2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.40 2000-11-29 15:21:31 adam
7 * Fixed problem with passwd db.
9 * Revision 1.39 2000/11/29 14:24:01 adam
10 * Script configure uses yaz pthreads options. Added locking for
11 * zebra_register_{lock,unlock}.
13 * Revision 1.38 2000/11/08 13:46:58 adam
14 * Fixed scan: server could break if bad attribute/database was selected.
15 * Work on remote update.
17 * Revision 1.37 2000/10/17 12:37:09 adam
18 * Fixed notification of live-updates. Fixed minor problem with mf_init
19 * where it didn't handle shadow area file names correctly.
21 * Revision 1.36 2000/09/06 08:59:36 adam
22 * Using read-only (for now) for server.
24 * Revision 1.35 2000/07/07 12:49:20 adam
25 * Optimized resultSetInsert{Rank,Sort}.
27 * Revision 1.34 2000/06/09 13:56:38 ian
28 * Added some logging on Authentication and searches.
30 * Revision 1.33 2000/05/18 12:01:36 adam
31 * System call times(2) used again. More 64-bit fixes.
33 * Revision 1.32 2000/04/19 14:35:59 adam
34 * WIN32 update (this version is known not to work on Windows).
36 * Revision 1.31 2000/04/05 10:07:02 adam
37 * Minor zebra compile fix.
39 * Revision 1.30 2000/04/05 09:49:35 adam
40 * On Unix, zebra/z'mbol uses automake.
42 * Revision 1.29 2000/03/20 19:08:36 adam
43 * Added remote record import using Z39.50 extended services and Segment
46 * Revision 1.28 2000/03/15 15:00:30 adam
47 * First work on threaded version.
49 * Revision 1.27 2000/02/24 12:31:17 adam
50 * Added zebra_string_norm.
52 * Revision 1.26 1999/11/30 13:48:03 adam
53 * Improved installation. Updated for inclusion of YAZ header files.
55 * Revision 1.25 1999/11/04 15:00:45 adam
56 * Implemented delete result set(s).
58 * Revision 1.24 1999/10/14 14:33:50 adam
59 * Added truncation 5=106.
61 * Revision 1.23 1999/09/07 11:36:32 adam
64 * Revision 1.22 1999/08/02 10:13:47 adam
65 * Fixed bug regarding zebra_hits.
67 * Revision 1.21 1999/07/14 10:59:26 adam
68 * Changed functions isc_getmethod, isams_getmethod.
69 * Improved fatal error handling (such as missing EXPLAIN schema).
71 * Revision 1.20 1999/07/06 12:28:04 adam
72 * Updated record index structure. Format includes version ID. Compression
73 * algorithm ID is stored for each record block.
75 * Revision 1.19 1999/05/26 07:49:13 adam
78 * Revision 1.18 1999/05/15 14:36:38 adam
79 * Updated dictionary. Implemented "compression" of dictionary.
81 * Revision 1.17 1999/05/12 13:08:06 adam
82 * First version of ISAMS.
84 * Revision 1.16 1999/02/19 10:38:30 adam
85 * Implemented chdir-setting.
87 * Revision 1.15 1999/02/17 12:18:12 adam
88 * Fixed zebra_close so that a NULL pointer is ignored.
90 * Revision 1.14 1999/02/02 14:51:11 adam
91 * Updated WIN32 code specific sections. Changed header.
93 * Revision 1.13 1998/12/16 12:23:30 adam
94 * Added facility for database name mapping using resource mapdb.
96 * Revision 1.12 1998/11/16 10:18:10 adam
97 * Better error reporting for result sets.
99 * Revision 1.11 1998/10/16 08:14:34 adam
100 * Updated record control system.
102 * Revision 1.10 1998/09/22 10:03:42 adam
103 * Changed result sets to be persistent in the sense that they can
104 * be re-searched if needed.
105 * Fixed memory leak in rsm_or.
107 * Revision 1.9 1998/09/02 13:53:17 adam
108 * Extra parameter decode added to search routines to implement
109 * persistent queries.
111 * Revision 1.8 1998/08/24 17:29:23 adam
114 * Revision 1.7 1998/06/24 12:16:13 adam
115 * Support for relations on text operands. Open range support in
116 * DFA module (i.e. [-j], [g-]).
118 * Revision 1.6 1998/06/22 11:36:47 adam
119 * Added authentication check facility to zebra.
