regcomp.c and more
[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 {
7423f6db 37 HEK *xhv_name; /* name, if a symbol table */
bfcb3514 38 HE *xhv_eiter; /* current entry of iterator */
39 I32 xhv_riter; /* current root of iterator */
40};
41
b79f7545 42#define HV_AUX_SIZE STRUCT_OFFSET(struct xpvhv_aux, xhv_array)
43
5cbe4eec 44/* hash structure: */
6ee623d5 45/* This structure must match the beginning of struct xpvmg in sv.h. */
79072805 46struct xpvhv {
311a25d9 47 NV xnv_nv; /* numeric value, if any */
463ee0b2 48 STRLEN xhv_fill; /* how full xhv_array currently is */
49 STRLEN xhv_max; /* subscript of last element of xhv_array */
e4305a63 50 union {
311a25d9 51 IV xivu_iv; /* integer value or pv offset */
52 UV xivu_uv;
53 void * xivu_p1;
54 } xiv_u;
79072805 55 MAGIC* xmg_magic; /* magic for scalar array */
56 HV* xmg_stash; /* class package */
79072805 57};
58
311a25d9 59#define xhv_keys xiv_u.xivu_iv
e4305a63 60
c5f87f51 61#if 0
59813432 62typedef struct xpvhv xpvhv_allocated;
9f1501b2 63#else
64typedef struct {
65 STRLEN xhv_fill; /* how full xhv_array currently is */
66 STRLEN xhv_max; /* subscript of last element of xhv_array */
67 union {
311a25d9 68 IV xivu_iv; /* integer value or pv offset */
69 UV xivu_uv;
70 void * xivu_p1;
71 } xiv_u;
9f1501b2 72 MAGIC* xmg_magic; /* magic for scalar array */
73 HV* xmg_stash; /* class package */
74} xpvhv_allocated;
75#endif
e4305a63 76
5cbe4eec 77/* hash a key */
5afd6d42 78/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
79 * from requirements by Colin Plumb.
80 * (http://burtleburtle.net/bob/hash/doobs.html) */
81/* The use of a temporary pointer and the casting games
82 * is needed to serve the dual purposes of
83 * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
84 * a "char" can be either signed or signed, depending on the compiler)
85 * (b) catering for old code that uses a "char"
830b38bd 86 *
504f80c1 87 * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
88 * to avoid "algorithmic complexity attacks".
830b38bd 89 *
90 * If USE_HASH_SEED is defined, hash randomisation is done by default
91 * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
92 * only if the environment variable PERL_HASH_SEED is set.
93 * For maximal control, one can define PERL_HASH_SEED.
90ce55a1 94 * (see also perl.c:perl_parse()).
5afd6d42 95 */
830b38bd 96#ifndef PERL_HASH_SEED
97# if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
98# define PERL_HASH_SEED PL_hash_seed
99# else
100# define PERL_HASH_SEED 0
101# endif
504f80c1 102#endif
bf6bd887 103#define PERL_HASH(hash,str,len) \
104 STMT_START { \
5afd6d42 105 register const char *s_PeRlHaSh_tmp = str; \
106 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
bf6bd887 107 register I32 i_PeRlHaSh = len; \
504f80c1 108 register U32 hash_PeRlHaSh = PERL_HASH_SEED; \
a6fe520e 109 while (i_PeRlHaSh--) { \
110 hash_PeRlHaSh += *s_PeRlHaSh++; \
111 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
112 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
113 } \
114 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
115 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
4fee5c71 116 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
bf6bd887 117 } STMT_END
118
3d78eb94 119/* Only hv.c and mod_perl should be doing this. */
120#ifdef PERL_HASH_INTERNAL_ACCESS
4b5190b5 121#define PERL_HASH_INTERNAL(hash,str,len) \
122 STMT_START { \
123 register const char *s_PeRlHaSh_tmp = str; \
124 register const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
125 register I32 i_PeRlHaSh = len; \
008fb0c0 126 register U32 hash_PeRlHaSh = PL_rehash_seed; \
4b5190b5 127 while (i_PeRlHaSh--) { \
128 hash_PeRlHaSh += *s_PeRlHaSh++; \
129 hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
130 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
131 } \
132 hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
133 hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
134 (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
135 } STMT_END
136#endif
137
954c1994 138/*
ccfc67b7 139=head1 Hash Manipulation Functions
140
954c1994 141=for apidoc AmU||HEf_SVKEY
142This flag, used in the length slot of hash entries and magic structures,
d1be9408 143specifies the structure contains an C<SV*> pointer where a C<char*> pointer
954c1994 144is to be expected. (For information only--not to be used).
