2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.37 2000-10-17 12:37:09 adam
7 * Fixed notification of live-updates. Fixed minor problem with mf_init
8 * where it didn't handle shadow area file names correctly.
10 * Revision 1.36 2000/09/06 08:59:36 adam
11 * Using read-only (for now) for server.
13 * Revision 1.35 2000/07/07 12:49:20 adam
14 * Optimized resultSetInsert{Rank,Sort}.
16 * Revision 1.34 2000/06/09 13:56:38 ian
17 * Added some logging on Authentication and searches.
19 * Revision 1.33 2000/05/18 12:01:36 adam
20 * System call times(2) used again. More 64-bit fixes.
22 * Revision 1.32 2000/04/19 14:35:59 adam
23 * WIN32 update (this version is known not to work on Windows).
25 * Revision 1.31 2000/04/05 10:07:02 adam
26 * Minor zebra compile fix.
28 * Revision 1.30 2000/04/05 09:49:35 adam
29 * On Unix, zebra/z'mbol uses automake.
31 * Revision 1.29 2000/03/20 19:08:36 adam
32 * Added remote record import using Z39.50 extended services and Segment
35 * Revision 1.28 2000/03/15 15:00:30 adam
36 * First work on threaded version.
38 * Revision 1.27 2000/02/24 12:31:17 adam
39 * Added zebra_string_norm.
41 * Revision 1.26 1999/11/30 13:48:03 adam
42 * Improved installation. Updated for inclusion of YAZ header files.
44 * Revision 1.25 1999/11/04 15:00:45 adam
45 * Implemented delete result set(s).
47 * Revision 1.24 1999/10/14 14:33:50 adam
48 * Added truncation 5=106.
50 * Revision 1.23 1999/09/07 11:36:32 adam
53 * Revision 1.22 1999/08/02 10:13:47 adam
54 * Fixed bug regarding zebra_hits.
56 * Revision 1.21 1999/07/14 10:59:26 adam
57 * Changed functions isc_getmethod, isams_getmethod.
58 * Improved fatal error handling (such as missing EXPLAIN schema).
60 * Revision 1.20 1999/07/06 12:28:04 adam
61 * Updated record index structure. Format includes version ID. Compression
62 * algorithm ID is stored for each record block.
64 * Revision 1.19 1999/05/26 07:49:13 adam
67 * Revision 1.18 1999/05/15 14:36:38 adam
68 * Updated dictionary. Implemented "compression" of dictionary.
70 * Revision 1.17 1999/05/12 13:08:06 adam
71 * First version of ISAMS.
73 * Revision 1.16 1999/02/19 10:38:30 adam
74 * Implemented chdir-setting.
76 * Revision 1.15 1999/02/17 12:18:12 adam
77 * Fixed zebra_close so that a NULL pointer is ignored.
79 * Revision 1.14 1999/02/02 14:51:11 adam
80 * Updated WIN32 code specific sections. Changed header.
82 * Revision 1.13 1998/12/16 12:23:30 adam
83 * Added facility for database name mapping using resource mapdb.
85 * Revision 1.12 1998/11/16 10:18:10 adam
86 * Better error reporting for result sets.
88 * Revision 1.11 1998/10/16 08:14:34 adam
89 * Updated record control system.
91 * Revision 1.10 1998/09/22 10:03:42 adam
92 * Changed result sets to be persistent in the sense that they can
93 * be re-searched if needed.
94 * Fixed memory leak in rsm_or.
96 * Revision 1.9 1998/09/02 13:53:17 adam
97 * Extra parameter decode added to search routines to implement
100 * Revision 1.8 1998/08/24 17:29:23 adam
103 * Revision 1.7 1998/06/24 12:16:13 adam
104 * Support for relations on text operands. Open range support in
105 * DFA module (i.e. [-j], [g-]).
107 * Revision 1.6 1998/06/22 11:36:47 adam
108 * Added authentication check facility to zebra.
110 * Revision 1.5 1998/06/13 00:14:08 adam
113 * Revision 1.4 1998/06/12 12:22:12 adam
116 * Revision 1.3 1998/05/27 16:57:44 adam
117 * Zebra returns surrogate diagnostic for single records when
120 * Revision 1.2 1998/05/20 10:12:19 adam
121 * Implemented automatic EXPLAIN database maintenance.
