np-ferret 0.11.6
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +24 -0
- data/MIT-LICENSE +20 -0
- data/README +102 -0
- data/Rakefile +338 -0
- data/TODO +17 -0
- data/TUTORIAL +231 -0
- data/bin/ferret-browser +79 -0
- data/ext/Makefile +218 -0
- data/ext/analysis.c +1584 -0
- data/ext/analysis.h +219 -0
- data/ext/analysis.o +0 -0
- data/ext/api.c +69 -0
- data/ext/api.h +27 -0
- data/ext/api.o +0 -0
- data/ext/array.c +123 -0
- data/ext/array.h +53 -0
- data/ext/array.o +0 -0
- data/ext/bitvector.c +540 -0
- data/ext/bitvector.h +272 -0
- data/ext/bitvector.o +0 -0
- data/ext/compound_io.c +383 -0
- data/ext/compound_io.o +0 -0
- data/ext/config.h +42 -0
- data/ext/document.c +156 -0
- data/ext/document.h +53 -0
- data/ext/document.o +0 -0
- data/ext/except.c +120 -0
- data/ext/except.h +168 -0
- data/ext/except.o +0 -0
- data/ext/extconf.rb +14 -0
- data/ext/ferret.c +402 -0
- data/ext/ferret.h +91 -0
- data/ext/ferret.o +0 -0
- data/ext/ferret_ext.bundle +0 -0
- data/ext/filter.c +156 -0
- data/ext/filter.o +0 -0
- data/ext/fs_store.c +484 -0
- data/ext/fs_store.o +0 -0
- data/ext/global.c +418 -0
- data/ext/global.h +117 -0
- data/ext/global.o +0 -0
- data/ext/hash.c +598 -0
- data/ext/hash.h +475 -0
- data/ext/hash.o +0 -0
- data/ext/hashset.c +170 -0
- data/ext/hashset.h +187 -0
- data/ext/hashset.o +0 -0
- data/ext/header.h +58 -0
- data/ext/helper.c +62 -0
- data/ext/helper.h +13 -0
- data/ext/helper.o +0 -0
- data/ext/inc/lang.h +48 -0
- data/ext/inc/threading.h +31 -0
- data/ext/index.c +6510 -0
- data/ext/index.h +964 -0
- data/ext/index.o +0 -0
- data/ext/lang.h +66 -0
- data/ext/libstemmer.c +92 -0
- data/ext/libstemmer.h +79 -0
- data/ext/libstemmer.o +0 -0
- data/ext/mempool.c +87 -0
- data/ext/mempool.h +35 -0
- data/ext/mempool.o +0 -0
- data/ext/modules.h +162 -0
- data/ext/multimapper.c +310 -0
- data/ext/multimapper.h +51 -0
- data/ext/multimapper.o +0 -0
- data/ext/posh.c +1006 -0
- data/ext/posh.h +1007 -0
- data/ext/posh.o +0 -0
- data/ext/priorityqueue.c +151 -0
- data/ext/priorityqueue.h +143 -0
- data/ext/priorityqueue.o +0 -0
- data/ext/q_boolean.c +1608 -0
- data/ext/q_boolean.o +0 -0
- data/ext/q_const_score.c +165 -0
- data/ext/q_const_score.o +0 -0
- data/ext/q_filtered_query.c +209 -0
- data/ext/q_filtered_query.o +0 -0
- data/ext/q_fuzzy.c +335 -0
- data/ext/q_fuzzy.o +0 -0
- data/ext/q_match_all.c +148 -0
- data/ext/q_match_all.o +0 -0
- data/ext/q_multi_term.c +677 -0
- data/ext/q_multi_term.o +0 -0
- data/ext/q_parser.c +2825 -0
- data/ext/q_parser.o +0 -0
- data/ext/q_phrase.c +1126 -0
- data/ext/q_phrase.o +0 -0
- data/ext/q_prefix.c +100 -0
- data/ext/q_prefix.o +0 -0
- data/ext/q_range.c +356 -0
- data/ext/q_range.o +0 -0
- data/ext/q_span.c +2402 -0
- data/ext/q_span.o +0 -0
- data/ext/q_term.c +337 -0
- data/ext/q_term.o +0 -0
- data/ext/q_wildcard.c +171 -0
- data/ext/q_wildcard.o +0 -0
- data/ext/r_analysis.c +2636 -0
- data/ext/r_analysis.o +0 -0
- data/ext/r_index.c +3509 -0
- data/ext/r_index.o +0 -0
- data/ext/r_qparser.c +585 -0
- data/ext/r_qparser.o +0 -0
- data/ext/r_search.c +4240 -0
- data/ext/r_search.o +0 -0
- data/ext/r_store.c +513 -0
- data/ext/r_store.o +0 -0
- data/ext/r_utils.c +963 -0
- data/ext/r_utils.o +0 -0
- data/ext/ram_store.c +471 -0
- data/ext/ram_store.o +0 -0
- data/ext/search.c +1743 -0
- data/ext/search.h +885 -0
- data/ext/search.o +0 -0
- data/ext/similarity.c +150 -0
- data/ext/similarity.h +82 -0
- data/ext/similarity.o +0 -0
- data/ext/sort.c +985 -0
- data/ext/sort.o +0 -0
- data/ext/stem_ISO_8859_1_danish.c +338 -0
- data/ext/stem_ISO_8859_1_danish.h +16 -0
- data/ext/stem_ISO_8859_1_danish.o +0 -0
- data/ext/stem_ISO_8859_1_dutch.c +635 -0
- data/ext/stem_ISO_8859_1_dutch.h +16 -0