145
ccfc67b7 146=head1 Handy Values
147
954c1994 148=for apidoc AmU||Nullhv
149Null HV pointer.
150
ccfc67b7 151=head1 Hash Manipulation Functions
152
954c1994 153=for apidoc Am|char*|HvNAME|HV* stash
154Returns the package name of a stash. See C<SvSTASH>, C<CvSTASH>.
155
156=for apidoc Am|void*|HeKEY|HE* he
157Returns the actual pointer stored in the key slot of the hash entry. The
158pointer may be either C<char*> or C<SV*>, depending on the value of
159C<HeKLEN()>. Can be assigned to. The C<HePV()> or C<HeSVKEY()> macros are
160usually preferable for finding the value of a key.
161
162=for apidoc Am|STRLEN|HeKLEN|HE* he
163If this is negative, and amounts to C<HEf_SVKEY>, it indicates the entry
164holds an C<SV*> key. Otherwise, holds the actual length of the key. Can
165be assigned to. The C<HePV()> macro is usually preferable for finding key
166lengths.
167
168=for apidoc Am|SV*|HeVAL|HE* he
169Returns the value slot (type C<SV*>) stored in the hash entry.
170
171=for apidoc Am|U32|HeHASH|HE* he
172Returns the computed hash stored in the hash entry.
173
174=for apidoc Am|char*|HePV|HE* he|STRLEN len
175Returns the key slot of the hash entry as a C<char*> value, doing any
176necessary dereferencing of possibly C<SV*> keys. The length of the string
177is placed in C<len> (this is a macro, so do I<not> use C<&len>). If you do
178not care about what the length of the key is, you may use the global
179variable C<PL_na>, though this is rather less efficient than using a local
180variable. Remember though, that hash keys in perl are free to contain
181embedded nulls, so using C<strlen()> or similar is not a good way to find
182the length of hash keys. This is very similar to the C<SvPV()> macro
183described elsewhere in this document.
184
185=for apidoc Am|SV*|HeSVKEY|HE* he
186Returns the key as an C<SV*>, or C<Nullsv> if the hash entry does not
187contain an C<SV*> key.
188
189=for apidoc Am|SV*|HeSVKEY_force|HE* he
190Returns the key as an C<SV*>. Will create and return a temporary mortal
191C<SV*> if the hash entry contains only a C<char*> key.
192
193=for apidoc Am|SV*|HeSVKEY_set|HE* he|SV* sv
194Sets the key to a given C<SV*>, taking care to set the appropriate flags to
195indicate the presence of an C<SV*> key, and returns the same
196C<SV*>.
