2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.33 2000-05-18 12:01:36 adam
7 * System call times(2) used again. More 64-bit fixes.
9 * Revision 1.32 2000/04/19 14:35:59 adam
10 * WIN32 update (this version is known not to work on Windows).
12 * Revision 1.31 2000/04/05 10:07:02 adam
13 * Minor zebra compile fix.
15 * Revision 1.30 2000/04/05 09:49:35 adam
16 * On Unix, zebra/z'mbol uses automake.
18 * Revision 1.29 2000/03/20 19:08:36 adam
19 * Added remote record import using Z39.50 extended services and Segment
22 * Revision 1.28 2000/03/15 15:00:30 adam
23 * First work on threaded version.
25 * Revision 1.27 2000/02/24 12:31:17 adam
26 * Added zebra_string_norm.
28 * Revision 1.26 1999/11/30 13:48:03 adam
29 * Improved installation. Updated for inclusion of YAZ header files.
31 * Revision 1.25 1999/11/04 15:00:45 adam
32 * Implemented delete result set(s).
34 * Revision 1.24 1999/10/14 14:33:50 adam
35 * Added truncation 5=106.
37 * Revision 1.23 1999/09/07 11:36:32 adam
40 * Revision 1.22 1999/08/02 10:13:47 adam
41 * Fixed bug regarding zebra_hits.
43 * Revision 1.21 1999/07/14 10:59:26 adam
44 * Changed functions isc_getmethod, isams_getmethod.
45 * Improved fatal error handling (such as missing EXPLAIN schema).
47 * Revision 1.20 1999/07/06 12:28:04 adam
48 * Updated record index structure. Format includes version ID. Compression
49 * algorithm ID is stored for each record block.
51 * Revision 1.19 1999/05/26 07:49:13 adam
54 * Revision 1.18 1999/05/15 14:36:38 adam
55 * Updated dictionary. Implemented "compression" of dictionary.
57 * Revision 1.17 1999/05/12 13:08:06 adam
58 * First version of ISAMS.
60 * Revision 1.16 1999/02/19 10:38:30 adam
61 * Implemented chdir-setting.
63 * Revision 1.15 1999/02/17 12:18:12 adam
64 * Fixed zebra_close so that a NULL pointer is ignored.
66 * Revision 1.14 1999/02/02 14:51:11 adam
67 * Updated WIN32 code specific sections. Changed header.
69 * Revision 1.13 1998/12/16 12:23:30 adam
70 * Added facility for database name mapping using resource mapdb.
72 * Revision 1.12 1998/11/16 10:18:10 adam
73 * Better error reporting for result sets.
75 * Revision 1.11 1998/10/16 08:14:34 adam
76 * Updated record control system.
78 * Revision 1.10 1998/09/22 10:03:42 adam
79 * Changed result sets to be persistent in the sense that they can
80 * be re-searched if needed.
81 * Fixed memory leak in rsm_or.
83 * Revision 1.9 1998/09/02 13:53:17 adam
84 * Extra parameter decode added to search routines to implement
87 * Revision 1.8 1998/08/24 17:29:23 adam
90 * Revision 1.7 1998/06/24 12:16:13 adam
91 * Support for relations on text operands. Open range support in
92 * DFA module (i.e. [-j], [g-]).
94 * Revision 1.6 1998/06/22 11:36:47 adam
95 * Added authentication check facility to zebra.
97 * Revision 1.5 1998/06/13 00:14:08 adam
100 * Revision 1.4 1998/06/12 12:22:12 adam
103 * Revision 1.3 1998/05/27 16:57:44 adam
104 * Zebra returns surrogate diagnostic for single records when
107 * Revision 1.2 1998/05/20 10:12:19 adam
108 * Implemented automatic EXPLAIN database maintenance.
109 * Modified Zebra to work with ASN.1 compiled version of YAZ.
111 * Revision 1.1 1998/03/05 08:45:13 adam
112 * New result set model and modular ranking system. Moved towards
113 * descent server API. System information stored as "SGML" records.
