This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
ExtUtils::CBuilder Tru64 support
[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/* typedefs to eliminate some typing */
79072805 12typedef struct he HE;
ff68c719 13typedef struct hek HEK;
79072805 14
5cbe4eec 15/* entry in hash value chain */
79072805 16struct he {
5cbe4eec
MF
17 HE *hent_next; /* next entry in chain */
18 HEK *hent_hek; /* hash key */
19 SV *hent_val; /* scalar value that was hashed */
bbce6d69
PP
20};
21
5cbe4eec 22/* hash key -- defined separately for use as shared pointer */
ff68c719 23struct hek {
5cbe4eec
MF
24 U32 hek_hash; /* hash of key */
25 I32 hek_len; /* length of hash key */
26 char hek_key[1]; /* variable-length hash key */
e05949c7 27 /* the hash-key is \0-terminated */
1e54db1a
JH
28 /* after the \0 there is a byte for flags, such as whether the key
29 is UTF-8 */
79072805
LW
30};
31
bfcb3514
NC
32
33/* Subject to change.
34 Don't access this directly.
35*/
36struct xpvhv_aux {
37 char *xhv_name; /* name, if a symbol table */
38 HE *xhv_eiter; /* current entry of iterator */
39 I32 xhv_riter; /* current root of iterator */
40};
41
5cbe4eec 42/* hash structure: */
6ee623d5 43/* This structure must match the beginning of struct xpvmg in sv.h. */
79072805 44struct xpvhv {
463ee0b2
LW
45 STRLEN xhv_fill; /* how full xhv_array currently is */
46 STRLEN xhv_max; /* subscript of last element of xhv_array */
6ee623d5 47 IV xhv_keys; /* how many elements in the array */
65202027 48 NV xnv_nv; /* numeric value, if any */
79072805
LW
49 MAGIC* xmg_magic; /* magic for scalar array */
50 HV* xmg_stash; /* class package */
51
bfcb3514 52 struct xpvhv_aux* xhv_aux;
8d2f4536 53 /* list of pm's for this package is now stored in symtab magic. */
79072805
LW
54};
55
5cbe4eec 56/* hash a key */
5afd6d42
JH
57/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
58 * from requirements by Colin Plumb.
59 * (http://burtleburtle.net/bob/hash/doobs.html) */
60/* The use of a temporary pointer and the casting games
61 * is needed to serve the dual purposes of
62 * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
63 * a "char" can be either signed or signed, depending on the compiler)
64 * (b) catering for old code that uses a "char"
830b38bd 65 *
504f80c1
JH
66 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
67 * to avoid "algorithmic complexity attacks".
830b38bd
JH
68 *
69 * If USE_HASH_SEED is defined, hash randomisation is done by default
70 * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
71 * only if the environment variable PERL_HASH_SEED is set.
72 * For maximal control, one can define PERL_HASH_SEED.
90ce55a1 73 * (see also perl.c:perl_parse()).
5afd6d42 74 */
830b38bd
JH
75#ifndef PERL_HASH_SEED
76# if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
77# define PERL_HASH_SEED PL_hash_seed
78# else
79# define PERL_HASH_SEED 0
80# endif
504f80c1 81#endif
bf6bd887
PP
82#define PERL_HASH(hash,str,len) \
83 STMT_START { \
5afd6d42
JH
84 register const char *s_PeRlHaSh_tmp = str; \
85 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
bf6bd887 86 register I32 i_PeRlHaSh = len; \
504f80c1 87 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
a6fe520e
MF
88 while (i_PeRlHaSh--) { \
89 hash_PeRlHaSh += *s_PeRlHaSh++; \
90 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
91 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
92 } \
93 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
94 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
4fee5c71 95 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
bf6bd887
PP
96 } STMT_END
97
3d78eb94
NC
98/* Only hv.c and mod_perl should be doing this. */
99#ifdef PERL_HASH_INTERNAL_ACCESS
4b5190b5
NC
100#define PERL_HASH_INTERNAL(hash,str,len) \
101 STMT_START { \
102 register const char *s_PeRlHaSh_tmp = str; \
103 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
104 register I32 i_PeRlHaSh = len; \
008fb0c0 105 register U32 hash_PeRlHaSh = PL_rehash_seed; \
4b5190b5
NC
106 while (i_PeRlHaSh--) { \
107 hash_PeRlHaSh += *s_PeRlHaSh++; \
108 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
109 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
110 } \
111 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
112 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
113 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
114 } STMT_END
115#endif
116
954c1994 117/*
ccfc67b7
JH
118=head1 Hash Manipulation Functions
119
954c1994
GS
120=for apidoc AmU||HEf_SVKEY
121This flag, used in the length slot of hash entries and magic structures,
d1be9408 122specifies the structure contains an C<SV*> pointer where a C<char*> pointer
954c1994
GS
123is to be expected. (For information only--not to be used).
