2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.75 1999-12-08 22:44:45 adam
8 * Zebra/Z'mbol dependencies added.
10 * Revision 1.74 1999/12/08 15:03:11 adam
11 * Implemented bf_reset.
14 * Revision 1.73 1999/11/30 13:48:03 adam
15 * Improved installation. Updated for inclusion of YAZ header files.
17 * Revision 1.72 1999/10/14 14:33:50 adam
18 * Added truncation 5=106.
20 * Revision 1.71 1999/09/08 12:12:06 adam
23 * Revision 1.70 1999/09/07 07:19:21 adam
24 * Work on character mapping. Implemented replace rules.
26 * Revision 1.69 1999/07/21 08:31:33 adam
27 * More version info on WIN32.
29 * Revision 1.68 1999/07/14 10:59:26 adam
30 * Changed functions isc_getmethod, isams_getmethod.
31 * Improved fatal error handling (such as missing EXPLAIN schema).
33 * Revision 1.67 1999/07/06 12:28:04 adam
34 * Updated record index structure. Format includes version ID. Compression
35 * algorithm ID is stored for each record block.
37 * Revision 1.66 1999/06/25 13:48:02 adam
38 * Updated MSVC project files.
39 * Added BZIP2 record compression (not very well tested).
41 * Revision 1.65 1999/05/20 12:57:18 adam
42 * Implemented TCL filter. Updated recctrl system.
44 * Revision 1.64 1999/05/15 14:36:38 adam
45 * Updated dictionary. Implemented "compression" of dictionary.
47 * Revision 1.63 1999/03/09 16:27:49 adam
48 * More work on SDRKit integration.
50 * Revision 1.62 1999/02/18 15:01:24 adam
53 * Revision 1.61 1999/02/02 14:51:00 adam
54 * Updated WIN32 code specific sections. Changed header.
56 * Revision 1.60 1998/10/28 10:54:39 adam
59 * Revision 1.59 1998/10/16 08:14:32 adam
60 * Updated record control system.
62 * Revision 1.58 1998/06/08 14:43:13 adam
63 * Added suport for EXPLAIN Proxy servers - added settings databasePath
64 * and explainDatabase to facilitate this. Increased maximum number
65 * of databases and attributes in one register.
67 * Revision 1.57 1998/05/20 10:12:18 adam
68 * Implemented automatic EXPLAIN database maintenance.
69 * Modified Zebra to work with ASN.1 compiled version of YAZ.
71 * Revision 1.56 1998/03/05 08:45:12 adam
72 * New result set model and modular ranking system. Moved towards
73 * descent server API. System information stored as "SGML" records.
75 * Revision 1.55 1998/01/26 10:37:34 adam
78 * Revision 1.54 1998/01/12 15:04:08 adam
79 * The test option (-s) only uses read-lock (and not write lock).
81 * Revision 1.53 1997/11/18 10:05:08 adam
82 * Changed character map facility so that admin can specify character
83 * mapping files for each register type, w, p, etc.
85 * Revision 1.52 1997/10/31 12:34:04 adam
86 * Added missing nmem_init.
88 * Revision 1.51 1997/10/27 14:33:05 adam
89 * Moved towards generic character mapping depending on "structure"
90 * field in abstract syntax file. Fixed a few memory leaks. Fixed
91 * bug with negative integers when doing searches with relational
94 * Revision 1.50 1997/09/25 14:55:52 adam
97 * Revision 1.49 1997/09/17 12:19:15 adam
98 * Zebra version corresponds to YAZ version 1.4.
99 * Changed Zebra server so that it doesn't depend on global common_resource.
101 * Revision 1.48 1997/09/09 13:38:08 adam
102 * Partial port to WIN95/NT.
104 * Revision 1.47 1997/02/12 20:39:46 adam
105 * Implemented options -f <n> that limits the log to the first <n>
107 * Changed some log messages also.
109 * Revision 1.46 1997/02/10 10:20:13 adam
110 * Flag fileVerboseFlag set to 0 (default).
112 * Revision 1.45 1996/11/08 11:10:26 adam
113 * Buffers used during file match got bigger.
114 * Compressed ISAM support everywhere.
115 * Bug fixes regarding masking characters in queries.
116 * Redesigned Regexp-2 queries.
118 * Revision 1.44 1996/10/29 14:09:48 adam
119 * Use of cisam system - enabled if setting isamc is 1.