127 #include <yaz/diagbib1.h>
131 static void zebra_chdir (ZebraService zh)
133 const char *dir = res_get (zh->res, "chdir");
136 logf (LOG_DEBUG, "chdir %s", dir);
144 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
145 const char *buf, size_t buf_size,
146 const char *databaseName, int delete_flag,
147 int test_mode, int *sysno,
148 int store_keys, int store_data,
149 const char *match_criteria);
151 static int explain_extract (void *handle, Record rec, data1_node *n);
152 static void extract_index (ZebraHandle zh);
154 static void zebra_register_unlock (ZebraHandle zh);
156 static int zebra_register_lock (ZebraHandle zh)
160 if (!zh->service->active)
171 static void zebra_register_unlock (ZebraHandle zh)
175 logf (LOG_LOG, "user/system: %ld/%ld",
176 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
177 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
182 ZebraHandle zebra_open (ZebraService zs)
190 zh = (ZebraHandle) xmalloc (sizeof(*zh));
199 zh->admin_databaseName = 0;
201 zebra_mutex_cond_lock (&zs->session_lock);
203 zh->next = zs->sessions;
206 zebra_mutex_cond_unlock (&zs->session_lock);
210 static int zebra_register_activate (ZebraService zh);
211 static int zebra_register_deactivate (ZebraService zh);
213 ZebraService zebra_start (const char *configName)
215 ZebraService zh = xmalloc (sizeof(*zh));
217 yaz_log (LOG_LOG, "zebra_start %s", configName);
219 zh->configName = xstrdup(configName);
223 zebra_mutex_cond_init (&zh->session_lock);
224 zebra_register_activate (zh);
228 static int zebra_register_activate (ZebraService zh)
232 yaz_log (LOG_LOG, "zebra_register_activate");
233 if (!(zh->res = res_open (zh->configName)))
235 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
239 zh->dh = data1_create ();
242 zh->bfs = bfs_create (res_get (zh->res, "register"));
245 data1_destroy(zh->dh);
248 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
249 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
250 zh->registerState = -1; /* trigger open of registers! */
251 zh->registerChange = 0;
252 zh->recTypes = recTypes_init (zh->dh);
253 recTypes_default_handlers (zh->recTypes);
256 zh->zebra_maps = zebra_maps_open (zh->res);
257 zh->rank_classes = NULL;
270 zebraRankInstall (zh, rank1_class);
272 if (!res_get (zh->res, "passwd"))
273 zh->passwd_db = NULL;
276 zh->passwd_db = passwd_db_open ();
278 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
280 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
283 if (!(zh->records = rec_open (zh->bfs, 1, 0)))
285 logf (LOG_WARN, "rec_open");
288 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, 1, 0)))
290 logf (LOG_WARN, "dict_open");
293 if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0)))
295 logf (LOG_WARN, "sortIdx_open");
298 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
300 struct ISAMS_M_s isams_m;
301 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, 1,
302 key_isams_m(zh->res, &isams_m))))
304 logf (LOG_WARN, "isams_open");
309 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
311 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, 1,
312 sizeof (struct it_key), zh->res)))
314 logf (LOG_WARN, "is_open");
318 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
320 struct ISAMC_M_s isamc_m;
321 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
322 1, key_isamc_m(zh->res, &isamc_m))))
324 logf (LOG_WARN, "isc_open");
328 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
330 struct ISAMD_M_s isamd_m;
332 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
333 1, key_isamd_m(zh->res, &isamd_m))))
335 logf (LOG_WARN, "isamd_open");
340 zh->zei = zebraExplain_open (zh->records, zh->dh,
341 zh->res, 1, 0 /* rGroup */,
345 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
349 yaz_log (LOG_LOG, "zebra_register_activate ok");
353 void zebra_admin_shutdown (ZebraHandle zh)
355 zebraExplain_flush (zh->service->zei, 1, zh);
358 zebra_mutex_cond_lock (&zh->service->session_lock);
359 zh->service->stop_flag = 1;
