Logo Search packages:      
Sourcecode: batmand version File versions  Download package

hash.c

/* Copyright (C) 2006 B.A.T.M.A.N. contributors:
 * Simon Wunderlich, Marek Lindner
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of version 2 of the GNU General Public
 * License as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
 * 02110-1301, USA
 *
 */


#include <stdio.h>            /* NULL */
#include "hash.h"
#include "allocate.h"


/* clears the hash */
void hash_init(struct hashtable_t *hash) {
      int i;
      hash->elements=0;
      for (i=0 ; i<hash->size ; i++) {
            hash->table[i] = NULL;
      }
}


/* remove the hash structure. if hashdata_free_cb != NULL,
 * this function will be called to remove the elements inside of the hash.
 * if you don't remove the elements, memory might be leaked. */
void hash_delete(struct hashtable_t *hash, hashdata_free_cb free_cb) {
      struct element_t *bucket, *last_bucket;
      int i;

      for (i=0; i<hash->size; i++) {

            bucket= hash->table[i];
            while (bucket != NULL) {

                  if (free_cb!=NULL)
                        free_cb( bucket->data );

                  last_bucket= bucket;
                  bucket= bucket->next;
                  debugFree(last_bucket, 1301);

            }

      }
      hash_destroy(hash);
}



/* free only the hashtable and the hash itself. */
void hash_destroy(struct hashtable_t *hash) {

      debugFree( hash->table, 1302 );
      debugFree( hash, 1303 );

}




/* iterate though the hash. first element is selected with iter_in NULL.
 * use the returned iterator to access the elements until hash_it_t returns NULL. */
struct hash_it_t *hash_iterate(struct hashtable_t *hash, struct hash_it_t *iter_in) {
      struct hash_it_t *iter;

      if (iter_in == NULL) {
            iter= debugMalloc(sizeof(struct hash_it_t), 301);
            iter->index =  -1;
            iter->bucket = NULL;
            iter->prev_bucket = NULL;
      } else
            iter= iter_in;

      /* sanity checks first (if our bucket got deleted in the last iteration): */
      if (iter->bucket!=NULL) {
            if (iter->first_bucket != NULL) {

                  /* we're on the first element and it got removed after the last iteration. */
                  if ((*iter->first_bucket) != iter->bucket) {

                        /* there are still other elements in the list */
                        if ( (*iter->first_bucket) != NULL ) {
                              iter->prev_bucket = NULL;
                              iter->bucket= (*iter->first_bucket);
                              iter->first_bucket = &hash->table[ iter->index ];
                              return(iter);
                        } else {
                              iter->bucket = NULL;
                        }

                  }

            } else if ( iter->prev_bucket != NULL ) {

                  /* we're not on the first element, and the bucket got removed after the last iteration.
                  * the last bucket's next pointer is not pointing to our actual bucket anymore.
                  * select the next. */
                  if ( iter->prev_bucket->next != iter->bucket )
                        iter->bucket= iter->prev_bucket;

            }

      }

      /* now as we are sane, select the next one if there is some */
      if (iter->bucket!=NULL) {
            if (iter->bucket->next!=NULL) {
                  iter->prev_bucket= iter->bucket;
                  iter->bucket= iter->bucket->next;
                  iter->first_bucket = NULL;
                  return(iter);
            }
      }
      /* if not returned yet, we've reached the last one on the index and have to search forward */

      iter->index++;
      while ( iter->index < hash->size ) {            /* go through the entries of the hash table */
            if ((hash->table[ iter->index ]) != NULL){
                  iter->prev_bucket = NULL;
                  iter->bucket = hash->table[ iter->index ];
                  iter->first_bucket = &hash->table[ iter->index ];
                  return(iter);                                   /* if this table entry is not null, return it */
            } else
                  iter->index++;                                  /* else, go to the next */
      }
      /* nothing to iterate over anymore */
      debugFree(iter, 1304);
      return(NULL);
}


/* allocates and clears the hash */
struct hashtable_t *hash_new(int size, hashdata_compare_cb compare, hashdata_choose_cb choose) {
      struct hashtable_t *hash;

      hash= debugMalloc( sizeof(struct hashtable_t) , 302);
      if ( hash == NULL )                 /* could not allocate the hash control structure */
            return (NULL);

      hash->size= size;
      hash->table= debugMalloc( sizeof(struct element_t *) * size, 303);
      if ( hash->table == NULL ) {  /* could not allocate the table */
            debugFree(hash, 1305);
            return(NULL);
      }
      hash_init(hash);
      hash->compare= compare;
      hash->choose= choose;
      return(hash);
}


