Commit | Line | Data |
a0d0e21e |
1 | /* hv.h |
79072805 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1996, 1997, 1998, 1999, |
4 | * 2000, 2001, 2002, 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 |
12 | typedef struct he HE; |
ff68c719 |
13 | typedef struct hek HEK; |
79072805 |
14 | |
5cbe4eec |
15 | /* entry in hash value chain */ |
79072805 |
16 | struct 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 |
23 | struct 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 | |
5cbe4eec |
32 | /* hash structure: */ |
6ee623d5 |
33 | /* This structure must match the beginning of struct xpvmg in sv.h. */ |
79072805 |
34 | struct xpvhv { |
463ee0b2 |
35 | char * xhv_array; /* pointer to malloced string */ |
36 | STRLEN xhv_fill; /* how full xhv_array currently is */ |
37 | STRLEN xhv_max; /* subscript of last element of xhv_array */ |
6ee623d5 |
38 | IV xhv_keys; /* how many elements in the array */ |
65202027 |
39 | NV xnv_nv; /* numeric value, if any */ |
8aacddc1 |
40 | #define xhv_placeholders xnv_nv |
79072805 |
41 | MAGIC* xmg_magic; /* magic for scalar array */ |
42 | HV* xmg_stash; /* class package */ |
43 | |
79072805 |
44 | I32 xhv_riter; /* current root of iterator */ |
45 | HE *xhv_eiter; /* current entry of iterator */ |
46 | PMOP *xhv_pmroot; /* list of pm's for this package */ |
47 | char *xhv_name; /* name, if a symbol table */ |
79072805 |
48 | }; |
49 | |
5cbe4eec |
50 | /* hash a key */ |
5afd6d42 |
51 | /* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins |
52 | * from requirements by Colin Plumb. |
53 | * (http://burtleburtle.net/bob/hash/doobs.html) */ |
54 | /* The use of a temporary pointer and the casting games |
55 | * is needed to serve the dual purposes of |
56 | * (a) the hashed data being interpreted as "unsigned char" (new since 5.8, |
57 | * a "char" can be either signed or signed, depending on the compiler) |
58 | * (b) catering for old code that uses a "char" |
830b38bd |
59 | * |
504f80c1 |
60 | * The "hash seed" feature was added in Perl 5.8.1 to perturb the results |
61 | * to avoid "algorithmic complexity attacks". |
830b38bd |
62 | * |
63 | * If USE_HASH_SEED is defined, hash randomisation is done by default |
64 | * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done |
65 | * only if the environment variable PERL_HASH_SEED is set. |
66 | * For maximal control, one can define PERL_HASH_SEED. |
67 | * (see also erl.c:perl_parse()). |
5afd6d42 |
68 | */ |
830b38bd |
69 | #ifndef PERL_HASH_SEED |
70 | # if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT) |
71 | # define PERL_HASH_SEED PL_hash_seed |
72 | # else |
73 | # define PERL_HASH_SEED 0 |
74 | # endif |
504f80c1 |
75 | #endif |
bf6bd887 |
76 | #define PERL_HASH(hash,str,len) \ |
77 | STMT_START { \ |
5afd6d42 |
78 | register const char *s_PeRlHaSh_tmp = str; \ |
79 | register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \ |
bf6bd887 |
80 | register I32 i_PeRlHaSh = len; \ |
504f80c1 |
81 | register U32 hash_PeRlHaSh = PERL_HASH_SEED; \ |
a6fe520e |
82 | while (i_PeRlHaSh--) { \ |
83 | hash_PeRlHaSh += *s_PeRlHaSh++; \ |
84 | hash_PeRlHaSh += (hash_PeRlHaSh << 10); \ |
85 | hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \ |
86 | } \ |
87 | hash_PeRlHaSh += (hash_PeRlHaSh << 3); \ |
88 | hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \ |
4fee5c71 |
89 | (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \ |
bf6bd887 |
90 | } STMT_END |
91 | |
3d78eb94 |
92 | /* Only hv.c and mod_perl should be doing this. */ |
93 | #ifdef PERL_HASH_INTERNAL_ACCESS |
4b5190b5 |
94 | #define PERL_HASH_INTERNAL(hash,str,len) \ |
95 | STMT_START { \ |
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; \ |
008fb0c0 |
99 | register U32 hash_PeRlHaSh = PL_rehash_seed; \ |
4b5190b5 |
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); \ |
107 | (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \ |
108 | } STMT_END |
109 | #endif |
110 | |
954c1994 |
111 | /* |
ccfc67b7 |
112 | =head1 Hash Manipulation Functions |
113 | |
954c1994 |
114 | =for apidoc AmU||HEf_SVKEY |
115 | This flag, used in the length slot of hash entries and magic structures, |
d1be9408 |
116 | specifies the structure contains an C<SV*> pointer where a C<char*> pointer |
954c1994 |
117 | is to be expected. (For information only--not to be used). |
118 | |
ccfc67b7 |
119 | =head1 Handy Values |
120 | |
954c1994 |
121 | =for apidoc AmU||Nullhv |
122 | Null HV pointer. |
123 | |
ccfc67b7 |
124 | =head1 Hash Manipulation Functions |
125 | |
954c1994 |
126 | =for apidoc Am|char*|HvNAME|HV* stash |
127 | Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>. |
128 | |
129 | =for apidoc Am|void*|HeKEY|HE* he |
130 | Returns the actual pointer stored in the key slot of the hash entry. The |
131 | pointer may be either C<char*> or C<SV*>, depending on the value of |
132 | C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are |
133 | usually preferable for finding the value of a key. |
134 | |
