Assorted 2.15 fixes.
[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 {
102d3b8a 45 IV for_rent;
463ee0b2 46 STRLEN xhv_fill; /* how full xhv_array currently is */
47 STRLEN xhv_max; /* subscript of last element of xhv_array */
e4305a63 48 union {
49 NV xnvu_nv; /* numeric value, if any */
50 struct {
51 void *xnv_p1;
52 union {
53 void *xnv_p2;
54 IV xnv_i2; /* how many elements in the array */
55 } xnv_u2;
56 } xnv_s;
57 } xnv_u;
79072805 58 MAGIC* xmg_magic; /* magic for scalar array */
59 HV* xmg_stash; /* class package */
8d2f4536 60 /* list of pm's for this package is now stored in symtab magic. */
79072805 61};
62
e4305a63 63#define xhv_aux xnv_u.xnv_s.xnv_p1
64#define xhv_keys xnv_u.xnv_s.xnv_u2.xnv_i2
65
59813432 66typedef struct xpvhv xpvhv_allocated;
e4305a63 67
5cbe4eec 68/* hash a key */
5afd6d42 69/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
70 * from requirements by Colin Plumb.
71 * (http://burtleburtle.net/bob/hash/doobs.html) */
72/* The use of a temporary pointer and the casting games
73 * is needed to serve the dual purposes of
74 * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
75 * a "char" can be either signed or signed, depending on the compiler)
76 * (b) catering for old code that uses a "char"
830b38bd 77 *
504f80c1 78 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
79 * to avoid "algorithmic complexity attacks".
830b38bd 80 *
81 * If USE_HASH_SEED is defined, hash randomisation is done by default
82 * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
83 * only if the environment variable PERL_HASH_SEED is set.
84 * For maximal control, one can define PERL_HASH_SEED.
90ce55a1 85 * (see also perl.c:perl_parse()).
5afd6d42 86 */
830b38bd 87#ifndef PERL_HASH_SEED
88# if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
89# define PERL_HASH_SEED PL_hash_seed
90# else
91# define PERL_HASH_SEED 0
92# endif
504f80c1 93#endif
bf6bd887 94#define PERL_HASH(hash,str,len) \
95 STMT_START { \
5afd6d42 96 register const char *s_PeRlHaSh_tmp = str; \
97 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
bf6bd887 98 register I32 i_PeRlHaSh = len; \
504f80c1 99 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
a6fe520e 100 while (i_PeRlHaSh--) { \
101 hash_PeRlHaSh += *s_PeRlHaSh++; \
102 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
103 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
104 } \
105 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
106 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
4fee5c71 107 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
bf6bd887 108 } STMT_END
109
3d78eb94 110/* Only hv.c and mod_perl should be doing this. */
111#ifdef PERL_HASH_INTERNAL_ACCESS
4b5190b5 112#define PERL_HASH_INTERNAL(hash,str,len) \
113 STMT_START { \
114 register const char *s_PeRlHaSh_tmp = str; \
115 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
116 register I32 i_PeRlHaSh = len; \
008fb0c0 117 register U32 hash_PeRlHaSh = PL_rehash_seed; \
4b5190b5 118 while (i_PeRlHaSh--) { \
119 hash_PeRlHaSh += *s_PeRlHaSh++; \
120 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
121 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
122 } \
123 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
124 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
125 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
126 } STMT_END
127#endif
128
954c1994 129/*
ccfc67b7 130=head1 Hash Manipulation Functions
131
954c1994 132=for apidoc AmU||HEf_SVKEY
133This flag, used in the length slot of hash entries and magic structures,
d1be9408 134specifies the structure contains an C<SV*> pointer where a C<char*> pointer
954c1994 135is to be expected. (For information only--not to be used).
136
ccfc67b7 137=head1 Handy Values
138
954c1994 139=for apidoc AmU||Nullhv
140Null HV pointer.
141
ccfc67b7 142=head1 Hash Manipulation Functions
143
954c1994 144=for apidoc Am|char*|HvNAME|HV* stash
145Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
146
147=for apidoc Am|void*|HeKEY|HE* he
148Returns the actual pointer stored in the key slot of the hash entry. The
149pointer may be either C<char*> or C<SV*>, depending on the value of
150C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
151usually preferable for finding the value of a key.
