* lib/hash.c: New file.

* lib/hash.h: New file.
* lib/Makefile.am: Update.

lib/Makefile.am

@@ -14,6 +14,7 @@ EXTRA_DIST = malloc.c realloc.c
 libbison_a_SOURCES = \
   error.c error.h \
   getopt.h getopt.c getopt1.c \
+  hash.h hash.c \
   obstack.h obstack.c \
   quote.h quote.c quotearg.h quotearg.c \
   xalloc.h xmalloc.c xstrdup.c

lib/hash.c

@@ -0,0 +1,293 @@
+/* hash.c -- hash table maintenance
+   Copyright (C) 1995 Free Software Foundation, Inc.
+   Written by Greg McGary <gkm@gnu.ai.mit.edu>
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2, or (at your option)
+   any later version.
+
+   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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+*/
+
+#include <config.h>
+#include <stdio.h>
+#include "hash.h"
+#include "error.h"
+#include "system.h"
+#include "xalloc.h"
+
+static void hash_rehash __P((struct hash_table* ht));
+static unsigned long round_up_2 __P((unsigned long rough));
+
+/* Implement double hashing with open addressing.  The table size is
+   always a power of two.  The secondary (`increment') hash function
+   is forced to return an odd-value, in order to be relatively prime
+   to the table size.  This guarantees that the increment can
+   potentially hit every slot in the table during collision
+   resolution.  */
+
+void *hash_deleted_item = &hash_deleted_item;
+
+/* Force the table size to be a power of two, possibly rounding up the
+   given size.  */
+
+void
+hash_init (struct hash_table* ht, unsigned long size,
+	   hash_func_t hash_1, hash_func_t hash_2, hash_cmp_func_t hash_cmp)
+{
+  ht->ht_size = round_up_2 (size);
+  if (ht->ht_size > (128 * 1024)) /* prevent size from getting out of hand */
+    ht->ht_size /= 2;
+  ht->ht_vec = (void**) XCALLOC (struct token *, ht->ht_size);
+  if (ht->ht_vec == 0)
+    error (1, 0, _("can't allocate %ld bytes for hash table: memory exhausted"),
+	   ht->ht_size * sizeof(struct token *));
+  ht->ht_capacity = ht->ht_size * 15 / 16; /* 93.75% loading factor */
+  ht->ht_fill = 0;
+  ht->ht_collisions = 0;
+  ht->ht_lookups = 0;
+  ht->ht_rehashes = 0;
+  ht->ht_hash_1 = hash_1;
+  ht->ht_hash_2 = hash_2;
+  ht->ht_compare = hash_cmp;
+}
+
+/* Load an array of items into `ht'.  */
+
+void
+hash_load (struct hash_table* ht, void *item_table, unsigned long cardinality, unsigned long size)
+{
+  char *items = (char *) item_table;
+  while (cardinality--)
+    {
+      hash_insert (ht, items);
+      items += size;
+    }
+}
+
+/* Returns the address of the table slot matching `key'.  If `key' is
+   not found, return the address of an empty slot suitable for
+   inserting `key'.  The caller is responsible for incrementing
+   ht_fill on insertion.  */
+
+void **
+hash_find_slot (struct hash_table* ht, void const *key)
+{
+  void **slot;
+  void **deleted_slot = 0;
+  unsigned int hash_2 = 0;
+  unsigned int hash_1 = (*ht->ht_hash_1) (key);
+
+  ht->ht_lookups++;
+  for (;;)
+    {
+      hash_1 %= ht->ht_size;
+      slot = &ht->ht_vec[hash_1];
+
+      if (*slot == 0)
+	return slot;
+      if (*slot == hash_deleted_item)
+	{
+	  if (deleted_slot == 0)
+	    deleted_slot = slot;
+	}
+      else
+	{
+	  if (key == *slot)
+	    return slot;
+	  if ((*ht->ht_compare) (key, *slot) == 0)
+	    return slot;
+	  ht->ht_collisions++;
+	}
+      if (!hash_2)
+	  hash_2 = (*ht->ht_hash_2) (key) | 1;
+      hash_1 += hash_2;
+    }
+}
+
+void *
+hash_find_item (struct hash_table* ht, void const *key)
+{
+  void **slot = hash_find_slot (ht, key);
+  return ((HASH_VACANT (*slot)) ? 0 : *slot);
+}
+
+void *
+hash_insert (struct hash_table* ht, void *item)
+{
