2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.74 1999-12-08 15:03:11 adam
8 * Implemented bf_reset.
11 * Revision 1.73 1999/11/30 13:48:03 adam
12 * Improved installation. Updated for inclusion of YAZ header files.
14 * Revision 1.72 1999/10/14 14:33:50 adam
15 * Added truncation 5=106.
17 * Revision 1.71 1999/09/08 12:12:06 adam
20 * Revision 1.70 1999/09/07 07:19:21 adam
21 * Work on character mapping. Implemented replace rules.
23 * Revision 1.69 1999/07/21 08:31:33 adam
24 * More version info on WIN32.
26 * Revision 1.68 1999/07/14 10:59:26 adam
27 * Changed functions isc_getmethod, isams_getmethod.
28 * Improved fatal error handling (such as missing EXPLAIN schema).
30 * Revision 1.67 1999/07/06 12:28:04 adam
31 * Updated record index structure. Format includes version ID. Compression
32 * algorithm ID is stored for each record block.
34 * Revision 1.66 1999/06/25 13:48:02 adam
35 * Updated MSVC project files.
36 * Added BZIP2 record compression (not very well tested).
38 * Revision 1.65 1999/05/20 12:57:18 adam
39 * Implemented TCL filter. Updated recctrl system.
41 * Revision 1.64 1999/05/15 14:36:38 adam
42 * Updated dictionary. Implemented "compression" of dictionary.
44 * Revision 1.63 1999/03/09 16:27:49 adam
45 * More work on SDRKit integration.
47 * Revision 1.62 1999/02/18 15:01:24 adam
50 * Revision 1.61 1999/02/02 14:51:00 adam
51 * Updated WIN32 code specific sections. Changed header.
53 * Revision 1.60 1998/10/28 10:54:39 adam
56 * Revision 1.59 1998/10/16 08:14:32 adam
57 * Updated record control system.
59 * Revision 1.58 1998/06/08 14:43:13 adam
60 * Added suport for EXPLAIN Proxy servers - added settings databasePath
61 * and explainDatabase to facilitate this. Increased maximum number
62 * of databases and attributes in one register.
64 * Revision 1.57 1998/05/20 10:12:18 adam
65 * Implemented automatic EXPLAIN database maintenance.
66 * Modified Zebra to work with ASN.1 compiled version of YAZ.
68 * Revision 1.56 1998/03/05 08:45:12 adam
69 * New result set model and modular ranking system. Moved towards
70 * descent server API. System information stored as "SGML" records.
72 * Revision 1.55 1998/01/26 10:37:34 adam
75 * Revision 1.54 1998/01/12 15:04:08 adam
76 * The test option (-s) only uses read-lock (and not write lock).
78 * Revision 1.53 1997/11/18 10:05:08 adam
79 * Changed character map facility so that admin can specify character
80 * mapping files for each register type, w, p, etc.
82 * Revision 1.52 1997/10/31 12:34:04 adam
83 * Added missing nmem_init.
85 * Revision 1.51 1997/10/27 14:33:05 adam
86 * Moved towards generic character mapping depending on "structure"
87 * field in abstract syntax file. Fixed a few memory leaks. Fixed
88 * bug with negative integers when doing searches with relational
91 * Revision 1.50 1997/09/25 14:55:52 adam
94 * Revision 1.49 1997/09/17 12:19:15 adam
95 * Zebra version corresponds to YAZ version 1.4.
96 * Changed Zebra server so that it doesn't depend on global common_resource.
98 * Revision 1.48 1997/09/09 13:38:08 adam
99 * Partial port to WIN95/NT.
101 * Revision 1.47 1997/02/12 20:39:46 adam
102 * Implemented options -f <n> that limits the log to the first <n>
104 * Changed some log messages also.
106 * Revision 1.46 1997/02/10 10:20:13 adam
107 * Flag fileVerboseFlag set to 0 (default).
109 * Revision 1.45 1996/11/08 11:10:26 adam
110 * Buffers used during file match got bigger.
111 * Compressed ISAM support everywhere.
112 * Bug fixes regarding masking characters in queries.
113 * Redesigned Regexp-2 queries.
115 * Revision 1.44 1996/10/29 14:09:48 adam
116 * Use of cisam system - enabled if setting isamc is 1.
