lib/hash.h

   1 /* hash - hashing table processing.
   2    Copyright (C) 1998-1999, 2001, 2003, 2009-2021 Free Software Foundation,
   3    Inc.
   4    Written by Jim Meyering <meyering@ascend.com>, 1998.
   5
   6    This program is free software: you can redistribute it and/or modify
   7    it under the terms of the GNU General Public License as published by
   8    the Free Software Foundation; either version 3 of the License, or
   9    (at your option) any later version.
  10
  11    This program is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14    GNU General Public License for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with this program.  If not, see <https://www.gnu.org/licenses/>.  */
  18
  19 /* A generic hash table package.  */
  20
  21 /* Make sure USE_OBSTACK is defined to 1 if you want the allocator to use
  22    obstacks instead of malloc, and recompile 'hash.c' with same setting.  */
  23
  24 #ifndef HASH_H_
  25 # define HASH_H_
  26
  27 # include <stdio.h>
  28 # include <stdbool.h>
  29
  30 # ifdef __cplusplus
  31 extern "C" {
  32 # endif
  33
  34 struct hash_tuning
  35   {
  36     /* This structure is mainly used for 'hash_initialize', see the block
  37        documentation of 'hash_reset_tuning' for more complete comments.  */
  38
  39     float shrink_threshold;     /* ratio of used buckets to trigger a shrink */
  40     float shrink_factor;        /* ratio of new smaller size to original size */
  41     float growth_threshold;     /* ratio of used buckets to trigger a growth */
  42     float growth_factor;        /* ratio of new bigger size to original size */
  43     bool is_n_buckets;          /* if CANDIDATE really means table size */
  44   };
  45
  46 typedef struct hash_tuning Hash_tuning;
  47
  48 struct hash_table;
  49
  50 typedef struct hash_table Hash_table;
  51
  52 /*
  53  * Information and lookup.
  54  */
  55
  56 /* The following few functions provide information about the overall hash
  57    table organization: the number of entries, number of buckets and maximum
  58    length of buckets.  */
  59
  60 /* Return the number of buckets in the hash table.  The table size, the total
  61    number of buckets (used plus unused), or the maximum number of slots, are
  62    the same quantity.  */
  63 extern size_t hash_get_n_buckets (const Hash_table *table)
  64        _GL_ATTRIBUTE_PURE;
  65
  66 /* Return the number of slots in use (non-empty buckets).  */
  67 extern size_t hash_get_n_buckets_used (const Hash_table *table)
  68        _GL_ATTRIBUTE_PURE;
  69
  70 /* Return the number of active entries.  */
  71 extern size_t hash_get_n_entries (const Hash_table *table)
  72        _GL_ATTRIBUTE_PURE;
  73
  74 /* Return the length of the longest chain (bucket).  */
  75 extern size_t hash_get_max_bucket_length (const Hash_table *table)
  76        _GL_ATTRIBUTE_PURE;
  77
  78 /* Do a mild validation of a hash table, by traversing it and checking two
  79    statistics.  */
  80 extern bool hash_table_ok (const Hash_table *table)
  81        _GL_ATTRIBUTE_PURE;
  82
  83 extern void hash_print_statistics (const Hash_table *table, FILE *stream);
  84
  85 /* If ENTRY matches an entry already in the hash table, return the
  86    entry from the table.  Otherwise, return NULL.  */
  87 extern void *hash_lookup (const Hash_table *table, const void *entry);
  88
  89 /*
  90  * Walking.
  91  */
  92
  93 /* The functions in this page traverse the hash table and process the
  94    contained entries.  For the traversal to work properly, the hash table
  95    should not be resized nor modified while any particular entry is being
  96    processed.  In particular, entries should not be added, and an entry
  97    may be removed only if there is no shrink threshold and the entry being
  98    removed has already been passed to hash_get_next.  */
  99
 100 /* Return the first data in the table, or NULL if the table is empty.  */
 101 extern void *hash_get_first (const Hash_table *table)
 102        _GL_ATTRIBUTE_PURE;
 103
 104 /* Return the user data for the entry following ENTRY, where ENTRY has been
 105    returned by a previous call to either 'hash_get_first' or 'hash_get_next'.
