2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.78 2000-10-17 12:37:09 adam
8 * Fixed notification of live-updates. Fixed minor problem with mf_init
9 * where it didn't handle shadow area file names correctly.
11 * Revision 1.77 2000/09/05 14:04:05 adam
12 * Updates for prefix 'yaz_' for YAZ log functions.
14 * Revision 1.76 2000/03/20 19:08:36 adam
15 * Added remote record import using Z39.50 extended services and Segment
18 * Revision 1.75 1999/12/08 22:44:45 adam
19 * Zebra/Z'mbol dependencies added.
21 * Revision 1.74 1999/12/08 15:03:11 adam
22 * Implemented bf_reset.
25 * Revision 1.73 1999/11/30 13:48:03 adam
26 * Improved installation. Updated for inclusion of YAZ header files.
28 * Revision 1.72 1999/10/14 14:33:50 adam
29 * Added truncation 5=106.
31 * Revision 1.71 1999/09/08 12:12:06 adam
34 * Revision 1.70 1999/09/07 07:19:21 adam
35 * Work on character mapping. Implemented replace rules.
37 * Revision 1.69 1999/07/21 08:31:33 adam
38 * More version info on WIN32.
40 * Revision 1.68 1999/07/14 10:59:26 adam
41 * Changed functions isc_getmethod, isams_getmethod.
42 * Improved fatal error handling (such as missing EXPLAIN schema).
44 * Revision 1.67 1999/07/06 12:28:04 adam
45 * Updated record index structure. Format includes version ID. Compression
46 * algorithm ID is stored for each record block.
48 * Revision 1.66 1999/06/25 13:48:02 adam
49 * Updated MSVC project files.
50 * Added BZIP2 record compression (not very well tested).
52 * Revision 1.65 1999/05/20 12:57:18 adam
53 * Implemented TCL filter. Updated recctrl system.
55 * Revision 1.64 1999/05/15 14:36:38 adam
56 * Updated dictionary. Implemented "compression" of dictionary.
58 * Revision 1.63 1999/03/09 16:27:49 adam
59 * More work on SDRKit integration.
61 * Revision 1.62 1999/02/18 15:01:24 adam
64 * Revision 1.61 1999/02/02 14:51:00 adam
65 * Updated WIN32 code specific sections. Changed header.
67 * Revision 1.60 1998/10/28 10:54:39 adam
70 * Revision 1.59 1998/10/16 08:14:32 adam
71 * Updated record control system.
73 * Revision 1.58 1998/06/08 14:43:13 adam
74 * Added suport for EXPLAIN Proxy servers - added settings databasePath
75 * and explainDatabase to facilitate this. Increased maximum number
76 * of databases and attributes in one register.
78 * Revision 1.57 1998/05/20 10:12:18 adam
79 * Implemented automatic EXPLAIN database maintenance.
80 * Modified Zebra to work with ASN.1 compiled version of YAZ.
82 * Revision 1.56 1998/03/05 08:45:12 adam
83 * New result set model and modular ranking system. Moved towards
84 * descent server API. System information stored as "SGML" records.
86 * Revision 1.55 1998/01/26 10:37:34 adam
89 * Revision 1.54 1998/01/12 15:04:08 adam
90 * The test option (-s) only uses read-lock (and not write lock).
92 * Revision 1.53 1997/11/18 10:05:08 adam
93 * Changed character map facility so that admin can specify character
94 * mapping files for each register type, w, p, etc.
96 * Revision 1.52 1997/10/31 12:34:04 adam
97 * Added missing nmem_init.
99 * Revision 1.51 1997/10/27 14:33:05 adam
100 * Moved towards generic character mapping depending on "structure"
101 * field in abstract syntax file. Fixed a few memory leaks. Fixed
102 * bug with negative integers when doing searches with relational
105 * Revision 1.50 1997/09/25 14:55:52 adam
108 * Revision 1.49 1997/09/17 12:19:15 adam
109 * Zebra version corresponds to YAZ version 1.4.
110 * Changed Zebra server so that it doesn't depend on global common_resource.
112 * Revision 1.48 1997/09/09 13:38:08 adam
113 * Partial port to WIN95/NT.
115 * Revision 1.47 1997/02/12 20:39:46 adam
116 * Implemented options -f <n> that limits the log to the first <n>
118 * Changed some log messages also.
120 * Revision 1.46 1997/02/10 10:20:13 adam
121 * Flag fileVerboseFlag set to 0 (default).
123 * Revision 1.45 1996/11/08 11:10:26 adam
124 * Buffers used during file match got bigger.
