2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.73 1999-11-30 13:48:03 adam
8 * Improved installation. Updated for inclusion of YAZ header files.
10 * Revision 1.72 1999/10/14 14:33:50 adam
11 * Added truncation 5=106.
13 * Revision 1.71 1999/09/08 12:12:06 adam
16 * Revision 1.70 1999/09/07 07:19:21 adam
17 * Work on character mapping. Implemented replace rules.
19 * Revision 1.69 1999/07/21 08:31:33 adam
20 * More version info on WIN32.
22 * Revision 1.68 1999/07/14 10:59:26 adam
23 * Changed functions isc_getmethod, isams_getmethod.
24 * Improved fatal error handling (such as missing EXPLAIN schema).
26 * Revision 1.67 1999/07/06 12:28:04 adam
27 * Updated record index structure. Format includes version ID. Compression
28 * algorithm ID is stored for each record block.
30 * Revision 1.66 1999/06/25 13:48:02 adam
31 * Updated MSVC project files.
32 * Added BZIP2 record compression (not very well tested).
34 * Revision 1.65 1999/05/20 12:57:18 adam
35 * Implemented TCL filter. Updated recctrl system.
37 * Revision 1.64 1999/05/15 14:36:38 adam
38 * Updated dictionary. Implemented "compression" of dictionary.
40 * Revision 1.63 1999/03/09 16:27:49 adam
41 * More work on SDRKit integration.
43 * Revision 1.62 1999/02/18 15:01:24 adam
46 * Revision 1.61 1999/02/02 14:51:00 adam
47 * Updated WIN32 code specific sections. Changed header.
49 * Revision 1.60 1998/10/28 10:54:39 adam
52 * Revision 1.59 1998/10/16 08:14:32 adam
53 * Updated record control system.
55 * Revision 1.58 1998/06/08 14:43:13 adam
56 * Added suport for EXPLAIN Proxy servers - added settings databasePath
57 * and explainDatabase to facilitate this. Increased maximum number
58 * of databases and attributes in one register.
60 * Revision 1.57 1998/05/20 10:12:18 adam
61 * Implemented automatic EXPLAIN database maintenance.
62 * Modified Zebra to work with ASN.1 compiled version of YAZ.
64 * Revision 1.56 1998/03/05 08:45:12 adam
65 * New result set model and modular ranking system. Moved towards
66 * descent server API. System information stored as "SGML" records.
68 * Revision 1.55 1998/01/26 10:37:34 adam
71 * Revision 1.54 1998/01/12 15:04:08 adam
72 * The test option (-s) only uses read-lock (and not write lock).
74 * Revision 1.53 1997/11/18 10:05:08 adam
75 * Changed character map facility so that admin can specify character
76 * mapping files for each register type, w, p, etc.
78 * Revision 1.52 1997/10/31 12:34:04 adam
79 * Added missing nmem_init.
81 * Revision 1.51 1997/10/27 14:33:05 adam
82 * Moved towards generic character mapping depending on "structure"
83 * field in abstract syntax file. Fixed a few memory leaks. Fixed
84 * bug with negative integers when doing searches with relational
87 * Revision 1.50 1997/09/25 14:55:52 adam
90 * Revision 1.49 1997/09/17 12:19:15 adam
91 * Zebra version corresponds to YAZ version 1.4.
92 * Changed Zebra server so that it doesn't depend on global common_resource.
94 * Revision 1.48 1997/09/09 13:38:08 adam
95 * Partial port to WIN95/NT.
97 * Revision 1.47 1997/02/12 20:39:46 adam
98 * Implemented options -f <n> that limits the log to the first <n>
100 * Changed some log messages also.
102 * Revision 1.46 1997/02/10 10:20:13 adam
103 * Flag fileVerboseFlag set to 0 (default).
105 * Revision 1.45 1996/11/08 11:10:26 adam
106 * Buffers used during file match got bigger.
107 * Compressed ISAM support everywhere.
108 * Bug fixes regarding masking characters in queries.
109 * Redesigned Regexp-2 queries.
111 * Revision 1.44 1996/10/29 14:09:48 adam
112 * Use of cisam system - enabled if setting isamc is 1.
114 * Revision 1.43 1996/06/06 12:08:42 quinn
115 * Added showRecord function
117 * Revision 1.42 1996/05/31 09:07:01 quinn
118 * Work on character-set handling
120 * Revision 1.41 1996/05/14 14:04:34 adam
121 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
124 * Revision 1.40 1996/04/26 10:00:23 adam
125 * Added option -V to zebraidx to display version information.
126 * Removed stupid warnings from file update.
128 * Revision 1.39 1996/04/09 10:05:21 adam
129 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
131 * Revision 1.38 1996/03/26 16:01:14 adam
132 * New setting lockPath: directory of various lock files.
134 * Revision 1.37 1996/03/19 12:43:26 adam
135 * Bug fix: File update traversal didn't handle trailing slashes correctly.
136 * Bug fix: Update of sub directory groups wasn't handled correctly.
138 * Revision 1.36 1996/02/12 18:45:37 adam
139 * New fileVerboseFlag in record group control.
141 * Revision 1.35 1996/02/12 15:56:11 adam
142 * New code command: unread.
144 * Revision 1.34 1996/02/07 14:06:39 adam
145 * Better progress report during register merge.
146 * New command: clean - removes temporary shadow files.
148 * Revision 1.33 1996/02/06 17:11:18 adam
151 * Revision 1.32 1996/02/01 20:50:04 adam
152 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
153 * was never called - happens when no commands are specified.
155 * Revision 1.31 1996/01/08 19:15:46 adam
156 * New input filter that works!
158 * Revision 1.30 1995/12/12 16:00:59 adam
159 * System call sync(2) used after update/commit.
160 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
163 * Revision 1.29 1995/12/11 11:43:30 adam
164 * Locking based on fcntl instead of flock.
165 * Setting commitEnable removed. Command line option -n can be used to
166 * prevent commit if commit setting is defined in the configuration file.
168 * Revision 1.28 1995/12/08 16:22:56 adam
169 * Work on update while servers are running. Three lock files introduced.
170 * The servers reload their registers when necessary, but they don't
171 * reestablish result sets yet.
173 * Revision 1.27 1995/12/07 17:38:47 adam
174 * Work locking mechanisms for concurrent updates/commit.
176 * Revision 1.26 1995/12/06 12:41:23 adam
177 * New command 'stat' for the index program.
178 * Filenames can be read from stdin by specifying '-'.
179 * Bug fix/enhancement of the transformation from terms to regular
180 * expressons in the search engine.
182 * Revision 1.25 1995/12/01 16:24:39 adam
183 * Commit files use separate meta file area.
185 * Revision 1.24 1995/11/30 17:01:38 adam
186 * New setting commitCache: points to commit directories/files.
187 * New command commit: commits at the end of a zebraidx run.
189 * Revision 1.23 1995/11/30 08:34:31 adam
190 * Started work on commit facility.
191 * Changed a few malloc/free to xmalloc/xfree.
193 * Revision 1.22 1995/11/28 09:09:42 adam
194 * Zebra config renamed.
195 * Use setting 'recordId' to identify record now.
196 * Bug fix in recindex.c: rec_release_blocks was invokeded even
197 * though the blocks were already released.
198 * File traversal properly deletes records when needed.
200 * Revision 1.21 1995/11/27 14:27:39 adam
201 * Renamed 'update' command to 'dir'.
203 * Revision 1.20 1995/11/27 13:58:53 adam
204 * New option -t. storeStore data implemented in server.
206 * Revision 1.19 1995/11/25 10:24:06 adam
207 * More record fields - they are enumerated now.
208 * New options: flagStoreData flagStoreKey.
210 * Revision 1.18 1995/11/22 17:19:17 adam
211 * Record management uses the bfile system.
213 * Revision 1.17 1995/11/21 15:01:16 adam
214 * New general match criteria implemented.
215 * New feature: document groups.
217 * Revision 1.16 1995/11/20 11:56:27 adam
218 * Work on new traversal.
220 * Revision 1.15 1995/11/01 16:25:51 quinn
221 * *** empty log message ***
223 * Revision 1.14 1995/10/17 18:02:09 adam
224 * New feature: databases. Implemented as prefix to words in dictionary.
