2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.72 1999-10-14 14:33:50 adam
8 * Added truncation 5=106.
10 * Revision 1.71 1999/09/08 12:12:06 adam
13 * Revision 1.70 1999/09/07 07:19:21 adam
14 * Work on character mapping. Implemented replace rules.
16 * Revision 1.69 1999/07/21 08:31:33 adam
17 * More version info on WIN32.
19 * Revision 1.68 1999/07/14 10:59:26 adam
20 * Changed functions isc_getmethod, isams_getmethod.
21 * Improved fatal error handling (such as missing EXPLAIN schema).
23 * Revision 1.67 1999/07/06 12:28:04 adam
24 * Updated record index structure. Format includes version ID. Compression
25 * algorithm ID is stored for each record block.
27 * Revision 1.66 1999/06/25 13:48:02 adam
28 * Updated MSVC project files.
29 * Added BZIP2 record compression (not very well tested).
31 * Revision 1.65 1999/05/20 12:57:18 adam
32 * Implemented TCL filter. Updated recctrl system.
34 * Revision 1.64 1999/05/15 14:36:38 adam
35 * Updated dictionary. Implemented "compression" of dictionary.
37 * Revision 1.63 1999/03/09 16:27:49 adam
38 * More work on SDRKit integration.
40 * Revision 1.62 1999/02/18 15:01:24 adam
43 * Revision 1.61 1999/02/02 14:51:00 adam
44 * Updated WIN32 code specific sections. Changed header.
46 * Revision 1.60 1998/10/28 10:54:39 adam
49 * Revision 1.59 1998/10/16 08:14:32 adam
50 * Updated record control system.
52 * Revision 1.58 1998/06/08 14:43:13 adam
53 * Added suport for EXPLAIN Proxy servers - added settings databasePath
54 * and explainDatabase to facilitate this. Increased maximum number
55 * of databases and attributes in one register.
57 * Revision 1.57 1998/05/20 10:12:18 adam
58 * Implemented automatic EXPLAIN database maintenance.
59 * Modified Zebra to work with ASN.1 compiled version of YAZ.
61 * Revision 1.56 1998/03/05 08:45:12 adam
62 * New result set model and modular ranking system. Moved towards
63 * descent server API. System information stored as "SGML" records.
65 * Revision 1.55 1998/01/26 10:37:34 adam
68 * Revision 1.54 1998/01/12 15:04:08 adam
69 * The test option (-s) only uses read-lock (and not write lock).
71 * Revision 1.53 1997/11/18 10:05:08 adam
72 * Changed character map facility so that admin can specify character
73 * mapping files for each register type, w, p, etc.
75 * Revision 1.52 1997/10/31 12:34:04 adam
76 * Added missing nmem_init.
78 * Revision 1.51 1997/10/27 14:33:05 adam
79 * Moved towards generic character mapping depending on "structure"
80 * field in abstract syntax file. Fixed a few memory leaks. Fixed
81 * bug with negative integers when doing searches with relational
84 * Revision 1.50 1997/09/25 14:55:52 adam
87 * Revision 1.49 1997/09/17 12:19:15 adam
88 * Zebra version corresponds to YAZ version 1.4.
89 * Changed Zebra server so that it doesn't depend on global common_resource.
91 * Revision 1.48 1997/09/09 13:38:08 adam
92 * Partial port to WIN95/NT.
94 * Revision 1.47 1997/02/12 20:39:46 adam
95 * Implemented options -f <n> that limits the log to the first <n>
97 * Changed some log messages also.
99 * Revision 1.46 1997/02/10 10:20:13 adam
100 * Flag fileVerboseFlag set to 0 (default).
102 * Revision 1.45 1996/11/08 11:10:26 adam
103 * Buffers used during file match got bigger.
104 * Compressed ISAM support everywhere.
105 * Bug fixes regarding masking characters in queries.
106 * Redesigned Regexp-2 queries.
108 * Revision 1.44 1996/10/29 14:09:48 adam
109 * Use of cisam system - enabled if setting isamc is 1.
