2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.38 2000-11-08 13:46:58 adam
7 * Fixed scan: server could break if bad attribute/database was selected.
8 * Work on remote update.
10 * Revision 1.37 2000/10/17 12:37:09 adam
11 * Fixed notification of live-updates. Fixed minor problem with mf_init
12 * where it didn't handle shadow area file names correctly.
14 * Revision 1.36 2000/09/06 08:59:36 adam
15 * Using read-only (for now) for server.
17 * Revision 1.35 2000/07/07 12:49:20 adam
18 * Optimized resultSetInsert{Rank,Sort}.
20 * Revision 1.34 2000/06/09 13:56:38 ian
21 * Added some logging on Authentication and searches.
23 * Revision 1.33 2000/05/18 12:01:36 adam
24 * System call times(2) used again. More 64-bit fixes.
26 * Revision 1.32 2000/04/19 14:35:59 adam
27 * WIN32 update (this version is known not to work on Windows).
29 * Revision 1.31 2000/04/05 10:07:02 adam
30 * Minor zebra compile fix.
32 * Revision 1.30 2000/04/05 09:49:35 adam
33 * On Unix, zebra/z'mbol uses automake.
35 * Revision 1.29 2000/03/20 19:08:36 adam
36 * Added remote record import using Z39.50 extended services and Segment
39 * Revision 1.28 2000/03/15 15:00:30 adam
40 * First work on threaded version.
42 * Revision 1.27 2000/02/24 12:31:17 adam
43 * Added zebra_string_norm.
45 * Revision 1.26 1999/11/30 13:48:03 adam
46 * Improved installation. Updated for inclusion of YAZ header files.
48 * Revision 1.25 1999/11/04 15:00:45 adam
49 * Implemented delete result set(s).
51 * Revision 1.24 1999/10/14 14:33:50 adam
52 * Added truncation 5=106.
54 * Revision 1.23 1999/09/07 11:36:32 adam
57 * Revision 1.22 1999/08/02 10:13:47 adam
58 * Fixed bug regarding zebra_hits.
60 * Revision 1.21 1999/07/14 10:59:26 adam
61 * Changed functions isc_getmethod, isams_getmethod.
62 * Improved fatal error handling (such as missing EXPLAIN schema).
64 * Revision 1.20 1999/07/06 12:28:04 adam
65 * Updated record index structure. Format includes version ID. Compression
66 * algorithm ID is stored for each record block.
68 * Revision 1.19 1999/05/26 07:49:13 adam
71 * Revision 1.18 1999/05/15 14:36:38 adam
72 * Updated dictionary. Implemented "compression" of dictionary.
74 * Revision 1.17 1999/05/12 13:08:06 adam
75 * First version of ISAMS.
77 * Revision 1.16 1999/02/19 10:38:30 adam
78 * Implemented chdir-setting.
80 * Revision 1.15 1999/02/17 12:18:12 adam
81 * Fixed zebra_close so that a NULL pointer is ignored.
83 * Revision 1.14 1999/02/02 14:51:11 adam
84 * Updated WIN32 code specific sections. Changed header.
86 * Revision 1.13 1998/12/16 12:23:30 adam
87 * Added facility for database name mapping using resource mapdb.
89 * Revision 1.12 1998/11/16 10:18:10 adam
90 * Better error reporting for result sets.
92 * Revision 1.11 1998/10/16 08:14:34 adam
93 * Updated record control system.
95 * Revision 1.10 1998/09/22 10:03:42 adam
96 * Changed result sets to be persistent in the sense that they can
97 * be re-searched if needed.
98 * Fixed memory leak in rsm_or.
100 * Revision 1.9 1998/09/02 13:53:17 adam
101 * Extra parameter decode added to search routines to implement
102 * persistent queries.
104 * Revision 1.8 1998/08/24 17:29:23 adam
107 * Revision 1.7 1998/06/24 12:16:13 adam
108 * Support for relations on text operands. Open range support in
109 * DFA module (i.e. [-j], [g-]).
111 * Revision 1.6 1998/06/22 11:36:47 adam
112 * Added authentication check facility to zebra.
114 * Revision 1.5 1998/06/13 00:14:08 adam
117 * Revision 1.4 1998/06/12 12:22:12 adam
120 * Revision 1.3 1998/05/27 16:57:44 adam
121 * Zebra returns surrogate diagnostic for single records when
124 * Revision 1.2 1998/05/20 10:12:19 adam
125 * Implemented automatic EXPLAIN database maintenance.
