2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.71 1999-09-08 12:12:06 adam
10 * Revision 1.70 1999/09/07 07:19:21 adam
11 * Work on character mapping. Implemented replace rules.
13 * Revision 1.69 1999/07/21 08:31:33 adam
14 * More version info on WIN32.
16 * Revision 1.68 1999/07/14 10:59:26 adam
17 * Changed functions isc_getmethod, isams_getmethod.
18 * Improved fatal error handling (such as missing EXPLAIN schema).
20 * Revision 1.67 1999/07/06 12:28:04 adam
21 * Updated record index structure. Format includes version ID. Compression
22 * algorithm ID is stored for each record block.
24 * Revision 1.66 1999/06/25 13:48:02 adam
25 * Updated MSVC project files.
26 * Added BZIP2 record compression (not very well tested).
28 * Revision 1.65 1999/05/20 12:57:18 adam
29 * Implemented TCL filter. Updated recctrl system.
31 * Revision 1.64 1999/05/15 14:36:38 adam
32 * Updated dictionary. Implemented "compression" of dictionary.
34 * Revision 1.63 1999/03/09 16:27:49 adam
35 * More work on SDRKit integration.
37 * Revision 1.62 1999/02/18 15:01:24 adam
40 * Revision 1.61 1999/02/02 14:51:00 adam
41 * Updated WIN32 code specific sections. Changed header.
43 * Revision 1.60 1998/10/28 10:54:39 adam
46 * Revision 1.59 1998/10/16 08:14:32 adam
47 * Updated record control system.
49 * Revision 1.58 1998/06/08 14:43:13 adam
50 * Added suport for EXPLAIN Proxy servers - added settings databasePath
51 * and explainDatabase to facilitate this. Increased maximum number
52 * of databases and attributes in one register.
54 * Revision 1.57 1998/05/20 10:12:18 adam
55 * Implemented automatic EXPLAIN database maintenance.
56 * Modified Zebra to work with ASN.1 compiled version of YAZ.
58 * Revision 1.56 1998/03/05 08:45:12 adam
59 * New result set model and modular ranking system. Moved towards
60 * descent server API. System information stored as "SGML" records.
62 * Revision 1.55 1998/01/26 10:37:34 adam
65 * Revision 1.54 1998/01/12 15:04:08 adam
66 * The test option (-s) only uses read-lock (and not write lock).
68 * Revision 1.53 1997/11/18 10:05:08 adam
69 * Changed character map facility so that admin can specify character
70 * mapping files for each register type, w, p, etc.
72 * Revision 1.52 1997/10/31 12:34:04 adam
73 * Added missing nmem_init.
75 * Revision 1.51 1997/10/27 14:33:05 adam
76 * Moved towards generic character mapping depending on "structure"
77 * field in abstract syntax file. Fixed a few memory leaks. Fixed
78 * bug with negative integers when doing searches with relational
81 * Revision 1.50 1997/09/25 14:55:52 adam
84 * Revision 1.49 1997/09/17 12:19:15 adam
85 * Zebra version corresponds to YAZ version 1.4.
86 * Changed Zebra server so that it doesn't depend on global common_resource.
88 * Revision 1.48 1997/09/09 13:38:08 adam
89 * Partial port to WIN95/NT.
91 * Revision 1.47 1997/02/12 20:39:46 adam
92 * Implemented options -f <n> that limits the log to the first <n>
94 * Changed some log messages also.
96 * Revision 1.46 1997/02/10 10:20:13 adam
97 * Flag fileVerboseFlag set to 0 (default).
99 * Revision 1.45 1996/11/08 11:10:26 adam
100 * Buffers used during file match got bigger.
101 * Compressed ISAM support everywhere.
102 * Bug fixes regarding masking characters in queries.
103 * Redesigned Regexp-2 queries.
105 * Revision 1.44 1996/10/29 14:09:48 adam
106 * Use of cisam system - enabled if setting isamc is 1.
108 * Revision 1.43 1996/06/06 12:08:42 quinn
109 * Added showRecord function
111 * Revision 1.42 1996/05/31 09:07:01 quinn
112 * Work on character-set handling
114 * Revision 1.41 1996/05/14 14:04:34 adam
115 * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
118 * Revision 1.40 1996/04/26 10:00:23 adam
119 * Added option -V to zebraidx to display version information.