360 if (!zh->service->sessions)
361 zebra_register_deactivate(zh->service);
362 zebra_mutex_cond_unlock (&zh->service->session_lock);
365 void zebra_admin_start (ZebraHandle zh)
367 ZebraService zs = zh->service;
369 zebra_mutex_cond_lock (&zs->session_lock);
371 zebra_register_activate(zs);
372 zebra_mutex_cond_unlock (&zs->session_lock);
375 static int zebra_register_deactivate (ZebraService zh)
380 yaz_log(LOG_LOG, "zebra_register_deactivate");
384 zebraExplain_close (zh->zei, 1);
385 dict_close (zh->dict);
386 sortIdx_close (zh->sortIdx);
388 isams_close (zh->isams);
393 isc_close (zh->isamc);
395 isamd_close (zh->isamd);
397 rec_close (&zh->records);
399 recTypes_destroy (zh->recTypes);
400 zebra_maps_close (zh->zebra_maps);
401 zebraRankDestroy (zh);
402 bfs_destroy (zh->bfs);
403 data1_destroy (zh->dh);
406 passwd_db_close (zh->passwd_db);
412 void zebra_stop(ZebraService zh)
416 yaz_log (LOG_LOG, "zebra_stop");
418 assert (!zh->sessions);
420 zebra_mutex_cond_destroy (&zh->session_lock);
422 zebra_register_deactivate(zh);
423 xfree (zh->configName);
427 void zebra_close (ZebraHandle zh)
429 ZebraService zs = zh->service;
430 struct zebra_session **sp;
433 resultSetDestroy (zh, -1, 0, 0);
440 xfree (zh->admin_databaseName);
441 zebra_mutex_cond_lock (&zs->session_lock);
453 if (!zs->sessions && zs->stop_flag)
454 zebra_register_deactivate(zs);
455 zebra_mutex_cond_unlock (&zs->session_lock);
459 struct map_baseinfo {
465 char **new_basenames;
469 void map_basenames_func (void *vp, const char *name, const char *value)
471 struct map_baseinfo *p = (struct map_baseinfo *) vp;
473 char fromdb[128], todb[8][128];
476 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
477 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
478 todb[5], todb[6], todb[7]);
482 for (i = 0; i<p->num_bases; i++)
483 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
486 for (i = 0; i < no; i++)
488 if (p->new_num_bases == p->new_num_max)
490 p->new_basenames[(p->new_num_bases)++] =
491 nmem_strdup (p->mem, todb[i]);
497 void map_basenames (ZebraHandle zh, ODR stream,
498 int *num_bases, char ***basenames)
500 struct map_baseinfo info;
501 struct map_baseinfo *p = &info;
505 info.num_bases = *num_bases;
506 info.basenames = *basenames;
507 info.new_num_max = 128;
508 info.new_num_bases = 0;
509 info.new_basenames = (char **)
510 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
511 info.mem = stream->mem;
513 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
515 for (i = 0; i<p->num_bases; i++)
516 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
518 p->new_basenames[(p->new_num_bases)++] =
519 nmem_strdup (p->mem, p->basenames[i]);
521 *num_bases = info.new_num_bases;
522 *basenames = info.new_basenames;
523 for (i = 0; i<*num_bases; i++)
524 logf (LOG_LOG, "base %s", (*basenames)[i]);
527 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
528 Z_RPNQuery *query, int num_bases, char **basenames,
532 if (zebra_register_lock (zh))
534 map_basenames (zh, stream, &num_bases, &basenames);
535 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
537 zebra_register_unlock (zh);
540 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
541 const char *setname, Z_RecordComposition *comp,
542 oid_value input_format, int num_recs,
543 ZebraRetrievalRecord *recs)
548 if (zebra_register_lock (zh))
550 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
551 for (i = 0; i<num_recs; i++)
552 pos_array[i] = recs[i].position;
553 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
556 logf (LOG_DEBUG, "zebraPosSetCreate error");
558 zh->errString = nmem_strdup (stream->mem, setname);
562 for (i = 0; i<num_recs; i++)
568 sprintf (num_str, "%d", pos_array[i]);
570 zh->errString = nmem_strdup (stream->mem, num_str);
576 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
577 stream, input_format, comp,
578 &recs[i].format, &recs[i].buf,
581 recs[i].errString = NULL;
584 zebraPosSetDestroy (zh, poset, num_recs);
586 zebra_register_unlock (zh);