- data/ext/stem_ISO_8859_1_dutch.o +0 -0
- data/ext/stem_ISO_8859_1_english.c +1156 -0
- data/ext/stem_ISO_8859_1_english.h +16 -0
- data/ext/stem_ISO_8859_1_english.o +0 -0
- data/ext/stem_ISO_8859_1_finnish.c +792 -0
- data/ext/stem_ISO_8859_1_finnish.h +16 -0
- data/ext/stem_ISO_8859_1_finnish.o +0 -0
- data/ext/stem_ISO_8859_1_french.c +1276 -0
- data/ext/stem_ISO_8859_1_french.h +16 -0
- data/ext/stem_ISO_8859_1_french.o +0 -0
- data/ext/stem_ISO_8859_1_german.c +512 -0
- data/ext/stem_ISO_8859_1_german.h +16 -0
- data/ext/stem_ISO_8859_1_german.o +0 -0
- data/ext/stem_ISO_8859_1_italian.c +1091 -0
- data/ext/stem_ISO_8859_1_italian.h +16 -0
- data/ext/stem_ISO_8859_1_italian.o +0 -0
- data/ext/stem_ISO_8859_1_norwegian.c +296 -0
- data/ext/stem_ISO_8859_1_norwegian.h +16 -0
- data/ext/stem_ISO_8859_1_norwegian.o +0 -0
- data/ext/stem_ISO_8859_1_porter.c +776 -0
- data/ext/stem_ISO_8859_1_porter.h +16 -0
- data/ext/stem_ISO_8859_1_porter.o +0 -0
- data/ext/stem_ISO_8859_1_portuguese.c +1035 -0
- data/ext/stem_ISO_8859_1_portuguese.h +16 -0
- data/ext/stem_ISO_8859_1_portuguese.o +0 -0
- data/ext/stem_ISO_8859_1_spanish.c +1119 -0
- data/ext/stem_ISO_8859_1_spanish.h +16 -0
- data/ext/stem_ISO_8859_1_spanish.o +0 -0
- data/ext/stem_ISO_8859_1_swedish.c +307 -0
- data/ext/stem_ISO_8859_1_swedish.h +16 -0
- data/ext/stem_ISO_8859_1_swedish.o +0 -0
- data/ext/stem_KOI8_R_russian.c +701 -0
- data/ext/stem_KOI8_R_russian.h +16 -0
- data/ext/stem_KOI8_R_russian.o +0 -0
- data/ext/stem_UTF_8_danish.c +344 -0
- data/ext/stem_UTF_8_danish.h +16 -0
- data/ext/stem_UTF_8_danish.o +0 -0
- data/ext/stem_UTF_8_dutch.c +653 -0
- data/ext/stem_UTF_8_dutch.h +16 -0
- data/ext/stem_UTF_8_dutch.o +0 -0
- data/ext/stem_UTF_8_english.c +1176 -0
- data/ext/stem_UTF_8_english.h +16 -0
- data/ext/stem_UTF_8_english.o +0 -0
- data/ext/stem_UTF_8_finnish.c +808 -0
- data/ext/stem_UTF_8_finnish.h +16 -0
- data/ext/stem_UTF_8_finnish.o +0 -0
- data/ext/stem_UTF_8_french.c +1296 -0
- data/ext/stem_UTF_8_french.h +16 -0
- data/ext/stem_UTF_8_french.o +0 -0
- data/ext/stem_UTF_8_german.c +526 -0
- data/ext/stem_UTF_8_german.h +16 -0
- data/ext/stem_UTF_8_german.o +0 -0
- data/ext/stem_UTF_8_italian.c +1113 -0
- data/ext/stem_UTF_8_italian.h +16 -0
- data/ext/stem_UTF_8_italian.o +0 -0
- data/ext/stem_UTF_8_norwegian.c +302 -0
- data/ext/stem_UTF_8_norwegian.h +16 -0
- data/ext/stem_UTF_8_norwegian.o +0 -0
- data/ext/stem_UTF_8_porter.c +794 -0
- data/ext/stem_UTF_8_porter.h +16 -0
- data/ext/stem_UTF_8_porter.o +0 -0
- data/ext/stem_UTF_8_portuguese.c +1055 -0
- data/ext/stem_UTF_8_portuguese.h +16 -0
- data/ext/stem_UTF_8_portuguese.o +0 -0
- data/ext/stem_UTF_8_russian.c +709 -0
- data/ext/stem_UTF_8_russian.h +16 -0
- data/ext/stem_UTF_8_russian.o +0 -0
- data/ext/stem_UTF_8_spanish.c +1137 -0
- data/ext/stem_UTF_8_spanish.h +16 -0
- data/ext/stem_UTF_8_spanish.o +0 -0
- data/ext/stem_UTF_8_swedish.c +313 -0
- data/ext/stem_UTF_8_swedish.h +16 -0
- data/ext/stem_UTF_8_swedish.o +0 -0
- data/ext/stopwords.c +401 -0
- data/ext/stopwords.o +0 -0
- data/ext/store.c +692 -0
- data/ext/store.h +777 -0
- data/ext/store.o +0 -0
- data/ext/term_vectors.c +352 -0
- data/ext/term_vectors.o +0 -0
- data/ext/threading.h +31 -0
- data/ext/utilities.c +446 -0
- data/ext/utilities.o +0 -0
- data/ext/win32.h +54 -0
- data/ferret.gemspec +39 -0
- data/lib/ferret.rb +29 -0
- data/lib/ferret/browser.rb +246 -0
- data/lib/ferret/browser/s/global.js +192 -0
- data/lib/ferret/browser/s/style.css +148 -0
- data/lib/ferret/browser/views/document/list.rhtml +49 -0
- data/lib/ferret/browser/views/document/show.rhtml +27 -0
- data/lib/ferret/browser/views/error/index.rhtml +7 -0
- data/lib/ferret/browser/views/help/index.rhtml +8 -0
