2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.76 2000-03-20 19:08:36 adam
8 * Added remote record import using Z39.50 extended services and Segment
11 * Revision 1.75 1999/12/08 22:44:45 adam
12 * Zebra/Z'mbol dependencies added.
14 * Revision 1.74 1999/12/08 15:03:11 adam
15 * Implemented bf_reset.
18 * Revision 1.73 1999/11/30 13:48:03 adam
19 * Improved installation. Updated for inclusion of YAZ header files.
21 * Revision 1.72 1999/10/14 14:33:50 adam
22 * Added truncation 5=106.
24 * Revision 1.71 1999/09/08 12:12:06 adam
27 * Revision 1.70 1999/09/07 07:19:21 adam
28 * Work on character mapping. Implemented replace rules.
30 * Revision 1.69 1999/07/21 08:31:33 adam
31 * More version info on WIN32.
33 * Revision 1.68 1999/07/14 10:59:26 adam
34 * Changed functions isc_getmethod, isams_getmethod.
35 * Improved fatal error handling (such as missing EXPLAIN schema).
37 * Revision 1.67 1999/07/06 12:28:04 adam
38 * Updated record index structure. Format includes version ID. Compression
39 * algorithm ID is stored for each record block.
41 * Revision 1.66 1999/06/25 13:48:02 adam
42 * Updated MSVC project files.
43 * Added BZIP2 record compression (not very well tested).
45 * Revision 1.65 1999/05/20 12:57:18 adam
46 * Implemented TCL filter. Updated recctrl system.
48 * Revision 1.64 1999/05/15 14:36:38 adam
49 * Updated dictionary. Implemented "compression" of dictionary.
51 * Revision 1.63 1999/03/09 16:27:49 adam
52 * More work on SDRKit integration.
54 * Revision 1.62 1999/02/18 15:01:24 adam
57 * Revision 1.61 1999/02/02 14:51:00 adam
58 * Updated WIN32 code specific sections. Changed header.
60 * Revision 1.60 1998/10/28 10:54:39 adam
63 * Revision 1.59 1998/10/16 08:14:32 adam
64 * Updated record control system.
66 * Revision 1.58 1998/06/08 14:43:13 adam
67 * Added suport for EXPLAIN Proxy servers - added settings databasePath
68 * and explainDatabase to facilitate this. Increased maximum number
69 * of databases and attributes in one register.
71 * Revision 1.57 1998/05/20 10:12:18 adam
72 * Implemented automatic EXPLAIN database maintenance.
73 * Modified Zebra to work with ASN.1 compiled version of YAZ.
75 * Revision 1.56 1998/03/05 08:45:12 adam
76 * New result set model and modular ranking system. Moved towards
77 * descent server API. System information stored as "SGML" records.
79 * Revision 1.55 1998/01/26 10:37:34 adam
82 * Revision 1.54 1998/01/12 15:04:08 adam
83 * The test option (-s) only uses read-lock (and not write lock).
85 * Revision 1.53 1997/11/18 10:05:08 adam
86 * Changed character map facility so that admin can specify character
87 * mapping files for each register type, w, p, etc.
89 * Revision 1.52 1997/10/31 12:34:04 adam
90 * Added missing nmem_init.
92 * Revision 1.51 1997/10/27 14:33:05 adam
93 * Moved towards generic character mapping depending on "structure"
94 * field in abstract syntax file. Fixed a few memory leaks. Fixed
95 * bug with negative integers when doing searches with relational
98 * Revision 1.50 1997/09/25 14:55:52 adam
101 * Revision 1.49 1997/09/17 12:19:15 adam
102 * Zebra version corresponds to YAZ version 1.4.
103 * Changed Zebra server so that it doesn't depend on global common_resource.
105 * Revision 1.48 1997/09/09 13:38:08 adam
106 * Partial port to WIN95/NT.
108 * Revision 1.47 1997/02/12 20:39:46 adam
109 * Implemented options -f <n> that limits the log to the first <n>
111 * Changed some log messages also.
113 * Revision 1.46 1997/02/10 10:20:13 adam
114 * Flag fileVerboseFlag set to 0 (default).
116 * Revision 1.45 1996/11/08 11:10:26 adam
117 * Buffers used during file match got bigger.
118 * Compressed ISAM support everywhere.
119 * Bug fixes regarding masking characters in queries.
120 * Redesigned Regexp-2 queries.