125 * Compressed ISAM support everywhere.
126 * Bug fixes regarding masking characters in queries.
127 * Redesigned Regexp-2 queries.
129 * Revision 1.44 1996/10/29 14:09:48 adam
130 * Use of cisam system - enabled if setting isamc is 1.
132 * Revision 1.43 1996/06/06 12:08:42 quinn
133 * Added showRecord function
135 * Revision 1.42 1996/05/31 09:07:01 quinn
136 * Work on character-set handling
138 * Revision 1.41 1996/05/14 14:04:34 adam
139 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
142 * Revision 1.40 1996/04/26 10:00:23 adam
143 * Added option -V to zebraidx to display version information.
144 * Removed stupid warnings from file update.
146 * Revision 1.39 1996/04/09 10:05:21 adam
147 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
149 * Revision 1.38 1996/03/26 16:01:14 adam
150 * New setting lockPath: directory of various lock files.
152 * Revision 1.37 1996/03/19 12:43:26 adam
153 * Bug fix: File update traversal didn't handle trailing slashes correctly.
154 * Bug fix: Update of sub directory groups wasn't handled correctly.
156 * Revision 1.36 1996/02/12 18:45:37 adam
157 * New fileVerboseFlag in record group control.
159 * Revision 1.35 1996/02/12 15:56:11 adam
160 * New code command: unread.
162 * Revision 1.34 1996/02/07 14:06:39 adam
163 * Better progress report during register merge.
164 * New command: clean - removes temporary shadow files.
166 * Revision 1.33 1996/02/06 17:11:18 adam
169 * Revision 1.32 1996/02/01 20:50:04 adam
170 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
171 * was never called - happens when no commands are specified.
173 * Revision 1.31 1996/01/08 19:15:46 adam
174 * New input filter that works!
176 * Revision 1.30 1995/12/12 16:00:59 adam
177 * System call sync(2) used after update/commit.
178 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
181 * Revision 1.29 1995/12/11 11:43:30 adam
182 * Locking based on fcntl instead of flock.
183 * Setting commitEnable removed. Command line option -n can be used to
184 * prevent commit if commit setting is defined in the configuration file.
186 * Revision 1.28 1995/12/08 16:22:56 adam
187 * Work on update while servers are running. Three lock files introduced.
188 * The servers reload their registers when necessary, but they don't
189 * reestablish result sets yet.
191 * Revision 1.27 1995/12/07 17:38:47 adam
192 * Work locking mechanisms for concurrent updates/commit.
194 * Revision 1.26 1995/12/06 12:41:23 adam
195 * New command 'stat' for the index program.
196 * Filenames can be read from stdin by specifying '-'.
197 * Bug fix/enhancement of the transformation from terms to regular
198 * expressons in the search engine.
200 * Revision 1.25 1995/12/01 16:24:39 adam
201 * Commit files use separate meta file area.
203 * Revision 1.24 1995/11/30 17:01:38 adam
204 * New setting commitCache: points to commit directories/files.
205 * New command commit: commits at the end of a zebraidx run.
207 * Revision 1.23 1995/11/30 08:34:31 adam
208 * Started work on commit facility.
209 * Changed a few malloc/free to xmalloc/xfree.
211 * Revision 1.22 1995/11/28 09:09:42 adam
212 * Zebra config renamed.
213 * Use setting 'recordId' to identify record now.
214 * Bug fix in recindex.c: rec_release_blocks was invokeded even
215 * though the blocks were already released.
216 * File traversal properly deletes records when needed.
218 * Revision 1.21 1995/11/27 14:27:39 adam
219 * Renamed 'update' command to 'dir'.
221 * Revision 1.20 1995/11/27 13:58:53 adam
222 * New option -t. storeStore data implemented in server.
224 * Revision 1.19 1995/11/25 10:24:06 adam
225 * More record fields - they are enumerated now.
226 * New options: flagStoreData flagStoreKey.
228 * Revision 1.18 1995/11/22 17:19:17 adam
229 * Record management uses the bfile system.
231 * Revision 1.17 1995/11/21 15:01:16 adam
232 * New general match criteria implemented.
233 * New feature: document groups.
235 * Revision 1.16 1995/11/20 11:56:27 adam
236 * Work on new traversal.
238 * Revision 1.15 1995/11/01 16:25:51 quinn
239 * *** empty log message ***
241 * Revision 1.14 1995/10/17 18:02:09 adam
242 * New feature: databases. Implemented as prefix to words in dictionary.
