X-Git-Url: http://erislabs.net/gitweb/?a=blobdiff_plain;f=lib%2Fhash.c;h=3948ce4cf69d3b4dfaeaa3e79c8f44883d8befec;hb=22c1c15d0a29a1b7ea75456abe3f7062f835c887;hp=e26d0d047602e92c841cc686a71150c89664820d;hpb=b6044c6ab01d1b8131d6310f6341b4bf1d052a75;p=gnulib.git diff --git a/lib/hash.c b/lib/hash.c index e26d0d047..3948ce4cf 100644 --- a/lib/hash.c +++ b/lib/hash.c @@ -1,5 +1,8 @@ /* hash - hashing table processing. - Copyright (C) 1998, 1999 Free Software Foundation, Inc. + + Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003 Free Software + Foundation, Inc. + Written by Jim Meyering, 1992. This program is free software; you can redistribute it and/or modify @@ -24,23 +27,11 @@ #if HAVE_CONFIG_H # include #endif -#if HAVE_STDLIB_H -# include -#endif -#if HAVE_STDBOOL_H -# include -#else -typedef enum {false = 0, true = 1} bool; -#endif -#include -#include -#if !HAVE_DECL_FREE -void free (); -#endif -#if !HAVE_DECL_MALLOC -char *malloc (); -#endif +#include +#include +#include +#include #if USE_OBSTACK # include "obstack.h" @@ -54,6 +45,40 @@ char *malloc (); #include "hash.h" +struct hash_table + { + /* The array of buckets starts at BUCKET and extends to BUCKET_LIMIT-1, + for a possibility of N_BUCKETS. Among those, N_BUCKETS_USED buckets + are not empty, there are N_ENTRIES active entries in the table. */ + struct hash_entry *bucket; + struct hash_entry *bucket_limit; + unsigned n_buckets; + unsigned n_buckets_used; + unsigned n_entries; + + /* Tuning arguments, kept in a physicaly separate structure. */ + const Hash_tuning *tuning; + + /* Three functions are given to `hash_initialize', see the documentation + block for this function. In a word, HASHER randomizes a user entry + into a number up from 0 up to some maximum minus 1; COMPARATOR returns + true if two user entries compare equally; and DATA_FREER is the cleanup + function for a user entry. */ + Hash_hasher hasher; + Hash_comparator comparator; + Hash_data_freer data_freer; + + /* A linked list of freed struct hash_entry structs. */ + struct hash_entry *free_entry_list; + +#if USE_OBSTACK + /* Whenever obstacks are used, it is possible to allocate all overflowed + entries into a single stack, so they all can be freed in a single + operation. It is not clear if the speedup is worth the trouble. */ + struct obstack entry_stack; +#endif + }; + /* A hash table contains many internal entries, each holding a pointer to some user provided data (also called a user entry). An entry indistinctly refers to both the internal entry and its associated user entry. A user @@ -222,7 +247,8 @@ hash_lookup (const Hash_table *table, const void *entry) = table->bucket + table->hasher (entry, table->n_buckets); struct hash_entry *cursor; - assert (bucket < table->bucket_limit); + if (! (bucket < table->bucket_limit)) + abort (); if (bucket->data == NULL) return NULL; @@ -251,16 +277,16 @@ hash_get_first (const Hash_table *table) if (table->n_entries == 0) return NULL; - for (bucket = table->bucket; bucket < table->bucket_limit; bucket++) - if (bucket->data) + for (bucket = table->bucket; ; bucket++) + if (! (bucket < table->bucket_limit)) + abort (); + else if (bucket->data) return bucket->data; - - assert (0); } /* Return the user data for the entry following ENTRY, where ENTRY has been returned by a previous call to either `hash_get_first' or `hash_get_next'. - Return NULL if there is no more entries. */ + Return NULL if there are no more entries. */ void * hash_get_next (const Hash_table *table, const void *entry) @@ -269,7 +295,8 @@ hash_get_next (const Hash_table *table, const void *entry) = table->bucket + table->hasher (entry, table->n_buckets); struct hash_entry *cursor; - assert (bucket < table->bucket_limit); + if (! (bucket < table->bucket_limit)) + abort (); /* Find next entry in the same bucket. */ for (cursor = bucket; cursor; cursor = cursor->next) @@ -277,7 +304,7 @@ hash_get_next (const Hash_table *table, const void *entry) return cursor->next->data; /* Find first entry in any subsequent bucket. */ - for (; bucket < table->bucket_limit; bucket++) + while (++bucket < table->bucket_limit) if (bucket->data) return bucket->data; @@ -361,9 +388,6 @@ hash_do_for_each (const Hash_table *table, Hash_processor processor, unsigned hash_string (const char *string, unsigned n_buckets) { -# ifndef CHAR_BIT -# define CHAR_BIT 8 -# endif # define ROTATE_LEFT(Value, Shift) \ ((Value) << (Shift) | (Value) >> ((sizeof (unsigned) * CHAR_BIT) - (Shift))) # define HASH_ONE_CHAR(Value, Byte) \ @@ -415,7 +439,7 @@ is_prime (unsigned long candidate) divisor++; } - return candidate % divisor != 0; + return (candidate % divisor ? true : false); } /* Round a given CANDIDATE number up to the nearest prime, and return that @@ -445,8 +469,8 @@ hash_reset_tuning (Hash_tuning *tuning) /* For the given hash TABLE, check the user supplied tuning structure for reasonable values, and return true if there is no gross error with it. - Otherwise, definitvely reset the TUNING field to some acceptable default in - the hash table (that is, the user loses the right of further modifying + Otherwise, definitively reset the TUNING