830b38bd 88 *
504f80c1
JH
89 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
90 * to avoid "algorithmic complexity attacks".
830b38bd
JH
91 *
92 * If USE_HASH_SEED is defined, hash randomisation is done by default
93 * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
94 * only if the environment variable PERL_HASH_SEED is set.
95 * For maximal control, one can define PERL_HASH_SEED.
90ce55a1 96 * (see also perl.c:perl_parse()).
5afd6d42 97 */
830b38bd
JH
98#ifndef PERL_HASH_SEED
99# if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
100# define PERL_HASH_SEED PL_hash_seed
101# else
102# define PERL_HASH_SEED 0
103# endif
504f80c1 104#endif
bf6bd887 105#define PERL_HASH(hash,str,len) \
106 STMT_START { \
0bd48802 107 register const char * const s_PeRlHaSh_tmp = str; \
5afd6d42 108 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
bf6bd887 109 register I32 i_PeRlHaSh = len; \
504f80c1 110 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
a6fe520e
MLF
111 while (i_PeRlHaSh--) { \
112 hash_PeRlHaSh += *s_PeRlHaSh++; \
113 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
114 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
115 } \
116 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
117 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
4fee5c71 118 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
bf6bd887 119 } STMT_END
120
3d78eb94
NC
121/* Only hv.c and mod_perl should be doing this. */
122#ifdef PERL_HASH_INTERNAL_ACCESS
4b5190b5
NC
123#define PERL_HASH_INTERNAL(hash,str,len) \
124 STMT_START { \
0bd48802 125 register const char * const s_PeRlHaSh_tmp = str; \
4b5190b5
NC
126 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
127 register I32 i_PeRlHaSh = len; \
008fb0c0 128 register U32 hash_PeRlHaSh = PL_rehash_seed; \
4b5190b5
NC
129 while (i_PeRlHaSh--) { \
130 hash_PeRlHaSh += *s_PeRlHaSh++; \
131 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
132 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
133 } \
134 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
135 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
136 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
137 } STMT_END
138#endif
139
954c1994 140/*
ccfc67b7
JH
141=head1 Hash Manipulation Functions
142
954c1994
GS
143=for apidoc AmU||HEf_SVKEY
144This flag, used in the length slot of hash entries and magic structures,
d1be9408 145specifies the structure contains an C<SV*> pointer where a C<char*> pointer
954c1994
GS
146is to be expected. (For information only--not to be used).
147
ccfc67b7
JH
148=head1 Handy Values
149
954c1994
GS
150=for apidoc AmU||Nullhv
151Null HV pointer.
152
ccfc67b7
JH
153=head1 Hash Manipulation Functions
154
954c1994 155=for apidoc Am|char*|HvNAME|HV* stash
9282b5fd
SH
156Returns the package name of a stash, or NULL if C<stash> isn't a stash.
157See C<SvSTASH>, C<CvSTASH>.
954c1994
GS
158
159=for apidoc Am|void*|HeKEY|HE* he
160Returns the actual pointer stored in the key slot of the hash entry. The
161pointer may be either C<char*> or C<SV*>, depending on the value of
162C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
163usually preferable for finding the value of a key.
164
165=for apidoc Am|STRLEN|HeKLEN|HE* he
166If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
167holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
168be assigned to. The C<HePV()> macro is usually preferable for finding key
169lengths.
170
171=for apidoc Am|SV*|HeVAL|HE* he
172Returns the value slot (type C<SV*>) stored in the hash entry.
173
174=for apidoc Am|U32|HeHASH|HE* he
175Returns the computed hash stored in the hash entry.
176
177=for apidoc Am|char*|HePV|HE* he|STRLEN len
178Returns the key slot of the hash entry as a C<char*> value, doing any
179necessary dereferencing of possibly C<SV*> keys. The length of the string
180is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
181not care about what the length of the key is, you may use the global
182variable C<PL_na>, though this is rather less efficient than using a local
183variable. Remember though, that hash keys in perl are free to contain
184embedded nulls, so using C<strlen()> or similar is not a good way to find
185the length of hash keys. This is very similar to the C<SvPV()> macro
186described elsewhere in this document.
187
188=for apidoc Am|SV*|HeSVKEY|HE* he
189Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
190contain an C<SV*> key.