122 * Modified Zebra to work with ASN.1 compiled version of YAZ.
124 * Revision 1.1 1998/03/05 08:45:13 adam
125 * New result set model and modular ranking system. Moved towards
126 * descent server API. System information stored as "SGML" records.
140 #include <yaz/diagbib1.h>
144 static void zebra_chdir (ZebraService zh)
146 const char *dir = res_get (zh->res, "chdir");
149 logf (LOG_DEBUG, "chdir %s", dir);
157 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
158 const char *buf, size_t buf_size,
159 const char *databaseName, int delete_flag,
160 int test_mode, int *sysno,
161 int store_keys, int store_data,
162 const char *match_criteria);
164 static int explain_extract (void *handle, Record rec, data1_node *n);
165 static void extract_index (ZebraHandle zh);
167 static void zebra_register_unlock (ZebraHandle zh);
169 static int zebra_register_activate (ZebraService zh, int rw);
170 static int zebra_register_deactivate (ZebraService zh);
172 static int zebra_register_lock (ZebraHandle zh)
178 if (!zh->service->active)
188 state = zebra_server_lock_get_state(zh->service, &lastChange);
190 zebra_server_lock (zh->service, state);
200 if (zh->service->registerState == state)
202 if (zh->service->registerChange >= lastChange)
204 logf (LOG_LOG, "Register completely updated since last access");
206 else if (zh->service->registerState == -1)
207 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
210 logf (LOG_LOG, "Register has changed state from %d to %d",
211 zh->service->registerState, state);
212 zh->service->registerChange = lastChange;
214 zebra_register_deactivate (zh->service);
216 zh->service->registerState = state;
218 zebra_register_activate (zh->service, 0);
223 static void zebra_register_unlock (ZebraHandle zh)
225 if (zh->service->registerState != -1)
226 zebra_server_unlock (zh->service, zh->service->registerState);
229 logf (LOG_LOG, "user/system: %ld/%ld",
230 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
231 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
236 ZebraHandle zebra_open (ZebraService zs)
244 zh = (ZebraHandle) xmalloc (sizeof(*zh));
253 zh->admin_databaseName = 0;
255 zebra_mutex_cond_lock (&zs->session_lock);
257 zh->next = zs->sessions;
260 zebra_mutex_cond_unlock (&zs->session_lock);
266 ZebraService zebra_start (const char *configName)
268 ZebraService zh = xmalloc (sizeof(*zh));
270 yaz_log (LOG_LOG, "zebra_start %s", configName);
272 zh->configName = xstrdup(configName);
277 zh->registerState = -1;
278 zh->registerChange = 0;
280 if (!(zh->res = res_open (zh->configName)))
282 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
286 zebra_server_lock_init (zh);
287 zebra_mutex_cond_init (&zh->session_lock);
291 static int zebra_register_activate (ZebraService zh, int rw)
295 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
296 zh->registerState ? "yes" : "no");
298 zh->dh = data1_create ();
301 zh->bfs = bfs_create (res_get (zh->res, "register"));
304 data1_destroy(zh->dh);
307 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
308 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
309 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
310 zh->recTypes = recTypes_init (zh->dh);
311 recTypes_default_handlers (zh->recTypes);
314 zh->zebra_maps = zebra_maps_open (zh->res);
315 zh->rank_classes = NULL;
328 zebraRankInstall (zh, rank1_class);
330 if (!res_get (zh->res, "passwd"))
331 zh->passwd_db = NULL;
334 zh->passwd_db = passwd_db_open ();
336 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
338 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
341 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
343 logf (LOG_WARN, "rec_open");
346 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
348 logf (LOG_WARN, "dict_open");
351 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
353 logf (LOG_WARN, "sortIdx_open");
356 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
358 struct ISAMS_M_s isams_m;
359 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
360 key_isams_m(zh->res, &isams_m))))
362 logf (LOG_WARN, "isams_open");
367 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
369 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