152
153=for apidoc Am|STRLEN|HeKLEN|HE* he
154If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
155holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
156be assigned to. The C<HePV()> macro is usually preferable for finding key
157lengths.
158
159=for apidoc Am|SV*|HeVAL|HE* he
160Returns the value slot (type C<SV*>) stored in the hash entry.
161
162=for apidoc Am|U32|HeHASH|HE* he
163Returns the computed hash stored in the hash entry.
164
165=for apidoc Am|char*|HePV|HE* he|STRLEN len
166Returns the key slot of the hash entry as a C<char*> value, doing any
167necessary dereferencing of possibly C<SV*> keys. The length of the string
168is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
169not care about what the length of the key is, you may use the global
170variable C<PL_na>, though this is rather less efficient than using a local
171variable. Remember though, that hash keys in perl are free to contain
172embedded nulls, so using C<strlen()> or similar is not a good way to find
173the length of hash keys. This is very similar to the C<SvPV()> macro
174described elsewhere in this document.
175
176=for apidoc Am|SV*|HeSVKEY|HE* he
177Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
178contain an C<SV*> key.
179
180=for apidoc Am|SV*|HeSVKEY_force|HE* he
181Returns the key as an C<SV*>. Will create and return a temporary mortal
182C<SV*> if the hash entry contains only a C<char*> key.
183
184=for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
185Sets the key to a given C<SV*>, taking care to set the appropriate flags to
186indicate the presence of an C<SV*> key, and returns the same
187C<SV*>.
188
189=cut
190*/
bf6bd887 191
bf5b86ae 192/* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
d1be9408 193#define HEf_SVKEY -2 /* hent_key is an SV* */
bf6bd887 194
195
79072805 196#define Nullhv Null(HV*)
7b2c381c 197#define HvARRAY(hv) (*(HE***)&((hv)->sv_u.sv_array))
79072805 198#define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
463ee0b2 199#define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
bfcb3514 200#define HvRITER(hv) (*Perl_hv_riter_p(aTHX_ (HV*)(hv)))
201#define HvEITER(hv) (*Perl_hv_eiter_p(aTHX_ (HV*)(hv)))
202#define HvRITER_set(hv,r) Perl_hv_riter_set(aTHX_ (HV*)(hv), r)
203#define HvEITER_set(hv,e) Perl_hv_eiter_set(aTHX_ (HV*)(hv), e)
204#define HvRITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
205 ((struct xpvhv_aux*)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_riter : -1)
206#define HvEITER_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
207 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_eiter : 0)
208#define HvNAME(hv) (*Perl_hv_name_p(aTHX_ (HV*)hv))
209/* FIXME - all of these should use a UTF8 aware API, which should also involve
210 getting the length. */
211#define HvNAME_get(hv) (((XPVHV *)SvANY(hv))->xhv_aux ? \
212 ((struct xpvhv_aux *)((XPVHV *)SvANY(hv))->xhv_aux)->xhv_name : 0)
a0d0e21e 213
8aacddc1 214/* the number of keys (including any placeholers) */
215#define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
216
8aacddc1 217/*
218 * HvKEYS gets the number of keys that actually exist(), and is provided
219 * for backwards compatibility with old XS code. The core uses HvUSEDKEYS
220 * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders)
221 */
ca732855 222#define HvKEYS(hv) HvUSEDKEYS(hv)
223#define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
8aacddc1 224#define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
ca732855 225#define HvPLACEHOLDERS(hv) (*Perl_hv_placeholders_p(aTHX_ (HV*)hv))
226#define HvPLACEHOLDERS_get(hv) (SvMAGIC(hv) ? Perl_hv_placeholders_get(aTHX_ (HV*)hv) : 0)
227#define HvPLACEHOLDERS_set(hv,p) Perl_hv_placeholders_set(aTHX_ (HV*)hv, p)
8aacddc1 228
bf6bd887 229#define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS)
230#define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS)
231#define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS)
232
19692e8d 233/* This is an optimisation flag. It won't be set if all hash keys have a 0
234 * flag. Currently the only flags relate to utf8.
235 * Hence it won't be set if all keys are 8 bit only. It will be set if any key
236 * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading
237 * when retrieved during iteration. It may still be set when there are no longer
238 * any utf8 keys.
fdcd69b6 239 * See HVhek_ENABLEHVKFLAGS for the trigger.