124
ccfc67b7
JH
125=head1 Handy Values
126
954c1994
GS
127=for apidoc AmU||Nullhv
128Null HV pointer.
129
ccfc67b7
JH
130=head1 Hash Manipulation Functions
131
954c1994
GS
132=for apidoc Am|char*|HvNAME|HV* stash
133Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
134
135=for apidoc Am|void*|HeKEY|HE* he
136Returns the actual pointer stored in the key slot of the hash entry. The
137pointer may be either C<char*> or C<SV*>, depending on the value of
138C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
139usually preferable for finding the value of a key.
140
141=for apidoc Am|STRLEN|HeKLEN|HE* he
142If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
143holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
144be assigned to. The C<HePV()> macro is usually preferable for finding key
145lengths.
146
147=for apidoc Am|SV*|HeVAL|HE* he
148Returns the value slot (type C<SV*>) stored in the hash entry.
149
150=for apidoc Am|U32|HeHASH|HE* he
151Returns the computed hash stored in the hash entry.
152
153=for apidoc Am|char*|HePV|HE* he|STRLEN len
154Returns the key slot of the hash entry as a C<char*> value, doing any
155necessary dereferencing of possibly C<SV*> keys. The length of the string
156is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
157not care about what the length of the key is, you may use the global
158variable C<PL_na>, though this is rather less efficient than using a local
159variable. Remember though, that hash keys in perl are free to contain
160embedded nulls, so using C<strlen()> or similar is not a good way to find
161the length of hash keys. This is very similar to the C<SvPV()> macro
162described elsewhere in this document.
163
164=for apidoc Am|SV*|HeSVKEY|HE* he
165Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
166contain an C<SV*> key.
167
168=for apidoc Am|SV*|HeSVKEY_force|HE* he
169Returns the key as an C<SV*>. Will create and return a temporary mortal
170C<SV*> if the hash entry contains only a C<char*> key.
171
172=for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
173Sets the key to a given C<SV*>, taking care to set the appropriate flags to
174indicate the presence of an C<SV*> key, and returns the same
175C<SV*>.
176
177=cut
178*/
bf6bd887 179
bf5b86ae 180/* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
d1be9408 181#define HEf_SVKEY -2 /* hent_key is an SV* */
bf6bd887
PP
182
183
79072805 184#define Nullhv Null(HV*)
7b2c381c 185#define HvARRAY(hv) (*(HE***)&((hv)->sv_u.sv_array))
79072805 186#define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
463ee0b2 187#define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
bfcb3514
NC
188#define HvRITER(hv) (*Perl_hv_riter_p(aTHX_ (HV*)(hv)))
189#define HvEITER(hv) (*Perl_hv_eiter_p(aTHX_ (HV*)(hv)))
190#define HvRITER_set(hv,r) Perl_hv_riter_set(aTHX_ (HV*)(hv), r)
191#define HvEITER_set(hv,e) Perl_hv_eiter_set(aTHX_ (HV*)(hv), e)
192#define HvRITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
193 ((struct xpvhv_aux*)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_riter : -1)
194#define HvEITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
195 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_eiter : 0)
196#define HvNAME(hv) (*Perl_hv_name_p(aTHX_ (HV*)hv))
197/* FIXME - all of these should use a UTF8 aware API, which should also involve
198 getting the length. */
199#define HvNAME_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
200 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_name : 0)
a0d0e21e 201
8aacddc1
NIS
202/* the number of keys (including any placeholers) */
203#define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
204
8aacddc1
NIS
205/*
206 * HvKEYS gets the number of keys that actually exist(), and is provided
207 * for backwards compatibility with old XS code. The core uses HvUSEDKEYS
208 * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders)
209 */
ca732855
NC
210#define HvKEYS(hv) HvUSEDKEYS(hv)
211#define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
8aacddc1 212#define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
ca732855
NC
213#define HvPLACEHOLDERS(hv) (*Perl_hv_placeholders_p(aTHX_ (HV*)hv))
214#define HvPLACEHOLDERS_get(hv) (SvMAGIC(hv) ? Perl_hv_placeholders_get(aTHX_ (HV*)hv) : 0)
215#define HvPLACEHOLDERS_set(hv,p) Perl_hv_placeholders_set(aTHX_ (HV*)hv, p)
8aacddc1 216
bf6bd887
PP
217#define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS)
218#define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS)
219#define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS)
220
19692e8d
NC
221/* This is an optimisation flag. It won't be set if all hash keys have a 0
222 * flag. Currently the only flags relate to utf8.