122 * Revision 1.44 1996/10/29 14:09:48 adam
123 * Use of cisam system - enabled if setting isamc is 1.
125 * Revision 1.43 1996/06/06 12:08:42 quinn
126 * Added showRecord function
128 * Revision 1.42 1996/05/31 09:07:01 quinn
129 * Work on character-set handling
131 * Revision 1.41 1996/05/14 14:04:34 adam
132 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
135 * Revision 1.40 1996/04/26 10:00:23 adam
136 * Added option -V to zebraidx to display version information.
137 * Removed stupid warnings from file update.
139 * Revision 1.39 1996/04/09 10:05:21 adam
140 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
142 * Revision 1.38 1996/03/26 16:01:14 adam
143 * New setting lockPath: directory of various lock files.
145 * Revision 1.37 1996/03/19 12:43:26 adam
146 * Bug fix: File update traversal didn't handle trailing slashes correctly.
147 * Bug fix: Update of sub directory groups wasn't handled correctly.
149 * Revision 1.36 1996/02/12 18:45:37 adam
150 * New fileVerboseFlag in record group control.
152 * Revision 1.35 1996/02/12 15:56:11 adam
153 * New code command: unread.
155 * Revision 1.34 1996/02/07 14:06:39 adam
156 * Better progress report during register merge.
157 * New command: clean - removes temporary shadow files.
159 * Revision 1.33 1996/02/06 17:11:18 adam
162 * Revision 1.32 1996/02/01 20:50:04 adam
163 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
164 * was never called - happens when no commands are specified.
166 * Revision 1.31 1996/01/08 19:15:46 adam
167 * New input filter that works!
169 * Revision 1.30 1995/12/12 16:00:59 adam
170 * System call sync(2) used after update/commit.
171 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
174 * Revision 1.29 1995/12/11 11:43:30 adam
175 * Locking based on fcntl instead of flock.
176 * Setting commitEnable removed. Command line option -n can be used to
177 * prevent commit if commit setting is defined in the configuration file.
179 * Revision 1.28 1995/12/08 16:22:56 adam
180 * Work on update while servers are running. Three lock files introduced.
181 * The servers reload their registers when necessary, but they don't
182 * reestablish result sets yet.
184 * Revision 1.27 1995/12/07 17:38:47 adam
185 * Work locking mechanisms for concurrent updates/commit.
187 * Revision 1.26 1995/12/06 12:41:23 adam
188 * New command 'stat' for the index program.
189 * Filenames can be read from stdin by specifying '-'.
190 * Bug fix/enhancement of the transformation from terms to regular
191 * expressons in the search engine.
193 * Revision 1.25 1995/12/01 16:24:39 adam
194 * Commit files use separate meta file area.
196 * Revision 1.24 1995/11/30 17:01:38 adam
197 * New setting commitCache: points to commit directories/files.
198 * New command commit: commits at the end of a zebraidx run.
200 * Revision 1.23 1995/11/30 08:34:31 adam
201 * Started work on commit facility.
202 * Changed a few malloc/free to xmalloc/xfree.
204 * Revision 1.22 1995/11/28 09:09:42 adam
205 * Zebra config renamed.
206 * Use setting 'recordId' to identify record now.
207 * Bug fix in recindex.c: rec_release_blocks was invokeded even
208 * though the blocks were already released.
209 * File traversal properly deletes records when needed.
211 * Revision 1.21 1995/11/27 14:27:39 adam
212 * Renamed 'update' command to 'dir'.
214 * Revision 1.20 1995/11/27 13:58:53 adam
215 * New option -t. storeStore data implemented in server.
217 * Revision 1.19 1995/11/25 10:24:06 adam
218 * More record fields - they are enumerated now.
219 * New options: flagStoreData flagStoreKey.
221 * Revision 1.18 1995/11/22 17:19:17 adam
222 * Record management uses the bfile system.
224 * Revision 1.17 1995/11/21 15:01:16 adam
225 * New general match criteria implemented.
226 * New feature: document groups.
228 * Revision 1.16 1995/11/20 11:56:27 adam
229 * Work on new traversal.
231 * Revision 1.15 1995/11/01 16:25:51 quinn
232 * *** empty log message ***
234 * Revision 1.14 1995/10/17 18:02:09 adam
235 * New feature: databases. Implemented as prefix to words in dictionary.
237 * Revision 1.13 1995/10/10 12:24:39 adam
238 * Temporary sort files are compressed.
