summaryrefslogtreecommitdiff
path: root/hashtbl.c
blob: 458bb25e4f50d3b9564206f7331052a0fe86dc3b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
/*
 * hashtbl.c
 *
 * Efficient dictionary hash table class.
 */

#include "compiler.h"

#include <inttypes.h>
#include <string.h>
#include "nasm.h"
#include "hashtbl.h"

#define HASH_INITIAL_SIZE	64
#define HASH_MAX_LOAD		2 /* Higher = more memory-efficient, slower */

static struct hash_tbl_node *alloc_table(size_t newsize)
{
    size_t bytes = newsize*sizeof(struct hash_tbl_node);
    struct hash_tbl_node *newtbl = nasm_zalloc(bytes);

    return newtbl;
}

struct hash_table *hash_init(void)
{
    struct hash_table *head = nasm_malloc(sizeof(struct hash_table));

    head->table    = alloc_table(HASH_INITIAL_SIZE);
    head->load     = 0;
    head->size     = HASH_INITIAL_SIZE;
    head->max_load = HASH_INITIAL_SIZE*(HASH_MAX_LOAD-1)/HASH_MAX_LOAD;

    return head;
}

/*
 * Find an entry in a hash table.
 *
 * On failure, if "insert" is non-NULL, store data in that structure
 * which can be used to insert that node using hash_add().
 *
 * WARNING: this data is only valid until the very next call of
 * hash_add(); it cannot be "saved" to a later date.
 *
 * On success, return a pointer to the "data" element of the hash
 * structure.
 */
void **hash_find(struct hash_table *head, const char *key,
		struct hash_insert *insert)
{
    struct hash_tbl_node *np;
    uint64_t hash = crc64(CRC64_INIT, key);
    struct hash_tbl_node *tbl = head->table;
    size_t mask = head->size-1;
    size_t pos  = hash & mask;
    size_t inc  = ((hash >> 32) & mask) | 1;	/* Always odd */

    while ((np = &tbl[pos])->key) {
	if (hash == np->hash && !strcmp(key, np->key))
	    return &np->data;
	pos = (pos+inc) & mask;
    }

    /* Not found.  Store info for insert if requested. */
    if (insert) {
	insert->head  = head;
	insert->hash  = hash;
	insert->where = np;
    }
    return NULL;
}

/*
 * Same as hash_find, but for case-insensitive hashing.
 */
void **hash_findi(struct hash_table *head, const char *key,
		  struct hash_insert *insert)
{
    struct hash_tbl_node *np;
    uint64_t hash = crc64i(CRC64_INIT, key);
    struct hash_tbl_node *tbl = head->table;
    size_t mask = head->size-1;
    size_t pos  = hash & mask;
    size_t inc  = ((hash >> 32) & mask) | 1;	/* Always odd */

    while ((np = &tbl[pos])->key) {
	if (hash == np->hash && !nasm_stricmp(key, np->key))
	    return &np->data;
	pos = (pos+inc) & mask;
    }

    /* Not found.  Store info for insert if requested. */
    if (insert) {
	insert->head  = head;
	insert->hash  = hash;
	insert->where = np;
    }
    return NULL;
}

/*
 * Insert node.  Return a pointer to the "data" element of the newly
 * created hash node.
 */
void **hash_add(struct hash_insert *insert, const char *key, void *data)
{
    struct hash_table *head  = insert->head;
    struct hash_tbl_node *np = insert->where;

    /* Insert node.  We can always do this, even if we need to
       rebalance immediately after. */
    np->hash = insert->hash;
    np->key  = key;
    np->data = data;

    if (++head->load > head->max_load) {
	/* Need to expand the table */
	size_t newsize = head->size << 1;
	struct hash_tbl_node *newtbl = alloc_table(newsize);
	size_t mask = newsize-1;

	if (head->table) {
	    struct hash_tbl_node *op, *xp;
	    size_t i;

	    /* Rebalance all the entries */
	    for (i = 0, op = head->table; i < head->size; i++, op++) {
		if (op->key) {
		    size_t pos = op->hash & mask;
		    size_t inc = ((op->hash >> 32) & mask) | 1;

		    while ((xp = &newtbl[pos])->key)
			pos = (pos+inc) & mask;

		    *xp = *op;
		    if (op == np)
			np = xp;
		}
	    }
	    nasm_free(head->table);
	}

	head->table    = newtbl;
	head->size     = newsize;
	head->max_load = newsize*(HASH_MAX_LOAD-1)/HASH_MAX_LOAD;
    }

    return &np->data;
}

/*
 * Iterate over all members of a hash set.  For the first call,
 * iterator should be initialized to NULL.  Returns the data pointer,
 * or NULL on failure.
 */
void *hash_iterate(const struct hash_table *head,
		   struct hash_tbl_node **iterator,
		   const char **key)
{
    struct hash_tbl_node *np = *iterator;
    struct hash_tbl_node *ep = head->table + head->size;

    if (!np)
	np = head->table;

    while (np < ep) {
	if (np->key) {
	    *iterator = np+1;
	    if (key)
		*key = np->key;
	    return np->data;
	}
	np++;
    }

    *iterator = NULL;
    if (key)
	*key = NULL;
    return NULL;
}

/*
 * Free the hash itself.  Doesn't free the data elements; use
 * hash_iterate() to do that first, if needed.
 */
void hash_free(struct hash_table *head)
{
    nasm_free(head->table);
    nasm_free(head);
}