135 | =for apidoc Am|STRLEN|HeKLEN|HE* he |
136 | If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry |
137 | holds an C<SV*> key. Otherwise, holds the actual length of the key. Can |
138 | be assigned to. The C<HePV()> macro is usually preferable for finding key |
139 | lengths. |
140 | |
141 | =for apidoc Am|SV*|HeVAL|HE* he |
142 | Returns the value slot (type C<SV*>) stored in the hash entry. |
143 | |
144 | =for apidoc Am|U32|HeHASH|HE* he |
145 | Returns the computed hash stored in the hash entry. |
146 | |
147 | =for apidoc Am|char*|HePV|HE* he|STRLEN len |
148 | Returns the key slot of the hash entry as a C<char*> value, doing any |
149 | necessary dereferencing of possibly C<SV*> keys. The length of the string |
150 | is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do |
151 | not care about what the length of the key is, you may use the global |
152 | variable C<PL_na>, though this is rather less efficient than using a local |
153 | variable. Remember though, that hash keys in perl are free to contain |
154 | embedded nulls, so using C<strlen()> or similar is not a good way to find |
155 | the length of hash keys. This is very similar to the C<SvPV()> macro |
156 | described elsewhere in this document. |
157 | |
158 | =for apidoc Am|SV*|HeSVKEY|HE* he |
159 | Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not |
160 | contain an C<SV*> key. |
161 | |
162 | =for apidoc Am|SV*|HeSVKEY_force|HE* he |
163 | Returns the key as an C<SV*>. Will create and return a temporary mortal |
164 | C<SV*> if the hash entry contains only a C<char*> key. |
165 | |
166 | =for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv |
167 | Sets the key to a given C<SV*>, taking care to set the appropriate flags to |
168 | indicate the presence of an C<SV*> key, and returns the same |
169 | C<SV*>. |
170 | |
171 | =cut |
172 | */ |
bf6bd887 |
173 | |
bf5b86ae |
174 | /* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */ |
d1be9408 |
175 | #define HEf_SVKEY -2 /* hent_key is an SV* */ |
bf6bd887 |
176 | |
177 | |
79072805 |
178 | #define Nullhv Null(HV*) |
4efcf9a2 |
179 | #define HvARRAY(hv) (*(HE***)&((XPVHV*) SvANY(hv))->xhv_array) |
79072805 |
180 | #define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill |
463ee0b2 |
181 | #define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max |
79072805 |
182 | #define HvRITER(hv) ((XPVHV*) SvANY(hv))->xhv_riter |
183 | #define HvEITER(hv) ((XPVHV*) SvANY(hv))->xhv_eiter |
184 | #define HvPMROOT(hv) ((XPVHV*) SvANY(hv))->xhv_pmroot |
185 | #define HvNAME(hv) ((XPVHV*) SvANY(hv))->xhv_name |
a0d0e21e |
186 | |
8aacddc1 |
187 | /* the number of keys (including any placeholers) */ |
188 | #define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys) |
189 | |
190 | /* The number of placeholders in the enumerated-keys hash */ |
205c8ad3 |
191 | #define XHvPLACEHOLDERS(xhv) ((xhv)->xhv_placeholders) |
8aacddc1 |
192 | |
dd2710a1 |
193 | /* the number of keys that exist() (i.e. excluding placeholders) */ |
194 | #define XHvUSEDKEYS(xhv) (XHvTOTALKEYS(xhv) - (IV)XHvPLACEHOLDERS(xhv)) |
8aacddc1 |
195 | |
196 | /* |
197 | * HvKEYS gets the number of keys that actually exist(), and is provided |
198 | * for backwards compatibility with old XS code. The core uses HvUSEDKEYS |
199 | * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders) |
200 | */ |
201 | #define HvKEYS(hv) XHvUSEDKEYS((XPVHV*) SvANY(hv)) |
202 | #define HvUSEDKEYS(hv) XHvUSEDKEYS((XPVHV*) SvANY(hv)) |
203 | #define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv)) |
204 | #define HvPLACEHOLDERS(hv) XHvPLACEHOLDERS((XPVHV*) SvANY(hv)) |
205 | |
bf6bd887 |
206 | #define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS) |
207 | #define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS) |
208 | #define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS) |
209 | |
19692e8d |
210 | /* This is an optimisation flag. It won't be set if all hash keys have a 0 |
211 | * flag. Currently the only flags relate to utf8. |
212 | * Hence it won't be set if all keys are 8 bit only. It will be set if any key |
213 | * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading |
214 | * when retrieved during iteration. It may still be set when there are no longer |
215 | * any utf8 keys. |
fdcd69b6 |
216 | * See HVhek_ENABLEHVKFLAGS for the trigger. |
19692e8d |
217 | */ |
218 | #define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS) |
219 | #define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS) |
220 | #define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS) |
574c8022 |
221 | |
bf5b86ae |
222 | #define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL) |
223 | #define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL) |
224 | #define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL) |
225 | |
4b5190b5 |
226 | #define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH) |
227 | #define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH) |