- data/lib/ferret/browser/views/home/index.rhtml +29 -0
- data/lib/ferret/browser/views/layout.rhtml +22 -0
- data/lib/ferret/browser/views/term-vector/index.rhtml +4 -0
- data/lib/ferret/browser/views/term/index.rhtml +199 -0
- data/lib/ferret/browser/views/term/termdocs.rhtml +1 -0
- data/lib/ferret/browser/webrick.rb +14 -0
- data/lib/ferret/document.rb +130 -0
- data/lib/ferret/field_infos.rb +44 -0
- data/lib/ferret/index.rb +786 -0
- data/lib/ferret/number_tools.rb +157 -0
- data/lib/ferret_ext.bundle +0 -0
- data/lib/ferret_version.rb +3 -0
- data/pkg/ferret-0.11.6.gem +0 -0
- data/pkg/ferret-0.11.6.tgz +0 -0
- data/pkg/ferret-0.11.6.zip +0 -0
- data/setup.rb +1555 -0
- data/test/test_all.rb +5 -0
- data/test/test_helper.rb +24 -0
- data/test/threading/number_to_spoken.rb +132 -0
- data/test/threading/thread_safety_index_test.rb +79 -0
- data/test/threading/thread_safety_read_write_test.rb +76 -0
- data/test/threading/thread_safety_test.rb +133 -0
- data/test/unit/analysis/tc_analyzer.rb +548 -0
- data/test/unit/analysis/tc_token_stream.rb +646 -0
- data/test/unit/index/tc_index.rb +762 -0
- data/test/unit/index/tc_index_reader.rb +699 -0
- data/test/unit/index/tc_index_writer.rb +437 -0
- data/test/unit/index/th_doc.rb +315 -0
- data/test/unit/largefile/tc_largefile.rb +46 -0
- data/test/unit/query_parser/tc_query_parser.rb +238 -0
- data/test/unit/search/tc_filter.rb +135 -0
- data/test/unit/search/tc_fuzzy_query.rb +147 -0
- data/test/unit/search/tc_index_searcher.rb +61 -0
- data/test/unit/search/tc_multi_searcher.rb +128 -0
- data/test/unit/search/tc_multiple_search_requests.rb +58 -0
- data/test/unit/search/tc_search_and_sort.rb +179 -0
- data/test/unit/search/tc_sort.rb +49 -0
- data/test/unit/search/tc_sort_field.rb +27 -0
- data/test/unit/search/tc_spans.rb +190 -0
- data/test/unit/search/tm_searcher.rb +384 -0
- data/test/unit/store/tc_fs_store.rb +77 -0
- data/test/unit/store/tc_ram_store.rb +35 -0
- data/test/unit/store/tm_store.rb +34 -0
- data/test/unit/store/tm_store_lock.rb +68 -0
- data/test/unit/tc_document.rb +81 -0
- data/test/unit/ts_analysis.rb +2 -0
- data/test/unit/ts_index.rb +2 -0
- data/test/unit/ts_largefile.rb +4 -0
- data/test/unit/ts_query_parser.rb +2 -0
- data/test/unit/ts_search.rb +2 -0
- data/test/unit/ts_store.rb +2 -0
- data/test/unit/ts_utils.rb +2 -0
- data/test/unit/utils/tc_bit_vector.rb +295 -0
- data/test/unit/utils/tc_number_tools.rb +117 -0
- data/test/unit/utils/tc_priority_queue.rb +106 -0
- metadata +392 -0
data/ext/global.o
ADDED
Binary file
|
data/ext/hash.c
ADDED
@@ -0,0 +1,598 @@
|
|
1
|
+
#include "hash.h"
|
2
|
+
#include "global.h"
|
3
|
+
#include <string.h>
|
4
|
+
|
5
|
+
|
6
|
+
/****************************************************************************
|
7
|
+
*
|
8
|
+
* HashTable
|
9
|
+
*
|
10
|
+
* This hash table is modeled after Python's dictobject and a description of
|
11
|
+
* the algorithm can be found in the file dictobject.c in Python's src
|
12
|
+
****************************************************************************/
|
13
|
+
|
14
|
+
static char *dummy_key = "";
|
15
|
+
|
16
|
+
#define PERTURB_SHIFT 5
|
17
|
+
#define MAX_FREE_HASH_TABLES 80
|
18
|
+
|
19
|
+
#define max(a,b) ((b < a) ? a : b)
|
20
|
+
|
21
|
+
static HashTable *free_hts[MAX_FREE_HASH_TABLES];
|
22
|
+
static int num_free_hts = 0;
|
23
|
+
|
24
|
+
unsigned long str_hash(const char *const str)
|
25
|
+
{
|
26
|
+
register unsigned long h = 0;
|
27
|
+
register unsigned char *p = (unsigned char *) str;
|
28
|
+
|
29
|
+
for (; *p; p++) {
|
30
|
+
h = 37 * h + *p;
|
31
|
+
}
|
32
|
+
|
33
|
+
return h;
|
34
|
+
}
|
35
|
+
|
36
|
+
unsigned long ptr_hash(const void *const ptr)
|
37
|
+
{
|
38
|
+
return (unsigned long)ptr;
|