126 * Modified Zebra to work with ASN.1 compiled version of YAZ.
128 * Revision 1.1 1998/03/05 08:45:13 adam
129 * New result set model and modular ranking system. Moved towards
130 * descent server API. System information stored as "SGML" records.
144 #include <yaz/diagbib1.h>
148 static void zebra_chdir (ZebraService zh)
150 const char *dir = res_get (zh->res, "chdir");
153 logf (LOG_DEBUG, "chdir %s", dir);
161 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
162 const char *buf, size_t buf_size,
163 const char *databaseName, int delete_flag,
164 int test_mode, int *sysno,
165 int store_keys, int store_data,
166 const char *match_criteria);
168 static int explain_extract (void *handle, Record rec, data1_node *n);
169 static void extract_index (ZebraHandle zh);
171 static void zebra_register_unlock (ZebraHandle zh);
173 static int zebra_register_activate (ZebraService zh, int rw);
174 static int zebra_register_deactivate (ZebraService zh);
176 static int zebra_register_lock (ZebraHandle zh, int rw)
182 if (!zh->service->active)
192 state = zebra_server_lock_get_state(zh->service, &lastChange);
194 zebra_server_lock (zh->service, state);
205 logf (LOG_LOG, "Register in read/write mode");
206 else if (zh->service->registerState == state)
208 if (zh->service->registerChange >= lastChange)
210 logf (LOG_LOG, "Register completely updated since last access");
212 else if (zh->service->registerState == -1)
213 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
216 logf (LOG_LOG, "Register has changed state from %d to %d",
217 zh->service->registerState, state);
218 zh->service->registerChange = lastChange;
220 zebra_register_deactivate (zh->service);
222 zh->service->registerState = state;
224 zebra_register_activate (zh->service, rw);
229 static void zebra_register_unlock (ZebraHandle zh)
231 if (zh->service->registerState != -1)
232 zebra_server_unlock (zh->service, zh->service->registerState);
235 logf (LOG_LOG, "user/system: %ld/%ld",
236 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
237 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
242 ZebraHandle zebra_open (ZebraService zs)
250 zh = (ZebraHandle) xmalloc (sizeof(*zh));
259 zh->admin_databaseName = 0;
261 zebra_mutex_cond_lock (&zs->session_lock);
263 zh->next = zs->sessions;
266 zebra_mutex_cond_unlock (&zs->session_lock);
272 ZebraService zebra_start (const char *configName)
274 ZebraService zh = xmalloc (sizeof(*zh));
276 yaz_log (LOG_LOG, "zebra_start %s", configName);
278 zh->configName = xstrdup(configName);
283 zh->registerState = -1;
284 zh->registerChange = 0;
286 if (!(zh->res = res_open (zh->configName)))
288 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
292 zebra_server_lock_init (zh);
293 zebra_mutex_cond_init (&zh->session_lock);
297 static int zebra_register_activate (ZebraService zh, int rw)
301 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
302 zh->registerState ? "yes" : "no");
304 zh->dh = data1_create ();
307 zh->bfs = bfs_create (res_get (zh->res, "register"));
310 data1_destroy(zh->dh);
313 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
314 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
315 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
316 zh->recTypes = recTypes_init (zh->dh);
317 recTypes_default_handlers (zh->recTypes);
320 zh->zebra_maps = zebra_maps_open (zh->res);
321 zh->rank_classes = NULL;
334 zebraRankInstall (zh, rank1_class);
336 if (!res_get (zh->res, "passwd"))
337 zh->passwd_db = NULL;
340 zh->passwd_db = passwd_db_open ();
342 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
344 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
347 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
349 logf (LOG_WARN, "rec_open");
352 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
354 logf (LOG_WARN, "dict_open");
357 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
359 logf (LOG_WARN, "sortIdx_open");
362 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
364 struct ISAMS_M_s isams_m;
365 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
366 key_isams_m(zh->res, &isams_m))))
368 logf (LOG_WARN, "isams_open");
373 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