370 sizeof (struct it_key), zh->res)))
372 logf (LOG_WARN, "is_open");
376 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
378 struct ISAMC_M_s isamc_m;
379 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
380 rw, key_isamc_m(zh->res, &isamc_m))))
382 logf (LOG_WARN, "isc_open");
386 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
388 struct ISAMD_M_s isamd_m;
390 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
391 rw, key_isamd_m(zh->res, &isamd_m))))
393 logf (LOG_WARN, "isamd_open");
398 zh->zei = zebraExplain_open (zh->records, zh->dh,
399 zh->res, rw, 0 /* rGroup */,
403 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
407 yaz_log (LOG_LOG, "zebra_register_activate ok");
411 void zebra_admin_shutdown (ZebraHandle zh)
413 zebra_register_lock (zh);
414 zebraExplain_flush (zh->service->zei, 1, zh);
417 zebra_register_unlock (zh);
418 zebra_mutex_cond_lock (&zh->service->session_lock);
419 zh->service->stop_flag = 1;
420 if (!zh->service->sessions)
421 zebra_register_deactivate(zh->service);
422 zh->service->active = 0;
423 zebra_mutex_cond_unlock (&zh->service->session_lock);
426 void zebra_admin_start (ZebraHandle zh)
428 ZebraService zs = zh->service;
430 zebra_mutex_cond_lock (&zs->session_lock);
432 zh->service->active = 1;
433 zebra_mutex_cond_unlock (&zs->session_lock);
436 static int zebra_register_deactivate (ZebraService zh)
441 yaz_log(LOG_LOG, "zebra_register_deactivate");
445 zebraExplain_close (zh->zei, 0);
446 dict_close (zh->dict);
447 sortIdx_close (zh->sortIdx);
449 isams_close (zh->isams);
454 isc_close (zh->isamc);
456 isamd_close (zh->isamd);
458 rec_close (&zh->records);
460 recTypes_destroy (zh->recTypes);
461 zebra_maps_close (zh->zebra_maps);
462 zebraRankDestroy (zh);
463 bfs_destroy (zh->bfs);
464 data1_destroy (zh->dh);
467 passwd_db_close (zh->passwd_db);
472 void zebra_stop(ZebraService zh)
476 yaz_log (LOG_LOG, "zebra_stop");
478 assert (!zh->sessions);
480 zebra_mutex_cond_destroy (&zh->session_lock);
482 zebra_register_deactivate(zh);
484 xfree (zh->configName);
488 void zebra_close (ZebraHandle zh)
490 ZebraService zs = zh->service;
491 struct zebra_session **sp;
494 resultSetDestroy (zh, -1, 0, 0);
501 xfree (zh->admin_databaseName);
502 zebra_mutex_cond_lock (&zs->session_lock);
514 if (!zs->sessions && zs->stop_flag)
515 zebra_register_deactivate(zs);
516 zebra_mutex_cond_unlock (&zs->session_lock);
520 struct map_baseinfo {
526 char **new_basenames;
530 void map_basenames_func (void *vp, const char *name, const char *value)
532 struct map_baseinfo *p = (struct map_baseinfo *) vp;
534 char fromdb[128], todb[8][128];
537 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
538 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
539 todb[5], todb[6], todb[7]);
543 for (i = 0; i<p->num_bases; i++)
544 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
547 for (i = 0; i < no; i++)
549 if (p->new_num_bases == p->new_num_max)
551 p->new_basenames[(p->new_num_bases)++] =
552 nmem_strdup (p->mem, todb[i]);
558 void map_basenames (ZebraHandle zh, ODR stream,
559 int *num_bases, char ***basenames)
561 struct map_baseinfo info;
562 struct map_baseinfo *p = &info;
566 info.num_bases = *num_bases;
567 info.basenames = *basenames;
568 info.new_num_max = 128;
569 info.new_num_bases = 0;
570 info.new_basenames = (char **)
571 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
572 info.mem = stream->mem;
574 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
576 for (i = 0; i<p->num_bases; i++)
577 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
579 p->new_basenames[(p->new_num_bases)++] =
580 nmem_strdup (p->mem, p->basenames[i]);
582 *num_bases = info.new_num_bases;
583 *basenames = info.new_basenames;
584 for (i = 0; i<*num_bases; i++)
585 logf (LOG_LOG, "base %s", (*basenames)[i]);
588 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
589 Z_RPNQuery *query, int num_bases, char **basenames,
593 if (zebra_register_lock (zh))
595 map_basenames (zh, stream, &num_bases, &basenames);
596 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
598 zebra_register_unlock (zh);
600 logf(LOG_APP,"SEARCH:%d:",zh->hits);
603 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