590 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
591 oid_value attributeset,
592 int num_bases, char **basenames,
593 int *position, int *num_entries, ZebraScanEntry **entries,
596 if (zebra_register_lock (zh))
602 map_basenames (zh, stream, &num_bases, &basenames);
603 rpn_scan (zh, stream, zapt, attributeset,
604 num_bases, basenames, position,
605 num_entries, entries, is_partial);
606 zebra_register_unlock (zh);
609 void zebra_sort (ZebraHandle zh, ODR stream,
610 int num_input_setnames, const char **input_setnames,
611 const char *output_setname, Z_SortKeySpecList *sort_sequence,
614 if (zebra_register_lock (zh))
616 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
617 output_setname, sort_sequence, sort_status);
618 zebra_register_unlock (zh);
621 int zebra_deleleResultSet(ZebraHandle zh, int function,
622 int num_setnames, char **setnames,
626 if (zebra_register_lock (zh))
627 return Z_DeleteStatus_systemProblemAtTarget;
630 case Z_DeleteRequest_list:
631 resultSetDestroy (zh, num_setnames, setnames, statuses);
633 case Z_DeleteRequest_all:
634 resultSetDestroy (zh, -1, 0, statuses);
637 zebra_register_unlock (zh);
638 status = Z_DeleteStatus_success;
639 for (i = 0; i<num_setnames; i++)
640 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
641 status = statuses[i];
645 int zebra_errCode (ZebraHandle zh)
650 const char *zebra_errString (ZebraHandle zh)
652 return diagbib1_str (zh->errCode);
655 char *zebra_errAdd (ZebraHandle zh)
657 return zh->errString;
660 int zebra_hits (ZebraHandle zh)
665 int zebra_auth (ZebraService zh, const char *user, const char *pass)
667 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
672 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
674 if (zebra_register_lock (zh))
676 xfree (zh->admin_databaseName);
677 zh->admin_databaseName = xstrdup(database);
678 zebra_register_unlock(zh);
681 void zebra_admin_import_end (ZebraHandle zh)
683 zebraExplain_flush (zh->service->zei, 1, zh);
687 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
691 if (zebra_register_lock (zh))
693 for (i = 0; i<segment->num_segmentRecords; i++)
695 Z_NamePlusRecord *npr = segment->segmentRecords[i];
696 const char *databaseName = npr->databaseName;
699 databaseName = zh->admin_databaseName;
700 printf ("--------------%d--------------------\n", i);
701 if (npr->which == Z_NamePlusRecord_intermediateFragment)
703 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
704 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
706 Odr_oct *oct = fragment->u.notExternallyTagged;
707 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
711 extract_rec_in_mem (zh, "grs.sgml",
719 0 /* match criteria */);
723 zebra_register_unlock(zh);
726 void zebra_admin_create (ZebraHandle zh, const char *database)
728 ZebraService zs = zh->service;
729 if (zebra_register_lock(zh))
734 /* announce database */
735 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
738 zh->errString = "Database already exist";
740 zebra_register_unlock(zh);
743 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
744 const char *input_str, int input_len,
745 char *output_str, int output_len)
748 if (!zh->service->zebra_maps)
750 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
751 input_str, input_len);
754 if (wrbuf_len(wrbuf) >= output_len)
756 if (wrbuf_len(wrbuf))
757 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
758 output_str[wrbuf_len(wrbuf)] = '\0';
759 return wrbuf_len(wrbuf);
762 static void extract_init (struct recExtractCtrl *p, RecWord *w)
764 w->zebra_maps = p->zebra_maps;
765 w->seqnos = p->seqno;
766 w->attrSet = VAL_BIB1;
772 static void extract_add_index_string (RecWord *p, const char *string,
776 unsigned char attrSet;
777 unsigned short attrUse;
780 int *pseqno = &p->seqnos[p->reg_type];
781 ZebraHandle zh = p->extractCtrl->handle;
782 struct recKeys *keys = &zh->keys;
784 if (keys->buf_used+1024 > keys->buf_max)
788 b = (char *) xmalloc (keys->buf_max += 128000);
789 if (keys->buf_used > 0)