+  void **slot = hash_find_slot (ht, item);
+  return hash_insert_at (ht, item, slot);
+}
+
+void *
+hash_insert_at (struct hash_table* ht, void *item, void const *slot)
+{
+  void *old_item = *(void **) slot;
+  if (HASH_VACANT (old_item))
+    {
+      ht->ht_fill++;
+      old_item = item;
+    }
+  *(void const **) slot = item;
+  if (ht->ht_fill >= ht->ht_capacity)
+    hash_rehash (ht);
+  return old_item;
+}
+
+void *
+hash_delete (struct hash_table* ht, void const *item)
+{
+  void **slot = hash_find_slot (ht, item);
+  return hash_delete_at (ht, slot);
+}
+
+void *
+hash_delete_at (struct hash_table* ht, void const *slot)
+{
+  void *item = *(void **) slot;
+  if (!HASH_VACANT (item))
+    {
+      *(void const **) slot = hash_deleted_item;
+      ht->ht_fill--;
+      return item;
+    }
+  else
+    return 0;
+}
+
+void
+hash_free_items (struct hash_table* ht)
+{
+  void **vec = ht->ht_vec;
+  void **end = &vec[ht->ht_size];
+  for (; vec < end; vec++)
+    {
+      void *item = *vec;
+      if (!HASH_VACANT (item))
+	free (item);
+      *vec = 0;
+    }
+  ht->ht_fill = 0;
+}
+
+void
+hash_delete_items (struct hash_table* ht)
+{
+  void **vec = ht->ht_vec;
+  void **end = &vec[ht->ht_size];
+  for (; vec < end; vec++)
+    *vec = 0;
+  ht->ht_fill = 0;
+  ht->ht_collisions = 0;
+  ht->ht_lookups = 0;
+  ht->ht_rehashes = 0;
+}
+
+void
+hash_free (struct hash_table* ht, int free_items)
+{
+  if (free_items)
+    hash_free_items (ht);
+  free (ht->ht_vec);
+  ht->ht_vec = 0;
+  ht->ht_fill = 0;
+  ht->ht_capacity = 0;
+}
+
+void
+hash_map (struct hash_table *ht, hash_map_func_t map)
+{
+  void **slot;
+  void **end = &ht->ht_vec[ht->ht_size];
+
+  for (slot = ht->ht_vec; slot < end; slot++)
+    {
+      if (!HASH_VACANT (*slot))
+	(*map) (*slot);
+    }
+}
+
+/* Double the size of the hash table in the event of overflow... */
+
+static void
+hash_rehash (struct hash_table* ht)
+{
+  unsigned long old_ht_size = ht->ht_size;
+  void **old_vec = ht->ht_vec;
+  void **ovp;
+  void **slot;
+
+  ht->ht_size *= 2;
+  ht->ht_rehashes++;
+  ht->ht_capacity = ht->ht_size - (ht->ht_size >> 4);
+  ht->ht_vec = (void **) XCALLOC (struct token *, ht->ht_size);
+
+  for (ovp = old_vec; ovp < &old_vec[old_ht_size]; ovp++)
+    {
+      if (*ovp == 0)
+	continue;
+      slot = hash_find_slot (ht, *ovp);
+      *slot = *ovp;
+    }
+  free (old_vec);
+}
+
+void
+hash_print_stats (struct hash_table *ht, FILE *out_FILE)
+{
+  fprintf (out_FILE, _("Load=%ld/%ld=%.0f%%, "), ht->ht_fill, ht->ht_size,
+	   100.0 * (double) ht->ht_fill / (double) ht->ht_size);
+  fprintf (out_FILE, _("Rehash=%d, "), ht->ht_rehashes);
+  fprintf (out_FILE, _("Collisions=%ld/%ld=%.0f%%"), ht->ht_collisions, ht->ht_lookups,
+	   (ht->ht_lookups
+	    ? (100.0 * (double) ht->ht_collisions / (double) ht->ht_lookups)
+	    : 0));
+}
+
+/* Dump all items into a NULL-terminated vector.  Use the
+   user-supplied vector, or malloc one.  */
+
+void**
+hash_dump (struct hash_table *ht, void **vector_0, qsort_cmp_t compare)
+{
+  void **vector;
+  void **slot;
+  void **end = &ht->ht_vec[ht->ht_size];
+
+  if (vector_0 == 0)
+    vector_0 = XMALLOC (void *, ht->ht_fill + 1);
+  vector = vector_0;
+
+  for (slot = ht->ht_vec; slot < end; slot++)
+    if (!HASH_VACANT (*slot))
+      *vector++ = *slot;
+  *vector = 0;
+
+  if (compare)
+    qsort (vector_0, ht->ht_fill, sizeof (void *), compare);
+  return vector_0;
+}
+
+/* Round a given number up to the nearest power of 2. */
+
+static unsigned long
+round_up_2 (unsigned long rough)
+{
+  int round;
+
+  round = 1;
+  while (rough)
+    {
+      round <<= 1;
+      rough >>= 1;
+    }
+  return round;
+}