121 * Revision 1.5 1998/06/13 00:14:08 adam
124 * Revision 1.4 1998/06/12 12:22:12 adam
127 * Revision 1.3 1998/05/27 16:57:44 adam
128 * Zebra returns surrogate diagnostic for single records when
131 * Revision 1.2 1998/05/20 10:12:19 adam
132 * Implemented automatic EXPLAIN database maintenance.
133 * Modified Zebra to work with ASN.1 compiled version of YAZ.
135 * Revision 1.1 1998/03/05 08:45:13 adam
136 * New result set model and modular ranking system. Moved towards
137 * descent server API. System information stored as "SGML" records.
151 #include <yaz/diagbib1.h>
155 static void zebra_chdir (ZebraService zh)
157 const char *dir = res_get (zh->res, "chdir");
160 logf (LOG_DEBUG, "chdir %s", dir);
168 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
169 const char *buf, size_t buf_size,
170 const char *databaseName, int delete_flag,
171 int test_mode, int *sysno,
172 int store_keys, int store_data,
173 const char *match_criteria);
175 static int explain_extract (void *handle, Record rec, data1_node *n);
176 static void extract_index (ZebraHandle zh);
178 static void zebra_register_unlock (ZebraHandle zh);
180 static int zebra_register_activate (ZebraService zh, int rw);
181 static int zebra_register_deactivate (ZebraService zh);
183 static int zebra_register_lock (ZebraHandle zh, int rw)
189 if (!zh->service->active)
199 zebra_mutex_cond_lock (&zh->service->session_lock);
201 state = zebra_server_lock_get_state(zh->service, &lastChange);
203 zebra_server_lock (zh->service, state);
214 logf (LOG_LOG, "Register in read/write mode");
215 else if (zh->service->registerState == state)
217 if (zh->service->registerChange >= lastChange)
219 logf (LOG_LOG, "Register completely updated since last access");
221 else if (zh->service->registerState == -1)
222 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
225 logf (LOG_LOG, "Register has changed state from %d to %d",
226 zh->service->registerState, state);
227 zh->service->registerChange = lastChange;
229 zebra_register_deactivate (zh->service);
231 zh->service->registerState = state;
233 zebra_register_activate (zh->service, rw);
238 static void zebra_register_unlock (ZebraHandle zh)
240 if (zh->service->registerState != -1)
241 zebra_server_unlock (zh->service, zh->service->registerState);
242 zebra_mutex_cond_unlock (&zh->service->session_lock);
245 logf (LOG_LOG, "user/system: %ld/%ld",
246 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
247 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
252 ZebraHandle zebra_open (ZebraService zs)
260 zh = (ZebraHandle) xmalloc (sizeof(*zh));
261 yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh);
270 zh->admin_databaseName = 0;
272 zebra_mutex_cond_lock (&zs->session_lock);
274 zh->next = zs->sessions;
277 zebra_mutex_cond_unlock (&zs->session_lock);
283 ZebraService zebra_start (const char *configName)
285 ZebraService zh = xmalloc (sizeof(*zh));
287 yaz_log (LOG_LOG, "zebra_start %s", configName);
289 zh->configName = xstrdup(configName);
294 zh->registerState = -1;
295 zh->registerChange = 0;
297 if (!(zh->res = res_open (zh->configName)))
299 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
303 zebra_server_lock_init (zh);
304 zebra_mutex_cond_init (&zh->session_lock);
305 if (!res_get (zh->res, "passwd"))
306 zh->passwd_db = NULL;
309 zh->passwd_db = passwd_db_open ();
311 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
313 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
319 static int zebra_register_activate (ZebraService zh, int rw)
323 yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)",
327 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
328 zh->registerState ? "yes" : "no");
330 zh->dh = data1_create ();
333 zh->bfs = bfs_create (res_get (zh->res, "register"));
336 data1_destroy(zh->dh);
339 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
340 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
341 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
342 zh->recTypes = recTypes_init (zh->dh);
343 recTypes_default_handlers (zh->recTypes);
346 zh->zebra_maps = zebra_maps_open (zh->res);
347 zh->rank_classes = NULL;
360 zebraRankInstall (zh, rank1_class);
362 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
364 logf (LOG_WARN, "rec_open");
