2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.77 2000-09-05 14:04:05 adam
8 * Updates for prefix 'yaz_' for YAZ log functions.
10 * Revision 1.76 2000/03/20 19:08:36 adam
11 * Added remote record import using Z39.50 extended services and Segment
14 * Revision 1.75 1999/12/08 22:44:45 adam
15 * Zebra/Z'mbol dependencies added.
17 * Revision 1.74 1999/12/08 15:03:11 adam
18 * Implemented bf_reset.
21 * Revision 1.73 1999/11/30 13:48:03 adam
22 * Improved installation. Updated for inclusion of YAZ header files.
24 * Revision 1.72 1999/10/14 14:33:50 adam
25 * Added truncation 5=106.
27 * Revision 1.71 1999/09/08 12:12:06 adam
30 * Revision 1.70 1999/09/07 07:19:21 adam
31 * Work on character mapping. Implemented replace rules.
33 * Revision 1.69 1999/07/21 08:31:33 adam
34 * More version info on WIN32.
36 * Revision 1.68 1999/07/14 10:59:26 adam
37 * Changed functions isc_getmethod, isams_getmethod.
38 * Improved fatal error handling (such as missing EXPLAIN schema).
40 * Revision 1.67 1999/07/06 12:28:04 adam
41 * Updated record index structure. Format includes version ID. Compression
42 * algorithm ID is stored for each record block.
44 * Revision 1.66 1999/06/25 13:48:02 adam
45 * Updated MSVC project files.
46 * Added BZIP2 record compression (not very well tested).
48 * Revision 1.65 1999/05/20 12:57:18 adam
49 * Implemented TCL filter. Updated recctrl system.
51 * Revision 1.64 1999/05/15 14:36:38 adam
52 * Updated dictionary. Implemented "compression" of dictionary.
54 * Revision 1.63 1999/03/09 16:27:49 adam
55 * More work on SDRKit integration.
57 * Revision 1.62 1999/02/18 15:01:24 adam
60 * Revision 1.61 1999/02/02 14:51:00 adam
61 * Updated WIN32 code specific sections. Changed header.
63 * Revision 1.60 1998/10/28 10:54:39 adam
66 * Revision 1.59 1998/10/16 08:14:32 adam
67 * Updated record control system.
69 * Revision 1.58 1998/06/08 14:43:13 adam
70 * Added suport for EXPLAIN Proxy servers - added settings databasePath
71 * and explainDatabase to facilitate this. Increased maximum number
72 * of databases and attributes in one register.
74 * Revision 1.57 1998/05/20 10:12:18 adam
75 * Implemented automatic EXPLAIN database maintenance.
76 * Modified Zebra to work with ASN.1 compiled version of YAZ.
78 * Revision 1.56 1998/03/05 08:45:12 adam
79 * New result set model and modular ranking system. Moved towards
80 * descent server API. System information stored as "SGML" records.
82 * Revision 1.55 1998/01/26 10:37:34 adam
85 * Revision 1.54 1998/01/12 15:04:08 adam
86 * The test option (-s) only uses read-lock (and not write lock).
88 * Revision 1.53 1997/11/18 10:05:08 adam
89 * Changed character map facility so that admin can specify character
90 * mapping files for each register type, w, p, etc.
92 * Revision 1.52 1997/10/31 12:34:04 adam
93 * Added missing nmem_init.
95 * Revision 1.51 1997/10/27 14:33:05 adam
96 * Moved towards generic character mapping depending on "structure"
97 * field in abstract syntax file. Fixed a few memory leaks. Fixed
98 * bug with negative integers when doing searches with relational
101 * Revision 1.50 1997/09/25 14:55:52 adam
104 * Revision 1.49 1997/09/17 12:19:15 adam
105 * Zebra version corresponds to YAZ version 1.4.
106 * Changed Zebra server so that it doesn't depend on global common_resource.
108 * Revision 1.48 1997/09/09 13:38:08 adam
109 * Partial port to WIN95/NT.
111 * Revision 1.47 1997/02/12 20:39:46 adam
112 * Implemented options -f <n> that limits the log to the first <n>
114 * Changed some log messages also.
116 * Revision 1.46 1997/02/10 10:20:13 adam
117 * Flag fileVerboseFlag set to 0 (default).
119 * Revision 1.45 1996/11/08 11:10:26 adam
120 * Buffers used during file match got bigger.
121 * Compressed ISAM support everywhere.
122 * Bug fixes regarding masking characters in queries.