39
|
+
}
|
40
|
+
|
41
|
+
int ptr_eq(const void *q1, const void *q2)
|
42
|
+
{
|
43
|
+
return q1 == q2;
|
44
|
+
}
|
45
|
+
|
46
|
+
static int int_eq(const void *q1, const void *q2)
|
47
|
+
{
|
48
|
+
(void)q1;
|
49
|
+
(void)q2;
|
50
|
+
return true;
|
51
|
+
}
|
52
|
+
|
53
|
+
static unsigned long int_hash(const void *i)
|
54
|
+
{
|
55
|
+
return *((unsigned long *)i);
|
56
|
+
}
|
57
|
+
|
58
|
+
typedef HashEntry *(*lookup_ft)(struct HashTable *ht, register const void *key);
|
59
|
+
|
60
|
+
/**
|
61
|
+
* Fast lookup function for resizing as we know there are no equal elements or
|
62
|
+
* deletes to worry about.
|
63
|
+
*
|
64
|
+
* @param ht the HashTable to do the fast lookup in
|
65
|
+
* @param the hashkey we are looking for
|
66
|
+
*/
|
67
|
+
static INLINE HashEntry *h_resize_lookup(HashTable *ht,
|
68
|
+
register const unsigned long hash)
|
69
|
+
{
|
70
|
+
register unsigned long perturb;
|
71
|
+
register int mask = ht->mask;
|
72
|
+
register HashEntry *he0 = ht->table;
|
73
|
+
register int i = hash & mask;
|
74
|
+
register HashEntry *he = &he0[i];
|
75
|
+
|
76
|
+
if (he->key == NULL) {
|
77
|
+
he->hash = hash;
|
78
|
+
return he;
|
79
|
+
}
|
80
|
+
|
81
|
+
for (perturb = hash;; perturb >>= PERTURB_SHIFT) {
|
82
|
+
i = (i << 2) + i + perturb + 1;
|
83
|
+
he = &he0[i & mask];
|
84
|
+
if (he->key == NULL) {
|
85
|
+
he->hash = hash;
|
86
|
+
return he;
|
87
|
+
}
|
88
|
+
}
|
89
|
+
}
|
90
|
+
|
91
|
+
HashEntry *h_lookup_int(HashTable *ht, const void *key)
|
92
|
+
{
|
93
|
+
register unsigned long hash = *((int *)key);
|
94
|
+
register unsigned long perturb;
|
95
|
+
register int mask = ht->mask;
|
96
|
+
register HashEntry *he0 = ht->table;
|
97
|
+
register int i = hash & mask;
|
98
|
+
register HashEntry *he = &he0[i];
|
99
|
+
register HashEntry *freeslot = NULL;
|
100
|
+
|
101
|
+
if (he->key == NULL || he->hash == hash) {
|
102
|
+
he->hash = hash;
|
103
|
+
return he;
|
104
|
+
}
|
105
|
+
if (he->key == dummy_key) {
|
106
|
+
freeslot = he;
|
107
|
+
}
|
108
|
+
|
109
|
+
for (perturb = hash;; perturb >>= PERTURB_SHIFT) {
|
110
|
+
i = (i << 2) + i + perturb + 1;
|
111
|
+
he = &he0[i & mask];
|
112
|
+
if (he->key == NULL) {
|
113
|
+
if (freeslot != NULL) {
|
114
|
+
he = freeslot;
|
115
|
+
}
|
116
|
+
he->hash = hash;
|
117
|
+
return he;
|
118
|
+
}
|
119
|
+
if (he->hash == hash) {
|
120
|
+
return he;
|
121
|
+
}
|
122
|
+
if (he->key == dummy_key && freeslot == NULL) {
|
123
|
+
freeslot = he;
|
124
|
+
}
|
125
|
+
}
|
126
|
+
}
|
127
|
+
|
128
|
+
HashEntry *h_lookup_str(HashTable *ht, register const char *key)
|
129
|
+
{
|
130
|
+
register unsigned long hash = str_hash(key);
|
131
|
+
register unsigned long perturb;
|
132
|
+
register int mask = ht->mask;
|
133
|
+
register HashEntry *he0 = ht->table;
|
134
|
+
register int i = hash & mask;
|
135
|
+
register HashEntry *he = &he0[i];
|
136
|
+
register HashEntry *freeslot;
|
137
|
+
|
138
|
+
if (he->key == NULL || he->key == key) {
|
139
|
+
he->hash = hash;
|
140
|
+
return he;
|
141
|
+
}
|
142
|
+
if (he->key == dummy_key) {
|
143
|
+
freeslot = he;
|
144
|
+
}
|
145
|
+
else {
|
146
|
+
if ((he->hash == hash) && (strcmp(he->key, key) == 0)) {
|
147
|
+
return he;
|
148
|
+
}
|
149
|
+
freeslot = NULL;
|
150
|
+
}
|
151
|
+
|
152
|
+
for (perturb = hash;; perturb >>= PERTURB_SHIFT) {
|
153
|
+
i = (i << 2) + i + perturb + 1;
|
154
|
+
he = &he0[i & mask];
|
155
|
+
if (he->key == NULL) {
|
156
|
+
if (freeslot != NULL) {
|
157
|
+
he = freeslot;
|
158
|
+
}
|
159
|
+
he->hash = hash;
|
160
|
+
return he;
|
161
|
+
}
|
162
|
+
if (he->key == key
|
163
|
+
|| (he->hash == hash
|
164
|
+
&& he->key != dummy_key && strcmp(he->key, key) == 0)) {
|
165
|
+
return he;
|
166