604 const char *setname, Z_RecordComposition *comp,
605 oid_value input_format, int num_recs,
606 ZebraRetrievalRecord *recs)
611 if (zebra_register_lock (zh))
613 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
614 for (i = 0; i<num_recs; i++)
615 pos_array[i] = recs[i].position;
616 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
619 logf (LOG_DEBUG, "zebraPosSetCreate error");
621 zh->errString = nmem_strdup (stream->mem, setname);
625 for (i = 0; i<num_recs; i++)
631 sprintf (num_str, "%d", pos_array[i]);
633 zh->errString = nmem_strdup (stream->mem, num_str);
639 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
640 stream, input_format, comp,
641 &recs[i].format, &recs[i].buf,
644 recs[i].errString = NULL;
647 zebraPosSetDestroy (zh, poset, num_recs);
649 zebra_register_unlock (zh);
653 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
654 oid_value attributeset,
655 int num_bases, char **basenames,
656 int *position, int *num_entries, ZebraScanEntry **entries,
659 if (zebra_register_lock (zh))
665 map_basenames (zh, stream, &num_bases, &basenames);
666 rpn_scan (zh, stream, zapt, attributeset,
667 num_bases, basenames, position,
668 num_entries, entries, is_partial);
669 zebra_register_unlock (zh);
672 void zebra_sort (ZebraHandle zh, ODR stream,
673 int num_input_setnames, const char **input_setnames,
674 const char *output_setname, Z_SortKeySpecList *sort_sequence,
677 if (zebra_register_lock (zh))
679 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
680 output_setname, sort_sequence, sort_status);
681 zebra_register_unlock (zh);
684 int zebra_deleleResultSet(ZebraHandle zh, int function,
685 int num_setnames, char **setnames,
689 if (zebra_register_lock (zh))
690 return Z_DeleteStatus_systemProblemAtTarget;
693 case Z_DeleteRequest_list:
694 resultSetDestroy (zh, num_setnames, setnames, statuses);
696 case Z_DeleteRequest_all:
697 resultSetDestroy (zh, -1, 0, statuses);
700 zebra_register_unlock (zh);
701 status = Z_DeleteStatus_success;
702 for (i = 0; i<num_setnames; i++)
703 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
704 status = statuses[i];
708 int zebra_errCode (ZebraHandle zh)
713 const char *zebra_errString (ZebraHandle zh)
715 return diagbib1_str (zh->errCode);
718 char *zebra_errAdd (ZebraHandle zh)
720 return zh->errString;
723 int zebra_hits (ZebraHandle zh)
728 int zebra_auth (ZebraService zh, const char *user, const char *pass)
730 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
732 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
736 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
740 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
742 if (zebra_register_lock (zh))
744 xfree (zh->admin_databaseName);
745 zh->admin_databaseName = xstrdup(database);
746 zebra_register_unlock(zh);
749 void zebra_admin_import_end (ZebraHandle zh)
751 zebraExplain_flush (zh->service->zei, 1, zh);
755 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
759 if (zebra_register_lock (zh))
761 for (i = 0; i<segment->num_segmentRecords; i++)
763 Z_NamePlusRecord *npr = segment->segmentRecords[i];
764 const char *databaseName = npr->databaseName;
767 databaseName = zh->admin_databaseName;
768 printf ("--------------%d--------------------\n", i);
769 if (npr->which == Z_NamePlusRecord_intermediateFragment)
771 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
772 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
774 Odr_oct *oct = fragment->u.notExternallyTagged;
775 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
779 extract_rec_in_mem (zh, "grs.sgml",
787 0 /* match criteria */);
791 zebra_register_unlock(zh);
794 void zebra_admin_create (ZebraHandle zh, const char *database)
796 ZebraService zs = zh->service;
797 if (zebra_register_lock(zh))
802 /* announce database */
803 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
806 zh->errString = "Database already exist";
808 zebra_register_unlock(zh);
811 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
812 const char *input_str, int input_len,
813 char *output_str, int output_len)
816 if (!zh->service->zebra_maps)
818 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
819 input_str, input_len);