/* adds data to the hashtable. returns 0 on success, -1 on error */
int hash_add(struct hashtable_t *hash, void *data) {
      int index;
      struct element_t *bucket, *prev_bucket = NULL;

      index = hash->choose( data, hash->size );
      bucket = hash->table[index];

      while ( bucket!=NULL ) {
            if (0 == hash->compare( bucket->data, data ))
                  return(-1);

            prev_bucket = bucket;
            bucket= bucket->next;
      }

      /* found the tail of the list, add new element */
      if (NULL == (bucket= debugMalloc(sizeof(struct element_t),304)))
            return(-1); /* debugMalloc failed */

      bucket->data= data;                       /* init the new bucket */
      bucket->next= NULL;

      /* and link it */
      if ( prev_bucket == NULL ) {
            hash->table[index] = bucket;
      } else {
            prev_bucket->next = bucket;
      }

      hash->elements++;
      return(0);

}
/* finds data, based on the key in keydata. returns the found data on success, or NULL on error */
void *hash_find(struct hashtable_t *hash, void *keydata) {
      int index;
      struct element_t *bucket;

      index = hash->choose( keydata , hash->size );
      bucket = hash->table[index];

      while ( bucket!=NULL ) {
            if (0 == hash->compare( bucket->data, keydata ))
                  return( bucket->data );

            bucket= bucket->next;
      }

      return(NULL);

}

/* remove bucket (this might be used in hash_iterate() if you already found the bucket
 * you want to delete and don't need the overhead to find it again with hash_remove().
 * But usually, you don't want to use this function, as it fiddles with hash-internals. */
void *hash_remove_bucket(struct hashtable_t *hash, struct hash_it_t *hash_it_t) {
      void *data_save;

      data_save = hash_it_t->bucket->data;      /* save the pointer to the data */

      if ( hash_it_t->prev_bucket != NULL ) {
            hash_it_t->prev_bucket->next = hash_it_t->bucket->next;
      } else if ( hash_it_t->first_bucket != NULL ) {
            (*hash_it_t->first_bucket) = hash_it_t->bucket->next;
      }

      debugFree(hash_it_t->bucket, 1306);

      hash->elements--;
      return( data_save );

}


/* removes data from hash, if found. returns pointer do data on success,
 * so you can remove the used structure yourself, or NULL on error .
 * data could be the structure you use with just the key filled,
 * we just need the key for comparing. */
void *hash_remove(struct hashtable_t *hash, void *data) {
      struct hash_it_t hash_it_t;

      hash_it_t.index = hash->choose( data, hash->size );
      hash_it_t.bucket = hash->table[hash_it_t.index];
      hash_it_t.prev_bucket = NULL;

      while ( hash_it_t.bucket!=NULL ) {
            if (0 == hash->compare( hash_it_t.bucket->data, data )) {
                  hash_it_t.first_bucket = (hash_it_t.bucket == hash->table[hash_it_t.index] ? &hash->table[ hash_it_t.index ] : NULL);
                  return( hash_remove_bucket(hash, &hash_it_t) );
            }

            hash_it_t.prev_bucket = hash_it_t.bucket;
            hash_it_t.bucket= hash_it_t.bucket->next;
      }

      return(NULL);

}


/* resize the hash, returns the pointer to the new hash or NULL on error. removes the old hash on success. */
struct hashtable_t *hash_resize(struct hashtable_t *hash, int size) {
      struct hashtable_t *new_hash;
      struct element_t *bucket;
      int i;

      /* initialize a new hash with the new size */
      if (NULL == (new_hash= hash_new(size, hash->compare, hash->choose)))
            return(NULL);

      /* copy the elements */
      for (i=0; i<hash->size; i++) {
            bucket= hash->table[i];
            while (bucket != NULL) {
                  hash_add( new_hash, bucket->data );
                  bucket= bucket->next;
            }
      }
      hash_delete(hash, NULL);      /* remove hash and eventual overflow buckets but not the content itself. */

      return( new_hash);

}


/* print the hash table for debugging */
void hash_debug(struct hashtable_t *hash) {
      int i;
      struct element_t *bucket;

      for (i=0; i<hash->size;i++) {
            printf("[%d] ",i);
            bucket= hash->table[i];

            while (bucket != NULL) {
                  printf("-> [%10p] ", bucket);
                  bucket= bucket->next;
            }

            printf("\n");

      }
      printf("\n");
}


Generated by  Doxygen 1.6.0   Back to index