 106    Return NULL if there are no more entries.  */
 107 extern void *hash_get_next (const Hash_table *table, const void *entry);
 108
 109 /* Fill BUFFER with pointers to active user entries in the hash table, then
 110    return the number of pointers copied.  Do not copy more than BUFFER_SIZE
 111    pointers.  */
 112 extern size_t hash_get_entries (const Hash_table *table, void **buffer,
 113                                 size_t buffer_size);
 114
 115 typedef bool (*Hash_processor) (void *entry, void *processor_data);
 116
 117 /* Call a PROCESSOR function for each entry of a hash table, and return the
 118    number of entries for which the processor function returned success.  A
 119    pointer to some PROCESSOR_DATA which will be made available to each call to
 120    the processor function.  The PROCESSOR accepts two arguments: the first is
 121    the user entry being walked into, the second is the value of PROCESSOR_DATA
 122    as received.  The walking continue for as long as the PROCESSOR function
 123    returns nonzero.  When it returns zero, the walking is interrupted.  */
 124 extern size_t hash_do_for_each (const Hash_table *table,
 125                                 Hash_processor processor, void *processor_data);
 126
 127 /*
 128  * Allocation and clean-up.
 129  */
 130
 131 /* Return a hash index for a NUL-terminated STRING between 0 and N_BUCKETS-1.
 132    This is a convenience routine for constructing other hashing functions.  */
 133 extern size_t hash_string (const char *string, size_t n_buckets)
 134        _GL_ATTRIBUTE_PURE;
 135
 136 extern void hash_reset_tuning (Hash_tuning *tuning);
 137
 138 typedef size_t (*Hash_hasher) (const void *entry, size_t table_size);
 139 typedef bool (*Hash_comparator) (const void *entry1, const void *entry2);
 140 typedef void (*Hash_data_freer) (void *entry);
 141
 142 /* Allocate and return a new hash table, or NULL upon failure.  The initial
 143    number of buckets is automatically selected so as to _guarantee_ that you
 144    may insert at least CANDIDATE different user entries before any growth of
 145    the hash table size occurs.  So, if have a reasonably tight a-priori upper
 146    bound on the number of entries you intend to insert in the hash table, you
 147    may save some table memory and insertion time, by specifying it here.  If
 148    the IS_N_BUCKETS field of the TUNING structure is true, the CANDIDATE
 149    argument has its meaning changed to the wanted number of buckets.
 150
 151    TUNING points to a structure of user-supplied values, in case some fine
 152    tuning is wanted over the default behavior of the hasher.  If TUNING is
 153    NULL, the default tuning parameters are used instead.  If TUNING is
 154    provided but the values requested are out of bounds or might cause
 155    rounding errors, return NULL.
 156
 157    The user-supplied HASHER function, when not NULL, accepts two
 158    arguments ENTRY and TABLE_SIZE.  It computes, by hashing ENTRY contents, a
 159    slot number for that entry which should be in the range 0..TABLE_SIZE-1.
 160    This slot number is then returned.
 161
 162    The user-supplied COMPARATOR function, when not NULL, accepts two
 163    arguments pointing to user data, it then returns true for a pair of entries
 164    that compare equal, or false otherwise.  This function is internally called
 165    on entries which are already known to hash to the same bucket index,
 166    but which are distinct pointers.
 167
 168    The user-supplied DATA_FREER function, when not NULL, may be later called
 169    with the user data as an argument, just before the entry containing the
 170    data gets freed.  This happens from within 'hash_free' or 'hash_clear'.