822 if (wrbuf_len(wrbuf) >= output_len)
824 if (wrbuf_len(wrbuf))
825 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
826 output_str[wrbuf_len(wrbuf)] = '\0';
827 return wrbuf_len(wrbuf);
830 static void extract_init (struct recExtractCtrl *p, RecWord *w)
832 w->zebra_maps = p->zebra_maps;
833 w->seqnos = p->seqno;
834 w->attrSet = VAL_BIB1;
840 static void extract_add_index_string (RecWord *p, const char *string,
844 unsigned char attrSet;
845 unsigned short attrUse;
848 int *pseqno = &p->seqnos[p->reg_type];
849 ZebraHandle zh = p->extractCtrl->handle;
850 struct recKeys *keys = &zh->keys;
852 if (keys->buf_used+1024 > keys->buf_max)
856 b = (char *) xmalloc (keys->buf_max += 128000);
857 if (keys->buf_used > 0)
858 memcpy (b, keys->buf, keys->buf_used);
862 dst = keys->buf + keys->buf_used;
864 attrSet = p->attrSet;
865 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
868 keys->prevAttrSet = attrSet;
869 attrUse = p->attrUse;
870 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
873 keys->prevAttrUse = attrUse;
875 diff = 1 + *pseqno - keys->prevSeqNo;
876 if (diff >= 1 && diff <= 15)
881 keys->prevSeqNo = *pseqno;
887 memcpy (dst, &attrSet, sizeof(attrSet));
888 dst += sizeof(attrSet);
892 memcpy (dst, &attrUse, sizeof(attrUse));
893 dst += sizeof(attrUse);
895 *dst++ = p->reg_type;
896 memcpy (dst, string, length);
902 memcpy (dst, pseqno, sizeof(*pseqno));
903 dst += sizeof(*pseqno);
905 keys->buf_used = dst - keys->buf;
910 static void extract_add_sort_string (RecWord *p, const char *string,
914 ZebraHandle zh = p->extractCtrl->handle;
915 struct sortKey *sortKeys = zh->sortKeys;
917 for (sk = sortKeys; sk; sk = sk->next)
918 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
921 sk = (struct sortKey *) xmalloc (sizeof(*sk));
925 sk->string = (char *) xmalloc (length);
927 memcpy (sk->string, string, length);
929 sk->attrSet = p->attrSet;
930 sk->attrUse = p->attrUse;
933 static void extract_add_string (RecWord *p, const char *string, int length)
936 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
937 extract_add_sort_string (p, string, length);
939 extract_add_index_string (p, string, length);
942 static void extract_add_incomplete_field (RecWord *p)
944 const char *b = p->string;
945 int remain = p->length;
946 const char **map = 0;
949 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
953 char buf[IT_MAX_WORD+1];
957 while (map && *map && **map == *CHR_SPACE)
959 remain = p->length - (b - p->string);
961 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
968 while (map && *map && **map != *CHR_SPACE)
970 const char *cp = *map;
972 while (i < IT_MAX_WORD && *cp)
974 remain = p->length - (b - p->string);
976 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
982 extract_add_string (p, buf, i);
984 (p->seqnos[p->reg_type])++; /* to separate this from next one */
987 static void extract_add_complete_field (RecWord *p)
989 const char *b = p->string;
990 char buf[IT_MAX_WORD+1];
991 const char **map = 0;
992 int i = 0, remain = p->length;
995 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
997 while (remain > 0 && i < IT_MAX_WORD)
999 while (map && *map && **map == *CHR_SPACE)
1001 remain = p->length - (b - p->string);
1003 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1010 if (i && i < IT_MAX_WORD)
1011 buf[i++] = *CHR_SPACE;
1012 while (map && *map && **map != *CHR_SPACE)
1014 const char *cp = *map;
1016 if (i >= IT_MAX_WORD)
1018 while (i < IT_MAX_WORD && *cp)
1020 remain = p->length - (b - p->string);
1022 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1030 extract_add_string (p, buf, i);
1033 static void extract_token_add (RecWord *p)
1036 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1037 p->string, p->length)))
1039 p->string = wrbuf_buf(wrbuf);
1040 p->length = wrbuf_len(wrbuf);
1042 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1043 extract_add_complete_field (p);
1045 extract_add_incomplete_field(p);
1048 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1050 ZebraHandle zh = (ZebraHandle) (p->handle);
1051 zebraExplain_addSchema (zh->service->zei, oid);
1054 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1055 int cmd, struct sortKey **skp)