120 * Removed stupid warnings from file update.
122 * Revision 1.39 1996/04/09 10:05:21 adam
123 * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
125 * Revision 1.38 1996/03/26 16:01:14 adam
126 * New setting lockPath: directory of various lock files.
128 * Revision 1.37 1996/03/19 12:43:26 adam
129 * Bug fix: File update traversal didn't handle trailing slashes correctly.
130 * Bug fix: Update of sub directory groups wasn't handled correctly.
132 * Revision 1.36 1996/02/12 18:45:37 adam
133 * New fileVerboseFlag in record group control.
135 * Revision 1.35 1996/02/12 15:56:11 adam
136 * New code command: unread.
138 * Revision 1.34 1996/02/07 14:06:39 adam
139 * Better progress report during register merge.
140 * New command: clean - removes temporary shadow files.
142 * Revision 1.33 1996/02/06 17:11:18 adam
145 * Revision 1.32 1996/02/01 20:50:04 adam
146 * Bug fix: zebraIndexUnlock was always called even though zebraIndexLock
147 * was never called - happens when no commands are specified.
149 * Revision 1.31 1996/01/08 19:15:46 adam
150 * New input filter that works!
152 * Revision 1.30 1995/12/12 16:00:59 adam
153 * System call sync(2) used after update/commit.
154 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
157 * Revision 1.29 1995/12/11 11:43:30 adam
158 * Locking based on fcntl instead of flock.
159 * Setting commitEnable removed. Command line option -n can be used to
160 * prevent commit if commit setting is defined in the configuration file.
162 * Revision 1.28 1995/12/08 16:22:56 adam
163 * Work on update while servers are running. Three lock files introduced.
164 * The servers reload their registers when necessary, but they don't
165 * reestablish result sets yet.
167 * Revision 1.27 1995/12/07 17:38:47 adam
168 * Work locking mechanisms for concurrent updates/commit.
170 * Revision 1.26 1995/12/06 12:41:23 adam
171 * New command 'stat' for the index program.
172 * Filenames can be read from stdin by specifying '-'.
173 * Bug fix/enhancement of the transformation from terms to regular
174 * expressons in the search engine.
176 * Revision 1.25 1995/12/01 16:24:39 adam
177 * Commit files use separate meta file area.
179 * Revision 1.24 1995/11/30 17:01:38 adam
180 * New setting commitCache: points to commit directories/files.
181 * New command commit: commits at the end of a zebraidx run.
183 * Revision 1.23 1995/11/30 08:34:31 adam
184 * Started work on commit facility.
185 * Changed a few malloc/free to xmalloc/xfree.
187 * Revision 1.22 1995/11/28 09:09:42 adam
188 * Zebra config renamed.
189 * Use setting 'recordId' to identify record now.
190 * Bug fix in recindex.c: rec_release_blocks was invokeded even
191 * though the blocks were already released.
192 * File traversal properly deletes records when needed.
194 * Revision 1.21 1995/11/27 14:27:39 adam
195 * Renamed 'update' command to 'dir'.
197 * Revision 1.20 1995/11/27 13:58:53 adam
198 * New option -t. storeStore data implemented in server.
200 * Revision 1.19 1995/11/25 10:24:06 adam
201 * More record fields - they are enumerated now.
202 * New options: flagStoreData flagStoreKey.
204 * Revision 1.18 1995/11/22 17:19:17 adam
205 * Record management uses the bfile system.
207 * Revision 1.17 1995/11/21 15:01:16 adam
208 * New general match criteria implemented.
209 * New feature: document groups.
211 * Revision 1.16 1995/11/20 11:56:27 adam
212 * Work on new traversal.
214 * Revision 1.15 1995/11/01 16:25:51 quinn
215 * *** empty log message ***
217 * Revision 1.14 1995/10/17 18:02:09 adam
218 * New feature: databases. Implemented as prefix to words in dictionary.
220 * Revision 1.13 1995/10/10 12:24:39 adam
221 * Temporary sort files are compressed.
223 * Revision 1.12 1995/10/04 16:57:20 adam
224 * Key input and merge sort in one pass.
226 * Revision 1.11 1995/09/29 14:01:45 adam
229 * Revision 1.10 1995/09/28 14:22:57 adam
230 * Sort uses smaller temporary files.