790 memcpy (b, keys->buf, keys->buf_used);
794 dst = keys->buf + keys->buf_used;
796 attrSet = p->attrSet;
797 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
800 keys->prevAttrSet = attrSet;
801 attrUse = p->attrUse;
802 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
805 keys->prevAttrUse = attrUse;
807 diff = 1 + *pseqno - keys->prevSeqNo;
808 if (diff >= 1 && diff <= 15)
813 keys->prevSeqNo = *pseqno;
819 memcpy (dst, &attrSet, sizeof(attrSet));
820 dst += sizeof(attrSet);
824 memcpy (dst, &attrUse, sizeof(attrUse));
825 dst += sizeof(attrUse);
827 *dst++ = p->reg_type;
828 memcpy (dst, string, length);
834 memcpy (dst, pseqno, sizeof(*pseqno));
835 dst += sizeof(*pseqno);
837 keys->buf_used = dst - keys->buf;
842 static void extract_add_sort_string (RecWord *p, const char *string,
846 ZebraHandle zh = p->extractCtrl->handle;
847 struct sortKey *sortKeys = zh->sortKeys;
849 for (sk = sortKeys; sk; sk = sk->next)
850 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
853 sk = (struct sortKey *) xmalloc (sizeof(*sk));
857 sk->string = (char *) xmalloc (length);
859 memcpy (sk->string, string, length);
861 sk->attrSet = p->attrSet;
862 sk->attrUse = p->attrUse;
865 static void extract_add_string (RecWord *p, const char *string, int length)
868 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
869 extract_add_sort_string (p, string, length);
871 extract_add_index_string (p, string, length);
874 static void extract_add_incomplete_field (RecWord *p)
876 const char *b = p->string;
877 int remain = p->length;
878 const char **map = 0;
881 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
885 char buf[IT_MAX_WORD+1];
889 while (map && *map && **map == *CHR_SPACE)
891 remain = p->length - (b - p->string);
893 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
900 while (map && *map && **map != *CHR_SPACE)
902 const char *cp = *map;
904 while (i < IT_MAX_WORD && *cp)
906 remain = p->length - (b - p->string);
908 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
914 extract_add_string (p, buf, i);
916 (p->seqnos[p->reg_type])++; /* to separate this from next one */
919 static void extract_add_complete_field (RecWord *p)
921 const char *b = p->string;
922 char buf[IT_MAX_WORD+1];
923 const char **map = 0;
924 int i = 0, remain = p->length;
927 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
929 while (remain > 0 && i < IT_MAX_WORD)
931 while (map && *map && **map == *CHR_SPACE)
933 remain = p->length - (b - p->string);
935 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
942 if (i && i < IT_MAX_WORD)
943 buf[i++] = *CHR_SPACE;
944 while (map && *map && **map != *CHR_SPACE)
946 const char *cp = *map;
948 if (i >= IT_MAX_WORD)
950 while (i < IT_MAX_WORD && *cp)
952 remain = p->length - (b - p->string);
954 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
962 extract_add_string (p, buf, i);
965 static void extract_token_add (RecWord *p)
968 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
969 p->string, p->length)))
971 p->string = wrbuf_buf(wrbuf);
972 p->length = wrbuf_len(wrbuf);
974 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
975 extract_add_complete_field (p);
977 extract_add_incomplete_field(p);
980 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
982 ZebraHandle zh = (ZebraHandle) (p->handle);
983 zebraExplain_addSchema (zh->service->zei, oid);
986 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
987 int cmd, struct sortKey **skp)
989 struct sortKey *sk = *skp;
990 SortIdx sortIdx = zh->service->sortIdx;
992 sortIdx_sysno (sortIdx, sysno);
995 struct sortKey *sk_next = sk->next;
996 sortIdx_type (sortIdx, sk->attrUse);
997 sortIdx_add (sortIdx, sk->string, sk->length);
1005 struct encode_info {
1012 void encode_key_init (struct encode_info *i)
1019 char *encode_key_int (int d, char *bp)
1023 else if (d <= 16383)
1025 *bp++ = 64 + (d>>8);
1028 else if (d <= 4194303)
1030 *bp++ = 128 + (d>>16);
1031 *bp++ = (d>>8) & 255;
1036 *bp++ = 192 + (d>>24);
1037 *bp++ = (d>>16) & 255;
1038 *bp++ = (d>>8) & 255;