|
+
}
|
167
|
+
if (he->key == dummy_key && freeslot == NULL) {
|
168
|
+
freeslot = he;
|
169
|
+
}
|
170
|
+
}
|
171
|
+
}
|
172
|
+
|
173
|
+
HashEntry *h_lookup(HashTable *ht, register const void *key)
|
174
|
+
{
|
175
|
+
register unsigned int hash = ht->hash_i(key);
|
176
|
+
register unsigned int perturb;
|
177
|
+
register int mask = ht->mask;
|
178
|
+
register HashEntry *he0 = ht->table;
|
179
|
+
register int i = hash & mask;
|
180
|
+
register HashEntry *he = &he0[i];
|
181
|
+
register HashEntry *freeslot;
|
182
|
+
eq_ft eq = ht->eq_i;
|
183
|
+
|
184
|
+
if (he->key == NULL || he->key == key) {
|
185
|
+
he->hash = hash;
|
186
|
+
return he;
|
187
|
+
}
|
188
|
+
if (he->key == dummy_key) {
|
189
|
+
freeslot = he;
|
190
|
+
}
|
191
|
+
else {
|
192
|
+
if ((he->hash == hash) && eq(he->key, key)) {
|
193
|
+
return he;
|
194
|
+
}
|
195
|
+
freeslot = NULL;
|
196
|
+
}
|
197
|
+
|
198
|
+
for (perturb = hash;; perturb >>= PERTURB_SHIFT) {
|
199
|
+
i = (i << 2) + i + perturb + 1;
|
200
|
+
he = &he0[i & mask];
|
201
|
+
if (he->key == NULL) {
|
202
|
+
if (freeslot != NULL) {
|
203
|
+
he = freeslot;
|
204
|
+
}
|
205
|
+
he->hash = hash;
|
206
|
+
return he;
|
207
|
+
}
|
208
|
+
if (he->key == key
|
209
|
+
|| (he->hash == hash
|
210
|
+
&& he->key != dummy_key && eq(he->key, key))) {
|
211
|
+
return he;
|
212
|
+
}
|
213
|
+
if (he->key == dummy_key && freeslot == NULL) {
|
214
|
+
freeslot = he;
|
215
|
+
}
|
216
|
+
}
|
217
|
+
}
|
218
|
+
|
219
|
+
HashTable *h_new_str(free_ft free_key, free_ft free_value)
|
220
|
+
{
|
221
|
+
HashTable *ht;
|
222
|
+
if (num_free_hts > 0) {
|
223
|
+
ht = free_hts[--num_free_hts];
|
224
|
+
// ht->max_mask is preserved...
|
225
|
+
// ht->table is preserved...
|
226
|
+
}
|
227
|
+
else {
|
228
|
+
ht = ALLOC(HashTable);
|
229
|
+
ht->max_mask = HASH_MINSIZE - 1;
|
230
|
+
ht->table = ht->smalltable;
|
231
|
+
}
|
232
|
+
ht->fill = 0;
|
233
|
+
ht->size = 0;
|
234
|
+
ht->mask = HASH_MINSIZE - 1;
|
235
|
+
// memset(ht->smalltable, 0, sizeof(ht->smalltable));
|
236
|
+
memset(ht->table, 0, sizeof(HashEntry)*(ht->mask + 1));
|
237
|
+
ht->lookup_i = (lookup_ft)&h_lookup_str;
|
238
|
+
|
239
|
+
ht->free_key_i = free_key != NULL ? free_key : &dummy_free;
|
240
|
+
ht->free_value_i = free_value != NULL ? free_value : &dummy_free;
|
241
|
+
ht->ref_cnt = 1;
|
242
|
+
return ht;
|
243
|
+
}
|
244
|
+
|
245
|
+
HashTable *h_new_int(free_ft free_value)
|
246
|
+
{
|
247
|
+
HashTable *ht = h_new_str(NULL, free_value);
|
248
|
+
ht->lookup_i = &h_lookup_int;
|
249
|
+
ht->eq_i = int_eq;
|
250
|
+
ht->hash_i = int_hash;
|
251
|
+
return ht;
|
252
|
+
}
|
253
|
+
|
254
|
+
HashTable *h_new(hash_ft hash, eq_ft eq, free_ft free_key, free_ft free_value)
|
255
|
+
{
|
256
|
+
HashTable *ht = h_new_str(free_key, free_value);
|
257
|
+
|
258
|
+
ht->lookup_i = &h_lookup;
|
259
|
+
ht->eq_i = eq;
|
260
|
+
ht->hash_i = hash;
|
261
|
+
return ht;
|
262
|
+
}
|
263
|
+
|
264
|
+
void h_clear(HashTable *ht)
|
265
|
+
{
|
266
|
+
int i;
|
267
|
+
HashEntry *he;
|
268
|
+
free_ft free_key = ht->free_key_i;
|
269
|
+
free_ft free_value = ht->free_value_i;
|
270
|
+
|
271
|
+
/* Clear all the hash values and keys as necessary */
|
272
|
+
if (free_key != dummy_free || free_value != dummy_free) {
|
273
|
+
for (i = 0; i <= ht->mask; i++) {
|
274
|
+
he = &ht->table[i];
|
275
|
+
if (he->key != NULL && he->key != dummy_key) {
|
276
|
+
free_value(he->value);
|
277
|
+
free_key(he->key);
|
278
|
+
}
|
279
|
+
he->key = NULL;
|
280
|
+
}
|
281
|
+
}
|
282
|
+
// Set cur_mask small but leave max_mask big... So that
|
283
|
+
// we don't end up having to scan/compress the entire space
|
284
|
+
// even when we're working with a small set of values.