367 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
369 logf (LOG_WARN, "dict_open");
372 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
374 logf (LOG_WARN, "sortIdx_open");
377 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
379 struct ISAMS_M_s isams_m;
380 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
381 key_isams_m(zh->res, &isams_m))))
383 logf (LOG_WARN, "isams_open");
388 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
390 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
391 sizeof (struct it_key), zh->res)))
393 logf (LOG_WARN, "is_open");
397 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
399 struct ISAMC_M_s isamc_m;
400 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
401 rw, key_isamc_m(zh->res, &isamc_m))))
403 logf (LOG_WARN, "isc_open");
407 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
409 struct ISAMD_M_s isamd_m;
411 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
412 rw, key_isamd_m(zh->res, &isamd_m))))
414 logf (LOG_WARN, "isamd_open");
419 zh->zei = zebraExplain_open (zh->records, zh->dh,
420 zh->res, rw, 0 /* rGroup */,
424 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
428 yaz_log (LOG_LOG, "zebra_register_activate ok");
432 void zebra_admin_shutdown (ZebraHandle zh)
434 zebra_mutex_cond_lock (&zh->service->session_lock);
435 zh->service->stop_flag = 1;
436 if (!zh->service->sessions)
437 zebra_register_deactivate(zh->service);
438 zh->service->active = 0;
439 zebra_mutex_cond_unlock (&zh->service->session_lock);
442 void zebra_admin_start (ZebraHandle zh)
444 ZebraService zs = zh->service;
446 zebra_mutex_cond_lock (&zs->session_lock);
448 zh->service->active = 1;
449 zebra_mutex_cond_unlock (&zs->session_lock);
452 static int zebra_register_deactivate (ZebraService zs)
457 yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)",
461 yaz_log(LOG_LOG, "zebra_register_deactivate");
465 zebraExplain_close (zs->zei, 0);
466 dict_close (zs->dict);
467 sortIdx_close (zs->sortIdx);
469 isams_close (zs->isams);
474 isc_close (zs->isamc);
476 isamd_close (zs->isamd);
478 rec_close (&zs->records);
480 recTypes_destroy (zs->recTypes);
481 zebra_maps_close (zs->zebra_maps);
482 zebraRankDestroy (zs);
483 bfs_destroy (zs->bfs);
484 data1_destroy (zs->dh);
487 passwd_db_close (zs->passwd_db);
492 void zebra_stop(ZebraService zs)
496 yaz_log (LOG_LOG, "zebra_stop");
498 assert (!zs->sessions);
500 zebra_mutex_cond_destroy (&zs->session_lock);
502 zebra_register_deactivate(zs);
504 xfree (zs->configName);
508 void zebra_close (ZebraHandle zh)
510 ZebraService zs = zh->service;
511 struct zebra_session **sp;
513 yaz_log (LOG_LOG, "zebra_close zh=%p", zh);
516 resultSetDestroy (zh, -1, 0, 0);
523 xfree (zh->admin_databaseName);
524 zebra_mutex_cond_lock (&zs->session_lock);
536 if (!zs->sessions && zs->stop_flag)
537 zebra_register_deactivate(zs);
538 zebra_mutex_cond_unlock (&zs->session_lock);
542 struct map_baseinfo {
548 char **new_basenames;
552 void map_basenames_func (void *vp, const char *name, const char *value)
554 struct map_baseinfo *p = (struct map_baseinfo *) vp;
556 char fromdb[128], todb[8][128];
559 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
560 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
561 todb[5], todb[6], todb[7]);
565 for (i = 0; i<p->num_bases; i++)
566 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
569 for (i = 0; i < no; i++)
571 if (p->new_num_bases == p->new_num_max)
573 p->new_basenames[(p->new_num_bases)++] =
574 nmem_strdup (p->mem, todb[i]);
580 void map_basenames (ZebraHandle zh, ODR stream,
581 int *num_bases, char ***basenames)
583 struct map_baseinfo info;
584 struct map_baseinfo *p = &info;
588 info.num_bases = *num_bases;
589 info.basenames = *basenames;
590 info.new_num_max = 128;
591 info.new_num_bases = 0;
592 info.new_basenames = (char **)
593 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
594 info.mem = stream->mem;
596 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
598 for (i = 0; i<p->num_bases; i++)
599 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
601 p->new_basenames[(p->new_num_bases)++] =
602 nmem_strdup (p->mem, p->basenames[i]);
604 *num_bases = info.new_num_bases;
605 *basenames = info.new_basenames;
606 for (i = 0; i<*num_bases; i++)