1044 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1049 while ((*bp++ = *k++))
1051 memcpy (&key, k+1, sizeof(struct it_key));
1052 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1053 if (i->sysno != key.sysno)
1055 i->sysno = key.sysno;
1058 else if (!i->seqno && !key.seqno && i->cmd == *k)
1060 bp = encode_key_int (key.seqno - i->seqno, bp);
1061 i->seqno = key.seqno;
1063 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1065 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1070 static void extract_flushWriteKeys (ZebraHandle zh)
1073 char out_fname[200];
1075 struct encode_info encode_info;
1076 int ptr_i = zh->ptr_i;
1080 if (!zh->key_buf || ptr_i <= 0)
1083 (zh->key_file_no)++;
1084 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1086 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1088 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1090 if (!(outf = fopen (out_fname, "wb")))
1092 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1095 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1096 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1098 encode_key_init (&encode_info);
1099 encode_key_write (cp, &encode_info, outf);
1103 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1104 if (strcmp (cp, prevcp))
1106 encode_key_init (&encode_info);
1107 encode_key_write (cp, &encode_info, outf);
1111 encode_key_write (cp + strlen(cp), &encode_info, outf);
1114 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1115 extract_get_fname_tmp (out_fname, key_file_no);
1117 if (!(outf = fopen (out_fname, "wb")))
1119 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1122 logf (LOG_LOG, "writing section %d", key_file_no);
1124 prevcp = key_buf[ptr_top-i];
1126 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1128 key_y_len = strlen(prevcp)+1;
1130 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1131 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1133 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1134 sizeof(char*), key_y_compare);
1135 cp = key_buf[ptr_top-ptr_i];
1137 encode_key_init (&encode_info);
1138 encode_key_write (cp, &encode_info, outf);
1141 cp = key_buf[ptr_top-ptr_i];
1142 encode_key_write (cp+key_y_len, &encode_info, outf);
1146 prevcp = key_buf[ptr_top-ptr_i];
1151 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1154 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1156 zh->key_buf_used = 0;
1159 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1160 int cmd, struct recKeys *reckeys)
1162 unsigned char attrSet = (unsigned char) -1;
1163 unsigned short attrUse = (unsigned short) -1;
1166 ZebraExplainInfo zei = zh->service->zei;
1170 int mem = 8*1024*1024;
1171 zh->key_buf = (char**) xmalloc (mem);
1172 zh->ptr_top = mem/sizeof(char*);
1174 zh->key_buf_used = 0;
1175 zh->key_file_no = 0;
1177 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1178 while (off < reckeys->buf_used)
1180 const char *src = reckeys->buf + off;
1188 memcpy (&attrSet, src, sizeof(attrSet));
1189 src += sizeof(attrSet);
1193 memcpy (&attrUse, src, sizeof(attrUse));
1194 src += sizeof(attrUse);
1196 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1197 extract_flushWriteKeys (zh);
1199 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1200 (char*)zh->key_buf + zh->key_buf_used;
1202 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1204 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1207 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1210 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1212 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1213 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1216 seqno += ((lead>>2) & 15)-1;
1219 memcpy (&seqno, src, sizeof(seqno));
1220 src += sizeof(seqno);
1224 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1225 (zh->key_buf_used) += sizeof(key);
1226 off = src - reckeys->buf;
1228 assert (off == reckeys->buf_used);
1231 static void extract_index (ZebraHandle zh)
1233 extract_flushWriteKeys (zh);