123 * Redesigned Regexp-2 queries.
125 * Revision 1.44 1996/10/29 14:09:48 adam
126 * Use of cisam system - enabled if setting isamc is 1.
128 * Revision 1.43 1996/06/06 12:08:42 quinn
129 * Added showRecord function
131 * Revision 1.42 1996/05/31 09:07:01 quinn
132 * Work on character-set handling
134 * Revision 1.41 1996/05/14 14:04:34 adam
135 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
138 * Revision 1.40 1996/04/26 10:00:23 adam
139 * Added option -V to zebraidx to display version information.
140 * Removed stupid warnings from file update.
142 * Revision 1.39 1996/04/09 10:05:21 adam
143 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
145 * Revision 1.38 1996/03/26 16:01:14 adam
146 * New setting lockPath: directory of various lock files.
148 * Revision 1.37 1996/03/19 12:43:26 adam
149 * Bug fix: File update traversal didn't handle trailing slashes correctly.
150 * Bug fix: Update of sub directory groups wasn't handled correctly.
152 * Revision 1.36 1996/02/12 18:45:37 adam
153 * New fileVerboseFlag in record group control.
155 * Revision 1.35 1996/02/12 15:56:11 adam
156 * New code command: unread.
158 * Revision 1.34 1996/02/07 14:06:39 adam
159 * Better progress report during register merge.
160 * New command: clean - removes temporary shadow files.
162 * Revision 1.33 1996/02/06 17:11:18 adam
165 * Revision 1.32 1996/02/01 20:50:04 adam
166 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
167 * was never called - happens when no commands are specified.
169 * Revision 1.31 1996/01/08 19:15:46 adam
170 * New input filter that works!
172 * Revision 1.30 1995/12/12 16:00:59 adam
173 * System call sync(2) used after update/commit.
174 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
177 * Revision 1.29 1995/12/11 11:43:30 adam
178 * Locking based on fcntl instead of flock.
179 * Setting commitEnable removed. Command line option -n can be used to
180 * prevent commit if commit setting is defined in the configuration file.
182 * Revision 1.28 1995/12/08 16:22:56 adam
183 * Work on update while servers are running. Three lock files introduced.
184 * The servers reload their registers when necessary, but they don't
185 * reestablish result sets yet.
187 * Revision 1.27 1995/12/07 17:38:47 adam
188 * Work locking mechanisms for concurrent updates/commit.
190 * Revision 1.26 1995/12/06 12:41:23 adam
191 * New command 'stat' for the index program.
192 * Filenames can be read from stdin by specifying '-'.
193 * Bug fix/enhancement of the transformation from terms to regular
194 * expressons in the search engine.
196 * Revision 1.25 1995/12/01 16:24:39 adam
197 * Commit files use separate meta file area.
199 * Revision 1.24 1995/11/30 17:01:38 adam
200 * New setting commitCache: points to commit directories/files.
201 * New command commit: commits at the end of a zebraidx run.
203 * Revision 1.23 1995/11/30 08:34:31 adam
204 * Started work on commit facility.
205 * Changed a few malloc/free to xmalloc/xfree.
207 * Revision 1.22 1995/11/28 09:09:42 adam
208 * Zebra config renamed.
209 * Use setting 'recordId' to identify record now.
210 * Bug fix in recindex.c: rec_release_blocks was invokeded even
211 * though the blocks were already released.
212 * File traversal properly deletes records when needed.
214 * Revision 1.21 1995/11/27 14:27:39 adam
215 * Renamed 'update' command to 'dir'.
217 * Revision 1.20 1995/11/27 13:58:53 adam
218 * New option -t. storeStore data implemented in server.
220 * Revision 1.19 1995/11/25 10:24:06 adam
221 * More record fields - they are enumerated now.
222 * New options: flagStoreData flagStoreKey.
224 * Revision 1.18 1995/11/22 17:19:17 adam
225 * Record management uses the bfile system.
227 * Revision 1.17 1995/11/21 15:01:16 adam
228 * New general match criteria implemented.
229 * New feature: document groups.
231 * Revision 1.16 1995/11/20 11:56:27 adam
232 * Work on new traversal.
234 * Revision 1.15 1995/11/01 16:25:51 quinn
235 * *** empty log message ***
237 * Revision 1.14 1995/10/17 18:02:09 adam
238 * New feature: databases. Implemented as prefix to words in dictionary.