19692e8d 240 */
241#define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS)
242#define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS)
243#define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS)
574c8022 244
bf5b86ae 245#define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL)
246#define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL)
247#define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL)
248
4b5190b5 249#define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH)
250#define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH)
251#define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH)
252
a0d0e21e 253/* Maybe amagical: */
254/* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
255
256#define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM)
257#define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM)
258#define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM)
259
260/*
261#define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM)
262#define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM)
263#define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM)
264*/
bf6bd887 265
266#define Nullhe Null(HE*)
267#define HeNEXT(he) (he)->hent_next
ff68c719 268#define HeKEY_hek(he) (he)->hent_hek
269#define HeKEY(he) HEK_KEY(HeKEY_hek(he))
bbce6d69 270#define HeKEY_sv(he) (*(SV**)HeKEY(he))
ff68c719 271#define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
da58a35d 272#define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
19692e8d 273#define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
4b5190b5 274#define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
da58a35d 275#define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
19692e8d 276#define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
bf6bd887 277#define HeVAL(he) (he)->hent_val
ff68c719 278#define HeHASH(he) HEK_HASH(HeKEY_hek(he))
1e422769 279#define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
280 SvPV(HeKEY_sv(he),lp) : \
281 (((lp = HeKLEN(he)) >= 0) ? \
282 HeKEY(he) : Nullch))
283
bbce6d69 284#define HeSVKEY(he) ((HeKEY(he) && \
285 HeKLEN(he) == HEf_SVKEY) ? \
286 HeKEY_sv(he) : Nullsv)
287
288#define HeSVKEY_force(he) (HeKEY(he) ? \
289 ((HeKLEN(he) == HEf_SVKEY) ? \
290 HeKEY_sv(he) : \
79cb57f6 291 sv_2mortal(newSVpvn(HeKEY(he), \
bbce6d69 292 HeKLEN(he)))) : \
3280af22 293 &PL_sv_undef)
1e422769 294#define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
bbce6d69 295
ff68c719 296#define Nullhek Null(HEK*)
71be2cbc 297#define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
ff68c719 298#define HEK_HASH(hek) (hek)->hek_hash
299#define HEK_LEN(hek) (hek)->hek_len
300#define HEK_KEY(hek) (hek)->hek_key
19692e8d 301#define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
302
303#define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
304#define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
4b5190b5 305#define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
19692e8d 306#define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
e16e2ff8 307#define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
308 * (may change, but Storable is a core module) */
19692e8d 309#define HVhek_MASK 0xFF
310
fdcd69b6 311/* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as
312 HVhek_REHASH is only needed because the rehash flag has to be duplicated
313 into all keys as hv_iternext has no access to the hash flags. At this
314 point Storable's tests get upset, because sometimes hashes are "keyed"
315 and sometimes not, depending on the order of data insertion, and whether
316 it triggered rehashing. So currently HVhek_REHAS is exempt.
317*/
318
319#define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
320
19692e8d 321#define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8)
322#define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8)
323#define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8)
324#define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8)
325#define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8)
326#define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8)
4b5190b5 327#define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
328#define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
d18c6117 329
5cbe4eec 330/* calculate HV array allocation */
36768cf4 331#ifndef PERL_USE_LARGE_HV_ALLOC
332/* Default to allocating the correct size - default to assuming that malloc()
333 is not broken and is efficient at allocating blocks sized at powers-of-two.
334*/
d18c6117 335# define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
336#else
337# define MALLOC_OVERHEAD 16
338# define PERL_HV_ARRAY_ALLOC_BYTES(size) \
339 (((size) < 64) \
340 ? (size) * sizeof(HE*) \
341 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
342#endif
37d85e3a 343
e16e2ff8 344/* Flags for hv_iternext_flags. */
345#define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
346
37d85e3a 347/* available as a function in hv.c */
348#define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash))
349#define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash)
bfcb3514 350
351/*
352 * Local variables:
353 * c-indentation-style: bsd
354 * c-basic-offset: 4
355 * indent-tabs-mode: t
356 * End:
357 *
358 * ex: set ts=8 sts=4 sw=4 noet:
359 */