607 logf (LOG_LOG, "base %s", (*basenames)[i]);
610 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
611 Z_RPNQuery *query, int num_bases, char **basenames,
615 if (zebra_register_lock (zh, 0))
617 map_basenames (zh, stream, &num_bases, &basenames);
618 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
620 zebra_register_unlock (zh);
622 logf(LOG_APP,"SEARCH:%d:",zh->hits);
625 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
626 const char *setname, Z_RecordComposition *comp,
627 oid_value input_format, int num_recs,
628 ZebraRetrievalRecord *recs)
633 if (zebra_register_lock (zh, 0))
635 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
636 for (i = 0; i<num_recs; i++)
637 pos_array[i] = recs[i].position;
638 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
641 logf (LOG_DEBUG, "zebraPosSetCreate error");
643 zh->errString = nmem_strdup (stream->mem, setname);
647 for (i = 0; i<num_recs; i++)
653 sprintf (num_str, "%d", pos_array[i]);
655 zh->errString = nmem_strdup (stream->mem, num_str);
661 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
662 stream, input_format, comp,
663 &recs[i].format, &recs[i].buf,
666 recs[i].errString = NULL;
669 zebraPosSetDestroy (zh, poset, num_recs);
671 zebra_register_unlock (zh);
675 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
676 oid_value attributeset,
677 int num_bases, char **basenames,
678 int *position, int *num_entries, ZebraScanEntry **entries,
681 if (zebra_register_lock (zh, 0))
687 map_basenames (zh, stream, &num_bases, &basenames);
688 rpn_scan (zh, stream, zapt, attributeset,
689 num_bases, basenames, position,
690 num_entries, entries, is_partial);
691 zebra_register_unlock (zh);
694 void zebra_sort (ZebraHandle zh, ODR stream,
695 int num_input_setnames, const char **input_setnames,
696 const char *output_setname, Z_SortKeySpecList *sort_sequence,
699 if (zebra_register_lock (zh, 0))
701 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
702 output_setname, sort_sequence, sort_status);
703 zebra_register_unlock (zh);
706 int zebra_deleleResultSet(ZebraHandle zh, int function,
707 int num_setnames, char **setnames,
711 if (zebra_register_lock (zh, 0))
712 return Z_DeleteStatus_systemProblemAtTarget;
715 case Z_DeleteRequest_list:
716 resultSetDestroy (zh, num_setnames, setnames, statuses);
718 case Z_DeleteRequest_all:
719 resultSetDestroy (zh, -1, 0, statuses);
722 zebra_register_unlock (zh);
723 status = Z_DeleteStatus_success;
724 for (i = 0; i<num_setnames; i++)
725 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
726 status = statuses[i];
730 int zebra_errCode (ZebraHandle zh)
735 const char *zebra_errString (ZebraHandle zh)
737 return diagbib1_str (zh->errCode);
740 char *zebra_errAdd (ZebraHandle zh)
742 return zh->errString;
745 int zebra_hits (ZebraHandle zh)
750 int zebra_auth (ZebraService zh, const char *user, const char *pass)
752 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
754 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
758 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
762 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
764 if (zebra_register_lock (zh, 1))
766 xfree (zh->admin_databaseName);
767 zh->admin_databaseName = xstrdup(database);
770 void zebra_admin_import_end (ZebraHandle zh)
772 zebraExplain_flush (zh->service->zei, 1, zh);
774 zebra_register_unlock (zh);
777 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
781 if (zh->service->active < 2)
783 for (i = 0; i<segment->num_segmentRecords; i++)
785 Z_NamePlusRecord *npr = segment->segmentRecords[i];
786 const char *databaseName = npr->databaseName;
789 databaseName = zh->admin_databaseName;
790 printf ("--------------%d--------------------\n", i);
791 if (npr->which == Z_NamePlusRecord_intermediateFragment)
793 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
794 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
796 Odr_oct *oct = fragment->u.notExternallyTagged;
797 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
801 extract_rec_in_mem (zh, "grs.sgml",
809 0 /* match criteria */);
815 void zebra_admin_create (ZebraHandle zh, const char *database)
817 ZebraService zs = zh->service;
818 if (zebra_register_lock(zh, 1))
823 /* announce database */
824 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
