Move hv_name, hv_eiter and hv_riter into a new aux structure.
[p5sagit/p5-mst-13.2.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 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 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 17 HE *hent_next; /* next entry in chain */
18 HEK *hent_hek; /* hash key */
19 SV *hent_val; /* scalar value that was hashed */
bbce6d69 20};
21
5cbe4eec 22/* hash key -- defined separately for use as shared pointer */
ff68c719 23struct hek {
5cbe4eec 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 28 /* after the \0 there is a byte for flags, such as whether the key
29 is UTF-8 */
79072805 30};
31
bfcb3514 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 45 char * xhv_array; /* pointer to malloced string */
46 STRLEN xhv_fill; /* how full xhv_array currently is */
47 STRLEN xhv_max; /* subscript of last element of xhv_array */
6ee623d5 48 IV xhv_keys; /* how many elements in the array */
65202027 49 NV xnv_nv; /* numeric value, if any */
79072805 50 MAGIC* xmg_magic; /* magic for scalar array */
51 HV* xmg_stash; /* class package */
52
bfcb3514 53 struct xpvhv_aux* xhv_aux;
8d2f4536 54 /* list of pm's for this package is now stored in symtab magic. */
79072805 55};
56
5cbe4eec 57/* hash a key */
5afd6d42 58/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
59 * from requirements by Colin Plumb.
60 * (http://burtleburtle.net/bob/hash/doobs.html) */
61/* The use of a temporary pointer and the casting games
62 * is needed to serve the dual purposes of
63 * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
64 * a "char" can be either signed or signed, depending on the compiler)
65 * (b) catering for old code that uses a "char"
830b38bd 66 *
504f80c1 67 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
68 * to avoid "algorithmic complexity attacks".
830b38bd 69 *
70 * If USE_HASH_SEED is defined, hash randomisation is done by default
71 * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
72 * only if the environment variable PERL_HASH_SEED is set.
73 * For maximal control, one can define PERL_HASH_SEED.
90ce55a1 74 * (see also perl.c:perl_parse()).
5afd6d42 75 */
830b38bd 76#ifndef PERL_HASH_SEED
77# if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
78# define PERL_HASH_SEED PL_hash_seed
79# else
80# define PERL_HASH_SEED 0
81# endif
504f80c1 82#endif
bf6bd887 83#define PERL_HASH(hash,str,len) \
84 STMT_START { \
5afd6d42 85 register const char *s_PeRlHaSh_tmp = str; \
86 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
bf6bd887 87 register I32 i_PeRlHaSh = len; \
504f80c1 88 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
a6fe520e 89 while (i_PeRlHaSh--) { \
90 hash_PeRlHaSh += *s_PeRlHaSh++; \
91 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
92 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
93 } \
94 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
95 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
4fee5c71 96 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
bf6bd887 97 } STMT_END
98
3d78eb94 99/* Only hv.c and mod_perl should be doing this. */
100#ifdef PERL_HASH_INTERNAL_ACCESS
4b5190b5 101#define PERL_HASH_INTERNAL(hash,str,len) \
102 STMT_START { \
103 register const char *s_PeRlHaSh_tmp = str; \
104 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
105 register I32 i_PeRlHaSh = len; \
008fb0c0 106 register U32 hash_PeRlHaSh = PL_rehash_seed; \
4b5190b5 107 while (i_PeRlHaSh--) { \
108 hash_PeRlHaSh += *s_PeRlHaSh++; \
109 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
110 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
111 } \
112 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
113 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
114 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
115 } STMT_END
116#endif
117
954c1994 118/*
ccfc67b7 119=head1 Hash Manipulation Functions
120
954c1994 121=for apidoc AmU||HEf_SVKEY
122This flag, used in the length slot of hash entries and magic structures,
d1be9408 123specifies the structure contains an C<SV*> pointer where a C<char*> pointer
954c1994 124is to be expected. (For information only--not to be used).
125
ccfc67b7 126=head1 Handy Values
127
954c1994 128=for apidoc AmU||Nullhv
129Null HV pointer.
130
ccfc67b7 131=head1 Hash Manipulation Functions
132
954c1994 133=for apidoc Am|char*|HvNAME|HV* stash
134Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
135
136=for apidoc Am|void*|HeKEY|HE* he
137Returns the actual pointer stored in the key slot of the hash entry. The
138pointer may be either C<char*> or C<SV*>, depending on the value of
139C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
140usually preferable for finding the value of a key.