191
192=for apidoc Am|SV*|HeSVKEY_force|HE* he
193Returns the key as an C<SV*>. Will create and return a temporary mortal
194C<SV*> if the hash entry contains only a C<char*> key.
195
196=for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
197Sets the key to a given C<SV*>, taking care to set the appropriate flags to
198indicate the presence of an C<SV*> key, and returns the same
199C<SV*>.
200
201=cut
202*/
bf6bd887 203
bf5b86ae 204/* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
d1be9408 205#define HEf_SVKEY -2 /* hent_key is an SV* */
bf6bd887 206
207
79072805 208#define Nullhv Null(HV*)
43e6e717 209#define HvARRAY(hv) ((hv)->sv_u.svu_hash)
79072805 210#define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
463ee0b2 211#define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
b79f7545
NC
212/* This quite intentionally does no flag checking first. That's your
213 responsibility. */
214#define HvAUX(hv) ((struct xpvhv_aux*)&(HvARRAY(hv)[HvMAX(hv)+1]))
bfcb3514
NC
215#define HvRITER(hv) (*Perl_hv_riter_p(aTHX_ (HV*)(hv)))
216#define HvEITER(hv) (*Perl_hv_eiter_p(aTHX_ (HV*)(hv)))
217#define HvRITER_set(hv,r) Perl_hv_riter_set(aTHX_ (HV*)(hv), r)
218#define HvEITER_set(hv,e) Perl_hv_eiter_set(aTHX_ (HV*)(hv), e)
b79f7545
NC
219#define HvRITER_get(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_riter : -1)
220#define HvEITER_get(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_eiter : 0)
7423f6db 221#define HvNAME(hv) HvNAME_get(hv)
bfcb3514
NC
222/* FIXME - all of these should use a UTF8 aware API, which should also involve
223 getting the length. */
9b9d0b15 224/* This macro may go away without notice. */
b79f7545
NC
225#define HvNAME_HEK(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_name : 0)
226#define HvNAME_get(hv) ((SvOOK(hv) && (HvAUX(hv)->xhv_name)) \
227 ? HEK_KEY(HvAUX(hv)->xhv_name) : 0)
228#define HvNAMELEN_get(hv) ((SvOOK(hv) && (HvAUX(hv)->xhv_name)) \
229 ? HEK_LEN(HvAUX(hv)->xhv_name) : 0)
a0d0e21e 230
8aacddc1
NIS
231/* the number of keys (including any placeholers) */
232#define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
233
8aacddc1
NIS
234/*
235 * HvKEYS gets the number of keys that actually exist(), and is provided
236 * for backwards compatibility with old XS code. The core uses HvUSEDKEYS
237 * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders)
238 */
ca732855
NC
239#define HvKEYS(hv) HvUSEDKEYS(hv)
240#define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
8aacddc1 241#define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
ca732855
NC
242#define HvPLACEHOLDERS(hv) (*Perl_hv_placeholders_p(aTHX_ (HV*)hv))
243#define HvPLACEHOLDERS_get(hv) (SvMAGIC(hv) ? Perl_hv_placeholders_get(aTHX_ (HV*)hv) : 0)
244#define HvPLACEHOLDERS_set(hv,p) Perl_hv_placeholders_set(aTHX_ (HV*)hv, p)
8aacddc1 245
bf6bd887 246#define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS)
247#define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS)
248#define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS)
249
19692e8d
NC
250/* This is an optimisation flag. It won't be set if all hash keys have a 0
251 * flag. Currently the only flags relate to utf8.
252 * Hence it won't be set if all keys are 8 bit only. It will be set if any key
253 * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading
254 * when retrieved during iteration. It may still be set when there are no longer
255 * any utf8 keys.
fdcd69b6 256 * See HVhek_ENABLEHVKFLAGS for the trigger.