197
198=cut
199*/
bf6bd887 200
bf5b86ae 201/* these hash entry flags ride on hent_klen (for use only in magic/tied HVs) */
d1be9408 202#define HEf_SVKEY -2 /* hent_key is an SV* */
bf6bd887 203
204
79072805 205#define Nullhv Null(HV*)
339049b0 206#define HvARRAY(hv) (*(HE***)&((hv)->sv_u.svu_array))
79072805 207#define HvFILL(hv) ((XPVHV*) SvANY(hv))->xhv_fill
463ee0b2 208#define HvMAX(hv) ((XPVHV*) SvANY(hv))->xhv_max
b79f7545 209/* This quite intentionally does no flag checking first. That's your
210 responsibility. */
211#define HvAUX(hv) ((struct xpvhv_aux*)&(HvARRAY(hv)[HvMAX(hv)+1]))
bfcb3514 212#define HvRITER(hv) (*Perl_hv_riter_p(aTHX_ (HV*)(hv)))
213#define HvEITER(hv) (*Perl_hv_eiter_p(aTHX_ (HV*)(hv)))
214#define HvRITER_set(hv,r) Perl_hv_riter_set(aTHX_ (HV*)(hv), r)
215#define HvEITER_set(hv,e) Perl_hv_eiter_set(aTHX_ (HV*)(hv), e)
b79f7545 216#define HvRITER_get(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_riter : -1)
217#define HvEITER_get(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_eiter : 0)
7423f6db 218#define HvNAME(hv) HvNAME_get(hv)
bfcb3514 219/* FIXME - all of these should use a UTF8 aware API, which should also involve
220 getting the length. */
9b9d0b15 221/* This macro may go away without notice. */
b79f7545 222#define HvNAME_HEK(hv) (SvOOK(hv) ? HvAUX(hv)->xhv_name : 0)
223#define HvNAME_get(hv) ((SvOOK(hv) && (HvAUX(hv)->xhv_name)) \
224 ? HEK_KEY(HvAUX(hv)->xhv_name) : 0)
225#define HvNAMELEN_get(hv) ((SvOOK(hv) && (HvAUX(hv)->xhv_name)) \
226 ? HEK_LEN(HvAUX(hv)->xhv_name) : 0)
a0d0e21e 227
8aacddc1 228/* the number of keys (including any placeholers) */
229#define XHvTOTALKEYS(xhv) ((xhv)->xhv_keys)
230
8aacddc1 231/*
232 * HvKEYS gets the number of keys that actually exist(), and is provided
233 * for backwards compatibility with old XS code. The core uses HvUSEDKEYS
234 * (keys, excluding placeholdes) and HvTOTALKEYS (including placeholders)
235 */
ca732855 236#define HvKEYS(hv) HvUSEDKEYS(hv)
237#define HvUSEDKEYS(hv) (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(hv))
8aacddc1 238#define HvTOTALKEYS(hv) XHvTOTALKEYS((XPVHV*) SvANY(hv))
ca732855 239#define HvPLACEHOLDERS(hv) (*Perl_hv_placeholders_p(aTHX_ (HV*)hv))
240#define HvPLACEHOLDERS_get(hv) (SvMAGIC(hv) ? Perl_hv_placeholders_get(aTHX_ (HV*)hv) : 0)
241#define HvPLACEHOLDERS_set(hv,p) Perl_hv_placeholders_set(aTHX_ (HV*)hv, p)
8aacddc1 242
bf6bd887 243#define HvSHAREKEYS(hv) (SvFLAGS(hv) & SVphv_SHAREKEYS)
244#define HvSHAREKEYS_on(hv) (SvFLAGS(hv) |= SVphv_SHAREKEYS)
245#define HvSHAREKEYS_off(hv) (SvFLAGS(hv) &= ~SVphv_SHAREKEYS)
246
19692e8d 247/* This is an optimisation flag. It won't be set if all hash keys have a 0
248 * flag. Currently the only flags relate to utf8.
249 * Hence it won't be set if all keys are 8 bit only. It will be set if any key
250 * is utf8 (including 8 bit keys that were entered as utf8, and need upgrading
251 * when retrieved during iteration. It may still be set when there are no longer
252 * any utf8 keys.
fdcd69b6 253 * See HVhek_ENABLEHVKFLAGS for the trigger.