|
285
|
+
ht->mask = HASH_MINSIZE - 1;
|
286
|
+
|
287
|
+
ZEROSET_N(ht->table, HashEntry, ht->mask + 1);
|
288
|
+
ht->size = 0;
|
289
|
+
ht->fill = 0;
|
290
|
+
}
|
291
|
+
|
292
|
+
void h_destroy(HashTable *ht)
|
293
|
+
{
|
294
|
+
if (--(ht->ref_cnt) <= 0) {
|
295
|
+
h_clear(ht);
|
296
|
+
|
297
|
+
#ifdef DEBUG
|
298
|
+
/* if a new table was created, be sure to free it */
|
299
|
+
if (ht->table != ht->smalltable) {
|
300
|
+
free(ht->table);
|
301
|
+
}
|
302
|
+
free(ht);
|
303
|
+
#else
|
304
|
+
if (num_free_hts < MAX_FREE_HASH_TABLES) {
|
305
|
+
free_hts[num_free_hts++] = ht;
|
306
|
+
}
|
307
|
+
else {
|
308
|
+
/* if a new table was created, be sure to free it */
|
309
|
+
if (ht->table != ht->smalltable) {
|
310
|
+
free(ht->table);
|
311
|
+
}
|
312
|
+
free(ht);
|
313
|
+
}
|
314
|
+
#endif
|
315
|
+
}
|
316
|
+
}
|
317
|
+
|
318
|
+
void *h_get(HashTable *ht, const void *key)
|
319
|
+
{
|
320
|
+
/* Note: lookup_i will never return NULL. */
|
321
|
+
return ht->lookup_i(ht, key)->value;
|
322
|
+
}
|
323
|
+
|
324
|
+
int h_del(HashTable *ht, const void *key)
|
325
|
+
{
|
326
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
327
|
+
|
328
|
+
if (he->key != NULL && he->key != dummy_key) {
|
329
|
+
ht->free_key_i(he->key);
|
330
|
+
ht->free_value_i(he->value);
|
331
|
+
he->key = dummy_key;
|
332
|
+
he->value = NULL;
|
333
|
+
ht->size--;
|
334
|
+
return true;
|
335
|
+
}
|
336
|
+
else {
|
337
|
+
return false;
|
338
|
+
}
|
339
|
+
}
|
340
|
+
|
341
|
+
void *h_rem(HashTable *ht, const void *key, bool destroy_key)
|
342
|
+
{
|
343
|
+
void *val;
|
344
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
345
|
+
|
346
|
+
if (he->key != NULL && he->key != dummy_key) {
|
347
|
+
if (destroy_key) {
|
348
|
+
ht->free_key_i(he->key);
|
349
|
+
}
|
350
|
+
|
351
|
+
he->key = dummy_key;
|
352
|
+
val = he->value;
|
353
|
+
he->value = NULL;
|
354
|
+
ht->size--;
|
355
|
+
return val;
|
356
|
+
}
|
357
|
+
else {
|
358
|
+
return NULL;
|
359
|
+
}
|
360
|
+
}
|
361
|
+
|
362
|
+
static int h_resize(HashTable *ht, int min_newsize)
|
363
|
+
{
|
364
|
+
HashEntry smallcopy[HASH_MINSIZE];
|
365
|
+
HashEntry *oldtable;
|
366
|
+
HashEntry *he_old, *he_new;
|
367
|
+
int newsize, num_active;
|
368
|
+
bool copied = false;
|
369
|
+
|
370
|
+
/* newsize will be a power of two */
|
371
|
+
for (newsize = HASH_MINSIZE; newsize < min_newsize; newsize <<= 1) {
|
372
|
+
}
|
373
|
+
|
374
|
+
oldtable = ht->table;
|
375
|
+
if (newsize == HASH_MINSIZE) {
|
376
|
+
if (ht->table == ht->smalltable) {
|
377
|
+
/* need to copy the data out so we can rebuild the table into
|
378
|
+
* the same space */
|
379
|
+
memcpy(smallcopy, ht->smalltable, sizeof(smallcopy));
|
380
|
+
oldtable = smallcopy;
|
381
|
+
}
|
382
|
+
else {
|
383
|
+
ht->table = ht->smalltable;
|
384
|
+
}
|
385
|
+
}
|
386
|
+
|
387
|
+
//
|
388
|
+
// TODO: Figure out why this can cause segmentation faults. :(
|
389
|
+
//
|
390
|
+
// else if (newsize + ht->mask <= ht->max_mask) {
|
391
|
+
// // Instead of re-allocating space every time, try to re-use the existing
|
392
|
+
// // space if we can. Requires that both the old and new tables fit in
|
393
|
+
// // allocated space, though.