lib/hash.h

@@ -0,0 +1,144 @@
+/* hash.h -- decls for hash table
+   Copyright (C) 1995 Free Software Foundation, Inc.
+   Written by Greg McGary <gkm@gnu.ai.mit.edu>
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2, or (at your option)
+   any later version.
+
+   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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
+
+#ifndef _hash_h_
+#define _hash_h_
+
+#include <stdio.h>
+
+typedef unsigned long (*hash_func_t) __P((void const *key));
+typedef int (*hash_cmp_func_t) __P((void const *x, void const *y));
+typedef void (*hash_map_func_t) __P((void const *item));
+
+struct hash_table
+{
+  void **ht_vec;
+  unsigned long ht_size;	/* total number of slots (power of 2) */
+  unsigned long ht_capacity;	/* usable slots, limited by loading-factor */
+  unsigned long ht_fill;	/* items in table */
+  unsigned long ht_collisions;	/* # of failed calls to comparison function */
+  unsigned long ht_lookups;	/* # of queries */
+  unsigned int ht_rehashes;	/* # of times we've expanded table */
+  hash_func_t ht_hash_1;	/* primary hash function */
+  hash_func_t ht_hash_2;	/* secondary hash function */
+  hash_cmp_func_t ht_compare;	/* comparison function */
+};
+
+typedef int (*qsort_cmp_t) __P((void const *, void const *));
+
+void hash_init __P((struct hash_table *ht, unsigned long size,
+		    hash_func_t hash_1, hash_func_t hash_2, hash_cmp_func_t hash_cmp));
+void hash_load __P((struct hash_table *ht, void *item_table,
+		    unsigned long cardinality, unsigned long size));
+void **hash_find_slot __P((struct hash_table *ht, void const *key));
+void *hash_find_item __P((struct hash_table *ht, void const *key));
+void *hash_insert __P((struct hash_table *ht, void *item));
+void *hash_insert_at __P((struct hash_table *ht, void *item, void const *slot));
+void *hash_delete __P((struct hash_table *ht, void const *item));
+void *hash_delete_at __P((struct hash_table *ht, void const *slot));
+void hash_delete_items __P((struct hash_table *ht));
+void hash_free_items __P((struct hash_table *ht));
+void hash_free __P((struct hash_table *ht, int free_items));
+void hash_map __P((struct hash_table *ht, hash_map_func_t map));
+void hash_print_stats __P((struct hash_table *ht, FILE *out_FILE));
+void **hash_dump __P((struct hash_table *ht, void **vector_0, qsort_cmp_t compare));
+
+extern void *hash_deleted_item;
+#define HASH_VACANT(item) ((item) == 0 || (void *) (item) == hash_deleted_item)
+
+
+/* hash and comparison macros for string keys. */
+
+#define STRING_HASH_1(_key_, _result_) { \