226 * Revision 1.13 1995/10/10 12:24:39 adam
227 * Temporary sort files are compressed.
229 * Revision 1.12 1995/10/04 16:57:20 adam
230 * Key input and merge sort in one pass.
232 * Revision 1.11 1995/09/29 14:01:45 adam
235 * Revision 1.10 1995/09/28 14:22:57 adam
236 * Sort uses smaller temporary files.
238 * Revision 1.9 1995/09/14 07:48:24 adam
239 * Record control management.
241 * Revision 1.8 1995/09/06 16:11:18 adam
242 * Option: only one word key per file.
244 * Revision 1.7 1995/09/05 15:28:39 adam
245 * More work on search engine.
247 * Revision 1.6 1995/09/04 12:33:43 adam
248 * Various cleanup. YAZ util used instead.
250 * Revision 1.5 1995/09/04 09:10:39 adam
251 * More work on index add/del/update.
252 * Merge sort implemented.
253 * Initial work on z39 server.
255 * Revision 1.4 1995/09/01 14:06:36 adam
256 * Split of work into more files.
258 * Revision 1.3 1995/09/01 10:57:07 adam
261 * Revision 1.2 1995/09/01 10:30:24 adam
262 * More work on indexing. Not working yet.
264 * Revision 1.1 1995/08/31 14:50:24 adam
265 * New simple file index tool.
276 #include <yaz/data1.h>
278 #include "recindex.h"
285 #include "zebrasdr.h"
290 Res common_resource = 0;
292 int main (int argc, char **argv)
297 char *configName = FNAME_CONFIG;
299 int disableCommit = 0;
302 struct recordGroup rGroupDef;
308 rGroupDef.useSDR = 0;
310 rGroupDef.groupName = NULL;
311 rGroupDef.databaseName = NULL;
312 rGroupDef.path = NULL;
313 rGroupDef.recordId = NULL;
314 rGroupDef.recordType = NULL;
315 rGroupDef.flagStoreData = -1;
316 rGroupDef.flagStoreKeys = -1;
317 rGroupDef.flagRw = 1;
318 rGroupDef.databaseNamePath = 0;
319 rGroupDef.explainDatabase = 0;
320 rGroupDef.fileVerboseLimit = 100000;
321 rGroupDef.zebra_maps = NULL;
322 rGroupDef.dh = data1_create ();
323 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
324 recTypes_default_handlers (rGroupDef.recTypes);
329 fprintf (stderr, "zebraidx [options] command <dir> ...\n"
331 " update <dir> Update index with files below <dir>.\n"
332 " If <dir> is empty filenames are read from stdin.\n"
333 " delete <dir> Delete index with files below <dir>.\n"
334 " commit Commit changes\n"
335 " clean Clean shadow files\n"
337 " -t <type> Index files as <type> (grs or text).\n"
338 " -c <config> Read configuration file <config>.\n"
339 " -g <group> Index files according to group settings.\n"
340 " -d <database> Records belong to Z39.50 database <database>.\n"
341 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
342 " -n Don't use shadow system.\n"
343 " -s Show analysis on stdout, but do no work.\n"
344 " -v <level> Set logging to <level>.\n"
345 " -l <file> Write log to <file>.\n"
346 " -f <n> Display information for the first <n> records.\n"
350 " -V Show version.\n"
354 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
358 , argv, argc, &arg)) != -2)
363 if(cmd == 0) /* command */
365 if (!common_resource)
367 logf (LOG_LOG, "zebra version %s %s",
368 ZEBRAVER, ZEBRADATE);
369 common_resource = res_open (configName ?