|
394
|
+
// // printf("Copying... From %d to %d, %d entries / %d bytes (%d each).\n", ht->table, &ht->table[newsize], (ht->mask + 1), sizeof(HashEntry)*(ht->mask + 1), sizeof(HashEntry));
|
395
|
+
// memcpy(&ht->table[newsize], ht->table, sizeof(HashEntry) * (ht->mask + 1));
|
396
|
+
// oldtable = &ht->table[newsize];
|
397
|
+
// copied = true;
|
398
|
+
// }
|
399
|
+
|
400
|
+
else {
|
401
|
+
ht->table = ALLOC_N(HashEntry, newsize);
|
402
|
+
ht->max_mask = newsize - 1;
|
403
|
+
}
|
404
|
+
memset(ht->table, 0, sizeof(HashEntry) * newsize);
|
405
|
+
ht->fill = ht->size;
|
406
|
+
ht->mask = newsize - 1;
|
407
|
+
|
408
|
+
for (num_active = ht->size, he_old = oldtable; num_active > 0; he_old++) {
|
409
|
+
if (he_old->key && he_old->key != dummy_key) { /* active entry */
|
410
|
+
/*he_new = ht->lookup_i(ht, he_old->key); */
|
411
|
+
he_new = h_resize_lookup(ht, he_old->hash);
|
412
|
+
he_new->key = he_old->key;
|
413
|
+
he_new->value = he_old->value;
|
414
|
+
num_active--;
|
415
|
+
} /* else empty entry so nothing to do */
|
416
|
+
}
|
417
|
+
if (oldtable != smallcopy && oldtable != ht->smalltable && !copied) {
|
418
|
+
free(oldtable);
|
419
|
+
}
|
420
|
+
return 0;
|
421
|
+
}
|
422
|
+
|
423
|
+
int h_set(HashTable *ht, const void *key, void *value)
|
424
|
+
{
|
425
|
+
int ret_val = HASH_KEY_DOES_NOT_EXIST;
|
426
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
427
|
+
if (he->key == NULL) {
|
428
|
+
if (ht->fill * 3 > ht->mask * 2) {
|
429
|
+
h_resize(ht, ht->size * ((ht->size > SLOW_DOWN) ? 4 : 2));
|
430
|
+
he = ht->lookup_i(ht, key);
|
431
|
+
}
|
432
|
+
ht->fill++;
|
433
|
+
ht->size++;
|
434
|
+
}
|
435
|
+
else if (he->key == dummy_key) {
|
436
|
+
ht->size++;
|
437
|
+
}
|
438
|
+
else if (he->key != key) {
|
439
|
+
ht->free_key_i(he->key);
|
440
|
+
if (he->value != value) {
|
441
|
+
ht->free_value_i(he->value);
|
442
|
+
}
|
443
|
+
ret_val = HASH_KEY_EQUAL;
|
444
|
+
}
|
445
|
+
else {
|
446
|
+
/* safety check. Only free old value if it isn't the new value */
|
447
|
+
if (he->value != value) {
|
448
|
+
ht->free_value_i(he->value);
|
449
|
+
}
|
450
|
+
ret_val = HASH_KEY_SAME;
|
451
|
+
}
|
452
|
+
he->key = (void *)key;
|
453
|
+
he->value = value;
|
454
|
+
|
455
|
+
/*
|
456
|
+
if ((ht->fill > fill) && (ht->fill * 3 > ht->mask * 2)) {
|
457
|
+
h_resize(ht, ht->size * ((ht->size > SLOW_DOWN) ? 4 : 2));
|
458
|
+
}
|
459
|
+
*/
|
460
|
+
return ret_val;
|
461
|
+
}
|
462
|
+
|
463
|
+
HashEntry *h_set_ext(HashTable *ht, const void *key)
|
464
|
+
{
|
465
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
466
|
+
if (he->key == NULL) {
|
467
|
+
if (ht->fill * 3 > ht->mask * 2) {
|
468
|
+
h_resize(ht, ht->size * ((ht->size > SLOW_DOWN) ? 4 : 2));
|
469
|
+
he = ht->lookup_i(ht, key);
|
470
|
+
}
|
471
|
+
ht->fill++;
|
472
|
+
ht->size++;
|
473
|
+
}
|
474
|
+
else if (he->key == dummy_key) {
|
475
|
+
ht->size++;
|
476
|
+
}
|
477
|
+
|
478
|
+
return he;
|
479
|
+
}
|
480
|
+
|
481
|
+
int h_set_safe(HashTable *ht, const void *key, void *value)
|
482
|
+
{
|
483
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
484
|
+
int fill = ht->fill;
|
485
|
+
if (he->key == NULL) {
|
486
|
+
ht->fill++;
|
487
|
+
ht->size++;
|
488
|
+
}
|
489
|
+