141
142=for apidoc Am|STRLEN|HeKLEN|HE* he
143If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
144holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
145be assigned to. The C<HePV()> macro is usually preferable for finding key
146lengths.
147
148=for apidoc Am|SV*|HeVAL|HE* he
149Returns the value slot (type C<SV*>) stored in the hash entry.
150
151=for apidoc Am|U32|HeHASH|HE* he
152Returns the computed hash stored in the hash entry.
153
154=for apidoc Am|char*|HePV|HE* he|STRLEN len
155Returns the key slot of the hash entry as a C<char*> value, doing any
156necessary dereferencing of possibly C<SV*> keys. The length of the string
157is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
158not care about what the length of the key is, you may use the global
159variable C<PL_na>, though this is rather less efficient than using a local
160variable. Remember though, that hash keys in perl are free to contain
161embedded nulls, so using C<strlen()> or similar is not a good way to find
162the length of hash keys. This is very similar to the C<SvPV()> macro
163described elsewhere in this document.
164
165=for apidoc Am|SV*|HeSVKEY|HE* he
166Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
167contain an C<SV*> key.
168
169=for apidoc Am|SV*|HeSVKEY_force|HE* he
170Returns the key as an C<SV*>. Will create and return a temporary mortal
171C<SV*> if the hash entry contains only a C<char*> key.
172
173=for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
174Sets the key to a given C<SV*>, taking care to set the appropriate flags to
175indicate the presence of an C<SV*> key, and returns the same
176C<SV*>.
177
178=cut
179*/
bf6bd887 180
bf5b86ae 181/* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
d1be9408 182#define HEf_SVKEY -2 /* hent_key is an SV* */
bf6bd887 183
184
79072805 185#define Nullhv Null(HV*)
4efcf9a2 186#define HvARRAY(hv) (*(HE***)&((XPVHV*) SvANY(hv))->xhv_array)
79072805 187#define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
463ee0b2 188#define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
bfcb3514 189#define HvRITER(hv) (*Perl_hv_riter_p(aTHX_ (HV*)(hv)))
190#define HvEITER(hv) (*Perl_hv_eiter_p(aTHX_ (HV*)(hv)))
191#define HvRITER_set(hv,r) Perl_hv_riter_set(aTHX_ (HV*)(hv), r)
192#define HvEITER_set(hv,e) Perl_hv_eiter_set(aTHX_ (HV*)(hv), e)
193#define HvRITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
194 ((struct xpvhv_aux*)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_riter : -1)
195#define HvEITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
196 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_eiter : 0)
197#define HvNAME(hv) (*Perl_hv_name_p(aTHX_ (HV*)hv))
198/* FIXME - all of these should use a UTF8 aware API, which should also involve
199 getting the length. */
200#define HvNAME_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
201 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_name : 0)
a0d0e21e 202
8aacddc1 203/* the number of keys (including any placeholers) */
204#define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
205
8aacddc1 206/*
207 * HvKEYS gets the number of keys that actually exist(), and is provided
208 * for backwards compatibility with old XS code. The core uses HvUSEDKEYS
209 * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders)
210 */
ca732855 211#define HvKEYS(hv) HvUSEDKEYS(hv)
212#define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
8aacddc1 213#define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
ca732855 214#define HvPLACEHOLDERS(hv) (*Perl_hv_placeholders_p(aTHX_ (HV*)hv))
215#define HvPLACEHOLDERS_get(hv) (SvMAGIC(hv) ? Perl_hv_placeholders_get(aTHX_ (HV*)hv) : 0)
216#define HvPLACEHOLDERS_set(hv,p) Perl_hv_placeholders_set(aTHX_ (HV*)hv, p)
8aacddc1 217
bf6bd887 218#define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS)
219#define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS)
220#define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS)
221
19692e8d 222/* This is an optimisation flag. It won't be set if all hash keys have a 0
223 * flag. Currently the only flags relate to utf8.
224 * Hence it won't be set if all keys are 8 bit only. It will be set if any key
225 * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading
226 * when retrieved during iteration. It may still be set when there are no longer
227 * any utf8 keys.
fdcd69b6 228 * See HVhek_ENABLEHVKFLAGS for the trigger.