field to some acceptable default + in the hash table (that is, the user loses the right of further modifying tuning arguments), and return false. */ static bool @@ -468,15 +492,14 @@ check_tuning (Hash_table *table) return false; } -/* Allocate and return a new hash table, or NULL upon failure. The - initial number of buckets is automatically selected so as to _guarantee_ that - you may insert at least CANDIDATE different user entries before any growth - of the hash table size occurs. So, if have a reasonably tight a-priori - upper bound on the - number of entries you intend to insert in the hash table, you may save some - table memory and insertion time, by specifying it here. If the - IS_N_BUCKETS field of the TUNING structure is true, the CANDIDATE argument - has its meaning changed to the wanted number of buckets. +/* Allocate and return a new hash table, or NULL upon failure. The initial + number of buckets is automatically selected so as to _guarantee_ that you + may insert at least CANDIDATE different user entries before any growth of + the hash table size occurs. So, if have a reasonably tight a-priori upper + bound on the number of entries you intend to insert in the hash table, you + may save some table memory and insertion time, by specifying it here. If + the IS_N_BUCKETS field of the TUNING structure is true, the CANDIDATE + argument has its meaning changed to the wanted number of buckets. TUNING points to a structure of user-supplied values, in case some fine tuning is wanted over the default behavior of the hasher. If TUNING is @@ -569,19 +592,22 @@ void hash_clear (Hash_table *table) { struct hash_entry *bucket; - struct hash_entry *cursor; for (bucket = table->bucket; bucket < table->bucket_limit; bucket++) { if (bucket->data) { + struct hash_entry *cursor; + struct hash_entry *next; + /* Free the bucket overflow. */ - for (cursor = bucket->next; cursor; cursor = cursor->next) + for (cursor = bucket->next; cursor; cursor = next) { if (table->data_freer) (*table->data_freer) (cursor->data); cursor->data = NULL; + next = cursor->next; /* Relinking is done one entry at a time, as it is to be expected that overflows are either rare or short. */ cursor->next = table->free_entry_list; @@ -710,14 +736,16 @@ hash_find_entry (Hash_table *table, const void *entry, = table->bucket + table->hasher (entry, table->n_buckets); struct hash_entry *cursor; - assert (bucket < table->bucket_limit); + if (! (bucket < table->bucket_limit)) + abort (); + *bucket_head = bucket; /* Test for empty bucket. */ if (bucket->data == NULL) return NULL; - /* Check if then entry is found as the bucket head. */ + /* See if the entry is the first in the bucket. */ if ((*table->comparator) (entry, bucket->data)) { void *data = bucket->data; @@ -769,8 +797,8 @@ hash_find_entry (Hash_table *table, const void *entry, /* For an already existing hash table, change the number of buckets through specifying CANDIDATE. The contents of the hash table are preserved. The - new number of buckets is automatically selected so as to _guarantee_ that the - table may receive at least CANDIDATE different user entries, including + new number of buckets is automatically selected so as to _guarantee_ that + the table may receive at least CANDIDATE different user entries, including those already in the table, before any other growth of the hash table size occurs. If TUNING->IS_N_BUCKETS is true, then CANDIDATE specifies the exact number of buckets desired. */ @@ -804,7 +832,9 @@ hash_rehash (Hash_table *table, unsigned candidate) = (new_table->bucket + new_table->hasher (data, new_table->n_buckets)); - assert (new_bucket < new_table->bucket_limit); + if (! (new_bucket < new_table->bucket_limit)) + abort (); + next = cursor->next; if (new_bucket->data) @@ -848,6 +878,7 @@ hash_rehash (Hash_table *table, unsigned candidate) table->bucket_limit = new_table->bucket_limit; table->n_buckets = new_table->n_buckets; table->n_buckets_used = new_table->n_buckets_used; + table->free_entry_list = new_table->free_entry_list; /* table->n_entries already holds its value. */ #if USE_OBSTACK table->entry_stack = new_table->entry_stack; @@ -867,7 +898,9 @@ hash_insert (Hash_table *table, const void *entry) void *data; struct hash_entry *bucket; - assert (entry); /* cannot insert a NULL entry */ + /* The caller cannot insert a NULL entry. */ + if (! entry) + abort (); /* If there's a matching entry already in the table, return that. */ if ((data = hash_find_entry (table, entry, &bucket, false)) != NULL) @@ -937,7 +970,8 @@ hash_delete (Hash_table *table, const void *entry) void *data; struct hash_entry *bucket; - if (data = hash_find_entry (table, entry, &bucket, true), !data) + data = hash_find_entry (table, entry, &bucket, true); + if (!data) return NULL; table->n_entries--; @@ -986,13 +1020,14 @@ hash_print (const Hash_table *table) struct hash_entry *cursor; if (bucket) - printf ("%d:\n", slot); + printf ("%d:\n", bucket - table->bucket); for (cursor = bucket; cursor; cursor = cursor->next) { char *s = (char *) cursor->data; /* FIXME */ - printf (" %s\n", s); + if (s) + printf (" %s\n", s); } } }