121 * Revision 1.43 1996/06/06 12:08:42 quinn
122 * Added showRecord function
124 * Revision 1.42 1996/05/31 09:07:01 quinn
125 * Work on character-set handling
127 * Revision 1.41 1996/05/14 14:04:34 adam
128 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
131 * Revision 1.40 1996/04/26 10:00:23 adam
132 * Added option -V to zebraidx to display version information.
133 * Removed stupid warnings from file update.
135 * Revision 1.39 1996/04/09 10:05:21 adam
136 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
138 * Revision 1.38 1996/03/26 16:01:14 adam
139 * New setting lockPath: directory of various lock files.
141 * Revision 1.37 1996/03/19 12:43:26 adam
142 * Bug fix: File update traversal didn't handle trailing slashes correctly.
143 * Bug fix: Update of sub directory groups wasn't handled correctly.
145 * Revision 1.36 1996/02/12 18:45:37 adam
146 * New fileVerboseFlag in record group control.
148 * Revision 1.35 1996/02/12 15:56:11 adam
149 * New code command: unread.
151 * Revision 1.34 1996/02/07 14:06:39 adam
152 * Better progress report during register merge.
153 * New command: clean - removes temporary shadow files.
155 * Revision 1.33 1996/02/06 17:11:18 adam
158 * Revision 1.32 1996/02/01 20:50:04 adam
159 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
160 * was never called - happens when no commands are specified.
162 * Revision 1.31 1996/01/08 19:15:46 adam
163 * New input filter that works!
165 * Revision 1.30 1995/12/12 16:00:59 adam
166 * System call sync(2) used after update/commit.
167 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
170 * Revision 1.29 1995/12/11 11:43:30 adam
171 * Locking based on fcntl instead of flock.
172 * Setting commitEnable removed. Command line option -n can be used to
173 * prevent commit if commit setting is defined in the configuration file.
175 * Revision 1.28 1995/12/08 16:22:56 adam
176 * Work on update while servers are running. Three lock files introduced.
177 * The servers reload their registers when necessary, but they don't
178 * reestablish result sets yet.
180 * Revision 1.27 1995/12/07 17:38:47 adam
181 * Work locking mechanisms for concurrent updates/commit.
183 * Revision 1.26 1995/12/06 12:41:23 adam
184 * New command 'stat' for the index program.
185 * Filenames can be read from stdin by specifying '-'.
186 * Bug fix/enhancement of the transformation from terms to regular
187 * expressons in the search engine.
189 * Revision 1.25 1995/12/01 16:24:39 adam
190 * Commit files use separate meta file area.
192 * Revision 1.24 1995/11/30 17:01:38 adam
193 * New setting commitCache: points to commit directories/files.
194 * New command commit: commits at the end of a zebraidx run.
196 * Revision 1.23 1995/11/30 08:34:31 adam
197 * Started work on commit facility.
198 * Changed a few malloc/free to xmalloc/xfree.
200 * Revision 1.22 1995/11/28 09:09:42 adam
201 * Zebra config renamed.
202 * Use setting 'recordId' to identify record now.
203 * Bug fix in recindex.c: rec_release_blocks was invokeded even
204 * though the blocks were already released.
205 * File traversal properly deletes records when needed.
207 * Revision 1.21 1995/11/27 14:27:39 adam
208 * Renamed 'update' command to 'dir'.
210 * Revision 1.20 1995/11/27 13:58:53 adam
211 * New option -t. storeStore data implemented in server.
213 * Revision 1.19 1995/11/25 10:24:06 adam
214 * More record fields - they are enumerated now.
215 * New options: flagStoreData flagStoreKey.
217 * Revision 1.18 1995/11/22 17:19:17 adam
218 * Record management uses the bfile system.
220 * Revision 1.17 1995/11/21 15:01:16 adam
221 * New general match criteria implemented.
222 * New feature: document groups.
224 * Revision 1.16 1995/11/20 11:56:27 adam
225 * Work on new traversal.
227 * Revision 1.15 1995/11/01 16:25:51 quinn
228 * *** empty log message ***
230 * Revision 1.14 1995/10/17 18:02:09 adam
231 * New feature: databases. Implemented as prefix to words in dictionary.
233 * Revision 1.13 1995/10/10 12:24:39 adam
234 * Temporary sort files are compressed.
236 * Revision 1.12 1995/10/04 16:57:20 adam
237 * Key input and merge sort in one pass.