232 * Revision 1.9 1995/09/14 07:48:24 adam
233 * Record control management.
235 * Revision 1.8 1995/09/06 16:11:18 adam
236 * Option: only one word key per file.
238 * Revision 1.7 1995/09/05 15:28:39 adam
239 * More work on search engine.
241 * Revision 1.6 1995/09/04 12:33:43 adam
242 * Various cleanup. YAZ util used instead.
244 * Revision 1.5 1995/09/04 09:10:39 adam
245 * More work on index add/del/update.
246 * Merge sort implemented.
247 * Initial work on z39 server.
249 * Revision 1.4 1995/09/01 14:06:36 adam
250 * Split of work into more files.
252 * Revision 1.3 1995/09/01 10:57:07 adam
255 * Revision 1.2 1995/09/01 10:30:24 adam
256 * More work on indexing. Not working yet.
258 * Revision 1.1 1995/08/31 14:50:24 adam
259 * New simple file index tool.
272 #include "recindex.h"
279 #include "zebrasdr.h"
284 Res common_resource = 0;
286 int main (int argc, char **argv)
291 char *configName = FNAME_CONFIG;
293 int disableCommit = 0;
296 struct recordGroup rGroupDef;
302 rGroupDef.useSDR = 0;
304 rGroupDef.groupName = NULL;
305 rGroupDef.databaseName = NULL;
306 rGroupDef.path = NULL;
307 rGroupDef.recordId = NULL;
308 rGroupDef.recordType = NULL;
309 rGroupDef.flagStoreData = -1;
310 rGroupDef.flagStoreKeys = -1;
311 rGroupDef.flagRw = 1;
312 rGroupDef.databaseNamePath = 0;
313 rGroupDef.explainDatabase = 0;
314 rGroupDef.fileVerboseLimit = 100000;
315 rGroupDef.zebra_maps = NULL;
316 rGroupDef.dh = data1_create ();
317 rGroupDef.recTypes = recTypes_init (rGroupDef.dh);
318 recTypes_default_handlers (rGroupDef.recTypes);
323 fprintf (stderr, "zebraidx [options] command <dir> ...\n"
325 " update <dir> Update index with files below <dir>.\n"
326 " If <dir> is empty filenames are read from stdin.\n"
327 " delete <dir> Delete index with files below <dir>.\n"
328 " commit Commit changes\n"
329 " clean Clean shadow files\n"
331 " -t <type> Index files as <type> (grs or text).\n"
332 " -c <config> Read configuration file <config>.\n"
333 " -g <group> Index files according to group settings.\n"
334 " -d <database> Records belong to Z39.50 database <database>.\n"
335 " -m <mbytes> Use <mbytes> before flushing keys to disk.\n"
336 " -n Don't use shadow system.\n"
337 " -s Show analysis on stdout, but do no work.\n"
338 " -v <level> Set logging to <level>.\n"
339 " -l <file> Write log to <file>.\n"
340 " -f <n> Display information for the first <n> records.\n"
344 " -V Show version.\n"
348 while ((ret = options ("sVt:c:g:d:m:v:nf:l:"
352 , argv, argc, &arg)) != -2)
357 if(cmd == 0) /* command */
359 if (!common_resource)
361 logf (LOG_LOG, "zebra version %s %s",
362 ZEBRAVER, ZEBRADATE);
363 common_resource = res_open (configName ?