244 * Revision 1.13 1995/10/10 12:24:39 adam
245 * Temporary sort files are compressed.
247 * Revision 1.12 1995/10/04 16:57:20 adam
248 * Key input and merge sort in one pass.
250 * Revision 1.11 1995/09/29 14:01:45 adam
253 * Revision 1.10 1995/09/28 14:22:57 adam
254 * Sort uses smaller temporary files.
256 * Revision 1.9 1995/09/14 07:48:24 adam
257 * Record control management.
259 * Revision 1.8 1995/09/06 16:11:18 adam
260 * Option: only one word key per file.
262 * Revision 1.7 1995/09/05 15:28:39 adam
263 * More work on search engine.
265 * Revision 1.6 1995/09/04 12:33:43 adam
266 * Various cleanup. YAZ util used instead.
268 * Revision 1.5 1995/09/04 09:10:39 adam
269 * More work on index add/del/update.
270 * Merge sort implemented.
271 * Initial work on z39 server.
273 * Revision 1.4 1995/09/01 14:06:36 adam
274 * Split of work into more files.
276 * Revision 1.3 1995/09/01 10:57:07 adam
279 * Revision 1.2 1995/09/01 10:30:24 adam
280 * More work on indexing. Not working yet.
282 * Revision 1.1 1995/08/31 14:50:24 adam
283 * New simple file index tool.
294 #include <yaz/data1.h>
296 #include "recindex.h"
303 #include "zebrasdr.h"
308 Res common_resource = 0;
311 int main (int argc, char **argv)
316 char *configName = FNAME_CONFIG;
318 int disableCommit = 0;
321 struct recordGroup rGroupDef;
327 rGroupDef.useSDR = 0;
329 rGroupDef.groupName = NULL;
330 rGroupDef.databaseName = NULL;
331 rGroupDef.path = NULL;
332 rGroupDef.recordId = NULL;
333 rGroupDef.recordType = NULL;
334 rGroupDef.flagStoreData = -1;
335 rGroupDef.flagStoreKeys = -1;
336 rGroupDef.flagRw = 1;
337 rGroupDef.databaseNamePath = 0;
338 rGroupDef.explainDatabase = 0;
339 rGroupDef.fileVerboseLimit = 100000;
340 rGroupDef.zebra_maps = NULL;
341 rGroupDef.dh = data1_create ();
342 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
343 recTypes_default_handlers (rGroupDef.recTypes);
348 fprintf (stderr, "%s [options] command <dir> ...\n"
350 " update <dir> Update index with files below <dir>.\n"
351 " If <dir> is empty filenames are read from stdin.\n"
352 " delete <dir> Delete index with files below <dir>.\n"
353 " commit Commit changes\n"
354 " clean Clean shadow files\n"
356 " -t <type> Index files as <type> (grs or text).\n"
357 " -c <config> Read configuration file <config>.\n"
358 " -g <group> Index files according to group settings.\n"
359 " -d <database> Records belong to Z39.50 database <database>.\n"
360 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
361 " -n Don't use shadow system.\n"
362 " -s Show analysis on stdout, but do no work.\n"
363 " -v <level> Set logging to <level>.\n"
364 " -l <file> Write log to <file>.\n"
365 " -f <n> Display information for the first <n> records.\n"
369 " -V Show version.\n", *argv
373 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
377 , argv, argc, &arg)) != -2)
382 if(cmd == 0) /* command */
384 if (!common_resource)
387 logf (LOG_LOG, "zmbol version %s %s",
388 ZEBRAVER, ZEBRADATE);
390 logf (LOG_LOG, "zebra version %s %s",
391 ZEBRAVER, ZEBRADATE);
393 common_resource = res_open (configName ?