827 zh->errString = "Database already exist";
829 zebraExplain_flush (zh->service->zei, 1, zh);
831 zebra_register_unlock(zh);
834 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
835 const char *input_str, int input_len,
836 char *output_str, int output_len)
839 if (!zh->service->zebra_maps)
841 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
842 input_str, input_len);
845 if (wrbuf_len(wrbuf) >= output_len)
847 if (wrbuf_len(wrbuf))
848 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
849 output_str[wrbuf_len(wrbuf)] = '\0';
850 return wrbuf_len(wrbuf);
853 static void extract_init (struct recExtractCtrl *p, RecWord *w)
855 w->zebra_maps = p->zebra_maps;
856 w->seqnos = p->seqno;
857 w->attrSet = VAL_BIB1;
863 static void extract_add_index_string (RecWord *p, const char *string,
867 unsigned char attrSet;
868 unsigned short attrUse;
871 int *pseqno = &p->seqnos[p->reg_type];
872 ZebraHandle zh = p->extractCtrl->handle;
873 struct recKeys *keys = &zh->keys;
875 if (keys->buf_used+1024 > keys->buf_max)
879 b = (char *) xmalloc (keys->buf_max += 128000);
880 if (keys->buf_used > 0)
881 memcpy (b, keys->buf, keys->buf_used);
885 dst = keys->buf + keys->buf_used;
887 attrSet = p->attrSet;
888 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
891 keys->prevAttrSet = attrSet;
892 attrUse = p->attrUse;
893 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
896 keys->prevAttrUse = attrUse;
898 diff = 1 + *pseqno - keys->prevSeqNo;
899 if (diff >= 1 && diff <= 15)
904 keys->prevSeqNo = *pseqno;
910 memcpy (dst, &attrSet, sizeof(attrSet));
911 dst += sizeof(attrSet);
915 memcpy (dst, &attrUse, sizeof(attrUse));
916 dst += sizeof(attrUse);
918 *dst++ = p->reg_type;
919 memcpy (dst, string, length);
925 memcpy (dst, pseqno, sizeof(*pseqno));
926 dst += sizeof(*pseqno);
928 keys->buf_used = dst - keys->buf;
933 static void extract_add_sort_string (RecWord *p, const char *string,
937 ZebraHandle zh = p->extractCtrl->handle;
938 struct sortKey *sortKeys = zh->sortKeys;
940 for (sk = sortKeys; sk; sk = sk->next)
941 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
944 sk = (struct sortKey *) xmalloc (sizeof(*sk));
948 sk->string = (char *) xmalloc (length);
950 memcpy (sk->string, string, length);
952 sk->attrSet = p->attrSet;
953 sk->attrUse = p->attrUse;
956 static void extract_add_string (RecWord *p, const char *string, int length)
959 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
960 extract_add_sort_string (p, string, length);
962 extract_add_index_string (p, string, length);
965 static void extract_add_incomplete_field (RecWord *p)
967 const char *b = p->string;
968 int remain = p->length;
969 const char **map = 0;
972 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
976 char buf[IT_MAX_WORD+1];
980 while (map && *map && **map == *CHR_SPACE)
982 remain = p->length - (b - p->string);
984 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
991 while (map && *map && **map != *CHR_SPACE)
993 const char *cp = *map;
995 while (i < IT_MAX_WORD && *cp)
997 remain = p->length - (b - p->string);
999 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1005 extract_add_string (p, buf, i);
1007 (p->seqnos[p->reg_type])++; /* to separate this from next one */
1010 static void extract_add_complete_field (RecWord *p)
1012 const char *b = p->string;
1013 char buf[IT_MAX_WORD+1];
1014 const char **map = 0;
1015 int i = 0, remain = p->length;
1018 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1020 while (remain > 0 && i < IT_MAX_WORD)
1022 while (map && *map && **map == *CHR_SPACE)
1024 remain = p->length - (b - p->string);
1026 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1033 if (i && i < IT_MAX_WORD)
1034 buf[i++] = *CHR_SPACE;
1035 while (map && *map && **map != *CHR_SPACE)
1037 const char *cp = *map;
1039 if (i >= IT_MAX_WORD)
1041 while (i < IT_MAX_WORD && *cp)
1043 remain = p->length - (b - p->string);
1045 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1053 extract_add_string (p, buf, i);
1056 static void extract_token_add (RecWord *p)
1059 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1060 p->string, p->length)))
1062 p->string = wrbuf_buf(wrbuf);
1063 p->length = wrbuf_len(wrbuf);
1065 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1066 extract_add_complete_field (p);