370 configName : FNAME_CONFIG);
371 if (!common_resource)
373 logf (LOG_FATAL, "cannot read file `%s'", configName);
376 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
380 bfs_create (res_get (common_resource, "register"));
383 logf (LOG_FATAL, "Cannot access register");
387 bf_lockDir (rGroupDef.bfs,
388 res_get (common_resource, "lockDir"));
389 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
391 if (!strcmp (arg, "update"))
393 else if (!strcmp (arg, "update1"))
395 else if (!strcmp (arg, "update2"))
397 else if (!strcmp (arg, "dump"))
399 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
401 else if (!strcmp (arg, "commit"))
403 rval = res_get (common_resource, "shadow");
404 zebraIndexLock (rGroupDef.bfs, 1, rval);
406 bf_cache (rGroupDef.bfs, rval);
409 logf (LOG_FATAL, "Cannot perform commit");
410 logf (LOG_FATAL, "No shadow area defined");
413 if (bf_commitExists (rGroupDef.bfs))
415 logf (LOG_LOG, "commit start");
416 zebraIndexLockMsg ("c");
418 logf (LOG_LOG, "commit execute");
419 bf_commitExec (rGroupDef.bfs);
423 zebraIndexLockMsg ("d");
425 logf (LOG_LOG, "commit clean");
426 bf_commitClean (rGroupDef.bfs, rval);
429 logf (LOG_LOG, "cothing to commit");
431 else if (!strcmp (arg, "clean"))
433 rval = res_get (common_resource, "shadow");
434 zebraIndexLock (rGroupDef.bfs, 1, rval);
435 if (bf_commitExists (rGroupDef.bfs))
437 zebraIndexLockMsg ("d");
439 logf (LOG_LOG, "commit clean");
440 bf_commitClean (rGroupDef.bfs, rval);
443 logf (LOG_LOG, "nothing to clean");
445 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
448 rval = res_get (common_resource, "shadow");
449 zebraIndexLock (rGroupDef.bfs, 0, rval);
452 bf_cache (rGroupDef.bfs, rval);
453 zebraIndexLockMsg ("r");
455 records = rec_open (rGroupDef.bfs, 0, 0);
456 rec_prstat (records);
457 rec_close (&records);
458 inv_prstat (rGroupDef.bfs);
460 else if (!strcmp (arg, "compact"))
462 rval = res_get (common_resource, "shadow");
463 zebraIndexLock (rGroupDef.bfs, 0, rval);
466 bf_cache (rGroupDef.bfs, rval);
467 zebraIndexLockMsg ("r");
469 inv_compact(rGroupDef.bfs);
473 logf (LOG_FATAL, "unknown command: %s", arg);
479 struct recordGroup rGroup;
481 rval = res_get (common_resource, "shadow");
482 zebraIndexLock (rGroupDef.bfs, 0, rval);
483 if (rGroupDef.flagRw)
485 if (rval && *rval && !disableCommit)
487 bf_cache (rGroupDef.bfs, rval);
488 zebraIndexLockMsg ("r");
492 bf_cache (rGroupDef.bfs, 0);
493 zebraIndexLockMsg ("w");
497 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
502 if (!key_open (&rGroup, mem_max))
504 logf (LOG_LOG, "updating %s", rGroup.path);
505 repositoryUpdate (&rGroup);
506 nsections = key_close (&rGroup);
510 if (!key_open (&rGroup, mem_max))
512 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
513 repositoryUpdate (&rGroup);
519 if (!key_open (&rGroup,mem_max))
521 logf (LOG_LOG, "deleting %s", rGroup.path);
522 repositoryDelete (&rGroup);
523 nsections = key_close (&rGroup);
527 logf (LOG_LOG, "dumping %s", rGroup.path);
528 repositoryShow (&rGroup);
540 logf (LOG_LOG, "merging with index");
541 key_input (rGroup.bfs, nsections, 60);
546 log_event_end (NULL, NULL);
551 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
552 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
555 fprintf (stderr, " WIN32 Debug\n");
557 fprintf (stderr, " WIN32 Release\n");
561 fprintf (stderr, "libbzip2\n"
562 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
566 log_init_level (log_mask_str(arg));
570 mem_max = 1024*1024*atoi(arg);
572 rGroupDef.databaseName = arg;
574 rGroupDef.flagRw = 0;
576 rGroupDef.groupName = arg;
578 rGroupDef.fileVerboseLimit = atoi(arg);
582 rGroupDef.recordType = arg;
587 rGroupDef.useSDR = 1;
590 logf (LOG_WARN, "unknown option '-%s'", arg);
592 recTypes_destroy (rGroupDef.recTypes);
596 bfs_destroy (rGroupDef.bfs);
598 data1_destroy (rGroupDef.dh);