239 * Revision 1.11 1995/09/29 14:01:45 adam
242 * Revision 1.10 1995/09/28 14:22:57 adam
243 * Sort uses smaller temporary files.
245 * Revision 1.9 1995/09/14 07:48:24 adam
246 * Record control management.
248 * Revision 1.8 1995/09/06 16:11:18 adam
249 * Option: only one word key per file.
251 * Revision 1.7 1995/09/05 15:28:39 adam
252 * More work on search engine.
254 * Revision 1.6 1995/09/04 12:33:43 adam
255 * Various cleanup. YAZ util used instead.
257 * Revision 1.5 1995/09/04 09:10:39 adam
258 * More work on index add/del/update.
259 * Merge sort implemented.
260 * Initial work on z39 server.
262 * Revision 1.4 1995/09/01 14:06:36 adam
263 * Split of work into more files.
265 * Revision 1.3 1995/09/01 10:57:07 adam
268 * Revision 1.2 1995/09/01 10:30:24 adam
269 * More work on indexing. Not working yet.
271 * Revision 1.1 1995/08/31 14:50:24 adam
272 * New simple file index tool.
283 #include <yaz/data1.h>
285 #include "recindex.h"
292 #include "zebrasdr.h"
297 Res common_resource = 0;
300 int main (int argc, char **argv)
305 char *configName = FNAME_CONFIG;
307 int disableCommit = 0;
310 struct recordGroup rGroupDef;
316 rGroupDef.useSDR = 0;
318 rGroupDef.groupName = NULL;
319 rGroupDef.databaseName = NULL;
320 rGroupDef.path = NULL;
321 rGroupDef.recordId = NULL;
322 rGroupDef.recordType = NULL;
323 rGroupDef.flagStoreData = -1;
324 rGroupDef.flagStoreKeys = -1;
325 rGroupDef.flagRw = 1;
326 rGroupDef.databaseNamePath = 0;
327 rGroupDef.explainDatabase = 0;
328 rGroupDef.fileVerboseLimit = 100000;
329 rGroupDef.zebra_maps = NULL;
330 rGroupDef.dh = data1_create ();
331 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
332 recTypes_default_handlers (rGroupDef.recTypes);
337 fprintf (stderr, "%s [options] command <dir> ...\n"
339 " update <dir> Update index with files below <dir>.\n"
340 " If <dir> is empty filenames are read from stdin.\n"
341 " delete <dir> Delete index with files below <dir>.\n"
342 " commit Commit changes\n"
343 " clean Clean shadow files\n"
345 " -t <type> Index files as <type> (grs or text).\n"
346 " -c <config> Read configuration file <config>.\n"
347 " -g <group> Index files according to group settings.\n"
348 " -d <database> Records belong to Z39.50 database <database>.\n"
349 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
350 " -n Don't use shadow system.\n"
351 " -s Show analysis on stdout, but do no work.\n"
352 " -v <level> Set logging to <level>.\n"
353 " -l <file> Write log to <file>.\n"
354 " -f <n> Display information for the first <n> records.\n"
358 " -V Show version.\n", *argv
362 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
366 , argv, argc, &arg)) != -2)
371 if(cmd == 0) /* command */
373 if (!common_resource)
376 logf (LOG_LOG, "zmbol version %s %s",
377 ZEBRAVER, ZEBRADATE);
379 logf (LOG_LOG, "zebra version %s %s",
380 ZEBRAVER, ZEBRADATE);
382 common_resource = res_open (configName ?
