3 * Copyright (C) 1991, 1992, 1993, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2005, by Larry Wall and others
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.
11 /* typedefs to eliminate some typing */
13 typedef struct hek HEK;
15 /* entry in hash value chain */
17 HE *hent_next; /* next entry in chain */
18 HEK *hent_hek; /* hash key */
19 SV *hent_val; /* scalar value that was hashed */
22 /* hash key -- defined separately for use as shared pointer */
24 U32 hek_hash; /* hash of key */
25 I32 hek_len; /* length of hash key */
26 char hek_key[1]; /* variable-length hash key */
27 /* the hash-key is \0-terminated */
28 /* after the \0 there is a byte for flags, such as whether the key
34 Don't access this directly.
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 */
43 /* This structure must match the beginning of struct xpvmg in sv.h. */
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 */
48 IV xhv_keys; /* how many elements in the array */
49 NV xnv_nv; /* numeric value, if any */
50 MAGIC* xmg_magic; /* magic for scalar array */
51 HV* xmg_stash; /* class package */
53 struct xpvhv_aux* xhv_aux;
54 /* list of pm's for this package is now stored in symtab magic. */
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"
67 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
68 * to avoid "algorithmic complexity attacks".
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.
74 * (see also perl.c:perl_parse()).
76 #ifndef PERL_HASH_SEED
77 # if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
78 # define PERL_HASH_SEED PL_hash_seed
80 # define PERL_HASH_SEED 0
83 #define PERL_HASH(hash,str,len) \
85 register const char *s_PeRlHaSh_tmp = str; \
86 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
87 register I32 i_PeRlHaSh = len; \
88 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
89 while (i_PeRlHaSh--) { \
90 hash_PeRlHaSh += *s_PeRlHaSh++; \
91 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
92 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
94 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
95 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
96 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
99 /* Only hv.c and mod_perl should be doing this. */
100 #ifdef PERL_HASH_INTERNAL_ACCESS
101 #define PERL_HASH_INTERNAL(hash,str,len) \
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; \
106 register U32 hash_PeRlHaSh = PL_rehash_seed; \
107 while (i_PeRlHaSh--) { \
108 hash_PeRlHaSh += *s_PeRlHaSh++; \
109 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
110 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
112 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
113 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
114 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
119 =head1 Hash Manipulation Functions
121 =for apidoc AmU||HEf_SVKEY
122 This flag, used in the length slot of hash entries and magic structures,
123 specifies the structure contains an C<SV*> pointer where a C<char*> pointer
124 is to be expected. (For information only--not to be used).
128 =for apidoc AmU||Nullhv
131 =head1 Hash Manipulation Functions
133 =for apidoc Am|char*|HvNAME|HV* stash
134 Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
136 =for apidoc Am|void*|HeKEY|HE* he
137 Returns the actual pointer stored in the key slot of the hash entry. The
138 pointer may be either C<char*> or C<SV*>, depending on the value of
139 C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
140 usually preferable for finding the value of a key.
142 =for apidoc Am|STRLEN|HeKLEN|HE* he
143 If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
144 holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
145 be assigned to. The C<HePV()> macro is usually preferable for finding key
148 =for apidoc Am|SV*|HeVAL|HE* he
149 Returns the value slot (type C<SV*>) stored in the hash entry.
151 =for apidoc Am|U32|HeHASH|HE* he
152 Returns the computed hash stored in the hash entry.
154 =for apidoc Am|char*|HePV|HE* he|STRLEN len
155 Returns the key slot of the hash entry as a C<char*> value, doing any
156 necessary dereferencing of possibly C<SV*> keys. The length of the string
157 is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
158 not care about what the length of the key is, you may use the global
159 variable C<PL_na>, though this is rather less efficient than using a local
160 variable. Remember though, that hash keys in perl are free to contain
161 embedded nulls, so using C<strlen()> or similar is not a good way to find
162 the length of hash keys. This is very similar to the C<SvPV()> macro
163 described elsewhere in this document.
165 =for apidoc Am|SV*|HeSVKEY|HE* he
166 Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
167 contain an C<SV*> key.
169 =for apidoc Am|SV*|HeSVKEY_force|HE* he
170 Returns the key as an C<SV*>. Will create and return a temporary mortal
171 C<SV*> if the hash entry contains only a C<char*> key.
173 =for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
174 Sets the key to a given C<SV*>, taking care to set the appropriate flags to
175 indicate the presence of an C<SV*> key, and returns the same
181 /* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
182 #define HEf_SVKEY -2 /* hent_key is an SV* */
185 #define Nullhv Null(HV*)
186 #define HvARRAY(hv) (*(HE***)&((XPVHV*) SvANY(hv))->xhv_array)
187 #define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
188 #define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
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)
203 /* the number of keys (including any placeholers) */
204 #define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
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)
211 #define HvKEYS(hv) HvUSEDKEYS(hv)
212 #define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
213 #define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
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)
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)
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
228 * See HVhek_ENABLEHVKFLAGS for the trigger.
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)
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)
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)
242 /* Maybe amagical: */
243 /* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
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)
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)
255 #define Nullhe Null(HE*)
256 #define HeNEXT(he) (he)->hent_next
257 #define HeKEY_hek(he) (he)->hent_hek
258 #define HeKEY(he) HEK_KEY(HeKEY_hek(he))
259 #define HeKEY_sv(he) (*(SV**)HeKEY(he))
260 #define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
261 #define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
262 #define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
263 #define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
264 #define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
265 #define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
266 #define HeVAL(he) (he)->hent_val
267 #define HeHASH(he) HEK_HASH(HeKEY_hek(he))
268 #define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
269 SvPV(HeKEY_sv(he),lp) : \
270 (((lp = HeKLEN(he)) >= 0) ? \
273 #define HeSVKEY(he) ((HeKEY(he) && \
274 HeKLEN(he) == HEf_SVKEY) ? \
275 HeKEY_sv(he) : Nullsv)
277 #define HeSVKEY_force(he) (HeKEY(he) ? \
278 ((HeKLEN(he) == HEf_SVKEY) ? \
280 sv_2mortal(newSVpvn(HeKEY(he), \
283 #define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
285 #define Nullhek Null(HEK*)
286 #define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
287 #define HEK_HASH(hek) (hek)->hek_hash
288 #define HEK_LEN(hek) (hek)->hek_len
289 #define HEK_KEY(hek) (hek)->hek_key
290 #define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
292 #define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
293 #define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
294 #define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
295 #define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
296 #define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
297 * (may change, but Storable is a core module) */
298 #define HVhek_MASK 0xFF
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.
308 #define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
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)
316 #define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
317 #define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
319 /* calculate HV array allocation */
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.
324 # define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
326 # define MALLOC_OVERHEAD 16
327 # define PERL_HV_ARRAY_ALLOC_BYTES(size) \
329 ? (size) * sizeof(HE*) \
330 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
333 /* Flags for hv_iternext_flags. */
334 #define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
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)
342 * c-indentation-style: bsd
344 * indent-tabs-mode: t
347 * ex: set ts=8 sts=4 sw=4 noet: