2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.79 2001-10-01 08:56:58 adam
8 * For UNIX process ID is logged.
10 * Revision 1.78 2000/10/17 12:37:09 adam
11 * Fixed notification of live-updates. Fixed minor problem with mf_init
12 * where it didn't handle shadow area file names correctly.
14 * Revision 1.77 2000/09/05 14:04:05 adam
15 * Updates for prefix 'yaz_' for YAZ log functions.
17 * Revision 1.76 2000/03/20 19:08:36 adam
18 * Added remote record import using Z39.50 extended services and Segment
21 * Revision 1.75 1999/12/08 22:44:45 adam
22 * Zebra/Z'mbol dependencies added.
24 * Revision 1.74 1999/12/08 15:03:11 adam
25 * Implemented bf_reset.
28 * Revision 1.73 1999/11/30 13:48:03 adam
29 * Improved installation. Updated for inclusion of YAZ header files.
31 * Revision 1.72 1999/10/14 14:33:50 adam
32 * Added truncation 5=106.
34 * Revision 1.71 1999/09/08 12:12:06 adam
37 * Revision 1.70 1999/09/07 07:19:21 adam
38 * Work on character mapping. Implemented replace rules.
40 * Revision 1.69 1999/07/21 08:31:33 adam
41 * More version info on WIN32.
43 * Revision 1.68 1999/07/14 10:59:26 adam
44 * Changed functions isc_getmethod, isams_getmethod.
45 * Improved fatal error handling (such as missing EXPLAIN schema).
47 * Revision 1.67 1999/07/06 12:28:04 adam
48 * Updated record index structure. Format includes version ID. Compression
49 * algorithm ID is stored for each record block.
51 * Revision 1.66 1999/06/25 13:48:02 adam
52 * Updated MSVC project files.
53 * Added BZIP2 record compression (not very well tested).
55 * Revision 1.65 1999/05/20 12:57:18 adam
56 * Implemented TCL filter. Updated recctrl system.
58 * Revision 1.64 1999/05/15 14:36:38 adam
59 * Updated dictionary. Implemented "compression" of dictionary.
61 * Revision 1.63 1999/03/09 16:27:49 adam
62 * More work on SDRKit integration.
64 * Revision 1.62 1999/02/18 15:01:24 adam
67 * Revision 1.61 1999/02/02 14:51:00 adam
68 * Updated WIN32 code specific sections. Changed header.
70 * Revision 1.60 1998/10/28 10:54:39 adam
73 * Revision 1.59 1998/10/16 08:14:32 adam
74 * Updated record control system.
76 * Revision 1.58 1998/06/08 14:43:13 adam
77 * Added suport for EXPLAIN Proxy servers - added settings databasePath
78 * and explainDatabase to facilitate this. Increased maximum number
79 * of databases and attributes in one register.
81 * Revision 1.57 1998/05/20 10:12:18 adam
82 * Implemented automatic EXPLAIN database maintenance.
83 * Modified Zebra to work with ASN.1 compiled version of YAZ.
85 * Revision 1.56 1998/03/05 08:45:12 adam
86 * New result set model and modular ranking system. Moved towards
87 * descent server API. System information stored as "SGML" records.
89 * Revision 1.55 1998/01/26 10:37:34 adam
92 * Revision 1.54 1998/01/12 15:04:08 adam
93 * The test option (-s) only uses read-lock (and not write lock).
95 * Revision 1.53 1997/11/18 10:05:08 adam
96 * Changed character map facility so that admin can specify character
97 * mapping files for each register type, w, p, etc.
99 * Revision 1.52 1997/10/31 12:34:04 adam
100 * Added missing nmem_init.
102 * Revision 1.51 1997/10/27 14:33:05 adam
103 * Moved towards generic character mapping depending on "structure"
104 * field in abstract syntax file. Fixed a few memory leaks. Fixed
105 * bug with negative integers when doing searches with relational
108 * Revision 1.50 1997/09/25 14:55:52 adam
111 * Revision 1.49 1997/09/17 12:19:15 adam
112 * Zebra version corresponds to YAZ version 1.4.
113 * Changed Zebra server so that it doesn't depend on global common_resource.
115 * Revision 1.48 1997/09/09 13:38:08 adam
116 * Partial port to WIN95/NT.
118 * Revision 1.47 1997/02/12 20:39:46 adam
119 * Implemented options -f <n> that limits the log to the first <n>
121 * Changed some log messages also.
123 * Revision 1.46 1997/02/10 10:20:13 adam
124 * Flag fileVerboseFlag set to 0 (default).
126 * Revision 1.45 1996/11/08 11:10:26 adam
127 * Buffers used during file match got bigger.
128 * Compressed ISAM support everywhere.
129 * Bug fixes regarding masking characters in queries.
130 * Redesigned Regexp-2 queries.
132 * Revision 1.44 1996/10/29 14:09:48 adam
133 * Use of cisam system - enabled if setting isamc is 1.
135 * Revision 1.43 1996/06/06 12:08:42 quinn
136 * Added showRecord function
138 * Revision 1.42 1996/05/31 09:07:01 quinn
139 * Work on character-set handling
141 * Revision 1.41 1996/05/14 14:04:34 adam
142 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
145 * Revision 1.40 1996/04/26 10:00:23 adam
146 * Added option -V to zebraidx to display version information.
147 * Removed stupid warnings from file update.
149 * Revision 1.39 1996/04/09 10:05:21 adam
150 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
152 * Revision 1.38 1996/03/26 16:01:14 adam
153 * New setting lockPath: directory of various lock files.
155 * Revision 1.37 1996/03/19 12:43:26 adam
156 * Bug fix: File update traversal didn't handle trailing slashes correctly.
157 * Bug fix: Update of sub directory groups wasn't handled correctly.
159 * Revision 1.36 1996/02/12 18:45:37 adam
160 * New fileVerboseFlag in record group control.
162 * Revision 1.35 1996/02/12 15:56:11 adam
163 * New code command: unread.
165 * Revision 1.34 1996/02/07 14:06:39 adam
166 * Better progress report during register merge.
167 * New command: clean - removes temporary shadow files.
169 * Revision 1.33 1996/02/06 17:11:18 adam
172 * Revision 1.32 1996/02/01 20:50:04 adam
173 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
174 * was never called - happens when no commands are specified.
176 * Revision 1.31 1996/01/08 19:15:46 adam
177 * New input filter that works!
179 * Revision 1.30 1995/12/12 16:00:59 adam
180 * System call sync(2) used after update/commit.
181 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
184 * Revision 1.29 1995/12/11 11:43:30 adam
185 * Locking based on fcntl instead of flock.
186 * Setting commitEnable removed. Command line option -n can be used to
187 * prevent commit if commit setting is defined in the configuration file.
189 * Revision 1.28 1995/12/08 16:22:56 adam
190 * Work on update while servers are running. Three lock files introduced.
191 * The servers reload their registers when necessary, but they don't
192 * reestablish result sets yet.
194 * Revision 1.27 1995/12/07 17:38:47 adam
195 * Work locking mechanisms for concurrent updates/commit.
197 * Revision 1.26 1995/12/06 12:41:23 adam
198 * New command 'stat' for the index program.
199 * Filenames can be read from stdin by specifying '-'.
200 * Bug fix/enhancement of the transformation from terms to regular
201 * expressons in the search engine.
203 * Revision 1.25 1995/12/01 16:24:39 adam
204 * Commit files use separate meta file area.
206 * Revision 1.24 1995/11/30 17:01:38 adam
207 * New setting commitCache: points to commit directories/files.
208 * New command commit: commits at the end of a zebraidx run.
210 * Revision 1.23 1995/11/30 08:34:31 adam
211 * Started work on commit facility.
212 * Changed a few malloc/free to xmalloc/xfree.
214 * Revision 1.22 1995/11/28 09:09:42 adam
215 * Zebra config renamed.
216 * Use setting 'recordId' to identify record now.
217 * Bug fix in recindex.c: rec_release_blocks was invokeded even
218 * though the blocks were already released.
219 * File traversal properly deletes records when needed.
221 * Revision 1.21 1995/11/27 14:27:39 adam
222 * Renamed 'update' command to 'dir'.
224 * Revision 1.20 1995/11/27 13:58:53 adam
225 * New option -t. storeStore data implemented in server.
227 * Revision 1.19 1995/11/25 10:24:06 adam
228 * More record fields - they are enumerated now.
229 * New options: flagStoreData flagStoreKey.
231 * Revision 1.18 1995/11/22 17:19:17 adam
232 * Record management uses the bfile system.
234 * Revision 1.17 1995/11/21 15:01:16 adam
235 * New general match criteria implemented.
236 * New feature: document groups.
238 * Revision 1.16 1995/11/20 11:56:27 adam
239 * Work on new traversal.
241 * Revision 1.15 1995/11/01 16:25:51 quinn
242 * *** empty log message ***
244 * Revision 1.14 1995/10/17 18:02:09 adam
245 * New feature: databases. Implemented as prefix to words in dictionary.
247 * Revision 1.13 1995/10/10 12:24:39 adam
248 * Temporary sort files are compressed.
250 * Revision 1.12 1995/10/04 16:57:20 adam
251 * Key input and merge sort in one pass.
253 * Revision 1.11 1995/09/29 14:01:45 adam
256 * Revision 1.10 1995/09/28 14:22:57 adam
257 * Sort uses smaller temporary files.
259 * Revision 1.9 1995/09/14 07:48:24 adam
260 * Record control management.
262 * Revision 1.8 1995/09/06 16:11:18 adam
263 * Option: only one word key per file.
265 * Revision 1.7 1995/09/05 15:28:39 adam
266 * More work on search engine.
268 * Revision 1.6 1995/09/04 12:33:43 adam
269 * Various cleanup. YAZ util used instead.
271 * Revision 1.5 1995/09/04 09:10:39 adam
272 * More work on index add/del/update.
273 * Merge sort implemented.
274 * Initial work on z39 server.
276 * Revision 1.4 1995/09/01 14:06:36 adam
277 * Split of work into more files.
279 * Revision 1.3 1995/09/01 10:57:07 adam
282 * Revision 1.2 1995/09/01 10:30:24 adam
283 * More work on indexing. Not working yet.
285 * Revision 1.1 1995/08/31 14:50:24 adam
286 * New simple file index tool.
298 #include <yaz/data1.h>
300 #include "recindex.h"
307 #include "zebrasdr.h"
312 Res common_resource = 0;
315 int main (int argc, char **argv)
320 char *configName = FNAME_CONFIG;
322 int disableCommit = 0;
325 struct recordGroup rGroupDef;
331 sprintf(nbuf, "%.40s(%d)", *argv, getpid());
332 yaz_log_init_prefix (nbuf);
337 rGroupDef.useSDR = 0;
339 rGroupDef.groupName = NULL;
340 rGroupDef.databaseName = NULL;
341 rGroupDef.path = NULL;
342 rGroupDef.recordId = NULL;
343 rGroupDef.recordType = NULL;
344 rGroupDef.flagStoreData = -1;
345 rGroupDef.flagStoreKeys = -1;
346 rGroupDef.flagRw = 1;
347 rGroupDef.databaseNamePath = 0;
348 rGroupDef.explainDatabase = 0;
349 rGroupDef.fileVerboseLimit = 100000;
350 rGroupDef.zebra_maps = NULL;
351 rGroupDef.dh = data1_create ();
352 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
353 recTypes_default_handlers (rGroupDef.recTypes);
358 fprintf (stderr, "%s [options] command <dir> ...\n"
360 " update <dir> Update index with files below <dir>.\n"
361 " If <dir> is empty filenames are read from stdin.\n"
362 " delete <dir> Delete index with files below <dir>.\n"
363 " commit Commit changes\n"
364 " clean Clean shadow files\n"
366 " -t <type> Index files as <type> (grs or text).\n"
367 " -c <config> Read configuration file <config>.\n"
368 " -g <group> Index files according to group settings.\n"
369 " -d <database> Records belong to Z39.50 database <database>.\n"
370 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
371 " -n Don't use shadow system.\n"
372 " -s Show analysis on stdout, but do no work.\n"
373 " -v <level> Set logging to <level>.\n"
374 " -l <file> Write log to <file>.\n"
375 " -f <n> Display information for the first <n> records.\n"
379 " -V Show version.\n", *argv
383 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
387 , argv, argc, &arg)) != -2)
392 if(cmd == 0) /* command */
394 if (!common_resource)
397 logf (LOG_LOG, "zmbol version %s %s",
398 ZEBRAVER, ZEBRADATE);
400 logf (LOG_LOG, "zebra version %s %s",
401 ZEBRAVER, ZEBRADATE);
403 common_resource = res_open (configName ?