+  unsigned char const *kk = (unsigned char const *) (_key_) - 1; \
+  while (*++kk) \
+    (_result_) += (*kk << (kk[1] & 0xf)); \
+} while (0)
+#define return_STRING_HASH_1(_key_) do { \
+  unsigned long result = 0; \
+  STRING_HASH_1 ((_key_), result); \
+  return result; \
+} while (0)
+
+#define STRING_HASH_2(_key_, _result_) do { \
+  unsigned char const *kk = (unsigned char const *) (_key_) - 1; \
+  while (*++kk) \
+    (_result_) += (*kk << (kk[1] & 0x7)); \
+} while (0)
+#define return_STRING_HASH_2(_key_) do { \
+  unsigned long result = 0; \
+  STRING_HASH_2 ((_key_), result); \
+  return result; \
+} while (0)
+
+#define STRING_COMPARE(_x_, _y_, _result_) do { \
+  unsigned char const *xx = (unsigned char const *) (_x_) - 1; \
+  unsigned char const *yy = (unsigned char const *) (_y_) - 1; \
+  do { \
+    if (*++xx == '\0') { \
+      yy++; \
+      break; \
+    } \
+  } while (*xx == *++yy); \
+  (_result_) = *xx - *yy; \
+} while (0)
+#define return_STRING_COMPARE(_x_, _y_) do { \
+  int result; \
+  STRING_COMPARE (_x_, _y_, result); \
+  return result; \
+} while (0)
+
+/* hash and comparison macros for integer keys. */
+
+#define INTEGER_HASH_1(_key_, _result_) do { \
+  (_result_) += ((unsigned long)(_key_)); \
+} while (0)
+#define return_INTEGER_HASH_1(_key_) do { \
+  unsigned long result = 0; \
+  INTEGER_HASH_1 ((_key_), result); \
+  return result; \
+} while (0)
+
+#define INTEGER_HASH_2(_key_, _result_) do { \
+  (_result_) += ~((unsigned long)(_key_)); \
+} while (0)
+#define return_INTEGER_HASH_2(_key_) do { \
+  unsigned long result = 0; \
+  INTEGER_HASH_2 ((_key_), result); \
+  return result; \
+} while (0)
+
+#define INTEGER_COMPARE(_x_, _y_, _result_) do { \
+  (_result_) = _x_ - _y_; \
+} while (0)
+#define return_INTEGER_COMPARE(_x_, _y_) do { \
+  int result; \
+  INTEGER_COMPARE (_x_, _y_, result); \
+  return result; \
+} while (0)
+
+/* hash and comparison macros for address keys. */
+
+#define ADDRESS_HASH_1(_key_, _result_) INTEGER_HASH_1 (((unsigned long)(_key_)) >> 3, (_result_))
+#define ADDRESS_HASH_2(_key_, _result_) INTEGER_HASH_2 (((unsigned long)(_key_)) >> 3, (_result_))
+#define ADDRESS_COMPARE(_x_, _y_, _result_) INTEGER_COMPARE ((_x_), (_y_), (_result_))
+#define return_ADDRESS_HASH_1(_key_) return_INTEGER_HASH_1 (((unsigned long)(_key_)) >> 3)
+#define return_ADDRESS_HASH_2(_key_) return_INTEGER_HASH_2 (((unsigned long)(_key_)) >> 3)
+#define return_ADDRESS_COMPARE(_x_, _y_) return_INTEGER_COMPARE ((_x_), (_y_))
+
+#endif /* not _hash_h_ */