19692e8d 229 */
230#define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS)
231#define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS)
232#define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS)
574c8022 233
bf5b86ae 234#define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL)
235#define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL)
236#define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL)
237
4b5190b5 238#define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH)
239#define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH)
240#define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH)
241
a0d0e21e 242/* Maybe amagical: */
243/* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
244
245#define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM)
246#define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM)
247#define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM)
248
249/*
250#define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM)
251#define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM)
252#define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM)
253*/
bf6bd887 254
255#define Nullhe Null(HE*)
256#define HeNEXT(he) (he)->hent_next
ff68c719 257#define HeKEY_hek(he) (he)->hent_hek
258#define HeKEY(he) HEK_KEY(HeKEY_hek(he))
bbce6d69 259#define HeKEY_sv(he) (*(SV**)HeKEY(he))
ff68c719 260#define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
da58a35d 261#define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
19692e8d 262#define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
4b5190b5 263#define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
da58a35d 264#define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
19692e8d 265#define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
bf6bd887 266#define HeVAL(he) (he)->hent_val
ff68c719 267#define HeHASH(he) HEK_HASH(HeKEY_hek(he))
1e422769 268#define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
269 SvPV(HeKEY_sv(he),lp) : \
270 (((lp = HeKLEN(he)) >= 0) ? \
271 HeKEY(he) : Nullch))
272
bbce6d69 273#define HeSVKEY(he) ((HeKEY(he) && \
274 HeKLEN(he) == HEf_SVKEY) ? \
275 HeKEY_sv(he) : Nullsv)
276
277#define HeSVKEY_force(he) (HeKEY(he) ? \
278 ((HeKLEN(he) == HEf_SVKEY) ? \
279 HeKEY_sv(he) : \
79cb57f6 280 sv_2mortal(newSVpvn(HeKEY(he), \
bbce6d69 281 HeKLEN(he)))) : \
3280af22 282 &PL_sv_undef)
1e422769 283#define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
bbce6d69 284
ff68c719 285#define Nullhek Null(HEK*)
71be2cbc 286#define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
ff68c719 287#define HEK_HASH(hek) (hek)->hek_hash
288#define HEK_LEN(hek) (hek)->hek_len
289#define HEK_KEY(hek) (hek)->hek_key
19692e8d 290#define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
291
292#define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
293#define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
4b5190b5 294#define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
19692e8d 295#define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
e16e2ff8 296#define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
297 * (may change, but Storable is a core module) */
19692e8d 298#define HVhek_MASK 0xFF
299
fdcd69b6 300/* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as
301 HVhek_REHASH is only needed because the rehash flag has to be duplicated
302 into all keys as hv_iternext has no access to the hash flags. At this
303 point Storable's tests get upset, because sometimes hashes are "keyed"
304 and sometimes not, depending on the order of data insertion, and whether
305 it triggered rehashing. So currently HVhek_REHAS is exempt.
306*/
307
308#define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
309
19692e8d 310#define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8)
311#define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8)
312#define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8)
313#define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8)
314#define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8)
315#define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8)
4b5190b5 316#define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
317#define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
d18c6117 318
5cbe4eec 319/* calculate HV array allocation */
36768cf4 320#ifndef PERL_USE_LARGE_HV_ALLOC
321/* Default to allocating the correct size - default to assuming that malloc()
322 is not broken and is efficient at allocating blocks sized at powers-of-two.
323*/
d18c6117 324# define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
325#else
326# define MALLOC_OVERHEAD 16
327# define PERL_HV_ARRAY_ALLOC_BYTES(size) \
328 (((size) < 64) \
329 ? (size) * sizeof(HE*) \
330 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
331#endif
37d85e3a 332
e16e2ff8 333/* Flags for hv_iternext_flags. */
334#define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
335
37d85e3a 336/* available as a function in hv.c */
337#define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash))
338#define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash)
bfcb3514 339
340/*
341 * Local variables:
342 * c-indentation-style: bsd
343 * c-basic-offset: 4
344 * indent-tabs-mode: t
345 * End:
346 *
347 * ex: set ts=8 sts=4 sw=4 noet:
348 */