19692e8d
NC
257 */
258#define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS)
259#define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS)
260#define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS)
574c8022 261
bf5b86ae
GS
262#define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL)
263#define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL)
264#define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL)
265
4b5190b5
NC
266#define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH)
267#define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH)
268#define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH)
269
a0d0e21e
LW
270/* Maybe amagical: */
271/* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
272
273#define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM)
274#define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM)
275#define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM)
276
277/*
278#define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM)
279#define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM)
280#define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM)
281*/
bf6bd887 282
283#define Nullhe Null(HE*)
284#define HeNEXT(he) (he)->hent_next
ff68c719 285#define HeKEY_hek(he) (he)->hent_hek
286#define HeKEY(he) HEK_KEY(HeKEY_hek(he))
bbce6d69 287#define HeKEY_sv(he) (*(SV**)HeKEY(he))
ff68c719 288#define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
da58a35d 289#define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
19692e8d 290#define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
4b5190b5 291#define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
da58a35d 292#define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
19692e8d 293#define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
bf6bd887 294#define HeVAL(he) (he)->hent_val
ff68c719 295#define HeHASH(he) HEK_HASH(HeKEY_hek(he))
1e422769 296#define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
297 SvPV(HeKEY_sv(he),lp) : \
298 (((lp = HeKLEN(he)) >= 0) ? \
299 HeKEY(he) : Nullch))
300
bbce6d69 301#define HeSVKEY(he) ((HeKEY(he) && \
302 HeKLEN(he) == HEf_SVKEY) ? \
303 HeKEY_sv(he) : Nullsv)
304
305#define HeSVKEY_force(he) (HeKEY(he) ? \
306 ((HeKLEN(he) == HEf_SVKEY) ? \
307 HeKEY_sv(he) : \
79cb57f6 308 sv_2mortal(newSVpvn(HeKEY(he), \
bbce6d69 309 HeKLEN(he)))) : \
3280af22 310 &PL_sv_undef)
1e422769 311#define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
bbce6d69 312
ff68c719 313#define Nullhek Null(HEK*)
71be2cbc 314#define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
ff68c719 315#define HEK_HASH(hek) (hek)->hek_hash
316#define HEK_LEN(hek) (hek)->hek_len
317#define HEK_KEY(hek) (hek)->hek_key
19692e8d
NC
318#define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
319
320#define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
321#define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
4b5190b5 322#define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
19692e8d 323#define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
e16e2ff8
NC
324#define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
325 * (may change, but Storable is a core module) */
19692e8d
NC
326#define HVhek_MASK 0xFF
327
fdcd69b6
NC
328/* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as
329 HVhek_REHASH is only needed because the rehash flag has to be duplicated
330 into all keys as hv_iternext has no access to the hash flags. At this
331 point Storable's tests get upset, because sometimes hashes are "keyed"
332 and sometimes not, depending on the order of data insertion, and whether
333 it triggered rehashing. So currently HVhek_REHAS is exempt.
334*/
335
336#define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
337
19692e8d
NC
338#define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8)
339#define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8)
340#define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8)
341#define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8)
342#define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8)
343#define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8)
4b5190b5
NC
344#define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
345#define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
d18c6117 346
5cbe4eec 347/* calculate HV array allocation */
36768cf4
NC
348#ifndef PERL_USE_LARGE_HV_ALLOC
349/* Default to allocating the correct size - default to assuming that malloc()
350 is not broken and is efficient at allocating blocks sized at powers-of-two.
351*/
d18c6117
GS
352# define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
353#else
354# define MALLOC_OVERHEAD 16
355# define PERL_HV_ARRAY_ALLOC_BYTES(size) \
356 (((size) < 64) \
357 ? (size) * sizeof(HE*) \
358 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
359#endif
37d85e3a 360
e16e2ff8
NC
361/* Flags for hv_iternext_flags. */
362#define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
363
7a7b9979 364#define hv_iternext(hv) hv_iternext_flags(hv, 0)
bc5cdc23 365#define hv_magic(hv, gv, how) sv_magic((SV*)(hv), (SV*)(gv), how, Nullch, 0)
7a7b9979 366
37d85e3a
JH
367/* available as a function in hv.c */
368#define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash))
369#define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash)
bfcb3514 370
d1db91c6
NC
371#define share_hek_hek(hek) \
372 (++(((struct shared_he *)(((char *)hek) \
373 - STRUCT_OFFSET(struct shared_he, \
374 shared_he_hek))) \
375 ->shared_he_he.hent_val), \
376 hek)
377
bfcb3514
NC
378/*
379 * Local variables:
380 * c-indentation-style: bsd
381 * c-basic-offset: 4
382 * indent-tabs-mode: t
383 * End:
384 *
385 * ex: set ts=8 sts=4 sw=4 noet:
386 */