 171    You should specify this function only if you want these functions to free
 172    all of your 'data' data.  This is typically the case when your data is
 173    simply an auxiliary struct that you have malloc'd to aggregate several
 174    values.  */
 175 extern Hash_table *hash_initialize (size_t candidate,
 176                                     const Hash_tuning *tuning,
 177                                     Hash_hasher hasher,
 178                                     Hash_comparator comparator,
 179                                     Hash_data_freer data_freer)
 180        _GL_ATTRIBUTE_NODISCARD;
 181
 182 /* Same as hash_initialize, but invokes xalloc_die on memory exhaustion.  */
 183 /* This function is defined by module 'xhash'.  */
 184 extern Hash_table *hash_xinitialize (size_t candidate,
 185                                      const Hash_tuning *tuning,
 186                                      Hash_hasher hasher,
 187                                      Hash_comparator comparator,
 188                                      Hash_data_freer data_freer)
 189        _GL_ATTRIBUTE_NODISCARD;
 190
 191 /* Make all buckets empty, placing any chained entries on the free list.
 192    Apply the user-specified function data_freer (if any) to the datas of any
 193    affected entries.  */
 194 extern void hash_clear (Hash_table *table);
 195
 196 /* Reclaim all storage associated with a hash table.  If a data_freer
 197    function has been supplied by the user when the hash table was created,
 198    this function applies it to the data of each entry before freeing that
 199    entry.  */
 200 extern void hash_free (Hash_table *table);
 201
 202 /*
 203  * Insertion and deletion.
 204  */
 205
 206 /* For an already existing hash table, change the number of buckets through
 207    specifying CANDIDATE.  The contents of the hash table are preserved.  The
 208    new number of buckets is automatically selected so as to _guarantee_ that
 209    the table may receive at least CANDIDATE different user entries, including
 210    those already in the table, before any other growth of the hash table size
 211    occurs.  If TUNING->IS_N_BUCKETS is true, then CANDIDATE specifies the
 212    exact number of buckets desired.  Return true iff the rehash succeeded.  */
 213 extern bool hash_rehash (Hash_table *table, size_t candidate)
 214        _GL_ATTRIBUTE_NODISCARD;
 215
 216 /* If ENTRY matches an entry already in the hash table, return the pointer
 217    to the entry from the table.  Otherwise, insert ENTRY and return ENTRY.
 218    Return NULL if the storage required for insertion cannot be allocated.
 219    This implementation does not support duplicate entries or insertion of
 220    NULL.  */
 221 extern void *hash_insert (Hash_table *table, const void *entry)
 222        _GL_ATTRIBUTE_NODISCARD;
 223
 224 /* Same as hash_insert, but invokes xalloc_die on memory exhaustion.  */
 225 /* This function is defined by module 'xhash'.  */
 226 extern void *hash_xinsert (Hash_table *table, const void *entry);
 227
 228 /* Insert ENTRY into hash TABLE if there is not already a matching entry.
 229
 230    Return -1 upon memory allocation failure.
 231    Return 1 if insertion succeeded.
 232    Return 0 if there is already a matching entry in the table,
 233    and in that case, if MATCHED_ENT is non-NULL, set *MATCHED_ENT
 234    to that entry.
 235
 236    This interface is easier to use than hash_insert when you must
 237    distinguish between the latter two cases.  More importantly,
 238    hash_insert is unusable for some types of ENTRY values.  When using
 239    hash_insert, the only way to distinguish those cases is to compare
 240    the return value and ENTRY.  That works only when you can have two
 241    different ENTRY values that point to data that compares "equal".  Thus,
 242    when the ENTRY value is a simple scalar, you must use
 243    hash_insert_if_absent.  ENTRY must not be NULL.  */
 244 extern int hash_insert_if_absent (Hash_table *table, const void *entry,
 245                                   const void **matched_ent);
 246
 247 /* If ENTRY is already in the table, remove it and return the just-deleted
 248    data (the user may want to deallocate its storage).  If ENTRY is not in the
 249    table, don't modify the table and return NULL.  */
 250 extern void *hash_remove (Hash_table *table, const void *entry);
 251
 252 /* Same as hash_remove.  This interface is deprecated.
 253    FIXME: Remove in 2022.  */
 254 extern void *hash_delete (Hash_table *table, const void *entry)
 255        _GL_ATTRIBUTE_DEPRECATED;
 256
 257 # ifdef __cplusplus
 258 }
 259 # endif
 260
 261 #endif