375 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
376 sizeof (struct it_key), zh->res)))
378 logf (LOG_WARN, "is_open");
382 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
384 struct ISAMC_M_s isamc_m;
385 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
386 rw, key_isamc_m(zh->res, &isamc_m))))
388 logf (LOG_WARN, "isc_open");
392 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
394 struct ISAMD_M_s isamd_m;
396 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
397 rw, key_isamd_m(zh->res, &isamd_m))))
399 logf (LOG_WARN, "isamd_open");
404 zh->zei = zebraExplain_open (zh->records, zh->dh,
405 zh->res, rw, 0 /* rGroup */,
409 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
413 yaz_log (LOG_LOG, "zebra_register_activate ok");
417 void zebra_admin_shutdown (ZebraHandle zh)
419 zebra_mutex_cond_lock (&zh->service->session_lock);
420 zh->service->stop_flag = 1;
421 if (!zh->service->sessions)
422 zebra_register_deactivate(zh->service);
423 zh->service->active = 0;
424 zebra_mutex_cond_unlock (&zh->service->session_lock);
427 void zebra_admin_start (ZebraHandle zh)
429 ZebraService zs = zh->service;
431 zebra_mutex_cond_lock (&zs->session_lock);
433 zh->service->active = 1;
434 zebra_mutex_cond_unlock (&zs->session_lock);
437 static int zebra_register_deactivate (ZebraService zh)
442 yaz_log(LOG_LOG, "zebra_register_deactivate");
446 zebraExplain_close (zh->zei, 0);
447 dict_close (zh->dict);
448 sortIdx_close (zh->sortIdx);
450 isams_close (zh->isams);
455 isc_close (zh->isamc);
457 isamd_close (zh->isamd);
459 rec_close (&zh->records);
461 recTypes_destroy (zh->recTypes);
462 zebra_maps_close (zh->zebra_maps);
463 zebraRankDestroy (zh);
464 bfs_destroy (zh->bfs);
465 data1_destroy (zh->dh);
468 passwd_db_close (zh->passwd_db);
473 void zebra_stop(ZebraService zh)
477 yaz_log (LOG_LOG, "zebra_stop");
479 assert (!zh->sessions);
481 zebra_mutex_cond_destroy (&zh->session_lock);
483 zebra_register_deactivate(zh);
485 xfree (zh->configName);
489 void zebra_close (ZebraHandle zh)
491 ZebraService zs = zh->service;
492 struct zebra_session **sp;
495 resultSetDestroy (zh, -1, 0, 0);
502 xfree (zh->admin_databaseName);
503 zebra_mutex_cond_lock (&zs->session_lock);
515 if (!zs->sessions && zs->stop_flag)
516 zebra_register_deactivate(zs);
517 zebra_mutex_cond_unlock (&zs->session_lock);
521 struct map_baseinfo {
527 char **new_basenames;
531 void map_basenames_func (void *vp, const char *name, const char *value)
533 struct map_baseinfo *p = (struct map_baseinfo *) vp;
535 char fromdb[128], todb[8][128];
538 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
539 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
540 todb[5], todb[6], todb[7]);
544 for (i = 0; i<p->num_bases; i++)
545 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
548 for (i = 0; i < no; i++)
550 if (p->new_num_bases == p->new_num_max)
552 p->new_basenames[(p->new_num_bases)++] =
553 nmem_strdup (p->mem, todb[i]);
559 void map_basenames (ZebraHandle zh, ODR stream,
560 int *num_bases, char ***basenames)
562 struct map_baseinfo info;
563 struct map_baseinfo *p = &info;
567 info.num_bases = *num_bases;
568 info.basenames = *basenames;
569 info.new_num_max = 128;
570 info.new_num_bases = 0;
571 info.new_basenames = (char **)
572 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
573 info.mem = stream->mem;
575 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
577 for (i = 0; i<p->num_bases; i++)
578 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
580 p->new_basenames[(p->new_num_bases)++] =
581 nmem_strdup (p->mem, p->basenames[i]);
583 *num_bases = info.new_num_bases;
584 *basenames = info.new_basenames;
585 for (i = 0; i<*num_bases; i++)
586 logf (LOG_LOG, "base %s", (*basenames)[i]);
589 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
590 Z_RPNQuery *query, int num_bases, char **basenames,
594 if (zebra_register_lock (zh, 0))
596 map_basenames (zh, stream, &num_bases, &basenames);
597 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
599 zebra_register_unlock (zh);
601 logf(LOG_APP,"SEARCH:%d:",zh->hits);
604 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
