jk-ferret 0.11.8.2
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 +90 -0
- data/RELEASE_CHANGES +137 -0
- data/RELEASE_NOTES +60 -0
- data/Rakefile +443 -0
- data/TODO +109 -0
- data/TUTORIAL +231 -0
- data/bin/ferret-browser +79 -0
- data/ext/BZLIB_blocksort.c +1094 -0
- data/ext/BZLIB_bzlib.c +1578 -0
- data/ext/BZLIB_compress.c +672 -0
- data/ext/BZLIB_crctable.c +104 -0
- data/ext/BZLIB_decompress.c +626 -0
- data/ext/BZLIB_huffman.c +205 -0
- data/ext/BZLIB_randtable.c +84 -0
- data/ext/STEMMER_api.c +66 -0
- data/ext/STEMMER_libstemmer.c +93 -0
- data/ext/STEMMER_stem_ISO_8859_1_danish.c +337 -0
- data/ext/STEMMER_stem_ISO_8859_1_dutch.c +624 -0
- data/ext/STEMMER_stem_ISO_8859_1_english.c +1117 -0
- data/ext/STEMMER_stem_ISO_8859_1_finnish.c +762 -0
- data/ext/STEMMER_stem_ISO_8859_1_french.c +1246 -0
- data/ext/STEMMER_stem_ISO_8859_1_german.c +503 -0
- data/ext/STEMMER_stem_ISO_8859_1_hungarian.c +1230 -0
- data/ext/STEMMER_stem_ISO_8859_1_italian.c +1065 -0
- data/ext/STEMMER_stem_ISO_8859_1_norwegian.c +297 -0
- data/ext/STEMMER_stem_ISO_8859_1_porter.c +749 -0
- data/ext/STEMMER_stem_ISO_8859_1_portuguese.c +1017 -0
- data/ext/STEMMER_stem_ISO_8859_1_spanish.c +1093 -0
- data/ext/STEMMER_stem_ISO_8859_1_swedish.c +307 -0
- data/ext/STEMMER_stem_ISO_8859_2_romanian.c +998 -0
- data/ext/STEMMER_stem_KOI8_R_russian.c +700 -0
- data/ext/STEMMER_stem_UTF_8_danish.c +339 -0
- data/ext/STEMMER_stem_UTF_8_dutch.c +634 -0
- data/ext/STEMMER_stem_UTF_8_english.c +1125 -0
- data/ext/STEMMER_stem_UTF_8_finnish.c +768 -0
- data/ext/STEMMER_stem_UTF_8_french.c +1256 -0
- data/ext/STEMMER_stem_UTF_8_german.c +509 -0
- data/ext/STEMMER_stem_UTF_8_hungarian.c +1234 -0
- data/ext/STEMMER_stem_UTF_8_italian.c +1073 -0
- data/ext/STEMMER_stem_UTF_8_norwegian.c +299 -0
- data/ext/STEMMER_stem_UTF_8_porter.c +755 -0
- data/ext/STEMMER_stem_UTF_8_portuguese.c +1023 -0
- data/ext/STEMMER_stem_UTF_8_romanian.c +1004 -0
- data/ext/STEMMER_stem_UTF_8_russian.c +694 -0
- data/ext/STEMMER_stem_UTF_8_spanish.c +1097 -0
- data/ext/STEMMER_stem_UTF_8_swedish.c +309 -0
- data/ext/STEMMER_stem_UTF_8_turkish.c +2205 -0
- data/ext/STEMMER_utilities.c +478 -0
- data/ext/analysis.c +1710 -0
- data/ext/analysis.h +266 -0
- data/ext/api.h +26 -0
- data/ext/array.c +125 -0
- data/ext/array.h +62 -0
- data/ext/bitvector.c +96 -0
- data/ext/bitvector.h +594 -0
- data/ext/bzlib.h +282 -0
- data/ext/bzlib_private.h +503 -0
- data/ext/compound_io.c +384 -0
- data/ext/config.h +52 -0
- data/ext/document.c +159 -0
- data/ext/document.h +63 -0
- data/ext/except.c +102 -0
- data/ext/except.h +176 -0
- data/ext/extconf.rb +15 -0
- data/ext/ferret.c +416 -0
- data/ext/ferret.h +94 -0
- data/ext/field_index.c +262 -0
- data/ext/field_index.h +52 -0
- data/ext/filter.c +157 -0
- data/ext/fs_store.c +493 -0
- data/ext/global.c +458 -0
- data/ext/global.h +302 -0
- data/ext/hash.c +524 -0
- data/ext/hash.h +515 -0
- data/ext/hashset.c +192 -0
- data/ext/hashset.h +215 -0
- data/ext/header.h +58 -0
- data/ext/helper.c +63 -0
- data/ext/helper.h +21 -0
- data/ext/index.c +6804 -0
- data/ext/index.h +935 -0
- data/ext/internal.h +1019 -0
- data/ext/lang.c +10 -0
- data/ext/lang.h +68 -0
- data/ext/libstemmer.h +79 -0
- data/ext/mempool.c +88 -0
- data/ext/mempool.h +43 -0
- data/ext/modules.h +190 -0
- data/ext/multimapper.c +351 -0
- data/ext/multimapper.h +60 -0
- data/ext/posh.c +1006 -0
- data/ext/posh.h +973 -0
- data/ext/priorityqueue.c +149 -0
- data/ext/priorityqueue.h +155 -0
- data/ext/q_boolean.c +1621 -0
- data/ext/q_const_score.c +162 -0
- data/ext/q_filtered_query.c +212 -0
- data/ext/q_fuzzy.c +280 -0
- data/ext/q_match_all.c +149 -0
- data/ext/q_multi_term.c +673 -0
- data/ext/q_parser.c +3103 -0
- data/ext/q_phrase.c +1206 -0
- data/ext/q_prefix.c +98 -0
- data/ext/q_range.c +682 -0
- data/ext/q_span.c +2390 -0
- data/ext/q_term.c +337 -0
- data/ext/q_wildcard.c +167 -0
- data/ext/r_analysis.c +2626 -0
- data/ext/r_index.c +3468 -0
- data/ext/r_qparser.c +635 -0
- data/ext/r_search.c +4490 -0
- data/ext/r_store.c +513 -0
- data/ext/r_utils.c +1131 -0
- data/ext/ram_store.c +476 -0
- data/ext/scanner.c +895 -0
- data/ext/scanner.h +36 -0
- data/ext/scanner_mb.c +6701 -0
- data/ext/scanner_utf8.c +4415 -0
- data/ext/search.c +1864 -0
- data/ext/search.h +953 -0
- data/ext/similarity.c +151 -0
- data/ext/similarity.h +89 -0
- data/ext/sort.c +786 -0
- data/ext/stem_ISO_8859_1_danish.h +16 -0
- data/ext/stem_ISO_8859_1_dutch.h +16 -0
- data/ext/stem_ISO_8859_1_english.h +16 -0
- data/ext/stem_ISO_8859_1_finnish.h +16 -0
- data/ext/stem_ISO_8859_1_french.h +16 -0
- data/ext/stem_ISO_8859_1_german.h +16 -0
- data/ext/stem_ISO_8859_1_hungarian.h +16 -0
- data/ext/stem_ISO_8859_1_italian.h +16 -0
- data/ext/stem_ISO_8859_1_norwegian.h +16 -0
- data/ext/stem_ISO_8859_1_porter.h +16 -0
- data/ext/stem_ISO_8859_1_portuguese.h +16 -0
- data/ext/stem_ISO_8859_1_spanish.h +16 -0
- data/ext/stem_ISO_8859_1_swedish.h +16 -0
- data/ext/stem_ISO_8859_2_romanian.h +16 -0
- data/ext/stem_KOI8_R_russian.h +16 -0
- data/ext/stem_UTF_8_danish.h +16 -0
- data/ext/stem_UTF_8_dutch.h +16 -0
- data/ext/stem_UTF_8_english.h +16 -0
- data/ext/stem_UTF_8_finnish.h +16 -0
- data/ext/stem_UTF_8_french.h +16 -0
- data/ext/stem_UTF_8_german.h +16 -0
- data/ext/stem_UTF_8_hungarian.h +16 -0
- data/ext/stem_UTF_8_italian.h +16 -0
- data/ext/stem_UTF_8_norwegian.h +16 -0
- data/ext/stem_UTF_8_porter.h +16 -0
- data/ext/stem_UTF_8_portuguese.h +16 -0
- data/ext/stem_UTF_8_romanian.h +16 -0
- data/ext/stem_UTF_8_russian.h +16 -0
- data/ext/stem_UTF_8_spanish.h +16 -0
- data/ext/stem_UTF_8_swedish.h +16 -0
- data/ext/stem_UTF_8_turkish.h +16 -0
- data/ext/stopwords.c +410 -0
- data/ext/store.c +698 -0
- data/ext/store.h +799 -0
- data/ext/symbol.c +10 -0
- data/ext/symbol.h +23 -0
- data/ext/term_vectors.c +73 -0
- data/ext/threading.h +31 -0
- data/ext/win32.h +62 -0
- data/lib/ferret.rb +30 -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/field_symbol.rb +87 -0
- data/lib/ferret/index.rb +973 -0
- data/lib/ferret/number_tools.rb +157 -0
- data/lib/ferret/version.rb +3 -0
- data/setup.rb +1555 -0
- data/test/long_running/largefile/tc_largefile.rb +46 -0
- data/test/test_all.rb +5 -0
- data/test/test_helper.rb +29 -0
- data/test/test_installed.rb +1 -0
- data/test/threading/number_to_spoken.rb +132 -0
- data/test/threading/thread_safety_index_test.rb +88 -0
- data/test/threading/thread_safety_read_write_test.rb +73 -0
- data/test/threading/thread_safety_test.rb +133 -0
- data/test/unit/analysis/tc_analyzer.rb +550 -0
- data/test/unit/analysis/tc_token_stream.rb +653 -0
- data/test/unit/index/tc_index.rb +867 -0
- data/test/unit/index/tc_index_reader.rb +699 -0
- data/test/unit/index/tc_index_writer.rb +447 -0
- data/test/unit/index/th_doc.rb +332 -0
- data/test/unit/query_parser/tc_query_parser.rb +238 -0
- data/test/unit/search/tc_filter.rb +156 -0
- data/test/unit/search/tc_fuzzy_query.rb +147 -0
- data/test/unit/search/tc_index_searcher.rb +67 -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 +436 -0
- data/test/unit/store/tc_fs_store.rb +115 -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/tc_field_symbol.rb +26 -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
- data/test/utils/content_generator.rb +226 -0
- metadata +319 -0
data/ext/hashset.c
ADDED
@@ -0,0 +1,192 @@
|
|
1
|
+
#include "hashset.h"
|
2
|
+
#include <string.h>
|
3
|
+
#include "internal.h"
|
4
|
+
|
5
|
+
/*
|
6
|
+
* The HashSet contains an array +elems+ of the elements that have been added.
|
7
|
+
* It always has +size+ elements so +size+ ane +elems+ can be used to iterate
|
8
|
+
* over all alements in the HashSet. It also uses a Hash to keep track of
|
9
|
+
* which elements have been added and their index in the +elems+ array.
|
10
|
+
*/
|
11
|
+
static HashSet *hs_alloc(free_ft free_func)
|
12
|
+
{
|
13
|
+
HashSet *hs = ALLOC(HashSet);
|
14
|
+
hs->size = 0;
|
15
|
+
hs->first = hs->last = NULL;
|
16
|
+
hs->free_elem_i = free_func ? free_func : &dummy_free;
|
17
|
+
return hs;
|
18
|
+
}
|
19
|
+
|
20
|
+
HashSet *hs_new(hash_ft hash_func, eq_ft eq_func, free_ft free_func)
|
21
|
+
{
|
22
|
+
HashSet *hs = hs_alloc(free_func);
|
23
|
+
hs->ht = h_new(hash_func, eq_func, NULL, NULL);
|
24
|
+
return hs;
|
25
|
+
}
|
26
|
+
|
27
|
+
HashSet *hs_new_str(free_ft free_func)
|
28
|
+
{
|
29
|
+
HashSet *hs = hs_alloc(free_func);
|
30
|
+
hs->ht = h_new_str((free_ft) NULL, NULL);
|
31
|
+
return hs;
|
32
|
+
}
|
33
|
+
|
34
|
+
HashSet *hs_new_ptr(free_ft free_func)
|
35
|
+
{
|
36
|
+
HashSet *hs = hs_alloc(free_func);
|
37
|
+
hs->ht = h_new_ptr(NULL);
|
38
|
+
return hs;
|
39
|
+
}
|
40
|
+
|
41
|
+
static INLINE void clear(HashSet *hs, bool destroy)
|
42
|
+
{
|
43
|
+
HashSetEntry *curr, *next = hs->first;
|
44
|
+
free_ft do_free = destroy ? hs->free_elem_i : &dummy_free;
|
45
|
+
while (NULL != (curr = next)) {
|
46
|
+
next = curr->next;
|
47
|
+
do_free(curr->elem);
|
48
|
+
free(curr);
|
49
|
+
}
|
50
|
+
hs->first = hs->last = NULL;
|
51
|
+
hs->size = 0;
|
52
|
+
}
|
53
|
+
|
54
|
+
void hs_clear(HashSet *hs)
|
55
|
+
{
|
56
|
+
clear(hs, true);
|
57
|
+
h_clear(hs->ht);
|
58
|
+
}
|
59
|
+
|
60
|
+
void hs_free(HashSet *hs)
|
61
|
+
{
|
62
|
+
clear(hs, false);
|
63
|
+
h_destroy(hs->ht);
|
64
|
+
free(hs);
|
65
|
+
}
|
66
|
+
|
67
|
+
void hs_destroy(HashSet *hs)
|
68
|
+
{
|
69
|
+
clear(hs, true);
|
70
|
+
h_destroy(hs->ht);
|
71
|
+
free(hs);
|
72
|
+
}
|
73
|
+
|
74
|
+
static INLINE void append(HashSet *hs, void *elem)
|
75
|
+
{
|
76
|
+
HashSetEntry *entry = ALLOC(HashSetEntry);
|
77
|
+
entry->elem = elem;
|
78
|
+
entry->prev = hs->last;
|
79
|
+
entry->next = NULL;
|
80
|
+
if (!hs->first) {
|
81
|
+
hs->first = hs->last = entry;
|
82
|
+
}
|
83
|
+
else {
|
84
|
+
hs->last->next = entry;
|
85
|
+
hs->last = entry;
|
86
|
+
}
|
87
|
+
h_set(hs->ht, elem, entry);
|
88
|
+
hs->size++;
|
89
|
+
}
|
90
|
+
|
91
|
+
HashKeyStatus hs_add(HashSet *hs, void *elem)
|
92
|
+
{
|
93
|
+
HashKeyStatus has_elem = h_has_key(hs->ht, elem);
|
94
|
+
switch (has_elem)
|
95
|
+
{
|
96
|
+
/* We don't want to keep two of the same elem so free if necessary */
|
97
|
+
case HASH_KEY_EQUAL:
|
98
|
+
hs->free_elem_i(elem);
|
99
|
+
return has_elem;
|
100
|
+
|
101
|
+
/* No need to do anything */
|
102
|
+
case HASH_KEY_SAME:
|
103
|
+
return has_elem;
|
104
|
+
|
105
|
+
/* add the elem to the array, resizing if necessary */
|
106
|
+
case HASH_KEY_DOES_NOT_EXIST:
|
107
|
+
break;
|
108
|
+
|
109
|
+
}
|
110
|
+
|
111
|
+
append(hs, elem);
|
112
|
+
return has_elem;
|
113
|
+
}
|
114
|
+
|
115
|
+
int hs_add_safe(HashSet *hs, void *elem)
|
116
|
+
{
|
117
|
+
switch(h_has_key(hs->ht, elem))
|
118
|
+
{
|
119
|
+
/* element can't be added */
|
120
|
+
case HASH_KEY_EQUAL: return false;
|
121
|
+
|
122
|
+
/* the exact same element has already been added */
|
123
|
+
case HASH_KEY_SAME : return true;
|
124
|
+
|
125
|
+
/* add the elem to the array, resizing if necessary */
|
126
|
+
case HASH_KEY_DOES_NOT_EXIST : break;
|
127
|
+
}
|
128
|
+
append(hs, elem);
|
129
|
+
return true;
|
130
|
+
}
|
131
|
+
|
132
|
+
void *hs_rem(HashSet *hs, const void *elem)
|
133
|
+
{
|
134
|
+
void *return_elem;
|
135
|
+
HashSetEntry *entry = (HashSetEntry *)h_get(hs->ht, elem);
|
136
|
+
if (entry == NULL) return NULL;
|
137
|
+
|
138
|
+
if (hs->first == hs->last) {
|
139
|
+
hs->first = hs->last = NULL;
|
140
|
+
}
|
141
|
+
else if (hs->first == entry) {
|
142
|
+
hs->first = entry->next;
|
143
|
+
hs->first->prev = NULL;
|
144
|
+
}
|
145
|
+
else if (hs->last == entry) {
|
146
|
+
hs->last = entry->prev;
|
147
|
+
hs->last->next = NULL;
|
148
|
+
}
|
149
|
+
else {
|
150
|
+
entry->prev->next = entry->next;
|
151
|
+
entry->next->prev = entry->prev;
|
152
|
+
}
|
153
|
+
return_elem = entry->elem;
|
154
|
+
h_del(hs->ht, return_elem);
|
155
|
+
free(entry);
|
156
|
+
hs->size--;
|
157
|
+
return return_elem;
|
158
|
+
}
|
159
|
+
|
160
|
+
int hs_del(HashSet *hs, const void *elem)
|
161
|
+
{
|
162
|
+
void *tmp_elem = hs_rem(hs, elem);
|
163
|
+
if (tmp_elem != NULL) {
|
164
|
+
hs->free_elem_i(tmp_elem);
|
165
|
+
return 1;
|
166
|
+
}
|
167
|
+
return 0;
|
168
|
+
}
|
169
|
+
|
170
|
+
HashKeyStatus hs_exists(HashSet *hs, const void *elem)
|
171
|
+
{
|
172
|
+
return h_has_key(hs->ht, elem);
|
173
|
+
}
|
174
|
+
|
175
|
+
HashSet *hs_merge(HashSet *hs, HashSet * other)
|
176
|
+
{
|
177
|
+
HashSetEntry *entry = other->first;
|
178
|
+
for (; entry != NULL; entry = entry->next) {
|
179
|
+
hs_add(hs, entry->elem);
|
180
|
+
}
|
181
|
+
/* Now free the other hashset. It is no longer needed. No need, however,
|
182
|
+
* to delete the elements as they were either destroyed or added to the
|
183
|
+
* new hashset. */
|
184
|
+
hs_free(other);
|
185
|
+
return hs;
|
186
|
+
}
|
187
|
+
|
188
|
+
void *hs_orig(HashSet *hs, const void *elem)
|
189
|
+
{
|
190
|
+
HashSetEntry *entry = (HashSetEntry *)h_get(hs->ht, elem);
|
191
|
+
return entry ? entry->elem : NULL;
|
192
|
+
}
|
data/ext/hashset.h
ADDED
@@ -0,0 +1,215 @@
|
|
1
|
+
#ifndef FRT_HASHSET_H
|
2
|
+
#define FRT_HASHSET_H
|
3
|
+
|
4
|
+
#ifdef __cplusplus
|
5
|
+
extern "C" {
|
6
|
+
#endif
|
7
|
+
|
8
|
+
#include "hash.h"
|
9
|
+
#include "global.h"
|
10
|
+
|
11
|
+
#define FRT_HS_MIN_SIZE 4
|
12
|
+
typedef struct FrtHashSetEntry {
|
13
|
+
void *elem;
|
14
|
+
struct FrtHashSetEntry *next;
|
15
|
+
struct FrtHashSetEntry *prev;
|
16
|
+
} FrtHashSetEntry;
|
17
|
+
|
18
|
+
typedef struct FrtHashSet
|
19
|
+
{
|
20
|
+
/* the number of elements in the instance */
|
21
|
+
int size;
|
22
|
+
|
23
|
+
/* the first element in the list of elements in the FrtHashSet. The elements
|
24
|
+
* will be listed in the order they were added and can be iterated over by
|
25
|
+
* following the ->next pointer */
|
26
|
+
FrtHashSetEntry *first;
|
27
|
+
|
28
|
+
/* the last element in the list of elements in the FrtHashSet. This is used
|
29
|
+
* internally to add elements to the list. */
|
30
|
+
FrtHashSetEntry *last;
|
31
|
+
|
32
|
+
/* Hash used internally */
|
33
|
+
FrtHash *ht;
|
34
|
+
|
35
|
+
/* Internal: Frees elements added to the FrtHashSet. Should never be NULL */
|
36
|
+
frt_free_ft free_elem_i;
|
37
|
+
} FrtHashSet;
|
38
|
+
|
39
|
+
/**
|
40
|
+
* Create a new FrtHashSet. The function will allocate a FrtHashSet Struct
|
41
|
+
* setting the functions used to hash the objects it will contain and the eq
|
42
|
+
* function. This should be used for non-string types.
|
43
|
+
*
|
44
|
+
* @param hash function to hash objects added to the FrtHashSet
|
45
|
+
* @param eq function to determine whether two items are equal
|
46
|
+
* @param free_elem function used to free elements as added to the FrtHashSet
|
47
|
+
* when the FrtHashSet if destroyed or duplicate elements are added to the Set
|
48
|
+
* @return a newly allocated FrtHashSet structure
|
49
|
+
*/
|
50
|
+
extern FrtHashSet *frt_hs_new(frt_hash_ft hash_func,
|
51
|
+
frt_eq_ft eq_func,
|
52
|
+
frt_free_ft free_func);
|
53
|
+
|
54
|
+
/**
|
55
|
+
* Create a new FrtHashSet specifically for strings. This will create a
|
56
|
+
* FrtHashSet as if you used frt_hs_new with the standard string hash and eq
|
57
|
+
* functions.
|
58
|
+
*
|
59
|
+
* @param free_elem function used to free elements as added to the FrtHashSet
|
60
|
+
* when the FrtHashSet if destroyed or duplicate elements are added to the Set
|
61
|
+
* @return a newly allocated FrtHashSet structure
|
62
|
+
*/
|
63
|
+
extern FrtHashSet *frt_hs_new_str(frt_free_ft free_func);
|
64
|
+
|
65
|
+
/**
|
66
|
+
* Create a new FrtHashSet specifically for pointers. Note that the only way
|
67
|
+
* two pointers will be considered equal is if they have the same address. So
|
68
|
+
* you can add the string "key" twice if it is stored at two different
|
69
|
+
* addresses.
|
70
|
+
*
|
71
|
+
* @param free_elem function used to free elements as added to the FrtHashSet
|
72
|
+
* when the FrtHashSet if destroyed or duplicate elements are added to the Set
|
73
|
+
* @return a newly allocated FrtHashSet structure
|
74
|
+
*/
|
75
|
+
extern FrtHashSet *frt_hs_new_ptr(frt_free_ft free_func);
|
76
|
+
|
77
|
+
/**
|
78
|
+
* Free the memory allocated by the FrtHashSet, but don't free the elements added
|
79
|
+
* to the FrtHashSet. If you'd like to free everything in the FrtHashSet you should
|
80
|
+
* use frt_hs_destroy
|
81
|
+
*
|
82
|
+
* @param hs the FrtHashSet to free
|
83
|
+
*/
|
84
|
+
extern void frt_hs_free(FrtHashSet *self);
|
85
|
+
|
86
|
+
/**
|
87
|
+
* Destroy the FrtHashSet including all elements added to the FrtHashSet. If you'd
|
88
|
+
* like to free the memory allocated to the FrtHashSet without touching the
|
89
|
+
* elements in the FrtHashSet then use frt_hs_free
|
90
|
+
*
|
91
|
+
* @param hs the FrtHashSet to destroy
|
92
|
+
*/
|
93
|
+
extern void frt_hs_destroy(FrtHashSet *self);
|
94
|
+
|
95
|
+
/**
|
96
|
+
* WARNING: this function may destroy some elements if you add them to a
|
97
|
+
* FrtHashSet were equivalent elements already exist, depending on how free_elem
|
98
|
+
* was set.
|
99
|
+
*
|
100
|
+
* Add the element to the FrtHashSet whether or not it was already in the
|
101
|
+
* FrtHashSet.
|
102
|
+
*
|
103
|
+
* When a element is added to the Hash where it already exists, free_elem
|
104
|
+
* is called on it, ie the element you tried to add might get destroyed.
|
105
|
+
*
|
106
|
+
* @param hs the FrtHashSet to add the element to
|
107
|
+
* @param elem the element to add to the FrtHashSet
|
108
|
+
* @return one of three values;
|
109
|
+
* <pre>
|
110
|
+
* HASH_KEY_DOES_NOT_EXIST the element was not already in the FrtHashSet.
|
111
|
+
* This value is equal to 0 or false
|
112
|
+
* HASH_KEY_SAME the element was identical (same memory
|
113
|
+
* pointer) to an existing element so no freeing
|
114
|
+
* was done
|
115
|
+
* HASH_KEY_EQUAL the element was equal to an element already in
|
116
|
+
* the FrtHashSet so the new_elem was freed if
|
117
|
+
* free_elem was set
|
118
|
+
* </pre>
|
119
|
+
*/
|
120
|
+
extern FrtHashKeyStatus frt_hs_add(FrtHashSet *self, void *elem);
|
121
|
+
|
122
|
+
/**
|
123
|
+
* Add element to the FrtHashSet. If the element already existed in the FrtHashSet
|
124
|
+
* and the new element was equal but not the same (same pointer/memory) then
|
125
|
+
* don't add the element and return false, otherwise return true.
|
126
|
+
*
|
127
|
+
* @param hs the FrtHashSet to add the element to
|
128
|
+
* @param elem the element to add to the FrtHashSet
|
129
|
+
* @return true if the element was successfully added or false otherwise
|
130
|
+
*/
|
131
|
+
extern int frt_hs_add_safe(FrtHashSet *self, void *elem);
|
132
|
+
|
133
|
+
/**
|
134
|
+
* Delete the element from the FrtHashSet. Returns true if the item was
|
135
|
+
* successfully deleted or false if the element never existed.
|
136
|
+
*
|
137
|
+
* @param hs the FrtHashSet to delete from
|
138
|
+
* @param elem the element to delete
|
139
|
+
* @return true if the element was deleted or false if the element never
|
140
|
+
* existed
|
141
|
+
*/
|
142
|
+
extern int frt_hs_del(FrtHashSet *self, const void *elem);
|
143
|
+
|
144
|
+
/**
|
145
|
+
* Remove an item from the FrtHashSet without actually freeing the item. This
|
146
|
+
* function should return the item itself so that it can be freed later if
|
147
|
+
* necessary.
|
148
|
+
*
|
149
|
+
* @param hs the FrtHashSet to remove the element from.
|
150
|
+
* @param elem the element to remove
|
151
|
+
* @param the element that was removed or NULL otherwise
|
152
|
+
*/
|
153
|
+
extern void *frt_hs_rem(FrtHashSet *self, const void *elem);
|
154
|
+
|
155
|
+
/**
|
156
|
+
* Check if the element exists and return the appropriate value described
|
157
|
+
* bellow.
|
158
|
+
*
|
159
|
+
* @param hs the FrtHashSet to check in
|
160
|
+
* @param elem the element to check for
|
161
|
+
* @return one of the following values
|
162
|
+
* <pre>
|
163
|
+
* HASH_KEY_DOES_NOT_EXIST the element was not already in the FrtHashSet.
|
164
|
+
* This value is equal to 0 or false
|
165
|
+
* HASH_KEY_SAME the element was identical (same memory
|
166
|
+
* pointer) to an existing element so no freeing
|
167
|
+
* was done
|
168
|
+
* HASH_KEY_EQUAL the element was equal to an element already in
|
169
|
+
* the FrtHashSet so the new_elem was freed if
|
170
|
+
* free_elem was set
|
171
|
+
* </pre>
|
172
|
+
*/
|
173
|
+
extern FrtHashKeyStatus frt_hs_exists(FrtHashSet *self, const void *elem);
|
174
|
+
|
175
|
+
/**
|
176
|
+
* Merge two HashSets. When a merge is done the merger (self) Hash is
|
177
|
+
* returned and the mergee is destroyed. All elements from mergee that were
|
178
|
+
* not found in merger (self) will be added to self, otherwise they will be
|
179
|
+
* destroyed.
|
180
|
+
*
|
181
|
+
* @param self the FrtHashSet to merge into
|
182
|
+
* @param other HastSet to be merged into self
|
183
|
+
* @return the merged FrtHashSet
|
184
|
+
*/
|
185
|
+
extern FrtHashSet *frt_hs_merge(FrtHashSet *self, FrtHashSet *other);
|
186
|
+
|
187
|
+
/**
|
188
|
+
* Return the original version of +elem+. So if you allocate two elements
|
189
|
+
* which are equal and add the first to the FrtHashSet, calling this function
|
190
|
+
* with the second element will return the first element from the FrtHashSet.
|
191
|
+
*/
|
192
|
+
extern void *frt_hs_orig(FrtHashSet *self, const void *elem);
|
193
|
+
|
194
|
+
/**
|
195
|
+
* Clear all elements from the FrtHashSet. If free_elem was set then use it to
|
196
|
+
* free all elements as they are cleared. After the method is called, the
|
197
|
+
* HashSets size will be 0.
|
198
|
+
*
|
199
|
+
* @param self the FrtHashSet to clear
|
200
|
+
*/
|
201
|
+
extern void frt_hs_clear(FrtHashSet *self);
|
202
|
+
|
203
|
+
/* TODO: finish implementing these functions FrtHashSet
|
204
|
+
int hs_osf(FrtHashSet *hs, void *elem);
|
205
|
+
FrtHashSet hs_or(FrtHashSet *hs1, FrtHashSet *h2);
|
206
|
+
FrtHashSet hs_excl_or(FrtHashSet *hs1, FrtHashSet *h2);
|
207
|
+
FrtHashSet hs_and(FrtHashSet *hs1, FrtHashSet *h2);
|
208
|
+
FrtHashSet hs_mask(FrtHashSet *hs1, FrtHashSet *h2);
|
209
|
+
*/
|
210
|
+
|
211
|
+
#ifdef __cplusplus
|
212
|
+
} // extern "C"
|
213
|
+
#endif
|
214
|
+
|
215
|
+
#endif
|
data/ext/header.h
ADDED
@@ -0,0 +1,58 @@
|
|
1
|
+
|
2
|
+
#include <limits.h>
|
3
|
+
|
4
|
+
#include "api.h"
|
5
|
+
|
6
|
+
#define MAXINT INT_MAX
|
7
|
+
#define MININT INT_MIN
|
8
|
+
|
9
|
+
#define HEAD 2*sizeof(int)
|
10
|
+
|
11
|
+
#define SIZE(p) ((int *)(p))[-1]
|
12
|
+
#define SET_SIZE(p, n) ((int *)(p))[-1] = n
|
13
|
+
#define CAPACITY(p) ((int *)(p))[-2]
|
14
|
+
|
15
|
+
struct among
|
16
|
+
{ int s_size; /* number of chars in string */
|
17
|
+
const symbol * s; /* search string */
|
18
|
+
int substring_i;/* index to longest matching substring */
|
19
|
+
int result; /* result of the lookup */
|
20
|
+
int (* function)(struct SN_env *);
|
21
|
+
};
|
22
|
+
|
23
|
+
extern symbol * create_s(void);
|
24
|
+
extern void lose_s(symbol * p);
|
25
|
+
|
26
|
+
extern int skip_utf8(const symbol * p, int c, int lb, int l, int n);
|
27
|
+
|
28
|
+
extern int in_grouping_U(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
29
|
+
extern int in_grouping_b_U(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
30
|
+
extern int out_grouping_U(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
31
|
+
extern int out_grouping_b_U(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
32
|
+
|
33
|
+
extern int in_grouping(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
34
|
+
extern int in_grouping_b(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
35
|
+
extern int out_grouping(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
36
|
+
extern int out_grouping_b(struct SN_env * z, const unsigned char * s, int min, int max, int repeat);
|
37
|
+
|
38
|
+
extern int eq_s(struct SN_env * z, int s_size, const symbol * s);
|
39
|
+
extern int eq_s_b(struct SN_env * z, int s_size, const symbol * s);
|
40
|
+
extern int eq_v(struct SN_env * z, const symbol * p);
|
41
|
+
extern int eq_v_b(struct SN_env * z, const symbol * p);
|
42
|
+
|
43
|
+
extern int find_among(struct SN_env * z, const struct among * v, int v_size);
|
44
|
+
extern int find_among_b(struct SN_env * z, const struct among * v, int v_size);
|
45
|
+
|
46
|
+
extern int replace_s(struct SN_env * z, int c_bra, int c_ket, int s_size, const symbol * s, int * adjustment);
|
47
|
+
extern int slice_from_s(struct SN_env * z, int s_size, const symbol * s);
|
48
|
+
extern int slice_from_v(struct SN_env * z, const symbol * p);
|
49
|
+
extern int slice_del(struct SN_env * z);
|
50
|
+
|
51
|
+
extern int insert_s(struct SN_env * z, int bra, int ket, int s_size, const symbol * s);
|
52
|
+
extern int insert_v(struct SN_env * z, int bra, int ket, const symbol * p);
|
53
|
+
|
54
|
+
extern symbol * slice_to(struct SN_env * z, symbol * p);
|
55
|
+
extern symbol * assign_to(struct SN_env * z, symbol * p);
|
56
|
+
|
57
|
+
extern void debug(struct SN_env * z, int number, int line_count);
|
58
|
+
|