240 * Revision 1.13 1995/10/10 12:24:39 adam
241 * Temporary sort files are compressed.
243 * Revision 1.12 1995/10/04 16:57:20 adam
244 * Key input and merge sort in one pass.
246 * Revision 1.11 1995/09/29 14:01:45 adam
249 * Revision 1.10 1995/09/28 14:22:57 adam
250 * Sort uses smaller temporary files.
252 * Revision 1.9 1995/09/14 07:48:24 adam
253 * Record control management.
255 * Revision 1.8 1995/09/06 16:11:18 adam
256 * Option: only one word key per file.
258 * Revision 1.7 1995/09/05 15:28:39 adam
259 * More work on search engine.
261 * Revision 1.6 1995/09/04 12:33:43 adam
262 * Various cleanup. YAZ util used instead.
264 * Revision 1.5 1995/09/04 09:10:39 adam
265 * More work on index add/del/update.
266 * Merge sort implemented.
267 * Initial work on z39 server.
269 * Revision 1.4 1995/09/01 14:06:36 adam
270 * Split of work into more files.
272 * Revision 1.3 1995/09/01 10:57:07 adam
275 * Revision 1.2 1995/09/01 10:30:24 adam
276 * More work on indexing. Not working yet.
278 * Revision 1.1 1995/08/31 14:50:24 adam
279 * New simple file index tool.
290 #include <yaz/data1.h>
292 #include "recindex.h"
299 #include "zebrasdr.h"
304 Res common_resource = 0;
307 int main (int argc, char **argv)
312 char *configName = FNAME_CONFIG;
314 int disableCommit = 0;
317 struct recordGroup rGroupDef;
323 rGroupDef.useSDR = 0;
325 rGroupDef.groupName = NULL;
326 rGroupDef.databaseName = NULL;
327 rGroupDef.path = NULL;
328 rGroupDef.recordId = NULL;
329 rGroupDef.recordType = NULL;
330 rGroupDef.flagStoreData = -1;
331 rGroupDef.flagStoreKeys = -1;
332 rGroupDef.flagRw = 1;
333 rGroupDef.databaseNamePath = 0;
334 rGroupDef.explainDatabase = 0;
335 rGroupDef.fileVerboseLimit = 100000;
336 rGroupDef.zebra_maps = NULL;
337 rGroupDef.dh = data1_create ();
338 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
339 recTypes_default_handlers (rGroupDef.recTypes);
344 fprintf (stderr, "%s [options] command <dir> ...\n"
346 " update <dir> Update index with files below <dir>.\n"
347 " If <dir> is empty filenames are read from stdin.\n"
348 " delete <dir> Delete index with files below <dir>.\n"
349 " commit Commit changes\n"
350 " clean Clean shadow files\n"
352 " -t <type> Index files as <type> (grs or text).\n"
353 " -c <config> Read configuration file <config>.\n"
354 " -g <group> Index files according to group settings.\n"
355 " -d <database> Records belong to Z39.50 database <database>.\n"
356 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
357 " -n Don't use shadow system.\n"
358 " -s Show analysis on stdout, but do no work.\n"
359 " -v <level> Set logging to <level>.\n"
360 " -l <file> Write log to <file>.\n"
361 " -f <n> Display information for the first <n> records.\n"
365 " -V Show version.\n", *argv
369 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
373 , argv, argc, &arg)) != -2)
378 if(cmd == 0) /* command */
380 if (!common_resource)
383 logf (LOG_LOG, "zmbol version %s %s",
384 ZEBRAVER, ZEBRADATE);
386 logf (LOG_LOG, "zebra version %s %s",
387 ZEBRAVER, ZEBRADATE);
389 common_resource = res_open (configName ?