1068 extract_add_incomplete_field(p);
1071 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1073 ZebraHandle zh = (ZebraHandle) (p->handle);
1074 zebraExplain_addSchema (zh->service->zei, oid);
1077 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1078 int cmd, struct sortKey **skp)
1080 struct sortKey *sk = *skp;
1081 SortIdx sortIdx = zh->service->sortIdx;
1083 sortIdx_sysno (sortIdx, sysno);
1086 struct sortKey *sk_next = sk->next;
1087 sortIdx_type (sortIdx, sk->attrUse);
1088 sortIdx_add (sortIdx, sk->string, sk->length);
1096 struct encode_info {
1103 void encode_key_init (struct encode_info *i)
1110 char *encode_key_int (int d, char *bp)
1114 else if (d <= 16383)
1116 *bp++ = 64 + (d>>8);
1119 else if (d <= 4194303)
1121 *bp++ = 128 + (d>>16);
1122 *bp++ = (d>>8) & 255;
1127 *bp++ = 192 + (d>>24);
1128 *bp++ = (d>>16) & 255;
1129 *bp++ = (d>>8) & 255;
1135 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1140 while ((*bp++ = *k++))
1142 memcpy (&key, k+1, sizeof(struct it_key));
1143 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1144 if (i->sysno != key.sysno)
1146 i->sysno = key.sysno;
1149 else if (!i->seqno && !key.seqno && i->cmd == *k)
1151 bp = encode_key_int (key.seqno - i->seqno, bp);
1152 i->seqno = key.seqno;
1154 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1156 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1161 static void extract_flushWriteKeys (ZebraHandle zh)
1164 char out_fname[200];
1166 struct encode_info encode_info;
1167 int ptr_i = zh->ptr_i;
1171 if (!zh->key_buf || ptr_i <= 0)
1174 (zh->key_file_no)++;
1175 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1177 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1179 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1181 if (!(outf = fopen (out_fname, "wb")))
1183 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1186 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1187 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1189 encode_key_init (&encode_info);
1190 encode_key_write (cp, &encode_info, outf);
1194 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1195 if (strcmp (cp, prevcp))
1197 encode_key_init (&encode_info);
1198 encode_key_write (cp, &encode_info, outf);
1202 encode_key_write (cp + strlen(cp), &encode_info, outf);
1205 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1206 extract_get_fname_tmp (out_fname, key_file_no);
1208 if (!(outf = fopen (out_fname, "wb")))
1210 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1213 logf (LOG_LOG, "writing section %d", key_file_no);
1215 prevcp = key_buf[ptr_top-i];
1217 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1219 key_y_len = strlen(prevcp)+1;
1221 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1222 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1224 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1225 sizeof(char*), key_y_compare);
1226 cp = key_buf[ptr_top-ptr_i];
1228 encode_key_init (&encode_info);
1229 encode_key_write (cp, &encode_info, outf);
1232 cp = key_buf[ptr_top-ptr_i];
1233 encode_key_write (cp+key_y_len, &encode_info, outf);
1237 prevcp = key_buf[ptr_top-ptr_i];
1242 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1245 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1247 zh->key_buf_used = 0;
1250 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1251 int cmd, struct recKeys *reckeys)
1253 unsigned char attrSet = (unsigned char) -1;
1254 unsigned short attrUse = (unsigned short) -1;
1257 ZebraExplainInfo zei = zh->service->zei;
1261 int mem = 8*1024*1024;
1262 zh->key_buf = (char**) xmalloc (mem);
1263 zh->ptr_top = mem/sizeof(char*);
1265 zh->key_buf_used = 0;
1266 zh->key_file_no = 0;
1268 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1269 while (off < reckeys->buf_used)
1271 const char *src = reckeys->buf + off;
1279 memcpy (&attrSet, src, sizeof(attrSet));
1280 src += sizeof(attrSet);
1284 memcpy (&attrUse, src, sizeof(attrUse));
1285 src += sizeof(attrUse);
1287 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1288 extract_flushWriteKeys (zh);
1290 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1291 (char*)zh->key_buf + zh->key_buf_used;