118 * Revision 1.43 1996/06/06 12:08:42 quinn
119 * Added showRecord function
121 * Revision 1.42 1996/05/31 09:07:01 quinn
122 * Work on character-set handling
124 * Revision 1.41 1996/05/14 14:04:34 adam
125 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
128 * Revision 1.40 1996/04/26 10:00:23 adam
129 * Added option -V to zebraidx to display version information.
130 * Removed stupid warnings from file update.
132 * Revision 1.39 1996/04/09 10:05:21 adam
133 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
135 * Revision 1.38 1996/03/26 16:01:14 adam
136 * New setting lockPath: directory of various lock files.
138 * Revision 1.37 1996/03/19 12:43:26 adam
139 * Bug fix: File update traversal didn't handle trailing slashes correctly.
140 * Bug fix: Update of sub directory groups wasn't handled correctly.
142 * Revision 1.36 1996/02/12 18:45:37 adam
143 * New fileVerboseFlag in record group control.
145 * Revision 1.35 1996/02/12 15:56:11 adam
146 * New code command: unread.
148 * Revision 1.34 1996/02/07 14:06:39 adam
149 * Better progress report during register merge.
150 * New command: clean - removes temporary shadow files.
152 * Revision 1.33 1996/02/06 17:11:18 adam
155 * Revision 1.32 1996/02/01 20:50:04 adam
156 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
157 * was never called - happens when no commands are specified.
159 * Revision 1.31 1996/01/08 19:15:46 adam
160 * New input filter that works!
162 * Revision 1.30 1995/12/12 16:00:59 adam
163 * System call sync(2) used after update/commit.
164 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
167 * Revision 1.29 1995/12/11 11:43:30 adam
168 * Locking based on fcntl instead of flock.
169 * Setting commitEnable removed. Command line option -n can be used to
170 * prevent commit if commit setting is defined in the configuration file.
172 * Revision 1.28 1995/12/08 16:22:56 adam
173 * Work on update while servers are running. Three lock files introduced.
174 * The servers reload their registers when necessary, but they don't
175 * reestablish result sets yet.
177 * Revision 1.27 1995/12/07 17:38:47 adam
178 * Work locking mechanisms for concurrent updates/commit.
180 * Revision 1.26 1995/12/06 12:41:23 adam
181 * New command 'stat' for the index program.
182 * Filenames can be read from stdin by specifying '-'.
183 * Bug fix/enhancement of the transformation from terms to regular
184 * expressons in the search engine.
186 * Revision 1.25 1995/12/01 16:24:39 adam
187 * Commit files use separate meta file area.
189 * Revision 1.24 1995/11/30 17:01:38 adam
190 * New setting commitCache: points to commit directories/files.
191 * New command commit: commits at the end of a zebraidx run.
193 * Revision 1.23 1995/11/30 08:34:31 adam
194 * Started work on commit facility.
195 * Changed a few malloc/free to xmalloc/xfree.
197 * Revision 1.22 1995/11/28 09:09:42 adam
198 * Zebra config renamed.
199 * Use setting 'recordId' to identify record now.
200 * Bug fix in recindex.c: rec_release_blocks was invokeded even
201 * though the blocks were already released.
202 * File traversal properly deletes records when needed.
204 * Revision 1.21 1995/11/27 14:27:39 adam
205 * Renamed 'update' command to 'dir'.
207 * Revision 1.20 1995/11/27 13:58:53 adam
208 * New option -t. storeStore data implemented in server.
210 * Revision 1.19 1995/11/25 10:24:06 adam
211 * More record fields - they are enumerated now.
212 * New options: flagStoreData flagStoreKey.
214 * Revision 1.18 1995/11/22 17:19:17 adam
215 * Record management uses the bfile system.
217 * Revision 1.17 1995/11/21 15:01:16 adam
218 * New general match criteria implemented.
219 * New feature: document groups.
221 * Revision 1.16 1995/11/20 11:56:27 adam
222 * Work on new traversal.
224 * Revision 1.15 1995/11/01 16:25:51 quinn
225 * *** empty log message ***
227 * Revision 1.14 1995/10/17 18:02:09 adam
228 * New feature: databases. Implemented as prefix to words in dictionary.
230 * Revision 1.13 1995/10/10 12:24:39 adam
231 * Temporary sort files are compressed.