383 configName : FNAME_CONFIG);
384 if (!common_resource)
386 logf (LOG_FATAL, "cannot read file `%s'", configName);
389 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
393 bfs_create (res_get (common_resource, "register"));
396 logf (LOG_FATAL, "Cannot access register");
400 bf_lockDir (rGroupDef.bfs,
401 res_get (common_resource, "lockDir"));
402 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
404 if (!strcmp (arg, "update"))
406 else if (!strcmp (arg, "update1"))
408 else if (!strcmp (arg, "update2"))
410 else if (!strcmp (arg, "dump"))
412 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
414 else if (!strcmp (arg, "init"))
417 rval = res_get (common_resource, "shadow");
418 zebraIndexLock (rGroupDef.bfs, 0, rval);
419 zebraIndexLockMsg ("w");
420 bf_reset (rGroupDef.bfs);
422 else if (!strcmp (arg, "commit"))
424 rval = res_get (common_resource, "shadow");
425 zebraIndexLock (rGroupDef.bfs, 1, rval);
427 bf_cache (rGroupDef.bfs, rval);
430 logf (LOG_FATAL, "Cannot perform commit");
431 logf (LOG_FATAL, "No shadow area defined");
434 if (bf_commitExists (rGroupDef.bfs))
436 logf (LOG_LOG, "commit start");
437 zebraIndexLockMsg ("c");
439 logf (LOG_LOG, "commit execute");
440 bf_commitExec (rGroupDef.bfs);
444 zebraIndexLockMsg ("d");
446 logf (LOG_LOG, "commit clean");
447 bf_commitClean (rGroupDef.bfs, rval);
450 logf (LOG_LOG, "cothing to commit");
452 else if (!strcmp (arg, "clean"))
454 rval = res_get (common_resource, "shadow");
455 zebraIndexLock (rGroupDef.bfs, 1, rval);
456 if (bf_commitExists (rGroupDef.bfs))
458 zebraIndexLockMsg ("d");
460 logf (LOG_LOG, "commit clean");
461 bf_commitClean (rGroupDef.bfs, rval);
464 logf (LOG_LOG, "nothing to clean");
466 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
469 rval = res_get (common_resource, "shadow");
470 zebraIndexLock (rGroupDef.bfs, 0, rval);
473 bf_cache (rGroupDef.bfs, rval);
474 zebraIndexLockMsg ("r");
476 records = rec_open (rGroupDef.bfs, 0, 0);
477 rec_prstat (records);
478 rec_close (&records);
479 inv_prstat (rGroupDef.bfs);
481 else if (!strcmp (arg, "compact"))
483 rval = res_get (common_resource, "shadow");
484 zebraIndexLock (rGroupDef.bfs, 0, rval);
487 bf_cache (rGroupDef.bfs, rval);
488 zebraIndexLockMsg ("r");
490 inv_compact(rGroupDef.bfs);
494 logf (LOG_FATAL, "unknown command: %s", arg);
500 struct recordGroup rGroup;
503 /* For zebra, delete lock file and reset register */
504 if (rGroupDef.flagRw)
507 bf_reset (rGroupDef.bfs);
510 rval = res_get (common_resource, "shadow");
511 zebraIndexLock (rGroupDef.bfs, 0, rval);
512 if (rGroupDef.flagRw)
514 if (rval && *rval && !disableCommit)
516 bf_cache (rGroupDef.bfs, rval);
517 zebraIndexLockMsg ("r");
521 bf_cache (rGroupDef.bfs, 0);
522 zebraIndexLockMsg ("w");
526 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
531 if (!key_open (&rGroup, mem_max))
533 logf (LOG_LOG, "updating %s", rGroup.path);
534 repositoryUpdate (&rGroup);
535 nsections = key_close (&rGroup);
539 if (!key_open (&rGroup, mem_max))
541 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
542 repositoryUpdate (&rGroup);
548 if (!key_open (&rGroup,mem_max))
550 logf (LOG_LOG, "deleting %s", rGroup.path);
551 repositoryDelete (&rGroup);
552 nsections = key_close (&rGroup);
556 logf (LOG_LOG, "dumping %s", rGroup.path);
557 repositoryShow (&rGroup);
569 logf (LOG_LOG, "merging with index");
570 key_input (rGroup.bfs, nsections, 60);
575 log_event_end (NULL, NULL);
581 fprintf (stderr, "Z'mbol %s %s\n", ZEBRAVER, ZEBRADATE);
583 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
585 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
588 fprintf (stderr, " WIN32 Debug\n");
590 fprintf (stderr, " WIN32 Release\n");
594 fprintf (stderr, "libbzip2\n"
595 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
599 log_init_level (log_mask_str(arg));
603 mem_max = 1024*1024*atoi(arg);
605 rGroupDef.databaseName = arg;
607 rGroupDef.flagRw = 0;
609 rGroupDef.groupName = arg;
611 rGroupDef.fileVerboseLimit = atoi(arg);
615 rGroupDef.recordType = arg;
620 rGroupDef.useSDR = 1;
623 logf (LOG_WARN, "unknown option '-%s'", arg);
625 recTypes_destroy (rGroupDef.recTypes);
629 bfs_destroy (rGroupDef.bfs);
631 data1_destroy (rGroupDef.dh);