else if (he->key == dummy_key) {
|
490
|
+
ht->size++;
|
491
|
+
}
|
492
|
+
else {
|
493
|
+
return false;
|
494
|
+
}
|
495
|
+
he->key = (void *)key;
|
496
|
+
he->value = value;
|
497
|
+
|
498
|
+
if ((ht->fill > fill) && (ht->fill * 3 > ht->mask * 2)) {
|
499
|
+
h_resize(ht, ht->size * ((ht->size > SLOW_DOWN) ? 4 : 2));
|
500
|
+
}
|
501
|
+
return true;
|
502
|
+
}
|
503
|
+
|
504
|
+
int h_has_key(HashTable *ht, const void *key)
|
505
|
+
{
|
506
|
+
HashEntry *he = ht->lookup_i(ht, key);
|
507
|
+
if (he->key == NULL || he->key == dummy_key) {
|
508
|
+
return HASH_KEY_DOES_NOT_EXIST;
|
509
|
+
}
|
510
|
+
else if (he->key == key) {
|
511
|
+
return HASH_KEY_SAME;
|
512
|
+
}
|
513
|
+
else {
|
514
|
+
return HASH_KEY_EQUAL;
|
515
|
+
}
|
516
|
+
}
|
517
|
+
|
518
|
+
void *h_get_int(HashTable *self, const unsigned long key)
|
519
|
+
{
|
520
|
+
return h_get(self, &key);
|
521
|
+
}
|
522
|
+
|
523
|
+
int h_del_int(HashTable *self, const unsigned long key)
|
524
|
+
{
|
525
|
+
return h_del(self, &key);
|
526
|
+
}
|
527
|
+
|
528
|
+
void *h_rem_int(HashTable *self, const unsigned long key)
|
529
|
+
{
|
530
|
+
return h_rem(self, &key, false);
|
531
|
+
}
|
532
|
+
|
533
|
+
int h_set_int(HashTable *self, const unsigned long key, void *value)
|
534
|
+
{
|
535
|
+
return h_set(self, &key, value);
|
536
|
+
}
|
537
|
+
|
538
|
+
int h_set_safe_int(HashTable *self, const unsigned long key, void *value)
|
539
|
+
{
|
540
|
+
return h_set_safe(self, &key, value);
|
541
|
+
}
|
542
|
+
|
543
|
+
int h_has_key_int(HashTable *self, const unsigned long key)
|
544
|
+
{
|
545
|
+
return h_has_key(self, &key);
|
546
|
+
}
|
547
|
+
|
548
|
+
void h_each(HashTable *ht,
|
549
|
+
void (*each_kv) (void *key, void *value, void *arg), void *arg)
|
550
|
+
{
|
551
|
+
HashEntry *he;
|
552
|
+
int i = ht->size;
|
553
|
+
for (he = ht->table; i > 0; he++) {
|
554
|
+
if (he->key && he->key != dummy_key) { /* active entry */
|
555
|
+
each_kv(he->key, he->value, arg);
|
556
|
+
i--;
|
557
|
+
}
|
558
|
+
}
|
559
|
+
}
|
560
|
+
|
561
|
+
HashTable *h_clone(HashTable *ht,
|
562
|
+
h_clone_func_t clone_key, h_clone_func_t clone_value)
|
563
|
+
{
|
564
|
+
void *key, *value;
|
565
|
+
HashEntry *he;
|
566
|
+
int i = ht->size;
|
567
|
+
HashTable *ht_clone;
|
568
|
+
|
569
|
+
if (ht->lookup_i == (lookup_ft)&h_lookup_str) {
|
570
|
+
ht_clone = h_new_str(ht->free_key_i, ht->free_value_i);
|
571
|
+
}
|
572
|
+
else {
|
573
|
+
ht_clone = h_new(ht->hash_i, ht->eq_i, ht->free_key_i, ht->free_value_i);
|
574
|
+
}
|
575
|
+
|
576
|
+
for (he = ht->table; i > 0; he++) {
|
577
|
+
if (he->key && he->key != dummy_key) { /* active entry */
|
578
|
+
key = clone_key ? clone_key(he->key) : he->key;
|
579
|
+
value = clone_value ? clone_value(he->value) : he->value;
|
580
|
+
h_set(ht_clone, key, value);
|
581
|
+
i--;
|
582
|
+
}
|
583
|
+
}
|
584
|
+
return ht_clone;
|
585
|
+
}
|
586
|
+
|
587
|
+
void h_str_print_keys(HashTable *ht)
|
588
|
+
{
|
589
|
+
HashEntry *he;
|
590
|
+
int i = ht->size;
|
591
|
+
printf("keys:\n");
|
592
|
+
for (he = ht->table; i > 0; he++) {
|
593
|
+
if (he->key && he->key != dummy_key) { /* active entry */
|
594
|
+
printf("\t%s\n", (char *)he->key);
|
595
|
+
i--;
|
596
|
+
}
|
597
|
+
}
|
598
|
+
}
|