233 * Revision 1.12 1995/10/04 16:57:20 adam
234 * Key input and merge sort in one pass.
236 * Revision 1.11 1995/09/29 14:01:45 adam
239 * Revision 1.10 1995/09/28 14:22:57 adam
240 * Sort uses smaller temporary files.
242 * Revision 1.9 1995/09/14 07:48:24 adam
243 * Record control management.
245 * Revision 1.8 1995/09/06 16:11:18 adam
246 * Option: only one word key per file.
248 * Revision 1.7 1995/09/05 15:28:39 adam
249 * More work on search engine.
251 * Revision 1.6 1995/09/04 12:33:43 adam
252 * Various cleanup. YAZ util used instead.
254 * Revision 1.5 1995/09/04 09:10:39 adam
255 * More work on index add/del/update.
256 * Merge sort implemented.
257 * Initial work on z39 server.
259 * Revision 1.4 1995/09/01 14:06:36 adam
260 * Split of work into more files.
262 * Revision 1.3 1995/09/01 10:57:07 adam
265 * Revision 1.2 1995/09/01 10:30:24 adam
266 * More work on indexing. Not working yet.
268 * Revision 1.1 1995/08/31 14:50:24 adam
269 * New simple file index tool.
280 #include <yaz/data1.h>
282 #include "recindex.h"
289 #include "zebrasdr.h"
294 Res common_resource = 0;
297 int main (int argc, char **argv)
302 char *configName = FNAME_CONFIG;
304 int disableCommit = 0;
307 struct recordGroup rGroupDef;
313 rGroupDef.useSDR = 0;
315 rGroupDef.groupName = NULL;
316 rGroupDef.databaseName = NULL;
317 rGroupDef.path = NULL;
318 rGroupDef.recordId = NULL;
319 rGroupDef.recordType = NULL;
320 rGroupDef.flagStoreData = -1;
321 rGroupDef.flagStoreKeys = -1;
322 rGroupDef.flagRw = 1;
323 rGroupDef.databaseNamePath = 0;
324 rGroupDef.explainDatabase = 0;
325 rGroupDef.fileVerboseLimit = 100000;
326 rGroupDef.zebra_maps = NULL;
327 rGroupDef.dh = data1_create ();
328 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
329 recTypes_default_handlers (rGroupDef.recTypes);
334 fprintf (stderr, "zebraidx [options] command <dir> ...\n"
336 " update <dir> Update index with files below <dir>.\n"
337 " If <dir> is empty filenames are read from stdin.\n"
338 " delete <dir> Delete index with files below <dir>.\n"
339 " commit Commit changes\n"
340 " clean Clean shadow files\n"
342 " -t <type> Index files as <type> (grs or text).\n"
343 " -c <config> Read configuration file <config>.\n"
344 " -g <group> Index files according to group settings.\n"
345 " -d <database> Records belong to Z39.50 database <database>.\n"
346 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
347 " -n Don't use shadow system.\n"
348 " -s Show analysis on stdout, but do no work.\n"
349 " -v <level> Set logging to <level>.\n"
350 " -l <file> Write log to <file>.\n"
351 " -f <n> Display information for the first <n> records.\n"
355 " -V Show version.\n"
359 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
363 , argv, argc, &arg)) != -2)
368 if(cmd == 0) /* command */
370 if (!common_resource)
372 logf (LOG_LOG, "zebra version %s %s",
373 ZEBRAVER, ZEBRADATE);
374 common_resource = res_open (configName ?