1293 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1295 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1298 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1301 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1303 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1304 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1307 seqno += ((lead>>2) & 15)-1;
1310 memcpy (&seqno, src, sizeof(seqno));
1311 src += sizeof(seqno);
1315 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1316 (zh->key_buf_used) += sizeof(key);
1317 off = src - reckeys->buf;
1319 assert (off == reckeys->buf_used);
1322 static void extract_index (ZebraHandle zh)
1324 extract_flushWriteKeys (zh);
1325 zebra_index_merge (zh);
1328 static int explain_extract (void *handle, Record rec, data1_node *n)
1330 ZebraHandle zh = (ZebraHandle) handle;
1331 struct recExtractCtrl extractCtrl;
1334 if (zebraExplain_curDatabase (zh->service->zei,
1335 rec->info[recInfo_databaseName]))
1338 if (zebraExplain_newDatabase (zh->service->zei,
1339 rec->info[recInfo_databaseName], 0))
1343 zh->keys.buf_used = 0;
1344 zh->keys.prevAttrUse = -1;
1345 zh->keys.prevAttrSet = -1;
1346 zh->keys.prevSeqNo = 0;
1349 extractCtrl.init = extract_init;
1350 extractCtrl.tokenAdd = extract_token_add;
1351 extractCtrl.schemaAdd = extract_schema_add;
1352 extractCtrl.dh = zh->service->dh;
1353 for (i = 0; i<256; i++)
1354 extractCtrl.seqno[i] = 0;
1355 extractCtrl.zebra_maps = zh->service->zebra_maps;
1356 extractCtrl.flagShowRecords = 0;
1357 extractCtrl.handle = handle;
1359 grs_extract_tree(&extractCtrl, n);
1361 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1363 if (rec->size[recInfo_delKeys])
1365 struct recKeys delkeys;
1366 struct sortKey *sortKeys = 0;
1368 delkeys.buf_used = rec->size[recInfo_delKeys];
1369 delkeys.buf = rec->info[recInfo_delKeys];
1370 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1371 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1373 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1374 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1376 xfree (rec->info[recInfo_delKeys]);
1377 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1378 rec->info[recInfo_delKeys] = zh->keys.buf;
1379 zh->keys.buf = NULL;
1380 zh->keys.buf_max = 0;
1384 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1385 const char *buf, size_t buf_size,
1386 const char *databaseName, int delete_flag,
1387 int test_mode, int *sysno,
1388 int store_keys, int store_data,
1389 const char *match_criteria)
1391 RecordAttr *recordAttr;
1392 struct recExtractCtrl extractCtrl;
1397 const char *fname = "<no file>";
1399 long recordOffset = 0;
1400 struct zebra_fetch_control fc;
1403 fc.record_int_buf = buf;
1404 fc.record_int_len = buf_size;
1405 fc.record_int_pos = 0;
1407 fc.record_offset = 0;
1409 extractCtrl.offset = 0;
1410 extractCtrl.readf = zebra_record_int_read;
1411 extractCtrl.seekf = zebra_record_int_seek;
1412 extractCtrl.tellf = zebra_record_int_tell;
1413 extractCtrl.endf = zebra_record_int_end;
1414 extractCtrl.fh = &fc;
1416 /* announce database */
1417 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1419 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1423 recType_byName (zh->service->recTypes, recordType, subType,
1426 logf (LOG_WARN, "No such record type: %s", recordType);
1430 zh->keys.buf_used = 0;
1431 zh->keys.prevAttrUse = -1;
1432 zh->keys.prevAttrSet = -1;
1433 zh->keys.prevSeqNo = 0;
1436 extractCtrl.subType = subType;
1437 extractCtrl.init = extract_init;
1438 extractCtrl.tokenAdd = extract_token_add;
1439 extractCtrl.schemaAdd = extract_schema_add;
1440 extractCtrl.dh = zh->service->dh;
1441 extractCtrl.handle = zh;
1442 extractCtrl.zebra_maps = zh->service->zebra_maps;
1443 extractCtrl.flagShowRecords = 0;
1444 for (i = 0; i<256; i++)
1446 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1447 extractCtrl.seqno[i] = 1;
1449 extractCtrl.seqno[i] = 0;
1452 r = (*recType->extract)(clientData, &extractCtrl);
1454 if (r == RECCTRL_EXTRACT_EOF)
1456 else if (r == RECCTRL_EXTRACT_ERROR)
1458 /* error occured during extraction ... */
1460 yaz_log (LOG_WARN, "extract error");
1462 if (rGroup->flagRw &&
1463 records_processed < rGroup->fileVerboseLimit)
1465 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1466 fname, (long) recordOffset);
1471 if (zh->keys.buf_used == 0)
1473 /* the extraction process returned no information - the record