223 * Hence it won't be set if all keys are 8 bit only. It will be set if any key
224 * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading
225 * when retrieved during iteration. It may still be set when there are no longer
226 * any utf8 keys.
fdcd69b6 227 * See HVhek_ENABLEHVKFLAGS for the trigger.
19692e8d
NC
228 */
229#define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS)
230#define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS)
231#define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS)
574c8022 232
bf5b86ae
GS
233#define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL)
234#define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL)
235#define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL)
236
4b5190b5
NC
237#define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH)
238#define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH)
239#define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH)
240
a0d0e21e
LW
241/* Maybe amagical: */
242/* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
243
244#define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM)
245#define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM)
246#define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM)
247
248/*
249#define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM)
250#define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM)
251#define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM)
252*/
bf6bd887
PP
253
254#define Nullhe Null(HE*)
255#define HeNEXT(he) (he)->hent_next
ff68c719
PP
256#define HeKEY_hek(he) (he)->hent_hek
257#define HeKEY(he) HEK_KEY(HeKEY_hek(he))
bbce6d69 258#define HeKEY_sv(he) (*(SV**)HeKEY(he))
ff68c719 259#define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
da58a35d 260#define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
19692e8d 261#define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
4b5190b5 262#define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
da58a35d 263#define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
19692e8d 264#define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
bf6bd887 265#define HeVAL(he) (he)->hent_val
ff68c719 266#define HeHASH(he) HEK_HASH(HeKEY_hek(he))
1e422769
PP
267#define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
268 SvPV(HeKEY_sv(he),lp) : \
269 (((lp = HeKLEN(he)) >= 0) ? \
270 HeKEY(he) : Nullch))
271
bbce6d69
PP
272#define HeSVKEY(he) ((HeKEY(he) && \
273 HeKLEN(he) == HEf_SVKEY) ? \
274 HeKEY_sv(he) : Nullsv)
275
276#define HeSVKEY_force(he) (HeKEY(he) ? \
277 ((HeKLEN(he) == HEf_SVKEY) ? \
278 HeKEY_sv(he) : \
79cb57f6 279 sv_2mortal(newSVpvn(HeKEY(he), \
bbce6d69 280 HeKLEN(he)))) : \
3280af22 281 &PL_sv_undef)
1e422769 282#define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
bbce6d69 283
ff68c719 284#define Nullhek Null(HEK*)
71be2cbc 285#define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
ff68c719
PP
286#define HEK_HASH(hek) (hek)->hek_hash
287#define HEK_LEN(hek) (hek)->hek_len
288#define HEK_KEY(hek) (hek)->hek_key
19692e8d
NC
289#define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
290
291#define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
292#define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
4b5190b5 293#define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
19692e8d 294#define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
e16e2ff8
NC
295#define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
296 * (may change, but Storable is a core module) */
19692e8d
NC
297#define HVhek_MASK 0xFF
298
fdcd69b6
NC
299/* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as
300 HVhek_REHASH is only needed because the rehash flag has to be duplicated
301 into all keys as hv_iternext has no access to the hash flags. At this
302 point Storable's tests get upset, because sometimes hashes are "keyed"
303 and sometimes not, depending on the order of data insertion, and whether
304 it triggered rehashing. So currently HVhek_REHAS is exempt.
305*/
306
307#define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
308
19692e8d
NC
309#define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8)
310#define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8)
311#define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8)
312#define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8)
313#define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8)
314#define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8)
4b5190b5
NC
315#define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
316#define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
d18c6117 317
5cbe4eec 318/* calculate HV array allocation */
36768cf4
NC
319#ifndef PERL_USE_LARGE_HV_ALLOC
320/* Default to allocating the correct size - default to assuming that malloc()
321 is not broken and is efficient at allocating blocks sized at powers-of-two.
322*/
d18c6117
GS
323# define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
324#else
325# define MALLOC_OVERHEAD 16
326# define PERL_HV_ARRAY_ALLOC_BYTES(size) \
327 (((size) < 64) \
328 ? (size) * sizeof(HE*) \
329 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
330#endif
37d85e3a 331
e16e2ff8
NC
332/* Flags for hv_iternext_flags. */
333#define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
334
37d85e3a
JH
335/* available as a function in hv.c */
336#define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash))
337#define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash)
bfcb3514
NC
338
339/*
340 * Local variables:
341 * c-indentation-style: bsd
342 * c-basic-offset: 4
343 * indent-tabs-mode: t
344 * End:
345 *
346 * ex: set ts=8 sts=4 sw=4 noet:
347 */