375 configName : FNAME_CONFIG);
376 if (!common_resource)
378 logf (LOG_FATAL, "cannot read file `%s'", configName);
381 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
385 bfs_create (res_get (common_resource, "register"));
388 logf (LOG_FATAL, "Cannot access register");
392 bf_lockDir (rGroupDef.bfs,
393 res_get (common_resource, "lockDir"));
394 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
396 if (!strcmp (arg, "update"))
398 else if (!strcmp (arg, "update1"))
400 else if (!strcmp (arg, "update2"))
402 else if (!strcmp (arg, "dump"))
404 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
406 else if (!strcmp (arg, "init"))
409 rval = res_get (common_resource, "shadow");
410 zebraIndexLock (rGroupDef.bfs, 0, rval);
411 zebraIndexLockMsg ("w");
412 bf_reset (rGroupDef.bfs);
414 else if (!strcmp (arg, "commit"))
416 rval = res_get (common_resource, "shadow");
417 zebraIndexLock (rGroupDef.bfs, 1, rval);
419 bf_cache (rGroupDef.bfs, rval);
422 logf (LOG_FATAL, "Cannot perform commit");
423 logf (LOG_FATAL, "No shadow area defined");
426 if (bf_commitExists (rGroupDef.bfs))
428 logf (LOG_LOG, "commit start");
429 zebraIndexLockMsg ("c");
431 logf (LOG_LOG, "commit execute");
432 bf_commitExec (rGroupDef.bfs);
436 zebraIndexLockMsg ("d");
438 logf (LOG_LOG, "commit clean");
439 bf_commitClean (rGroupDef.bfs, rval);
442 logf (LOG_LOG, "cothing to commit");
444 else if (!strcmp (arg, "clean"))
446 rval = res_get (common_resource, "shadow");
447 zebraIndexLock (rGroupDef.bfs, 1, rval);
448 if (bf_commitExists (rGroupDef.bfs))
450 zebraIndexLockMsg ("d");
452 logf (LOG_LOG, "commit clean");
453 bf_commitClean (rGroupDef.bfs, rval);
456 logf (LOG_LOG, "nothing to clean");
458 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
461 rval = res_get (common_resource, "shadow");
462 zebraIndexLock (rGroupDef.bfs, 0, rval);
465 bf_cache (rGroupDef.bfs, rval);
466 zebraIndexLockMsg ("r");
468 records = rec_open (rGroupDef.bfs, 0, 0);
469 rec_prstat (records);
470 rec_close (&records);
471 inv_prstat (rGroupDef.bfs);
473 else if (!strcmp (arg, "compact"))
475 rval = res_get (common_resource, "shadow");
476 zebraIndexLock (rGroupDef.bfs, 0, rval);
479 bf_cache (rGroupDef.bfs, rval);
480 zebraIndexLockMsg ("r");
482 inv_compact(rGroupDef.bfs);
486 logf (LOG_FATAL, "unknown command: %s", arg);
492 struct recordGroup rGroup;
495 /* For zebra, delete lock file and reset register */
496 if (rGroupDef.flagRw)
499 bf_reset (rGroupDef.bfs);
502 rval = res_get (common_resource, "shadow");
503 zebraIndexLock (rGroupDef.bfs, 0, rval);
504 if (rGroupDef.flagRw)
506 if (rval && *rval && !disableCommit)
508 bf_cache (rGroupDef.bfs, rval);
509 zebraIndexLockMsg ("r");
513 bf_cache (rGroupDef.bfs, 0);
514 zebraIndexLockMsg ("w");
518 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
523 if (!key_open (&rGroup, mem_max))
525 logf (LOG_LOG, "updating %s", rGroup.path);
526 repositoryUpdate (&rGroup);
527 nsections = key_close (&rGroup);
531 if (!key_open (&rGroup, mem_max))
533 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
534 repositoryUpdate (&rGroup);
540 if (!key_open (&rGroup,mem_max))
542 logf (LOG_LOG, "deleting %s", rGroup.path);
543 repositoryDelete (&rGroup);
544 nsections = key_close (&rGroup);
548 logf (LOG_LOG, "dumping %s", rGroup.path);
549 repositoryShow (&rGroup);
561 logf (LOG_LOG, "merging with index");
562 key_input (rGroup.bfs, nsections, 60);
567 log_event_end (NULL, NULL);
573 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
575 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
577 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
580 fprintf (stderr, " WIN32 Debug\n");
582 fprintf (stderr, " WIN32 Release\n");
586 fprintf (stderr, "libbzip2\n"
587 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
591 log_init_level (log_mask_str(arg));
595 mem_max = 1024*1024*atoi(arg);
597 rGroupDef.databaseName = arg;
599 rGroupDef.flagRw = 0;
601 rGroupDef.groupName = arg;
603 rGroupDef.fileVerboseLimit = atoi(arg);
607 rGroupDef.recordType = arg;
612 rGroupDef.useSDR = 1;
615 logf (LOG_WARN, "unknown option '-%s'", arg);
617 recTypes_destroy (rGroupDef.recTypes);
621 bfs_destroy (rGroupDef.bfs);
623 data1_destroy (rGroupDef.dh);