19692e8d 254 */
255#define HvHASKFLAGS(hv) (SvFLAGS(hv) & SVphv_HASKFLAGS)
256#define HvHASKFLAGS_on(hv) (SvFLAGS(hv) |= SVphv_HASKFLAGS)
257#define HvHASKFLAGS_off(hv) (SvFLAGS(hv) &= ~SVphv_HASKFLAGS)
574c8022 258
bf5b86ae 259#define HvLAZYDEL(hv) (SvFLAGS(hv) & SVphv_LAZYDEL)
260#define HvLAZYDEL_on(hv) (SvFLAGS(hv) |= SVphv_LAZYDEL)
261#define HvLAZYDEL_off(hv) (SvFLAGS(hv) &= ~SVphv_LAZYDEL)
262
4b5190b5 263#define HvREHASH(hv) (SvFLAGS(hv) & SVphv_REHASH)
264#define HvREHASH_on(hv) (SvFLAGS(hv) |= SVphv_REHASH)
265#define HvREHASH_off(hv) (SvFLAGS(hv) &= ~SVphv_REHASH)
266
a0d0e21e 267/* Maybe amagical: */
268/* #define HV_AMAGICmb(hv) (SvFLAGS(hv) & (SVpgv_badAM | SVpgv_AM)) */
269
270#define HV_AMAGIC(hv) (SvFLAGS(hv) & SVpgv_AM)
271#define HV_AMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_AM)
272#define HV_AMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_AM)
273
274/*
275#define HV_AMAGICbad(hv) (SvFLAGS(hv) & SVpgv_badAM)
276#define HV_badAMAGIC_on(hv) (SvFLAGS(hv) |= SVpgv_badAM)
277#define HV_badAMAGIC_off(hv) (SvFLAGS(hv) &= ~SVpgv_badAM)
278*/
bf6bd887 279
280#define Nullhe Null(HE*)
281#define HeNEXT(he) (he)->hent_next
ff68c719 282#define HeKEY_hek(he) (he)->hent_hek
283#define HeKEY(he) HEK_KEY(HeKEY_hek(he))
bbce6d69 284#define HeKEY_sv(he) (*(SV**)HeKEY(he))
ff68c719 285#define HeKLEN(he) HEK_LEN(HeKEY_hek(he))
da58a35d 286#define HeKUTF8(he) HEK_UTF8(HeKEY_hek(he))
19692e8d 287#define HeKWASUTF8(he) HEK_WASUTF8(HeKEY_hek(he))
4b5190b5 288#define HeKREHASH(he) HEK_REHASH(HeKEY_hek(he))
da58a35d 289#define HeKLEN_UTF8(he) (HeKUTF8(he) ? -HeKLEN(he) : HeKLEN(he))
19692e8d 290#define HeKFLAGS(he) HEK_FLAGS(HeKEY_hek(he))
bf6bd887 291#define HeVAL(he) (he)->hent_val
ff68c719 292#define HeHASH(he) HEK_HASH(HeKEY_hek(he))
1e422769 293#define HePV(he,lp) ((HeKLEN(he) == HEf_SVKEY) ? \
294 SvPV(HeKEY_sv(he),lp) : \
295 (((lp = HeKLEN(he)) >= 0) ? \
296 HeKEY(he) : Nullch))
297
bbce6d69 298#define HeSVKEY(he) ((HeKEY(he) && \
299 HeKLEN(he) == HEf_SVKEY) ? \
300 HeKEY_sv(he) : Nullsv)
301
302#define HeSVKEY_force(he) (HeKEY(he) ? \
303 ((HeKLEN(he) == HEf_SVKEY) ? \
304 HeKEY_sv(he) : \
79cb57f6 305 sv_2mortal(newSVpvn(HeKEY(he), \
bbce6d69 306 HeKLEN(he)))) : \
3280af22 307 &PL_sv_undef)
1e422769 308#define HeSVKEY_set(he,sv) ((HeKLEN(he) = HEf_SVKEY), (HeKEY_sv(he) = sv))
bbce6d69 309
ff68c719 310#define Nullhek Null(HEK*)
71be2cbc 311#define HEK_BASESIZE STRUCT_OFFSET(HEK, hek_key[0])
ff68c719 312#define HEK_HASH(hek) (hek)->hek_hash
313#define HEK_LEN(hek) (hek)->hek_len
314#define HEK_KEY(hek) (hek)->hek_key
19692e8d 315#define HEK_FLAGS(hek) (*((unsigned char *)(HEK_KEY(hek))+HEK_LEN(hek)+1))
316
317#define HVhek_UTF8 0x01 /* Key is utf8 encoded. */
318#define HVhek_WASUTF8 0x02 /* Key is bytes here, but was supplied as utf8. */
4b5190b5 319#define HVhek_REHASH 0x04 /* This key is in an hv using a custom HASH . */
19692e8d 320#define HVhek_FREEKEY 0x100 /* Internal flag to say key is malloc()ed. */
e16e2ff8 321#define HVhek_PLACEHOLD 0x200 /* Internal flag to create placeholder.