605 const char *setname, Z_RecordComposition *comp,
606 oid_value input_format, int num_recs,
607 ZebraRetrievalRecord *recs)
612 if (zebra_register_lock (zh, 0))
614 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
615 for (i = 0; i<num_recs; i++)
616 pos_array[i] = recs[i].position;
617 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
620 logf (LOG_DEBUG, "zebraPosSetCreate error");
622 zh->errString = nmem_strdup (stream->mem, setname);
626 for (i = 0; i<num_recs; i++)
632 sprintf (num_str, "%d", pos_array[i]);
634 zh->errString = nmem_strdup (stream->mem, num_str);
640 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
641 stream, input_format, comp,
642 &recs[i].format, &recs[i].buf,
645 recs[i].errString = NULL;
648 zebraPosSetDestroy (zh, poset, num_recs);
650 zebra_register_unlock (zh);
654 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
655 oid_value attributeset,
656 int num_bases, char **basenames,
657 int *position, int *num_entries, ZebraScanEntry **entries,
660 if (zebra_register_lock (zh, 0))
666 map_basenames (zh, stream, &num_bases, &basenames);
667 rpn_scan (zh, stream, zapt, attributeset,
668 num_bases, basenames, position,
669 num_entries, entries, is_partial);
670 zebra_register_unlock (zh);
673 void zebra_sort (ZebraHandle zh, ODR stream,
674 int num_input_setnames, const char **input_setnames,
675 const char *output_setname, Z_SortKeySpecList *sort_sequence,
678 if (zebra_register_lock (zh, 0))
680 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
681 output_setname, sort_sequence, sort_status);
682 zebra_register_unlock (zh);
685 int zebra_deleleResultSet(ZebraHandle zh, int function,
686 int num_setnames, char **setnames,
690 if (zebra_register_lock (zh, 0))
691 return Z_DeleteStatus_systemProblemAtTarget;
694 case Z_DeleteRequest_list:
695 resultSetDestroy (zh, num_setnames, setnames, statuses);
697 case Z_DeleteRequest_all:
698 resultSetDestroy (zh, -1, 0, statuses);
701 zebra_register_unlock (zh);
702 status = Z_DeleteStatus_success;
703 for (i = 0; i<num_setnames; i++)
704 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
705 status = statuses[i];
709 int zebra_errCode (ZebraHandle zh)
714 const char *zebra_errString (ZebraHandle zh)
716 return diagbib1_str (zh->errCode);
719 char *zebra_errAdd (ZebraHandle zh)
721 return zh->errString;
724 int zebra_hits (ZebraHandle zh)
729 int zebra_auth (ZebraService zh, const char *user, const char *pass)
731 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
733 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
737 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
741 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
743 if (zebra_register_lock (zh, 1))
745 xfree (zh->admin_databaseName);
746 zh->admin_databaseName = xstrdup(database);
749 void zebra_admin_import_end (ZebraHandle zh)
751 zebraExplain_flush (zh->service->zei, 1, zh);
753 zebra_register_unlock (zh);
756 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
760 if (zh->service->active < 2)
762 for (i = 0; i<segment->num_segmentRecords; i++)
764 Z_NamePlusRecord *npr = segment->segmentRecords[i];
765 const char *databaseName = npr->databaseName;
768 databaseName = zh->admin_databaseName;
769 printf ("--------------%d--------------------\n", i);
770 if (npr->which == Z_NamePlusRecord_intermediateFragment)
772 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
773 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
775 Odr_oct *oct = fragment->u.notExternallyTagged;
776 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
780 extract_rec_in_mem (zh, "grs.sgml",
788 0 /* match criteria */);
794 void zebra_admin_create (ZebraHandle zh, const char *database)
796 ZebraService zs = zh->service;
797 if (zebra_register_lock(zh, 1))
802 /* announce database */
803 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
806 zh->errString = "Database already exist";
808 zebraExplain_flush (zh->service->zei, 1, zh);
810 zebra_register_unlock(zh);
813 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
814 const char *input_str, int input_len,
815 char *output_str, int output_len)
818 if (!zh->service->zebra_maps)
820 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