111 * Revision 1.43 1996/06/06 12:08:42 quinn
112 * Added showRecord function
114 * Revision 1.42 1996/05/31 09:07:01 quinn
115 * Work on character-set handling
117 * Revision 1.41 1996/05/14 14:04:34 adam
118 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
121 * Revision 1.40 1996/04/26 10:00:23 adam
122 * Added option -V to zebraidx to display version information.
123 * Removed stupid warnings from file update.
125 * Revision 1.39 1996/04/09 10:05:21 adam
126 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
128 * Revision 1.38 1996/03/26 16:01:14 adam
129 * New setting lockPath: directory of various lock files.
131 * Revision 1.37 1996/03/19 12:43:26 adam
132 * Bug fix: File update traversal didn't handle trailing slashes correctly.
133 * Bug fix: Update of sub directory groups wasn't handled correctly.
135 * Revision 1.36 1996/02/12 18:45:37 adam
136 * New fileVerboseFlag in record group control.
138 * Revision 1.35 1996/02/12 15:56:11 adam
139 * New code command: unread.
141 * Revision 1.34 1996/02/07 14:06:39 adam
142 * Better progress report during register merge.
143 * New command: clean - removes temporary shadow files.
145 * Revision 1.33 1996/02/06 17:11:18 adam
148 * Revision 1.32 1996/02/01 20:50:04 adam
149 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
150 * was never called - happens when no commands are specified.
152 * Revision 1.31 1996/01/08 19:15:46 adam
153 * New input filter that works!
155 * Revision 1.30 1995/12/12 16:00:59 adam
156 * System call sync(2) used after update/commit.
157 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
160 * Revision 1.29 1995/12/11 11:43:30 adam
161 * Locking based on fcntl instead of flock.
162 * Setting commitEnable removed. Command line option -n can be used to
163 * prevent commit if commit setting is defined in the configuration file.
165 * Revision 1.28 1995/12/08 16:22:56 adam
166 * Work on update while servers are running. Three lock files introduced.
167 * The servers reload their registers when necessary, but they don't
168 * reestablish result sets yet.
170 * Revision 1.27 1995/12/07 17:38:47 adam
171 * Work locking mechanisms for concurrent updates/commit.
173 * Revision 1.26 1995/12/06 12:41:23 adam
174 * New command 'stat' for the index program.
175 * Filenames can be read from stdin by specifying '-'.
176 * Bug fix/enhancement of the transformation from terms to regular
177 * expressons in the search engine.
179 * Revision 1.25 1995/12/01 16:24:39 adam
180 * Commit files use separate meta file area.
182 * Revision 1.24 1995/11/30 17:01:38 adam
183 * New setting commitCache: points to commit directories/files.
184 * New command commit: commits at the end of a zebraidx run.
186 * Revision 1.23 1995/11/30 08:34:31 adam
187 * Started work on commit facility.
188 * Changed a few malloc/free to xmalloc/xfree.
190 * Revision 1.22 1995/11/28 09:09:42 adam
191 * Zebra config renamed.
192 * Use setting 'recordId' to identify record now.
193 * Bug fix in recindex.c: rec_release_blocks was invokeded even
194 * though the blocks were already released.
195 * File traversal properly deletes records when needed.
197 * Revision 1.21 1995/11/27 14:27:39 adam
198 * Renamed 'update' command to 'dir'.
200 * Revision 1.20 1995/11/27 13:58:53 adam
201 * New option -t. storeStore data implemented in server.
203 * Revision 1.19 1995/11/25 10:24:06 adam
204 * More record fields - they are enumerated now.
205 * New options: flagStoreData flagStoreKey.
207 * Revision 1.18 1995/11/22 17:19:17 adam
208 * Record management uses the bfile system.
210 * Revision 1.17 1995/11/21 15:01:16 adam
211 * New general match criteria implemented.
212 * New feature: document groups.
214 * Revision 1.16 1995/11/20 11:56:27 adam
215 * Work on new traversal.
217 * Revision 1.15 1995/11/01 16:25:51 quinn
218 * *** empty log message ***
220 * Revision 1.14 1995/10/17 18:02:09 adam
221 * New feature: databases. Implemented as prefix to words in dictionary.
223 * Revision 1.13 1995/10/10 12:24:39 adam
224 * Temporary sort files are compressed.
226 * Revision 1.12 1995/10/04 16:57:20 adam
227 * Key input and merge sort in one pass.
229 * Revision 1.11 1995/09/29 14:01:45 adam
232 * Revision 1.10 1995/09/28 14:22:57 adam
233 * Sort uses smaller temporary files.
235 * Revision 1.9 1995/09/14 07:48:24 adam
236 * Record control management.
238 * Revision 1.8 1995/09/06 16:11:18 adam
239 * Option: only one word key per file.
241 * Revision 1.7 1995/09/05 15:28:39 adam
242 * More work on search engine.
244 * Revision 1.6 1995/09/04 12:33:43 adam
245 * Various cleanup. YAZ util used instead.
247 * Revision 1.5 1995/09/04 09:10:39 adam
248 * More work on index add/del/update.
249 * Merge sort implemented.
250 * Initial work on z39 server.
252 * Revision 1.4 1995/09/01 14:06:36 adam
253 * Split of work into more files.
255 * Revision 1.3 1995/09/01 10:57:07 adam
258 * Revision 1.2 1995/09/01 10:30:24 adam
259 * More work on indexing. Not working yet.
261 * Revision 1.1 1995/08/31 14:50:24 adam
262 * New simple file index tool.
275 #include "recindex.h"
282 #include "zebrasdr.h"
287 Res common_resource = 0;
289 int main (int argc, char **argv)
294 char *configName = FNAME_CONFIG;
296 int disableCommit = 0;
299 struct recordGroup rGroupDef;
305 rGroupDef.useSDR = 0;
307 rGroupDef.groupName = NULL;
308 rGroupDef.databaseName = NULL;
309 rGroupDef.path = NULL;
310 rGroupDef.recordId = NULL;
311 rGroupDef.recordType = NULL;
312 rGroupDef.flagStoreData = -1;
313 rGroupDef.flagStoreKeys = -1;
314 rGroupDef.flagRw = 1;
315 rGroupDef.databaseNamePath = 0;
316 rGroupDef.explainDatabase = 0;
317 rGroupDef.fileVerboseLimit = 100000;
318 rGroupDef.zebra_maps = NULL;
319 rGroupDef.dh = data1_create ();
320 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
321 recTypes_default_handlers (rGroupDef.recTypes);
326 fprintf (stderr, "zebraidx [options] command <dir> ...\n"
328 " update <dir> Update index with files below <dir>.\n"
329 " If <dir> is empty filenames are read from stdin.\n"
330 " delete <dir> Delete index with files below <dir>.\n"
331 " commit Commit changes\n"
332 " clean Clean shadow files\n"
334 " -t <type> Index files as <type> (grs or text).\n"
335 " -c <config> Read configuration file <config>.\n"
336 " -g <group> Index files according to group settings.\n"
337 " -d <database> Records belong to Z39.50 database <database>.\n"
338 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
339 " -n Don't use shadow system.\n"
340 " -s Show analysis on stdout, but do no work.\n"
341 " -v <level> Set logging to <level>.\n"
342 " -l <file> Write log to <file>.\n"
343 " -f <n> Display information for the first <n> records.\n"
347 " -V Show version.\n"
351 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
355 , argv, argc, &arg)) != -2)
360 if(cmd == 0) /* command */
362 if (!common_resource)
364 logf (LOG_LOG, "zebra version %s %s",
365 ZEBRAVER, ZEBRADATE);
366 common_resource = res_open (configName ?
