2 * Copyright (C) 1995-1996, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.48 1997-02-12 20:39:45 adam
8 * Implemented options -f <n> that limits the log to the first <n>
10 * Changed some log messages also.
12 * Revision 1.47 1996/12/23 15:30:44 adam
14 * Bug fix: result sets weren't deleted after server shut down.
16 * Revision 1.46 1996/11/08 11:10:19 adam
17 * Buffers used during file match got bigger.
18 * Compressed ISAM support everywhere.
19 * Bug fixes regarding masking characters in queries.
20 * Redesigned Regexp-2 queries.
22 * Revision 1.45 1996/10/29 14:09:42 adam
23 * Use of cisam system - enabled if setting isamc is 1.
25 * Revision 1.44 1996/06/06 12:08:40 quinn
26 * Added showRecord function
28 * Revision 1.43 1996/06/04 10:18:12 adam
29 * Search/scan uses character mapping module.
31 * Revision 1.42 1996/06/04 08:20:16 quinn
34 * Revision 1.41 1996/06/04 07:54:55 quinn
37 * Revision 1.40 1996/05/31 09:06:58 quinn
38 * Work on character-set handling
40 * Revision 1.39 1996/05/14 14:04:33 adam
41 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
44 * Revision 1.38 1996/04/12 07:02:23 adam
45 * File update of single files.
47 * Revision 1.37 1996/03/26 16:01:13 adam
48 * New setting lockPath: directory of various lock files.
50 * Revision 1.36 1996/03/21 14:50:09 adam
51 * File update uses modify-time instead of change-time.
53 * Revision 1.35 1996/02/12 18:45:36 adam
54 * New fileVerboseFlag in record group control.
56 * Revision 1.34 1995/12/11 11:43:29 adam
57 * Locking based on fcntl instead of flock.
58 * Setting commitEnable removed. Command line option -n can be used to
59 * prevent commit if commit setting is defined in the configuration file.
61 * Revision 1.33 1995/12/08 16:22:53 adam
62 * Work on update while servers are running. Three lock files introduced.
63 * The servers reload their registers when necessary, but they don't
64 * reestablish result sets yet.
66 * Revision 1.32 1995/12/07 17:38:46 adam
67 * Work locking mechanisms for concurrent updates/commit.
69 * Revision 1.31 1995/12/06 12:41:22 adam
70 * New command 'stat' for the index program.
71 * Filenames can be read from stdin by specifying '-'.
72 * Bug fix/enhancement of the transformation from terms to regular
73 * expressons in the search engine.
75 * Revision 1.30 1995/12/05 11:25:02 adam
76 * Include of zebraver.h.
78 * Revision 1.29 1995/11/28 09:09:40 adam
79 * Zebra config renamed.
80 * Use setting 'recordId' to identify record now.
81 * Bug fix in recindex.c: rec_release_blocks was invokeded even
82 * though the blocks were already released.
83 * File traversal properly deletes records when needed.
85 * Revision 1.28 1995/11/27 13:58:53 adam
86 * New option -t. storeStore data implemented in server.
88 * Revision 1.27 1995/11/25 10:24:06 adam
89 * More record fields - they are enumerated now.
90 * New options: flagStoreData flagStoreKey.
92 * Revision 1.26 1995/11/22 17:19:17 adam
93 * Record management uses the bfile system.
95 * Revision 1.25 1995/11/21 15:29:12 adam
96 * Config file 'base' read by default by both indexer and server.
98 * Revision 1.24 1995/11/21 15:01:15 adam
99 * New general match criteria implemented.
100 * New feature: document groups.
102 * Revision 1.23 1995/11/20 16:59:45 adam
103 * New update method: the 'old' keys are saved for each records.
105 * Revision 1.22 1995/11/20 11:56:26 adam
106 * Work on new traversal.
108 * Revision 1.21 1995/11/16 15:34:55 adam
109 * Uses new record management system in both indexer and server.
111 * Revision 1.20 1995/11/15 14:46:18 adam
112 * Started work on better record management system.
114 * Revision 1.19 1995/10/27 14:00:11 adam
115 * Implemented detection of database availability.
117 * Revision 1.18 1995/10/17 18:02:08 adam
118 * New feature: databases. Implemented as prefix to words in dictionary.
120 * Revision 1.17 1995/10/13 16:01:49 adam
123 * Revision 1.16 1995/10/10 12:24:38 adam
124 * Temporary sort files are compressed.
126 * Revision 1.15 1995/10/04 16:57:19 adam
127 * Key input and merge sort in one pass.
129 * Revision 1.14 1995/09/29 14:01:40 adam
132 * Revision 1.13 1995/09/28 14:22:56 adam
133 * Sort uses smaller temporary files.
135 * Revision 1.12 1995/09/28 12:10:32 adam
136 * Bug fixes. Field prefix used in queries.