1474 is probably empty - unless flagShowRecords is in use */
1477 logf (LOG_WARN, "No keys generated for record");
1478 logf (LOG_WARN, " The file is probably empty");
1481 /* match criteria */
1488 logf (LOG_LOG, "delete %s %s %ld", recordType,
1489 fname, (long) recordOffset);
1490 logf (LOG_WARN, "cannot delete record above (seems new)");
1493 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1494 (long) recordOffset);
1495 rec = rec_new (zh->service->records);
1497 *sysno = rec->sysno;
1499 recordAttr = rec_init_attr (zh->service->zei, rec);
1504 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1507 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1508 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1512 /* record already exists */
1513 struct recKeys delkeys;
1515 rec = rec_get (zh->service->records, *sysno);
1518 recordAttr = rec_init_attr (zh->service->zei, rec);
1520 if (recordAttr->runNumber ==
1521 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1523 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1524 fname, (long) recordOffset);
1528 delkeys.buf_used = rec->size[recInfo_delKeys];
1529 delkeys.buf = rec->info[recInfo_delKeys];
1530 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1531 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1534 /* record going to be deleted */
1535 if (!delkeys.buf_used)
1537 logf (LOG_LOG, "delete %s %s %ld", recordType,
1538 fname, (long) recordOffset);
1539 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1543 logf (LOG_LOG, "delete %s %s %ld", recordType,
1544 fname, (long) recordOffset);
1547 dict_delete (matchDict, matchStr);
1549 rec_del (zh->service->records, &rec);
1556 /* record going to be updated */
1557 if (!delkeys.buf_used)
1559 logf (LOG_LOG, "update %s %s %ld", recordType,
1560 fname, (long) recordOffset);
1561 logf (LOG_WARN, "cannot update file above, storeKeys false");
1565 logf (LOG_LOG, "update %s %s %ld", recordType,
1566 fname, (long) recordOffset);
1567 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1571 /* update file type */
1572 xfree (rec->info[recInfo_fileType]);
1573 rec->info[recInfo_fileType] =
1574 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1576 /* update filename */
1577 xfree (rec->info[recInfo_filename]);
1578 rec->info[recInfo_filename] =
1579 rec_strdup (fname, &rec->size[recInfo_filename]);
1581 /* update delete keys */
1582 xfree (rec->info[recInfo_delKeys]);
1583 if (zh->keys.buf_used > 0 && store_keys == 1)
1585 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1586 rec->info[recInfo_delKeys] = zh->keys.buf;
1587 zh->keys.buf = NULL;
1588 zh->keys.buf_max = 0;
1592 rec->info[recInfo_delKeys] = NULL;
1593 rec->size[recInfo_delKeys] = 0;
1596 /* save file size of original record */
1597 zebraExplain_recordBytesIncrement (zh->service->zei,
1598 - recordAttr->recordSize);
1600 recordAttr->recordSize = fi->file_moffset - recordOffset;
1601 if (!recordAttr->recordSize)
1602 recordAttr->recordSize = fi->file_max - recordOffset;
1604 recordAttr->recordSize = buf_size;
1606 zebraExplain_recordBytesIncrement (zh->service->zei,
1607 recordAttr->recordSize);
1609 /* set run-number for this record */
1610 recordAttr->runNumber =
1611 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1613 /* update store data */
1614 xfree (rec->info[recInfo_storeData]);
1615 if (store_data == 1)
1617 rec->size[recInfo_storeData] = recordAttr->recordSize;
1618 rec->info[recInfo_storeData] = (char *)
1619 xmalloc (recordAttr->recordSize);
1621 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1623 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1625 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1626 (long) recordOffset, fname);
1629 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1630 < recordAttr->recordSize)
1632 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1633 recordAttr->recordSize, fname);
1640 rec->info[recInfo_storeData] = NULL;
1641 rec->size[recInfo_storeData] = 0;
1643 /* update database name */
1644 xfree (rec->info[recInfo_databaseName]);
1645 rec->info[recInfo_databaseName] =
1646 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1649 recordAttr->recordOffset = recordOffset;
1651 /* commit this record */
1652 rec_put (zh->service->records, &rec);