240 * Revision 1.12 1995/10/04 16:57:20 adam
241 * Key input and merge sort in one pass.
243 * Revision 1.11 1995/09/29 14:01:45 adam
246 * Revision 1.10 1995/09/28 14:22:57 adam
247 * Sort uses smaller temporary files.
249 * Revision 1.9 1995/09/14 07:48:24 adam
250 * Record control management.
252 * Revision 1.8 1995/09/06 16:11:18 adam
253 * Option: only one word key per file.
255 * Revision 1.7 1995/09/05 15:28:39 adam
256 * More work on search engine.
258 * Revision 1.6 1995/09/04 12:33:43 adam
259 * Various cleanup. YAZ util used instead.
261 * Revision 1.5 1995/09/04 09:10:39 adam
262 * More work on index add/del/update.
263 * Merge sort implemented.
264 * Initial work on z39 server.
266 * Revision 1.4 1995/09/01 14:06:36 adam
267 * Split of work into more files.
269 * Revision 1.3 1995/09/01 10:57:07 adam
272 * Revision 1.2 1995/09/01 10:30:24 adam
273 * More work on indexing. Not working yet.
275 * Revision 1.1 1995/08/31 14:50:24 adam
276 * New simple file index tool.
287 #include <yaz/data1.h>
289 #include "recindex.h"
296 #include "zebrasdr.h"
301 Res common_resource = 0;
304 int main (int argc, char **argv)
309 char *configName = FNAME_CONFIG;
311 int disableCommit = 0;
314 struct recordGroup rGroupDef;
320 rGroupDef.useSDR = 0;
322 rGroupDef.groupName = NULL;
323 rGroupDef.databaseName = NULL;
324 rGroupDef.path = NULL;
325 rGroupDef.recordId = NULL;
326 rGroupDef.recordType = NULL;
327 rGroupDef.flagStoreData = -1;
328 rGroupDef.flagStoreKeys = -1;
329 rGroupDef.flagRw = 1;
330 rGroupDef.databaseNamePath = 0;
331 rGroupDef.explainDatabase = 0;
332 rGroupDef.fileVerboseLimit = 100000;
333 rGroupDef.zebra_maps = NULL;
334 rGroupDef.dh = data1_create ();
335 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
336 recTypes_default_handlers (rGroupDef.recTypes);
341 fprintf (stderr, "%s [options] command <dir> ...\n"
343 " update <dir> Update index with files below <dir>.\n"
344 " If <dir> is empty filenames are read from stdin.\n"
345 " delete <dir> Delete index with files below <dir>.\n"
346 " commit Commit changes\n"
347 " clean Clean shadow files\n"
349 " -t <type> Index files as <type> (grs or text).\n"
350 " -c <config> Read configuration file <config>.\n"
351 " -g <group> Index files according to group settings.\n"
352 " -d <database> Records belong to Z39.50 database <database>.\n"
353 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
354 " -n Don't use shadow system.\n"
355 " -s Show analysis on stdout, but do no work.\n"
356 " -v <level> Set logging to <level>.\n"
357 " -l <file> Write log to <file>.\n"
358 " -f <n> Display information for the first <n> records.\n"
362 " -V Show version.\n", *argv
366 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
370 , argv, argc, &arg)) != -2)
375 if(cmd == 0) /* command */
377 if (!common_resource)
380 logf (LOG_LOG, "zmbol version %s %s",
381 ZEBRAVER, ZEBRADATE);
383 logf (LOG_LOG, "zebra version %s %s",
384 ZEBRAVER, ZEBRADATE);
386 common_resource = res_open (configName ?