367 configName : FNAME_CONFIG);
368 if (!common_resource)
370 logf (LOG_FATAL, "cannot read file `%s'", configName);
373 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
377 bfs_create (res_get (common_resource, "register"));
380 logf (LOG_FATAL, "Cannot access register");
384 bf_lockDir (rGroupDef.bfs,
385 res_get (common_resource, "lockDir"));
386 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
388 if (!strcmp (arg, "update"))
390 else if (!strcmp (arg, "update1"))
392 else if (!strcmp (arg, "update2"))
394 else if (!strcmp (arg, "dump"))
396 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
398 else if (!strcmp (arg, "commit"))
400 rval = res_get (common_resource, "shadow");
401 zebraIndexLock (rGroupDef.bfs, 1, rval);
403 bf_cache (rGroupDef.bfs, rval);
406 logf (LOG_FATAL, "Cannot perform commit");
407 logf (LOG_FATAL, "No shadow area defined");
410 if (bf_commitExists (rGroupDef.bfs))
412 logf (LOG_LOG, "commit start");
413 zebraIndexLockMsg ("c");
415 logf (LOG_LOG, "commit execute");
416 bf_commitExec (rGroupDef.bfs);
420 zebraIndexLockMsg ("d");
422 logf (LOG_LOG, "commit clean");
423 bf_commitClean (rGroupDef.bfs, rval);
426 logf (LOG_LOG, "cothing to commit");
428 else if (!strcmp (arg, "clean"))
430 rval = res_get (common_resource, "shadow");
431 zebraIndexLock (rGroupDef.bfs, 1, rval);
432 if (bf_commitExists (rGroupDef.bfs))
434 zebraIndexLockMsg ("d");
436 logf (LOG_LOG, "commit clean");
437 bf_commitClean (rGroupDef.bfs, rval);
440 logf (LOG_LOG, "nothing to clean");
442 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
445 rval = res_get (common_resource, "shadow");
446 zebraIndexLock (rGroupDef.bfs, 0, rval);
449 bf_cache (rGroupDef.bfs, rval);
450 zebraIndexLockMsg ("r");
452 records = rec_open (rGroupDef.bfs, 0, 0);
453 rec_prstat (records);
454 rec_close (&records);
455 inv_prstat (rGroupDef.bfs);
457 else if (!strcmp (arg, "compact"))
459 rval = res_get (common_resource, "shadow");
460 zebraIndexLock (rGroupDef.bfs, 0, rval);
463 bf_cache (rGroupDef.bfs, rval);
464 zebraIndexLockMsg ("r");
466 inv_compact(rGroupDef.bfs);
470 logf (LOG_FATAL, "unknown command: %s", arg);
476 struct recordGroup rGroup;
478 rval = res_get (common_resource, "shadow");
479 zebraIndexLock (rGroupDef.bfs, 0, rval);
480 if (rGroupDef.flagRw)
482 if (rval && *rval && !disableCommit)
484 bf_cache (rGroupDef.bfs, rval);
485 zebraIndexLockMsg ("r");
489 bf_cache (rGroupDef.bfs, 0);
490 zebraIndexLockMsg ("w");
494 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
499 if (!key_open (&rGroup, mem_max))
501 logf (LOG_LOG, "updating %s", rGroup.path);
502 repositoryUpdate (&rGroup);
503 nsections = key_close (&rGroup);
507 if (!key_open (&rGroup, mem_max))
509 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
510 repositoryUpdate (&rGroup);
516 if (!key_open (&rGroup,mem_max))
518 logf (LOG_LOG, "deleting %s", rGroup.path);
519 repositoryDelete (&rGroup);
520 nsections = key_close (&rGroup);
524 logf (LOG_LOG, "dumping %s", rGroup.path);
525 repositoryShow (&rGroup);
537 logf (LOG_LOG, "merging with index");
538 key_input (rGroup.bfs, nsections, 60);
543 log_event_end (NULL, NULL);
548 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
549 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
552 fprintf (stderr, " WIN32 Debug\n");
554 fprintf (stderr, " WIN32 Release\n");
558 fprintf (stderr, "libbzip2\n"
559 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
563 log_init_level (log_mask_str(arg));
567 mem_max = 1024*1024*atoi(arg);
569 rGroupDef.databaseName = arg;
571 rGroupDef.flagRw = 0;
573 rGroupDef.groupName = arg;
575 rGroupDef.fileVerboseLimit = atoi(arg);
579 rGroupDef.recordType = arg;
584 rGroupDef.useSDR = 1;
587 logf (LOG_WARN, "unknown option '-%s'", arg);
589 recTypes_destroy (rGroupDef.recTypes);
593 bfs_destroy (rGroupDef.bfs);
595 data1_destroy (rGroupDef.dh);