1057 struct sortKey *sk = *skp;
1058 SortIdx sortIdx = zh->service->sortIdx;
1060 sortIdx_sysno (sortIdx, sysno);
1063 struct sortKey *sk_next = sk->next;
1064 sortIdx_type (sortIdx, sk->attrUse);
1065 sortIdx_add (sortIdx, sk->string, sk->length);
1073 struct encode_info {
1080 void encode_key_init (struct encode_info *i)
1087 char *encode_key_int (int d, char *bp)
1091 else if (d <= 16383)
1093 *bp++ = 64 + (d>>8);
1096 else if (d <= 4194303)
1098 *bp++ = 128 + (d>>16);
1099 *bp++ = (d>>8) & 255;
1104 *bp++ = 192 + (d>>24);
1105 *bp++ = (d>>16) & 255;
1106 *bp++ = (d>>8) & 255;
1112 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1117 while ((*bp++ = *k++))
1119 memcpy (&key, k+1, sizeof(struct it_key));
1120 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1121 if (i->sysno != key.sysno)
1123 i->sysno = key.sysno;
1126 else if (!i->seqno && !key.seqno && i->cmd == *k)
1128 bp = encode_key_int (key.seqno - i->seqno, bp);
1129 i->seqno = key.seqno;
1131 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1133 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1138 static void extract_flushWriteKeys (ZebraHandle zh)
1141 char out_fname[200];
1143 struct encode_info encode_info;
1144 int ptr_i = zh->ptr_i;
1148 if (!zh->key_buf || ptr_i <= 0)
1151 (zh->key_file_no)++;
1152 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1154 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1156 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1158 if (!(outf = fopen (out_fname, "wb")))
1160 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1163 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1164 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1166 encode_key_init (&encode_info);
1167 encode_key_write (cp, &encode_info, outf);
1171 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1172 if (strcmp (cp, prevcp))
1174 encode_key_init (&encode_info);
1175 encode_key_write (cp, &encode_info, outf);
1179 encode_key_write (cp + strlen(cp), &encode_info, outf);
1182 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1183 extract_get_fname_tmp (out_fname, key_file_no);
1185 if (!(outf = fopen (out_fname, "wb")))
1187 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1190 logf (LOG_LOG, "writing section %d", key_file_no);
1192 prevcp = key_buf[ptr_top-i];
1194 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1196 key_y_len = strlen(prevcp)+1;
1198 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1199 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1201 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1202 sizeof(char*), key_y_compare);
1203 cp = key_buf[ptr_top-ptr_i];
1205 encode_key_init (&encode_info);
1206 encode_key_write (cp, &encode_info, outf);
1209 cp = key_buf[ptr_top-ptr_i];
1210 encode_key_write (cp+key_y_len, &encode_info, outf);
1214 prevcp = key_buf[ptr_top-ptr_i];
1219 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1222 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1224 zh->key_buf_used = 0;
1227 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1228 int cmd, struct recKeys *reckeys)
1230 unsigned char attrSet = (unsigned char) -1;
1231 unsigned short attrUse = (unsigned short) -1;
1234 ZebraExplainInfo zei = zh->service->zei;
1238 int mem = 8*1024*1024;
1239 zh->key_buf = (char**) xmalloc (mem);
1240 zh->ptr_top = mem/sizeof(char*);
1242 zh->key_buf_used = 0;
1243 zh->key_file_no = 0;
1245 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1246 while (off < reckeys->buf_used)
1248 const char *src = reckeys->buf + off;
1256 memcpy (&attrSet, src, sizeof(attrSet));
1257 src += sizeof(attrSet);
1261 memcpy (&attrUse, src, sizeof(attrUse));
1262 src += sizeof(attrUse);
1264 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1265 extract_flushWriteKeys (zh);
1267 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1268 (char*)zh->key_buf + zh->key_buf_used;
1270 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1272 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1275 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1278 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1280 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1281 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1284 seqno += ((lead>>2) & 15)-1;