322 * (may change, but Storable is a core module) */
19692e8d 323#define HVhek_MASK 0xFF
324
fdcd69b6 325/* Which flags enable HvHASKFLAGS? Somewhat a hack on a hack, as
326 HVhek_REHASH is only needed because the rehash flag has to be duplicated
327 into all keys as hv_iternext has no access to the hash flags. At this
328 point Storable's tests get upset, because sometimes hashes are "keyed"
329 and sometimes not, depending on the order of data insertion, and whether
330 it triggered rehashing. So currently HVhek_REHAS is exempt.
331*/
332
333#define HVhek_ENABLEHVKFLAGS (HVhek_MASK - HVhek_REHASH)
334
19692e8d 335#define HEK_UTF8(hek) (HEK_FLAGS(hek) & HVhek_UTF8)
336#define HEK_UTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_UTF8)
337#define HEK_UTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_UTF8)
338#define HEK_WASUTF8(hek) (HEK_FLAGS(hek) & HVhek_WASUTF8)
339#define HEK_WASUTF8_on(hek) (HEK_FLAGS(hek) |= HVhek_WASUTF8)
340#define HEK_WASUTF8_off(hek) (HEK_FLAGS(hek) &= ~HVhek_WASUTF8)
4b5190b5 341#define HEK_REHASH(hek) (HEK_FLAGS(hek) & HVhek_REHASH)
342#define HEK_REHASH_on(hek) (HEK_FLAGS(hek) |= HVhek_REHASH)
d18c6117 343
5cbe4eec 344/* calculate HV array allocation */
36768cf4 345#ifndef PERL_USE_LARGE_HV_ALLOC
346/* Default to allocating the correct size - default to assuming that malloc()
347 is not broken and is efficient at allocating blocks sized at powers-of-two.
348*/
d18c6117 349# define PERL_HV_ARRAY_ALLOC_BYTES(size) ((size) * sizeof(HE*))
350#else
351# define MALLOC_OVERHEAD 16
352# define PERL_HV_ARRAY_ALLOC_BYTES(size) \
353 (((size) < 64) \
354 ? (size) * sizeof(HE*) \
355 : (size) * sizeof(HE*) * 2 - MALLOC_OVERHEAD)
356#endif
37d85e3a 357
e16e2ff8 358/* Flags for hv_iternext_flags. */
359#define HV_ITERNEXT_WANTPLACEHOLDERS 0x01 /* Don't skip placeholders. */
360
37d85e3a 361/* available as a function in hv.c */
362#define Perl_sharepvn(sv, len, hash) HEK_KEY(share_hek(sv, len, hash))
363#define sharepvn(sv, len, hash) Perl_sharepvn(sv, len, hash)
bfcb3514 364
365/*
366 * Local variables:
367 * c-indentation-style: bsd
368 * c-basic-offset: 4
369 * indent-tabs-mode: t
370 * End:
371 *
372 * ex: set ts=8 sts=4 sw=4 noet:
373 */