404 configName : FNAME_CONFIG);
405 if (!common_resource)
407 logf (LOG_FATAL, "cannot read file `%s'", configName);
410 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
414 bfs_create (res_get (common_resource, "register"));
417 logf (LOG_FATAL, "Cannot access register");
421 bf_lockDir (rGroupDef.bfs,
422 res_get (common_resource, "lockDir"));
423 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
425 if (!strcmp (arg, "update"))
427 else if (!strcmp (arg, "update1"))
429 else if (!strcmp (arg, "update2"))
431 else if (!strcmp (arg, "dump"))
433 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
435 else if (!strcmp (arg, "init"))
438 rval = res_get (common_resource, "shadow");
439 zebraIndexLock (rGroupDef.bfs, 0, rval);
441 bf_cache (rGroupDef.bfs, rval);
442 zebraIndexLockMsg ("w");
443 bf_reset (rGroupDef.bfs);
445 else if (!strcmp (arg, "commit"))
447 rval = res_get (common_resource, "shadow");
448 zebraIndexLock (rGroupDef.bfs, 1, rval);
450 bf_cache (rGroupDef.bfs, rval);
453 logf (LOG_FATAL, "Cannot perform commit");
454 logf (LOG_FATAL, "No shadow area defined");
457 if (bf_commitExists (rGroupDef.bfs))
459 logf (LOG_LOG, "commit start");
460 zebraIndexLockMsg ("c");
462 logf (LOG_LOG, "commit execute");
463 bf_commitExec (rGroupDef.bfs);
467 zebraIndexLockMsg ("d");
469 logf (LOG_LOG, "commit clean");
470 bf_commitClean (rGroupDef.bfs, rval);
473 logf (LOG_LOG, "nothing to commit");
475 else if (!strcmp (arg, "clean"))
477 rval = res_get (common_resource, "shadow");
478 zebraIndexLock (rGroupDef.bfs, 1, rval);
479 if (bf_commitExists (rGroupDef.bfs))
481 zebraIndexLockMsg ("d");
483 logf (LOG_LOG, "commit clean");
484 bf_commitClean (rGroupDef.bfs, rval);
487 logf (LOG_LOG, "nothing to clean");
489 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
492 rval = res_get (common_resource, "shadow");
493 zebraIndexLock (rGroupDef.bfs, 0, rval);
496 bf_cache (rGroupDef.bfs, rval);
497 zebraIndexLockMsg ("r");
499 records = rec_open (rGroupDef.bfs, 0, 0);
500 rec_prstat (records);
501 rec_close (&records);
502 inv_prstat (rGroupDef.bfs);
504 else if (!strcmp (arg, "compact"))
506 rval = res_get (common_resource, "shadow");
507 zebraIndexLock (rGroupDef.bfs, 0, rval);
510 bf_cache (rGroupDef.bfs, rval);
511 zebraIndexLockMsg ("r");
513 inv_compact(rGroupDef.bfs);
517 logf (LOG_FATAL, "unknown command: %s", arg);
523 struct recordGroup rGroup;
526 /* For zebra, delete lock file and reset register */
527 if (rGroupDef.flagRw)
530 bf_reset (rGroupDef.bfs);
533 rval = res_get (common_resource, "shadow");
534 zebraIndexLock (rGroupDef.bfs, 0, rval);
535 if (rGroupDef.flagRw)
537 if (rval && *rval && !disableCommit)
539 bf_cache (rGroupDef.bfs, rval);
540 zebraIndexLockMsg ("r");
544 bf_cache (rGroupDef.bfs, 0);
545 zebraIndexLockMsg ("w");
549 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
554 if (!key_open (&rGroup, mem_max))
556 logf (LOG_LOG, "updating %s", rGroup.path);
557 repositoryUpdate (&rGroup);
558 nsections = key_close (&rGroup);
562 if (!key_open (&rGroup, mem_max))
564 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
565 repositoryUpdate (&rGroup);
571 if (!key_open (&rGroup,mem_max))
573 logf (LOG_LOG, "deleting %s", rGroup.path);
574 repositoryDelete (&rGroup);
575 nsections = key_close (&rGroup);
579 logf (LOG_LOG, "dumping %s", rGroup.path);
580 repositoryShow (&rGroup);
592 logf (LOG_LOG, "merging with index");
593 key_input (rGroup.bfs, nsections, 60, common_resource);
598 log_event_end (NULL, NULL);
604 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
606 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
608 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
611 fprintf (stderr, " WIN32 Debug\n");
613 fprintf (stderr, " WIN32 Release\n");
617 fprintf (stderr, "libbzip2\n"
618 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
622 yaz_log_init_level (yaz_log_mask_str(arg));
624 yaz_log_init_file (arg);
626 mem_max = 1024*1024*atoi(arg);
628 rGroupDef.databaseName = arg;
630 rGroupDef.flagRw = 0;
632 rGroupDef.groupName = arg;
634 rGroupDef.fileVerboseLimit = atoi(arg);
638 rGroupDef.recordType = arg;
643 rGroupDef.useSDR = 1;
646 logf (LOG_WARN, "unknown option '-%s'", arg);
648 recTypes_destroy (rGroupDef.recTypes);
652 bfs_destroy (rGroupDef.bfs);
654 data1_destroy (rGroupDef.dh);