1287 memcpy (&seqno, src, sizeof(seqno));
1288 src += sizeof(seqno);
1292 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1293 (zh->key_buf_used) += sizeof(key);
1294 off = src - reckeys->buf;
1296 assert (off == reckeys->buf_used);
1299 static void extract_index (ZebraHandle zh)
1301 extract_flushWriteKeys (zh);
1302 zebra_index_merge (zh);
1305 static int explain_extract (void *handle, Record rec, data1_node *n)
1307 ZebraHandle zh = (ZebraHandle) handle;
1308 struct recExtractCtrl extractCtrl;
1311 if (zebraExplain_curDatabase (zh->service->zei,
1312 rec->info[recInfo_databaseName]))
1315 if (zebraExplain_newDatabase (zh->service->zei,
1316 rec->info[recInfo_databaseName], 0))
1320 zh->keys.buf_used = 0;
1321 zh->keys.prevAttrUse = -1;
1322 zh->keys.prevAttrSet = -1;
1323 zh->keys.prevSeqNo = 0;
1326 extractCtrl.init = extract_init;
1327 extractCtrl.tokenAdd = extract_token_add;
1328 extractCtrl.schemaAdd = extract_schema_add;
1329 extractCtrl.dh = zh->service->dh;
1330 for (i = 0; i<256; i++)
1331 extractCtrl.seqno[i] = 0;
1332 extractCtrl.zebra_maps = zh->service->zebra_maps;
1333 extractCtrl.flagShowRecords = 0;
1334 extractCtrl.handle = handle;
1336 grs_extract_tree(&extractCtrl, n);
1338 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1340 if (rec->size[recInfo_delKeys])
1342 struct recKeys delkeys;
1343 struct sortKey *sortKeys = 0;
1345 delkeys.buf_used = rec->size[recInfo_delKeys];
1346 delkeys.buf = rec->info[recInfo_delKeys];
1347 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1348 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1350 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1351 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1353 xfree (rec->info[recInfo_delKeys]);
1354 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1355 rec->info[recInfo_delKeys] = zh->keys.buf;
1356 zh->keys.buf = NULL;
1357 zh->keys.buf_max = 0;
1361 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1362 const char *buf, size_t buf_size,
1363 const char *databaseName, int delete_flag,
1364 int test_mode, int *sysno,
1365 int store_keys, int store_data,
1366 const char *match_criteria)
1368 RecordAttr *recordAttr;
1369 struct recExtractCtrl extractCtrl;
1374 const char *fname = "<no file>";
1376 long recordOffset = 0;
1377 struct zebra_fetch_control fc;
1380 fc.record_int_buf = buf;
1381 fc.record_int_len = buf_size;
1382 fc.record_int_pos = 0;
1384 fc.record_offset = 0;
1386 extractCtrl.offset = 0;
1387 extractCtrl.readf = zebra_record_int_read;
1388 extractCtrl.seekf = zebra_record_int_seek;
1389 extractCtrl.tellf = zebra_record_int_tell;
1390 extractCtrl.endf = zebra_record_int_end;
1391 extractCtrl.fh = &fc;
1393 /* announce database */
1394 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1396 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1400 recType_byName (zh->service->recTypes, recordType, subType,
1403 logf (LOG_WARN, "No such record type: %s", recordType);
1407 zh->keys.buf_used = 0;
1408 zh->keys.prevAttrUse = -1;
1409 zh->keys.prevAttrSet = -1;
1410 zh->keys.prevSeqNo = 0;
1413 extractCtrl.subType = subType;
1414 extractCtrl.init = extract_init;
1415 extractCtrl.tokenAdd = extract_token_add;
1416 extractCtrl.schemaAdd = extract_schema_add;
1417 extractCtrl.dh = zh->service->dh;
1418 extractCtrl.handle = zh;
1419 extractCtrl.zebra_maps = zh->service->zebra_maps;
1420 extractCtrl.flagShowRecords = 0;
1421 for (i = 0; i<256; i++)
1423 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1424 extractCtrl.seqno[i] = 1;
1426 extractCtrl.seqno[i] = 0;
1429 r = (*recType->extract)(clientData, &extractCtrl);
1431 if (r == RECCTRL_EXTRACT_EOF)
1433 else if (r == RECCTRL_EXTRACT_ERROR)
1435 /* error occured during extraction ... */
1437 yaz_log (LOG_WARN, "extract error");
1439 if (rGroup->flagRw &&
1440 records_processed < rGroup->fileVerboseLimit)
1442 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1443 fname, (long) recordOffset);
1448 if (zh->keys.buf_used == 0)
1450 /* the extraction process returned no information - the record
1451 is probably empty - unless flagShowRecords is in use */