387 configName : FNAME_CONFIG);
388 if (!common_resource)
390 logf (LOG_FATAL, "cannot read file `%s'", configName);
393 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
397 bfs_create (res_get (common_resource, "register"));
400 logf (LOG_FATAL, "Cannot access register");
404 bf_lockDir (rGroupDef.bfs,
405 res_get (common_resource, "lockDir"));
406 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
408 if (!strcmp (arg, "update"))
410 else if (!strcmp (arg, "update1"))
412 else if (!strcmp (arg, "update2"))
414 else if (!strcmp (arg, "dump"))
416 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
418 else if (!strcmp (arg, "init"))
421 rval = res_get (common_resource, "shadow");
422 zebraIndexLock (rGroupDef.bfs, 0, rval);
423 zebraIndexLockMsg ("w");
424 bf_reset (rGroupDef.bfs);
426 else if (!strcmp (arg, "commit"))
428 rval = res_get (common_resource, "shadow");
429 zebraIndexLock (rGroupDef.bfs, 1, rval);
431 bf_cache (rGroupDef.bfs, rval);
434 logf (LOG_FATAL, "Cannot perform commit");
435 logf (LOG_FATAL, "No shadow area defined");
438 if (bf_commitExists (rGroupDef.bfs))
440 logf (LOG_LOG, "commit start");
441 zebraIndexLockMsg ("c");
443 logf (LOG_LOG, "commit execute");
444 bf_commitExec (rGroupDef.bfs);
448 zebraIndexLockMsg ("d");
450 logf (LOG_LOG, "commit clean");
451 bf_commitClean (rGroupDef.bfs, rval);
454 logf (LOG_LOG, "cothing to commit");
456 else if (!strcmp (arg, "clean"))
458 rval = res_get (common_resource, "shadow");
459 zebraIndexLock (rGroupDef.bfs, 1, rval);
460 if (bf_commitExists (rGroupDef.bfs))
462 zebraIndexLockMsg ("d");
464 logf (LOG_LOG, "commit clean");
465 bf_commitClean (rGroupDef.bfs, rval);
468 logf (LOG_LOG, "nothing to clean");
470 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
473 rval = res_get (common_resource, "shadow");
474 zebraIndexLock (rGroupDef.bfs, 0, rval);
477 bf_cache (rGroupDef.bfs, rval);
478 zebraIndexLockMsg ("r");
480 records = rec_open (rGroupDef.bfs, 0, 0);
481 rec_prstat (records);
482 rec_close (&records);
483 inv_prstat (rGroupDef.bfs);
485 else if (!strcmp (arg, "compact"))
487 rval = res_get (common_resource, "shadow");
488 zebraIndexLock (rGroupDef.bfs, 0, rval);
491 bf_cache (rGroupDef.bfs, rval);
492 zebraIndexLockMsg ("r");
494 inv_compact(rGroupDef.bfs);
498 logf (LOG_FATAL, "unknown command: %s", arg);
504 struct recordGroup rGroup;
507 /* For zebra, delete lock file and reset register */
508 if (rGroupDef.flagRw)
511 bf_reset (rGroupDef.bfs);
514 rval = res_get (common_resource, "shadow");
515 zebraIndexLock (rGroupDef.bfs, 0, rval);
516 if (rGroupDef.flagRw)
518 if (rval && *rval && !disableCommit)
520 bf_cache (rGroupDef.bfs, rval);
521 zebraIndexLockMsg ("r");
525 bf_cache (rGroupDef.bfs, 0);
526 zebraIndexLockMsg ("w");
530 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
535 if (!key_open (&rGroup, mem_max))
537 logf (LOG_LOG, "updating %s", rGroup.path);
538 repositoryUpdate (&rGroup);
539 nsections = key_close (&rGroup);
543 if (!key_open (&rGroup, mem_max))
545 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
546 repositoryUpdate (&rGroup);
552 if (!key_open (&rGroup,mem_max))
554 logf (LOG_LOG, "deleting %s", rGroup.path);
555 repositoryDelete (&rGroup);
556 nsections = key_close (&rGroup);
560 logf (LOG_LOG, "dumping %s", rGroup.path);
561 repositoryShow (&rGroup);
573 logf (LOG_LOG, "merging with index");
574 key_input (rGroup.bfs, nsections, 60, common_resource);
579 log_event_end (NULL, NULL);
585 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
587 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
589 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
592 fprintf (stderr, " WIN32 Debug\n");
594 fprintf (stderr, " WIN32 Release\n");
598 fprintf (stderr, "libbzip2\n"
599 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
603 log_init_level (log_mask_str(arg));
607 mem_max = 1024*1024*atoi(arg);
609 rGroupDef.databaseName = arg;
611 rGroupDef.flagRw = 0;
613 rGroupDef.groupName = arg;
615 rGroupDef.fileVerboseLimit = atoi(arg);
619 rGroupDef.recordType = arg;
624 rGroupDef.useSDR = 1;
627 logf (LOG_WARN, "unknown option '-%s'", arg);
629 recTypes_destroy (rGroupDef.recTypes);
633 bfs_destroy (rGroupDef.bfs);
635 data1_destroy (rGroupDef.dh);