138 * Revision 1.11 1995/09/27 12:22:28 adam
139 * More work on extract in record control.
140 * Field name is not in isam keys but in prefix in dictionary words.
142 * Revision 1.10 1995/09/14 07:48:23 adam
143 * Record control management.
145 * Revision 1.9 1995/09/11 13:09:33 adam
146 * More work on relevance feedback.
148 * Revision 1.8 1995/09/08 14:52:27 adam
149 * Minor changes. Dictionary is lower case now.
151 * Revision 1.7 1995/09/06 16:11:16 adam
152 * Option: only one word key per file.
154 * Revision 1.6 1995/09/05 15:28:39 adam
155 * More work on search engine.
157 * Revision 1.5 1995/09/04 12:33:42 adam
158 * Various cleanup. YAZ util used instead.
160 * Revision 1.4 1995/09/04 09:10:35 adam
161 * More work on index add/del/update.
162 * Merge sort implemented.
163 * Initial work on z39 server.
165 * Revision 1.3 1995/09/01 14:06:35 adam
166 * Split of work into more files.
168 * Revision 1.2 1995/09/01 10:30:24 adam
169 * More work on indexing. Not working yet.
171 * Revision 1.1 1995/08/31 14:50:24 adam
172 * New simple file index tool.
176 #include <zebraver.h>
177 #include <zebrautl.h>
183 #define IT_MAX_WORD 256
184 #define IT_KEY_HAVE_SEQNO 1
185 #define IT_KEY_HAVE_FIELD 0
194 enum dirsKind { dirs_dir, dirs_file };
218 int fileVerboseLimit;
221 void getFnameTmp (char *fname, int no);
223 struct dirs_info *dirs_open (Dict dict, const char *rep);
224 struct dirs_info *dirs_fopen (Dict dict, const char *path);
225 struct dirs_entry *dirs_read (struct dirs_info *p);
226 struct dirs_entry *dirs_last (struct dirs_info *p);
227 void dirs_mkdir (struct dirs_info *p, const char *src, time_t mtime);
228 void dirs_rmdir (struct dirs_info *p, const char *src);
229 void dirs_add (struct dirs_info *p, const char *src, int sysno, time_t mtime);
230 void dirs_del (struct dirs_info *p, const char *src);
231 void dirs_free (struct dirs_info **pp);
233 struct dir_entry *dir_open (const char *rep);
234 void dir_sort (struct dir_entry *e);
235 void dir_free (struct dir_entry **e_p);
237 void repositoryUpdate (struct recordGroup *rGroup);
238 void repositoryAdd (struct recordGroup *rGroup);
239 void repositoryDelete (struct recordGroup *rGroup);
240 void repositoryShow (struct recordGroup *rGroup);
242 void key_open (int mem);
243 int key_close (void);
244 int key_compare (const void *p1, const void *p2);
245 int key_compare_it (const void *p1, const void *p2);
246 int key_qsort_compare (const void *p1, const void *p2);
247 void key_logdump (int mask, const void *p);
248 void inv_prstat (void);
249 void key_input (int nkeys, int cache);
250 ISAMC_M key_isamc_m (void);
251 int merge_sort (char **buf, int from, int to);
253 #define FNAME_DICT "dict"
254 #define FNAME_ISAM "isam"
255 #define FNAME_ISAMC "isamc"
256 #define FNAME_CONFIG "zebra.cfg"
258 #define GMATCH_DICT "gmatch"
259 #define FMATCH_DICT "fmatch"
261 struct strtab *strtab_mk (void);
262 int strtab_src (struct strtab *t, const char *name, void ***infop);
263 void strtab_del (struct strtab *t,
264 void (*func)(const char *name, void *info, void *data),
266 int index_char_cvt (int c);
267 int index_word_prefix (char *string, int attset_ordinal,
268 int local_attribute, const char *databaseName);
270 int fileExtract (SYSNO *sysno, const char *fname,
271 const struct recordGroup *rGroup, int deleteFlag);
273 void rec_prstat (void);
275 void zebraLockPrefix (char *pathPrefix);
277 void zebraIndexLockMsg (const char *str);
278 void zebraIndexUnlock (void);
279 void zebraIndexLock (int commitNow);
280 int zebraIndexWait (int commitPhase);
282 #define FNAME_MAIN_LOCK "zebraidx.LCK"
283 #define FNAME_COMMIT_LOCK "zebracmt.LCK"
284 #define FNAME_ORG_LOCK "zebraorg.LCK"
285 #define FNAME_TOUCH_TIME "zebraidx.time"
287 int zebraLock (int fd, int wr);
288 int zebraLockNB (int fd, int wr);
289 int zebraUnlock (int fd);
291 void init_charmap(void);
292 char **map_chrs_input(char **from, int len);
293 const char *map_chrs_output(const char **from);