390 configName : FNAME_CONFIG);
391 if (!common_resource)
393 logf (LOG_FATAL, "cannot read file `%s'", configName);
396 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
400 bfs_create (res_get (common_resource, "register"));
403 logf (LOG_FATAL, "Cannot access register");
407 bf_lockDir (rGroupDef.bfs,
408 res_get (common_resource, "lockDir"));
409 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
411 if (!strcmp (arg, "update"))
413 else if (!strcmp (arg, "update1"))
415 else if (!strcmp (arg, "update2"))
417 else if (!strcmp (arg, "dump"))
419 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
421 else if (!strcmp (arg, "init"))
424 rval = res_get (common_resource, "shadow");
425 zebraIndexLock (rGroupDef.bfs, 0, rval);
426 zebraIndexLockMsg ("w");
427 bf_reset (rGroupDef.bfs);
429 else if (!strcmp (arg, "commit"))
431 rval = res_get (common_resource, "shadow");
432 zebraIndexLock (rGroupDef.bfs, 1, rval);
434 bf_cache (rGroupDef.bfs, rval);
437 logf (LOG_FATAL, "Cannot perform commit");
438 logf (LOG_FATAL, "No shadow area defined");
441 if (bf_commitExists (rGroupDef.bfs))
443 logf (LOG_LOG, "commit start");
444 zebraIndexLockMsg ("c");
446 logf (LOG_LOG, "commit execute");
447 bf_commitExec (rGroupDef.bfs);
451 zebraIndexLockMsg ("d");
453 logf (LOG_LOG, "commit clean");
454 bf_commitClean (rGroupDef.bfs, rval);
457 logf (LOG_LOG, "cothing to commit");
459 else if (!strcmp (arg, "clean"))
461 rval = res_get (common_resource, "shadow");
462 zebraIndexLock (rGroupDef.bfs, 1, rval);
463 if (bf_commitExists (rGroupDef.bfs))
465 zebraIndexLockMsg ("d");
467 logf (LOG_LOG, "commit clean");
468 bf_commitClean (rGroupDef.bfs, rval);
471 logf (LOG_LOG, "nothing to clean");
473 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
476 rval = res_get (common_resource, "shadow");
477 zebraIndexLock (rGroupDef.bfs, 0, rval);
480 bf_cache (rGroupDef.bfs, rval);
481 zebraIndexLockMsg ("r");
483 records = rec_open (rGroupDef.bfs, 0, 0);
484 rec_prstat (records);
485 rec_close (&records);
486 inv_prstat (rGroupDef.bfs);
488 else if (!strcmp (arg, "compact"))
490 rval = res_get (common_resource, "shadow");
491 zebraIndexLock (rGroupDef.bfs, 0, rval);
494 bf_cache (rGroupDef.bfs, rval);
495 zebraIndexLockMsg ("r");
497 inv_compact(rGroupDef.bfs);
501 logf (LOG_FATAL, "unknown command: %s", arg);
507 struct recordGroup rGroup;
510 /* For zebra, delete lock file and reset register */
511 if (rGroupDef.flagRw)
514 bf_reset (rGroupDef.bfs);
517 rval = res_get (common_resource, "shadow");
518 zebraIndexLock (rGroupDef.bfs, 0, rval);
519 if (rGroupDef.flagRw)
521 if (rval && *rval && !disableCommit)
523 bf_cache (rGroupDef.bfs, rval);
524 zebraIndexLockMsg ("r");
528 bf_cache (rGroupDef.bfs, 0);
529 zebraIndexLockMsg ("w");
533 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
538 if (!key_open (&rGroup, mem_max))
540 logf (LOG_LOG, "updating %s", rGroup.path);
541 repositoryUpdate (&rGroup);
542 nsections = key_close (&rGroup);
546 if (!key_open (&rGroup, mem_max))
548 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
549 repositoryUpdate (&rGroup);
555 if (!key_open (&rGroup,mem_max))
557 logf (LOG_LOG, "deleting %s", rGroup.path);
558 repositoryDelete (&rGroup);
559 nsections = key_close (&rGroup);
563 logf (LOG_LOG, "dumping %s", rGroup.path);
564 repositoryShow (&rGroup);
576 logf (LOG_LOG, "merging with index");
577 key_input (rGroup.bfs, nsections, 60, common_resource);
582 log_event_end (NULL, NULL);
588 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
590 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
592 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
595 fprintf (stderr, " WIN32 Debug\n");
597 fprintf (stderr, " WIN32 Release\n");
601 fprintf (stderr, "libbzip2\n"
602 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
606 yaz_log_init_level (yaz_log_mask_str(arg));
608 yaz_log_init_file (arg);
610 mem_max = 1024*1024*atoi(arg);
612 rGroupDef.databaseName = arg;
614 rGroupDef.flagRw = 0;
616 rGroupDef.groupName = arg;
618 rGroupDef.fileVerboseLimit = atoi(arg);
622 rGroupDef.recordType = arg;
627 rGroupDef.useSDR = 1;
630 logf (LOG_WARN, "unknown option '-%s'", arg);
632 recTypes_destroy (rGroupDef.recTypes);
636 bfs_destroy (rGroupDef.bfs);
638 data1_destroy (rGroupDef.dh);