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. */
46 STRLEN xhv_fill; /* how full xhv_array currently is */
47 STRLEN xhv_max; /* subscript of last element of xhv_array */
49 NV xnvu_nv; /* numeric value, if any */
54 IV xnv_i2; /* how many elements in the array */
58 MAGIC* xmg_magic; /* magic for scalar array */
59 HV* xmg_stash; /* class package */
60 /* list of pm's for this package is now stored in symtab magic. */
63 #define xhv_aux xnv_u.xnv_s.xnv_p1
64 #define xhv_keys xnv_u.xnv_s.xnv_u2.xnv_i2
66 typedef struct xpvhv xpvhv_allocated;
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"
78 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
79 * to avoid "algorithmic complexity attacks".
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.
85 * (see also perl.c:perl_parse()).
87 #ifndef PERL_HASH_SEED
88 # if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
89 # define PERL_HASH_SEED PL_hash_seed
91 # define PERL_HASH_SEED 0
94 #define PERL_HASH(hash,str,len) \
96 register const char *s_PeRlHaSh_tmp = str; \
97 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
98 register I32 i_PeRlHaSh = len; \
99 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
100 while (i_PeRlHaSh--) { \
101 hash_PeRlHaSh += *s_PeRlHaSh++; \
102 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
103 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
105 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
106 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
107 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
110 /* Only hv.c and mod_perl should be doing this. */
111 #ifdef PERL_HASH_INTERNAL_ACCESS
112 #define PERL_HASH_INTERNAL(hash,str,len) \
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; \
117 register U32 hash_PeRlHaSh = PL_rehash_seed; \
118 while (i_PeRlHaSh--) { \
119 hash_PeRlHaSh += *s_PeRlHaSh++; \
120 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
121 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
123 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
124 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
125 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
130 =head1 Hash Manipulation Functions
132 =for apidoc AmU||HEf_SVKEY
133 This flag, used in the length slot of hash entries and magic structures,
134 specifies the structure contains an C<SV*> pointer where a C<char*> pointer
135 is to be expected. (For information only--not to be used).
139 =for apidoc AmU||Nullhv
142 =head1 Hash Manipulation Functions
144 =for apidoc Am|char*|HvNAME|HV* stash
145 Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
147 =for apidoc Am|void*|HeKEY|HE* he
148 Returns the actual pointer stored in the key slot of the hash entry. The
149 pointer may be either C<char*> or C<SV*>, depending on the value of
150 C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
151 usually preferable for finding the value of a key.
153 =for apidoc Am|STRLEN|HeKLEN|HE* he
154 If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
155 holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
156 be assigned to. The C<HePV()> macro is usually preferable for finding key
159 =for apidoc Am|SV*|HeVAL|HE* he
160 Returns the value slot (type C<SV*>) stored in the hash entry.
162 =for apidoc Am|U32|HeHASH|HE* he
163 Returns the computed hash stored in the hash entry.
165 =for apidoc Am|char*|HePV|HE* he|STRLEN len
166 Returns the key slot of the hash entry as a C<char*> value, doing any
167 necessary dereferencing of possibly C<SV*> keys. The length of the string
168 is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
169 not care about what the length of the key is, you may use the global
170 variable C<PL_na>, though this is rather less efficient than using a local
171 variable. Remember though, that hash keys in perl are free to contain
172 embedded nulls, so using C<strlen()> or similar is not a good way to find
173 the length of hash keys. This is very similar to the C<SvPV()> macro
174 described elsewhere in this document.
176 =for apidoc Am|SV*|HeSVKEY|HE* he
177 Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
178 contain an C<SV*> key.
180 =for apidoc Am|SV*|HeSVKEY_force|HE* he
181 Returns the key as an C<SV*>. Will create and return a temporary mortal
182 C<SV*> if the hash entry contains only a C<char*> key.
184 =for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
185 Sets the key to a given C<SV*>, taking care to set the appropriate flags to
186 indicate the presence of an C<SV*> key, and returns the same
192 /* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
193 #define HEf_SVKEY -2 /* hent_key is an SV* */
196 #define Nullhv Null(HV*)
197 #define HvARRAY(hv) (*(HE***)&((hv)->sv_u.sv_array))
198 #define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
199 #define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
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)
214 /* the number of keys (including any placeholers) */
215 #define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
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)
222 #define HvKEYS(hv) HvUSEDKEYS(hv)
223 #define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
224 #define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
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)
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)
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
239 * See HVhek_ENABLEHVKFLAGS for the trigger.
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)
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)
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)
253 /* Maybe amagical: */
254 /* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
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)
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)
266 #define Nullhe Null(HE*)
267 #define HeNEXT(he) (he)->hent_next
268 #define HeKEY_hek(he) (he)->hent_hek
269 #define HeKEY(he) HEK_KEY(HeKEY_hek(he))
270 #define HeKEY_sv(he) (*(SV**)HeKEY(he))
271 #define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
272 #define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
273 #define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
274 #define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
275 #define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
276 #define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
277 #define HeVAL(he) (he)->hent_val
278 #define HeHASH(he) HEK_HASH(HeKEY_hek(he))
279 #define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
280 SvPV(HeKEY_sv(he),lp) : \
281 (((lp = HeKLEN(he)) >= 0) ? \
284 #define HeSVKEY(he) ((HeKEY(he) && \
285 HeKLEN(he) == HEf_SVKEY) ? \
286 HeKEY_sv(he) : Nullsv)
288 #define HeSVKEY_force(he) (HeKEY(he) ? \
289 ((HeKLEN(he) == HEf_SVKEY) ? \
291 sv_2mortal(newSVpvn(HeKEY(he), \
294 #define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
296 #define Nullhek Null(HEK*)
297 #define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
298 #define HEK_HASH(hek) (hek)->hek_hash
299 #define HEK_LEN(hek) (hek)->hek_len
300 #define HEK_KEY(hek) (hek)->hek_key
301 #define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
303 #define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
304 #define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
305 #define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
306 #define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
307 #define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
308 * (may change, but Storable is a core module) */
309 #define HVhek_MASK 0xFF
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.
319 #define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
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)
327 #define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
328 #define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
330 /* calculate HV array allocation */
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.
335 # define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
337 # define MALLOC_OVERHEAD 16
338 # define PERL_HV_ARRAY_ALLOC_BYTES(size) \
340 ? (size) * sizeof(HE*) \
341 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
344 /* Flags for hv_iternext_flags. */
345 #define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
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)
353 * c-indentation-style: bsd
355 * indent-tabs-mode: t
358 * ex: set ts=8 sts=4 sw=4 noet: