2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.28 1999-03-09 13:07:06 adam
8 * Work on dict_compact routine.
10 * Revision 1.27 1999/02/02 14:50:32 adam
11 * Updated WIN32 code specific sections. Changed header.
13 * Revision 1.26 1997/09/18 08:59:18 adam
14 * Extra generic handle for the character mapping routines.
16 * Revision 1.25 1997/09/17 12:19:09 adam
17 * Zebra version corresponds to YAZ version 1.4.
18 * Changed Zebra server so that it doesn't depend on global common_resource.
20 * Revision 1.24 1997/09/05 15:30:00 adam
21 * Changed prototype for chr_map_input - added const.
22 * Added support for C++, headers uses extern "C" for public definitions.
24 * Revision 1.23 1996/10/29 13:45:33 adam
25 * Changed definition of DICT_DEFAULT_PAGESIZE.
27 * Revision 1.22 1996/06/04 10:20:10 adam
28 * Added support for character mapping.
30 * Revision 1.21 1996/05/24 14:46:07 adam
31 * Added dict_grep_cmap function to define user-mapping in grep lookups.
33 * Revision 1.20 1996/03/20 09:35:23 adam
34 * Function dict_lookup_grep got extra parameter, init_pos, which marks
35 * from which position in pattern approximate pattern matching should occur.
37 * Revision 1.19 1996/02/02 13:43:54 adam
38 * The public functions simply use char instead of Dict_char to represent
39 * search strings. Dict_char is used internally only.
41 * Revision 1.18 1996/02/01 20:41:06 adam
42 * Bug fix: insert didn't work on 8-bit characters due to unsigned char
43 * compares in dict_strcmp (strcmp) and signed Dict_char. Dict_char is
46 * Revision 1.17 1995/12/07 11:47:04 adam
47 * Default pagesize is 4k instead of 8k.
49 * Revision 1.16 1995/12/06 14:41:13 adam
50 * New function: dict_delete.
52 * Revision 1.15 1995/10/27 13:59:17 adam
53 * Function dict_look_grep got extra parameter max_pos that upon return
54 * hold length of longest prefix that matches pattern.
56 * Revision 1.14 1995/10/09 16:18:35 adam
57 * Function dict_lookup_grep got extra client data parameter.
59 * Revision 1.13 1995/10/06 09:03:51 adam
60 * First version of scan.
62 * Revision 1.12 1995/09/14 11:53:02 adam
63 * Grep handle function parameter info is const now.
65 * Revision 1.11 1995/09/04 09:09:51 adam
66 * String arg in dict lookup is const.
69 * Revision 1.10 1994/10/05 12:16:58 adam
70 * Pagesize is a resource now.
72 * Revision 1.9 1994/10/04 12:08:19 adam
75 * Revision 1.8 1994/10/03 17:23:11 adam
76 * First version of dictionary lookup with regular expressions and errors.
78 * Revision 1.7 1994/09/22 10:44:47 adam
79 * Don't remember what changed!!
81 * Revision 1.6 1994/09/16 15:39:21 adam
82 * Initial code of lookup - not tested yet.
84 * Revision 1.5 1994/09/06 13:05:29 adam
85 * Further development of insertion. Some special cases are
86 * not properly handled yet! assert(0) are put here. The
87 * binary search in each page definitely reduce usr CPU.
89 * Revision 1.4 1994/09/01 17:44:40 adam
90 * Work on insertion in dictionary. Not finished yet.
92 * Revision 1.3 1994/08/18 12:41:12 adam
93 * Some development of dictionary. Not finished at all!
95 * Revision 1.2 1994/08/17 13:32:33 adam
96 * Use cache in dict - not in bfile.
98 * Revision 1.1 1994/08/16 16:26:53 adam
113 typedef unsigned Dict_ptr;
114 typedef unsigned char Dict_char;
119 Dict_ptr free_list, last;
122 struct Dict_file_block
124 struct Dict_file_block *h_next, **h_prev;
125 struct Dict_file_block *lru_next, *lru_prev;
131 typedef struct Dict_file_struct
136 struct Dict_file_block *all_blocks;
137 struct Dict_file_block *free_list;
138 struct Dict_file_block **hash_array;
140 struct Dict_file_block *lru_back, *lru_front;
149 typedef struct Dict_struct {
152 const char **(*grep_cmap)(void *vp, const char **from, int len);
153 void *grep_cmap_data;
154 struct Dict_head head;
157 #define DICT_MAGIC "dict00"
159 #define DICT_DEFAULT_PAGESIZE 4096
161 int dict_bf_readp (Dict_BFile bf, int no, void **bufp);
162 int dict_bf_newp (Dict_BFile bf, int no, void **bufp);
163 int dict_bf_touch (Dict_BFile bf, int no);
164 void dict_bf_flush_blocks (Dict_BFile bf, int no_to_flush);
165 Dict_BFile dict_bf_open (BFiles bfs, const char *name, int block_size,
167 int dict_bf_close (Dict_BFile dbf);
169 Dict dict_open (BFiles bfs, const char *name, int cache, int rw);
170 int dict_close (Dict dict);
171 int dict_insert (Dict dict, const char *p, int userlen, void *userinfo);
172 int dict_delete (Dict dict, const char *p);
173 char *dict_lookup (Dict dict, const char *p);
174 int dict_lookup_ec (Dict dict, char *p, int range,
175 int (*f)(char *name));
176 int dict_lookup_grep (Dict dict, const char *p, int range, void *client,
177 int *max_pos, int init_pos,
178 int (*f)(char *name, const char *info,
180 int dict_strcmp (const Dict_char *s1, const Dict_char *s2);
181 int dict_strlen (const Dict_char *s);
182 int dict_scan (Dict dict, char *str,
183 int *before, int *after, void *client,
184 int (*f)(char *name, const char *info, int pos,
187 void dict_grep_cmap (Dict dict, void *vp,
188 const char **(*cmap)(void *vp,
189 const char **from, int len));
190 int dict_compact (BFiles bfs, const char *from, const char *to);
194 #define DICT_type(x) 0[(Dict_ptr*) x]
195 #define DICT_backptr(x) 1[(Dict_ptr*) x]
196 #define DICT_nextptr(x) 2[(Dict_ptr*) x]
197 #define DICT_nodir(x) 0[(short*)((char*)(x)+3*sizeof(Dict_ptr))]
198 #define DICT_size(x) 1[(short*)((char*)(x)+3*sizeof(Dict_ptr))]
199 #define DICT_infoffset (3*sizeof(Dict_ptr)+2*sizeof(short))
200 #define DICT_pagesize(x) ((x)->head.page_size)
202 #define DICT_to_str(x) sizeof(Dict_info)+sizeof(Dict_ptr)
206 backptr pointer to parent
207 nextptr pointer to next page (if any)
209 size size of strings,info,ptr entries