394 configName : FNAME_CONFIG);
395 if (!common_resource)
397 logf (LOG_FATAL, "cannot read file `%s'", configName);
400 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
404 bfs_create (res_get (common_resource, "register"));
407 logf (LOG_FATAL, "Cannot access register");
411 bf_lockDir (rGroupDef.bfs,
412 res_get (common_resource, "lockDir"));
413 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
415 if (!strcmp (arg, "update"))
417 else if (!strcmp (arg, "update1"))
419 else if (!strcmp (arg, "update2"))
421 else if (!strcmp (arg, "dump"))
423 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
425 else if (!strcmp (arg, "init"))
428 rval = res_get (common_resource, "shadow");
429 zebraIndexLock (rGroupDef.bfs, 0, rval);
431 bf_cache (rGroupDef.bfs, rval);
432 zebraIndexLockMsg ("w");
433 bf_reset (rGroupDef.bfs);
435 else if (!strcmp (arg, "commit"))
437 rval = res_get (common_resource, "shadow");
438 zebraIndexLock (rGroupDef.bfs, 1, rval);
440 bf_cache (rGroupDef.bfs, rval);
443 logf (LOG_FATAL, "Cannot perform commit");
444 logf (LOG_FATAL, "No shadow area defined");
447 if (bf_commitExists (rGroupDef.bfs))
449 logf (LOG_LOG, "commit start");
450 zebraIndexLockMsg ("c");
452 logf (LOG_LOG, "commit execute");
453 bf_commitExec (rGroupDef.bfs);
457 zebraIndexLockMsg ("d");
459 logf (LOG_LOG, "commit clean");
460 bf_commitClean (rGroupDef.bfs, rval);
463 logf (LOG_LOG, "nothing to commit");
465 else if (!strcmp (arg, "clean"))
467 rval = res_get (common_resource, "shadow");
468 zebraIndexLock (rGroupDef.bfs, 1, rval);
469 if (bf_commitExists (rGroupDef.bfs))
471 zebraIndexLockMsg ("d");
473 logf (LOG_LOG, "commit clean");
474 bf_commitClean (rGroupDef.bfs, rval);
477 logf (LOG_LOG, "nothing to clean");
479 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
482 rval = res_get (common_resource, "shadow");
483 zebraIndexLock (rGroupDef.bfs, 0, rval);
486 bf_cache (rGroupDef.bfs, rval);
487 zebraIndexLockMsg ("r");
489 records = rec_open (rGroupDef.bfs, 0, 0);
490 rec_prstat (records);
491 rec_close (&records);
492 inv_prstat (rGroupDef.bfs);
494 else if (!strcmp (arg, "compact"))
496 rval = res_get (common_resource, "shadow");
497 zebraIndexLock (rGroupDef.bfs, 0, rval);
500 bf_cache (rGroupDef.bfs, rval);
501 zebraIndexLockMsg ("r");
503 inv_compact(rGroupDef.bfs);
507 logf (LOG_FATAL, "unknown command: %s", arg);
513 struct recordGroup rGroup;
516 /* For zebra, delete lock file and reset register */
517 if (rGroupDef.flagRw)
520 bf_reset (rGroupDef.bfs);
523 rval = res_get (common_resource, "shadow");
524 zebraIndexLock (rGroupDef.bfs, 0, rval);
525 if (rGroupDef.flagRw)
527 if (rval && *rval && !disableCommit)
529 bf_cache (rGroupDef.bfs, rval);
530 zebraIndexLockMsg ("r");
534 bf_cache (rGroupDef.bfs, 0);
535 zebraIndexLockMsg ("w");
539 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
544 if (!key_open (&rGroup, mem_max))
546 logf (LOG_LOG, "updating %s", rGroup.path);
547 repositoryUpdate (&rGroup);
548 nsections = key_close (&rGroup);
552 if (!key_open (&rGroup, mem_max))
554 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
555 repositoryUpdate (&rGroup);
561 if (!key_open (&rGroup,mem_max))
563 logf (LOG_LOG, "deleting %s", rGroup.path);
564 repositoryDelete (&rGroup);
565 nsections = key_close (&rGroup);
569 logf (LOG_LOG, "dumping %s", rGroup.path);
570 repositoryShow (&rGroup);
582 logf (LOG_LOG, "merging with index");
583 key_input (rGroup.bfs, nsections, 60, common_resource);
588 log_event_end (NULL, NULL);
594 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
596 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
598 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
601 fprintf (stderr, " WIN32 Debug\n");
603 fprintf (stderr, " WIN32 Release\n");
607 fprintf (stderr, "libbzip2\n"
608 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
612 yaz_log_init_level (yaz_log_mask_str(arg));
614 yaz_log_init_file (arg);
616 mem_max = 1024*1024*atoi(arg);
618 rGroupDef.databaseName = arg;
620 rGroupDef.flagRw = 0;
622 rGroupDef.groupName = arg;
624 rGroupDef.fileVerboseLimit = atoi(arg);
628 rGroupDef.recordType = arg;
633 rGroupDef.useSDR = 1;
636 logf (LOG_WARN, "unknown option '-%s'", arg);
638 recTypes_destroy (rGroupDef.recTypes);
642 bfs_destroy (rGroupDef.bfs);
644 data1_destroy (rGroupDef.dh);