228 | #define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH) |
229 | |
a0d0e21e |
230 | /* Maybe amagical: */ |
231 | /* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */ |
232 | |
233 | #define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM) |
234 | #define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM) |
235 | #define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM) |
236 | |
237 | /* |
238 | #define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM) |
239 | #define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM) |
240 | #define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM) |
241 | */ |
bf6bd887 |
242 | |
243 | #define Nullhe Null(HE*) |
244 | #define HeNEXT(he) (he)->hent_next |
ff68c719 |
245 | #define HeKEY_hek(he) (he)->hent_hek |
246 | #define HeKEY(he) HEK_KEY(HeKEY_hek(he)) |
bbce6d69 |
247 | #define HeKEY_sv(he) (*(SV**)HeKEY(he)) |
ff68c719 |
248 | #define HeKLEN(he) HEK_LEN(HeKEY_hek(he)) |
da58a35d |
249 | #define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he)) |
19692e8d |
250 | #define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he)) |
4b5190b5 |
251 | #define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he)) |
da58a35d |
252 | #define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he)) |
19692e8d |
253 | #define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he)) |
bf6bd887 |
254 | #define HeVAL(he) (he)->hent_val |
ff68c719 |
255 | #define HeHASH(he) HEK_HASH(HeKEY_hek(he)) |
1e422769 |
256 | #define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \ |
257 | SvPV(HeKEY_sv(he),lp) : \ |
258 | (((lp = HeKLEN(he)) >= 0) ? \ |
259 | HeKEY(he) : Nullch)) |
260 | |
bbce6d69 |
261 | #define HeSVKEY(he) ((HeKEY(he) && \ |
262 | HeKLEN(he) == HEf_SVKEY) ? \ |
263 | HeKEY_sv(he) : Nullsv) |
264 | |
265 | #define HeSVKEY_force(he) (HeKEY(he) ? \ |
266 | ((HeKLEN(he) == HEf_SVKEY) ? \ |
267 | HeKEY_sv(he) : \ |
79cb57f6 |
268 | sv_2mortal(newSVpvn(HeKEY(he), \ |
bbce6d69 |
269 | HeKLEN(he)))) : \ |
3280af22 |
270 | &PL_sv_undef) |
1e422769 |
271 | #define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv)) |
bbce6d69 |
272 | |
ff68c719 |
273 | #define Nullhek Null(HEK*) |
71be2cbc |
274 | #define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0]) |
ff68c719 |
275 | #define HEK_HASH(hek) (hek)->hek_hash |
276 | #define HEK_LEN(hek) (hek)->hek_len |
277 | #define HEK_KEY(hek) (hek)->hek_key |
19692e8d |
278 | #define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1)) |
279 | |
280 | #define HVhek_UTF8 0x01 /* Key is utf8 encoded. */ |
281 | #define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */ |
4b5190b5 |
282 | #define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */ |
19692e8d |
283 | #define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */ |
e16e2ff8 |
284 | #define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder. |
285 | * (may change, but Storable is a core module) */ |
19692e8d |
286 | #define HVhek_MASK 0xFF |
287 | |
fdcd69b6 |
288 | /* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as |
289 | HVhek_REHASH is only needed because the rehash flag has to be duplicated |
290 | into all keys as hv_iternext has no access to the hash flags. At this |
291 | point Storable's tests get upset, because sometimes hashes are "keyed" |
292 | and sometimes not, depending on the order of data insertion, and whether |
293 | it triggered rehashing. So currently HVhek_REHAS is exempt. |
294 | */ |
295 | |
296 | #define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH) |
297 | |
19692e8d |
298 | #define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8) |
299 | #define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8) |
300 | #define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8) |
301 | #define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8) |
302 | #define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8) |
303 | #define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8) |
4b5190b5 |
304 | #define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH) |
305 | #define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH) |
d18c6117 |
306 | |
5cbe4eec |
307 | /* calculate HV array allocation */ |
d18c6117 |
308 | #if defined(STRANGE_MALLOC) || defined(MYMALLOC) |
309 | # define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*)) |
310 | #else |
311 | # define MALLOC_OVERHEAD 16 |
312 | # define PERL_HV_ARRAY_ALLOC_BYTES(size) \ |
313 | (((size) < 64) \ |
314 | ? (size) * sizeof(HE*) \ |
315 | : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD) |
316 | #endif |
37d85e3a |
317 | |
e16e2ff8 |
318 | /* Flags for hv_iternext_flags. */ |
319 | #define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */ |
320 | |
37d85e3a |
321 | /* available as a function in hv.c */ |
322 | #define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash)) |
323 | #define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash) |