1454 logf (LOG_WARN, "No keys generated for record");
1455 logf (LOG_WARN, " The file is probably empty");
1458 /* match criteria */
1465 logf (LOG_LOG, "delete %s %s %ld", recordType,
1466 fname, (long) recordOffset);
1467 logf (LOG_WARN, "cannot delete record above (seems new)");
1470 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1471 (long) recordOffset);
1472 rec = rec_new (zh->service->records);
1474 *sysno = rec->sysno;
1476 recordAttr = rec_init_attr (zh->service->zei, rec);
1481 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1484 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1485 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1489 /* record already exists */
1490 struct recKeys delkeys;
1492 rec = rec_get (zh->service->records, *sysno);
1495 recordAttr = rec_init_attr (zh->service->zei, rec);
1497 if (recordAttr->runNumber ==
1498 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1500 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1501 fname, (long) recordOffset);
1505 delkeys.buf_used = rec->size[recInfo_delKeys];
1506 delkeys.buf = rec->info[recInfo_delKeys];
1507 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1508 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1511 /* record going to be deleted */
1512 if (!delkeys.buf_used)
1514 logf (LOG_LOG, "delete %s %s %ld", recordType,
1515 fname, (long) recordOffset);
1516 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1520 logf (LOG_LOG, "delete %s %s %ld", recordType,
1521 fname, (long) recordOffset);
1524 dict_delete (matchDict, matchStr);
1526 rec_del (zh->service->records, &rec);
1533 /* record going to be updated */
1534 if (!delkeys.buf_used)
1536 logf (LOG_LOG, "update %s %s %ld", recordType,
1537 fname, (long) recordOffset);
1538 logf (LOG_WARN, "cannot update file above, storeKeys false");
1542 logf (LOG_LOG, "update %s %s %ld", recordType,
1543 fname, (long) recordOffset);
1544 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1548 /* update file type */
1549 xfree (rec->info[recInfo_fileType]);
1550 rec->info[recInfo_fileType] =
1551 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1553 /* update filename */
1554 xfree (rec->info[recInfo_filename]);
1555 rec->info[recInfo_filename] =
1556 rec_strdup (fname, &rec->size[recInfo_filename]);
1558 /* update delete keys */
1559 xfree (rec->info[recInfo_delKeys]);
1560 if (zh->keys.buf_used > 0 && store_keys == 1)
1562 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1563 rec->info[recInfo_delKeys] = zh->keys.buf;
1564 zh->keys.buf = NULL;
1565 zh->keys.buf_max = 0;
1569 rec->info[recInfo_delKeys] = NULL;
1570 rec->size[recInfo_delKeys] = 0;
1573 /* save file size of original record */
1574 zebraExplain_recordBytesIncrement (zh->service->zei,
1575 - recordAttr->recordSize);
1577 recordAttr->recordSize = fi->file_moffset - recordOffset;
1578 if (!recordAttr->recordSize)
1579 recordAttr->recordSize = fi->file_max - recordOffset;
1581 recordAttr->recordSize = buf_size;
1583 zebraExplain_recordBytesIncrement (zh->service->zei,
1584 recordAttr->recordSize);
1586 /* set run-number for this record */
1587 recordAttr->runNumber =
1588 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1590 /* update store data */
1591 xfree (rec->info[recInfo_storeData]);
1592 if (store_data == 1)
1594 rec->size[recInfo_storeData] = recordAttr->recordSize;
1595 rec->info[recInfo_storeData] = (char *)
1596 xmalloc (recordAttr->recordSize);
1598 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1600 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1602 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1603 (long) recordOffset, fname);
1606 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1607 < recordAttr->recordSize)
1609 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1610 recordAttr->recordSize, fname);
1617 rec->info[recInfo_storeData] = NULL;
1618 rec->size[recInfo_storeData] = 0;
1620 /* update database name */
1621 xfree (rec->info[recInfo_databaseName]);
1622 rec->info[recInfo_databaseName] =
1623 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1626 recordAttr->recordOffset = recordOffset;
1628 /* commit this record */
1629 rec_put (zh->service->records, &rec);