1234 zebra_index_merge (zh);
1237 static int explain_extract (void *handle, Record rec, data1_node *n)
1239 ZebraHandle zh = (ZebraHandle) handle;
1240 struct recExtractCtrl extractCtrl;
1243 if (zebraExplain_curDatabase (zh->service->zei,
1244 rec->info[recInfo_databaseName]))
1247 if (zebraExplain_newDatabase (zh->service->zei,
1248 rec->info[recInfo_databaseName], 0))
1252 zh->keys.buf_used = 0;
1253 zh->keys.prevAttrUse = -1;
1254 zh->keys.prevAttrSet = -1;
1255 zh->keys.prevSeqNo = 0;
1258 extractCtrl.init = extract_init;
1259 extractCtrl.tokenAdd = extract_token_add;
1260 extractCtrl.schemaAdd = extract_schema_add;
1261 extractCtrl.dh = zh->service->dh;
1262 for (i = 0; i<256; i++)
1263 extractCtrl.seqno[i] = 0;
1264 extractCtrl.zebra_maps = zh->service->zebra_maps;
1265 extractCtrl.flagShowRecords = 0;
1266 extractCtrl.handle = handle;
1268 grs_extract_tree(&extractCtrl, n);
1270 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1272 if (rec->size[recInfo_delKeys])
1274 struct recKeys delkeys;
1275 struct sortKey *sortKeys = 0;
1277 delkeys.buf_used = rec->size[recInfo_delKeys];
1278 delkeys.buf = rec->info[recInfo_delKeys];
1279 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1280 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1282 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1283 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1285 xfree (rec->info[recInfo_delKeys]);
1286 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1287 rec->info[recInfo_delKeys] = zh->keys.buf;
1288 zh->keys.buf = NULL;
1289 zh->keys.buf_max = 0;
1293 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1294 const char *buf, size_t buf_size,
1295 const char *databaseName, int delete_flag,
1296 int test_mode, int *sysno,
1297 int store_keys, int store_data,
1298 const char *match_criteria)
1300 RecordAttr *recordAttr;
1301 struct recExtractCtrl extractCtrl;
1306 const char *fname = "<no file>";
1308 long recordOffset = 0;
1309 struct zebra_fetch_control fc;
1312 fc.record_int_buf = buf;
1313 fc.record_int_len = buf_size;
1314 fc.record_int_pos = 0;
1316 fc.record_offset = 0;
1318 extractCtrl.offset = 0;
1319 extractCtrl.readf = zebra_record_int_read;
1320 extractCtrl.seekf = zebra_record_int_seek;
1321 extractCtrl.tellf = zebra_record_int_tell;
1322 extractCtrl.endf = zebra_record_int_end;
1323 extractCtrl.fh = &fc;
1325 /* announce database */
1326 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1328 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1332 recType_byName (zh->service->recTypes, recordType, subType,
1335 logf (LOG_WARN, "No such record type: %s", recordType);
1339 zh->keys.buf_used = 0;
1340 zh->keys.prevAttrUse = -1;
1341 zh->keys.prevAttrSet = -1;
1342 zh->keys.prevSeqNo = 0;
1345 extractCtrl.subType = subType;
1346 extractCtrl.init = extract_init;
1347 extractCtrl.tokenAdd = extract_token_add;
1348 extractCtrl.schemaAdd = extract_schema_add;
1349 extractCtrl.dh = zh->service->dh;
1350 extractCtrl.handle = zh;
1351 extractCtrl.zebra_maps = zh->service->zebra_maps;
1352 extractCtrl.flagShowRecords = 0;
1353 for (i = 0; i<256; i++)
1355 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1356 extractCtrl.seqno[i] = 1;
1358 extractCtrl.seqno[i] = 0;
1361 r = (*recType->extract)(clientData, &extractCtrl);
1363 if (r == RECCTRL_EXTRACT_EOF)
1365 else if (r == RECCTRL_EXTRACT_ERROR)
1367 /* error occured during extraction ... */
1369 yaz_log (LOG_WARN, "extract error");
1371 if (rGroup->flagRw &&
1372 records_processed < rGroup->fileVerboseLimit)
1374 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1375 fname, (long) recordOffset);
1380 if (zh->keys.buf_used == 0)
1382 /* the extraction process returned no information - the record
1383 is probably empty - unless flagShowRecords is in use */
1386 logf (LOG_WARN, "No keys generated for record");
1387 logf (LOG_WARN, " The file is probably empty");
1390 /* match criteria */
1397 logf (LOG_LOG, "delete %s %s %ld", recordType,
1398 fname, (long) recordOffset);