364 configName : FNAME_CONFIG);
365 if (!common_resource)
367 logf (LOG_FATAL, "cannot read file `%s'", configName);
370 data1_set_tabpath (rGroupDef.dh, res_get (common_resource,
374 bfs_create (res_get (common_resource, "register"));
376 bf_lockDir (rGroupDef.bfs,
377 res_get (common_resource, "lockDir"));
378 rGroupDef.zebra_maps = zebra_maps_open (common_resource);
380 if (!strcmp (arg, "update"))
382 else if (!strcmp (arg, "update1"))
384 else if (!strcmp (arg, "update2"))
386 else if (!strcmp (arg, "dump"))
388 else if (!strcmp (arg, "del") || !strcmp(arg, "delete"))
390 else if (!strcmp (arg, "commit"))
392 rval = res_get (common_resource, "shadow");
393 zebraIndexLock (rGroupDef.bfs, 1, rval);
395 bf_cache (rGroupDef.bfs, rval);
398 logf (LOG_FATAL, "Cannot perform commit");
399 logf (LOG_FATAL, "No shadow area defined");
402 if (bf_commitExists (rGroupDef.bfs))
404 logf (LOG_LOG, "commit start");
405 zebraIndexLockMsg ("c");
407 logf (LOG_LOG, "commit execute");
408 bf_commitExec (rGroupDef.bfs);
412 zebraIndexLockMsg ("d");
414 logf (LOG_LOG, "commit clean");
415 bf_commitClean (rGroupDef.bfs, rval);
418 logf (LOG_LOG, "cothing to commit");
420 else if (!strcmp (arg, "clean"))
422 rval = res_get (common_resource, "shadow");
423 zebraIndexLock (rGroupDef.bfs, 1, rval);
424 if (bf_commitExists (rGroupDef.bfs))
426 zebraIndexLockMsg ("d");
428 logf (LOG_LOG, "commit clean");
429 bf_commitClean (rGroupDef.bfs, rval);
432 logf (LOG_LOG, "nothing to clean");
434 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
437 rval = res_get (common_resource, "shadow");
438 zebraIndexLock (rGroupDef.bfs, 0, rval);
441 bf_cache (rGroupDef.bfs, rval);
442 zebraIndexLockMsg ("r");
444 records = rec_open (rGroupDef.bfs, 0, 0);
445 rec_prstat (records);
446 rec_close (&records);
447 inv_prstat (rGroupDef.bfs);
449 else if (!strcmp (arg, "compact"))
451 rval = res_get (common_resource, "shadow");
452 zebraIndexLock (rGroupDef.bfs, 0, rval);
455 bf_cache (rGroupDef.bfs, rval);
456 zebraIndexLockMsg ("r");
458 inv_compact(rGroupDef.bfs);
462 logf (LOG_FATAL, "unknown command: %s", arg);
468 struct recordGroup rGroup;
470 rval = res_get (common_resource, "shadow");
471 zebraIndexLock (rGroupDef.bfs, 0, rval);
472 if (rGroupDef.flagRw)
474 if (rval && *rval && !disableCommit)
476 bf_cache (rGroupDef.bfs, rval);
477 zebraIndexLockMsg ("r");
481 bf_cache (rGroupDef.bfs, 0);
482 zebraIndexLockMsg ("w");
486 memcpy (&rGroup, &rGroupDef, sizeof(rGroup));
491 if (!key_open (&rGroup, mem_max))
493 logf (LOG_LOG, "updating %s", rGroup.path);
494 repositoryUpdate (&rGroup);
495 nsections = key_close (&rGroup);
499 if (!key_open (&rGroup, mem_max))
501 logf (LOG_LOG, "updating (pass 1) %s", rGroup.path);
502 repositoryUpdate (&rGroup);
508 if (!key_open (&rGroup,mem_max))
510 logf (LOG_LOG, "deleting %s", rGroup.path);
511 repositoryDelete (&rGroup);
512 nsections = key_close (&rGroup);
516 logf (LOG_LOG, "dumping %s", rGroup.path);
517 repositoryShow (&rGroup);
529 logf (LOG_LOG, "merging with index");
530 key_input (rGroup.bfs, nsections, 60);
535 log_event_end (NULL, NULL);
540 fprintf (stderr, "Zebra %s %s\n", ZEBRAVER, ZEBRADATE);
541 fprintf (stderr, " (C) 1994-1999, Index Data ApS\n");
544 fprintf (stderr, " WIN32 Debug\n");
546 fprintf (stderr, " WIN32 Release\n");
550 fprintf (stderr, "libbzip2\n"
551 " (C) 1996-1999 Julian R Seward. All rights reserved.\n");
555 log_init_level (log_mask_str(arg));
559 mem_max = 1024*1024*atoi(arg);
561 rGroupDef.databaseName = arg;
563 rGroupDef.flagRw = 0;
565 rGroupDef.groupName = arg;
567 rGroupDef.fileVerboseLimit = atoi(arg);
571 rGroupDef.recordType = arg;
576 rGroupDef.useSDR = 1;
579 logf (LOG_WARN, "unknown option '-%s'", arg);
581 recTypes_destroy (rGroupDef.recTypes);
585 bfs_destroy (rGroupDef.bfs);
587 data1_destroy (rGroupDef.dh);