2 * Copyright (C) 1994-2000, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.31 2000-12-05 09:59:10 adam
8 * Work on dict_delete_subtree.
10 * Revision 1.30 1999/11/30 13:48:03 adam
11 * Improved installation. Updated for inclusion of YAZ header files.
13 * Revision 1.29 1999/05/15 14:36:37 adam
14 * Updated dictionary. Implemented "compression" of dictionary.
16 * Revision 1.28 1999/03/09 13:07:06 adam
17 * Work on dict_compact routine.
19 * Revision 1.27 1999/02/02 14:50:32 adam
20 * Updated WIN32 code specific sections. Changed header.
22 * Revision 1.26 1997/09/18 08:59:18 adam
23 * Extra generic handle for the character mapping routines.
25 * Revision 1.25 1997/09/17 12:19:09 adam
26 * Zebra version corresponds to YAZ version 1.4.
27 * Changed Zebra server so that it doesn't depend on global common_resource.
29 * Revision 1.24 1997/09/05 15:30:00 adam
30 * Changed prototype for chr_map_input - added const.
31 * Added support for C++, headers uses extern "C" for public definitions.
33 * Revision 1.23 1996/10/29 13:45:33 adam
34 * Changed definition of DICT_DEFAULT_PAGESIZE.
36 * Revision 1.22 1996/06/04 10:20:10 adam
37 * Added support for character mapping.
39 * Revision 1.21 1996/05/24 14:46:07 adam
40 * Added dict_grep_cmap function to define user-mapping in grep lookups.
42 * Revision 1.20 1996/03/20 09:35:23 adam
43 * Function dict_lookup_grep got extra parameter, init_pos, which marks
44 * from which position in pattern approximate pattern matching should occur.
46 * Revision 1.19 1996/02/02 13:43:54 adam
47 * The public functions simply use char instead of Dict_char to represent
48 * search strings. Dict_char is used internally only.
50 * Revision 1.18 1996/02/01 20:41:06 adam
51 * Bug fix: insert didn't work on 8-bit characters due to unsigned char
52 * compares in dict_strcmp (strcmp) and signed Dict_char. Dict_char is
55 * Revision 1.17 1995/12/07 11:47:04 adam
56 * Default pagesize is 4k instead of 8k.
58 * Revision 1.16 1995/12/06 14:41:13 adam
59 * New function: dict_delete.
61 * Revision 1.15 1995/10/27 13:59:17 adam
62 * Function dict_look_grep got extra parameter max_pos that upon return
63 * hold length of longest prefix that matches pattern.
65 * Revision 1.14 1995/10/09 16:18:35 adam
66 * Function dict_lookup_grep got extra client data parameter.
68 * Revision 1.13 1995/10/06 09:03:51 adam
69 * First version of scan.
71 * Revision 1.12 1995/09/14 11:53:02 adam
72 * Grep handle function parameter info is const now.
74 * Revision 1.11 1995/09/04 09:09:51 adam
75 * String arg in dict lookup is const.
78 * Revision 1.10 1994/10/05 12:16:58 adam
79 * Pagesize is a resource now.
81 * Revision 1.9 1994/10/04 12:08:19 adam
84 * Revision 1.8 1994/10/03 17:23:11 adam
85 * First version of dictionary lookup with regular expressions and errors.
87 * Revision 1.7 1994/09/22 10:44:47 adam
88 * Don't remember what changed!!
90 * Revision 1.6 1994/09/16 15:39:21 adam
91 * Initial code of lookup - not tested yet.
93 * Revision 1.5 1994/09/06 13:05:29 adam
94 * Further development of insertion. Some special cases are
95 * not properly handled yet! assert(0) are put here. The
96 * binary search in each page definitely reduce usr CPU.
98 * Revision 1.4 1994/09/01 17:44:40 adam
99 * Work on insertion in dictionary. Not finished yet.
101 * Revision 1.3 1994/08/18 12:41:12 adam
102 * Some development of dictionary. Not finished at all!
104 * Revision 1.2 1994/08/17 13:32:33 adam
105 * Use cache in dict - not in bfile.
107 * Revision 1.1 1994/08/16 16:26:53 adam
122 typedef unsigned Dict_ptr;
123 typedef unsigned char Dict_char;
129 Dict_ptr root, last, freelist;
132 struct Dict_file_block
134 struct Dict_file_block *h_next, **h_prev;
135 struct Dict_file_block *lru_next, *lru_prev;
142 typedef struct Dict_file_struct
147 struct Dict_file_block *all_blocks;
148 struct Dict_file_block *free_list;
149 struct Dict_file_block **hash_array;
151 struct Dict_file_block *lru_back, *lru_front;
161 typedef struct Dict_struct {
164 const char **(*grep_cmap)(void *vp, const char **from, int len);
165 void *grep_cmap_data;
166 struct Dict_head head;
169 #define DICT_MAGIC "dict01"
171 #define DICT_DEFAULT_PAGESIZE 4096
173 int dict_bf_readp (Dict_BFile bf, int no, void **bufp);
174 int dict_bf_newp (Dict_BFile bf, int no, void **bufp, int nbytes);
175 int dict_bf_touch (Dict_BFile bf, int no);
176 void dict_bf_flush_blocks (Dict_BFile bf, int no_to_flush);
177 Dict_BFile dict_bf_open (BFiles bfs, const char *name, int block_size,
179 int dict_bf_close (Dict_BFile dbf);
180 void dict_bf_compact (Dict_BFile dbf);
182 Dict dict_open (BFiles bfs, const char *name, int cache, int rw,
184 int dict_close (Dict dict);
185 int dict_insert (Dict dict, const char *p, int userlen, void *userinfo);
186 int dict_delete (Dict dict, const char *p);
187 int dict_delete_subtree (Dict dict, const char *p, void *client,
188 int (*f)(const char *info, void *client));
189 char *dict_lookup (Dict dict, const char *p);
190 int dict_lookup_ec (Dict dict, char *p, int range,
191 int (*f)(char *name));
192 int dict_lookup_grep (Dict dict, const char *p, int range, void *client,
193 int *max_pos, int init_pos,
194 int (*f)(char *name, const char *info,
196 int dict_strcmp (const Dict_char *s1, const Dict_char *s2);
197 int dict_strncmp (const Dict_char *s1, const Dict_char *s2, size_t n);
198 int dict_strlen (const Dict_char *s);
199 int dict_scan (Dict dict, char *str,
200 int *before, int *after, void *client,
201 int (*f)(char *name, const char *info, int pos,
204 void dict_grep_cmap (Dict dict, void *vp,
205 const char **(*cmap)(void *vp,
206 const char **from, int len));
207 int dict_copy_compact (BFiles bfs, const char *from, const char *to);
211 #define DICT_type(x) 0[(Dict_ptr*) x]
212 #define DICT_backptr(x) 1[(Dict_ptr*) x]
213 #define DICT_bsize(x) 2[(short*)((char*)(x)+2*sizeof(Dict_ptr))]
214 #define DICT_nodir(x) 0[(short*)((char*)(x)+2*sizeof(Dict_ptr))]
215 #define DICT_size(x) 1[(short*)((char*)(x)+2*sizeof(Dict_ptr))]
216 #define DICT_infoffset (2*sizeof(Dict_ptr)+3*sizeof(short))
217 #define DICT_xxxxpagesize(x) ((x)->head.page_size)
219 #define DICT_to_str(x) sizeof(Dict_info)+sizeof(Dict_ptr)
223 backptr pointer to parent
224 nextptr pointer to next page (if any)
226 size size of strings,info,ptr entries