1399 logf (LOG_WARN, "cannot delete record above (seems new)");
1402 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1403 (long) recordOffset);
1404 rec = rec_new (zh->service->records);
1406 *sysno = rec->sysno;
1408 recordAttr = rec_init_attr (zh->service->zei, rec);
1413 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1416 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1417 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1421 /* record already exists */
1422 struct recKeys delkeys;
1424 rec = rec_get (zh->service->records, *sysno);
1427 recordAttr = rec_init_attr (zh->service->zei, rec);
1429 if (recordAttr->runNumber ==
1430 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1432 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1433 fname, (long) recordOffset);
1437 delkeys.buf_used = rec->size[recInfo_delKeys];
1438 delkeys.buf = rec->info[recInfo_delKeys];
1439 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1440 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1443 /* record going to be deleted */
1444 if (!delkeys.buf_used)
1446 logf (LOG_LOG, "delete %s %s %ld", recordType,
1447 fname, (long) recordOffset);
1448 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1452 logf (LOG_LOG, "delete %s %s %ld", recordType,
1453 fname, (long) recordOffset);
1456 dict_delete (matchDict, matchStr);
1458 rec_del (zh->service->records, &rec);
1465 /* record going to be updated */
1466 if (!delkeys.buf_used)
1468 logf (LOG_LOG, "update %s %s %ld", recordType,
1469 fname, (long) recordOffset);
1470 logf (LOG_WARN, "cannot update file above, storeKeys false");
1474 logf (LOG_LOG, "update %s %s %ld", recordType,
1475 fname, (long) recordOffset);
1476 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1480 /* update file type */
1481 xfree (rec->info[recInfo_fileType]);
1482 rec->info[recInfo_fileType] =
1483 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1485 /* update filename */
1486 xfree (rec->info[recInfo_filename]);
1487 rec->info[recInfo_filename] =
1488 rec_strdup (fname, &rec->size[recInfo_filename]);
1490 /* update delete keys */
1491 xfree (rec->info[recInfo_delKeys]);
1492 if (zh->keys.buf_used > 0 && store_keys == 1)
1494 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1495 rec->info[recInfo_delKeys] = zh->keys.buf;
1496 zh->keys.buf = NULL;
1497 zh->keys.buf_max = 0;
1501 rec->info[recInfo_delKeys] = NULL;
1502 rec->size[recInfo_delKeys] = 0;
1505 /* save file size of original record */
1506 zebraExplain_recordBytesIncrement (zh->service->zei,
1507 - recordAttr->recordSize);
1509 recordAttr->recordSize = fi->file_moffset - recordOffset;
1510 if (!recordAttr->recordSize)
1511 recordAttr->recordSize = fi->file_max - recordOffset;
1513 recordAttr->recordSize = buf_size;
1515 zebraExplain_recordBytesIncrement (zh->service->zei,
1516 recordAttr->recordSize);
1518 /* set run-number for this record */
1519 recordAttr->runNumber =
1520 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1522 /* update store data */
1523 xfree (rec->info[recInfo_storeData]);
1524 if (store_data == 1)
1526 rec->size[recInfo_storeData] = recordAttr->recordSize;
1527 rec->info[recInfo_storeData] = (char *)
1528 xmalloc (recordAttr->recordSize);
1530 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1532 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1534 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1535 (long) recordOffset, fname);
1538 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1539 < recordAttr->recordSize)
1541 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1542 recordAttr->recordSize, fname);
1549 rec->info[recInfo_storeData] = NULL;
1550 rec->size[recInfo_storeData] = 0;
1552 /* update database name */
1553 xfree (rec->info[recInfo_databaseName]);
1554 rec->info[recInfo_databaseName] =
1555 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1558 recordAttr->recordOffset = recordOffset;
1560 /* commit this record */
1561 rec_put (zh->service->records, &rec);