821 input_str, input_len);
824 if (wrbuf_len(wrbuf) >= output_len)
826 if (wrbuf_len(wrbuf))
827 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
828 output_str[wrbuf_len(wrbuf)] = '\0';
829 return wrbuf_len(wrbuf);
832 static void extract_init (struct recExtractCtrl *p, RecWord *w)
834 w->zebra_maps = p->zebra_maps;
835 w->seqnos = p->seqno;
836 w->attrSet = VAL_BIB1;
842 static void extract_add_index_string (RecWord *p, const char *string,
846 unsigned char attrSet;
847 unsigned short attrUse;
850 int *pseqno = &p->seqnos[p->reg_type];
851 ZebraHandle zh = p->extractCtrl->handle;
852 struct recKeys *keys = &zh->keys;
854 if (keys->buf_used+1024 > keys->buf_max)
858 b = (char *) xmalloc (keys->buf_max += 128000);
859 if (keys->buf_used > 0)
860 memcpy (b, keys->buf, keys->buf_used);
864 dst = keys->buf + keys->buf_used;
866 attrSet = p->attrSet;
867 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
870 keys->prevAttrSet = attrSet;
871 attrUse = p->attrUse;
872 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
875 keys->prevAttrUse = attrUse;
877 diff = 1 + *pseqno - keys->prevSeqNo;
878 if (diff >= 1 && diff <= 15)
883 keys->prevSeqNo = *pseqno;
889 memcpy (dst, &attrSet, sizeof(attrSet));
890 dst += sizeof(attrSet);
894 memcpy (dst, &attrUse, sizeof(attrUse));
895 dst += sizeof(attrUse);
897 *dst++ = p->reg_type;
898 memcpy (dst, string, length);
904 memcpy (dst, pseqno, sizeof(*pseqno));
905 dst += sizeof(*pseqno);
907 keys->buf_used = dst - keys->buf;
912 static void extract_add_sort_string (RecWord *p, const char *string,
916 ZebraHandle zh = p->extractCtrl->handle;
917 struct sortKey *sortKeys = zh->sortKeys;
919 for (sk = sortKeys; sk; sk = sk->next)
920 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
923 sk = (struct sortKey *) xmalloc (sizeof(*sk));
927 sk->string = (char *) xmalloc (length);
929 memcpy (sk->string, string, length);
931 sk->attrSet = p->attrSet;
932 sk->attrUse = p->attrUse;
935 static void extract_add_string (RecWord *p, const char *string, int length)
938 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
939 extract_add_sort_string (p, string, length);
941 extract_add_index_string (p, string, length);
944 static void extract_add_incomplete_field (RecWord *p)
946 const char *b = p->string;
947 int remain = p->length;
948 const char **map = 0;
951 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
955 char buf[IT_MAX_WORD+1];
959 while (map && *map && **map == *CHR_SPACE)
961 remain = p->length - (b - p->string);
963 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
970 while (map && *map && **map != *CHR_SPACE)
972 const char *cp = *map;
974 while (i < IT_MAX_WORD && *cp)
976 remain = p->length - (b - p->string);
978 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
984 extract_add_string (p, buf, i);
986 (p->seqnos[p->reg_type])++; /* to separate this from next one */
989 static void extract_add_complete_field (RecWord *p)
991 const char *b = p->string;
992 char buf[IT_MAX_WORD+1];
993 const char **map = 0;
994 int i = 0, remain = p->length;
997 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
999 while (remain > 0 && i < IT_MAX_WORD)
1001 while (map && *map && **map == *CHR_SPACE)
1003 remain = p->length - (b - p->string);
1005 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1012 if (i && i < IT_MAX_WORD)
1013 buf[i++] = *CHR_SPACE;
1014 while (map && *map && **map != *CHR_SPACE)
1016 const char *cp = *map;
1018 if (i >= IT_MAX_WORD)
1020 while (i < IT_MAX_WORD && *cp)
1022 remain = p->length - (b - p->string);
1024 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1032 extract_add_string (p, buf, i);
1035 static void extract_token_add (RecWord *p)
1038 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1039 p->string, p->length)))
1041 p->string = wrbuf_buf(wrbuf);
1042 p->length = wrbuf_len(wrbuf);
1044 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1045 extract_add_complete_field (p);
1047 extract_add_incomplete_field(p);
1050 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1052 ZebraHandle zh = (ZebraHandle) (p->handle);
1053 zebraExplain_addSchema (zh->service->zei, oid);
1056 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1057 int cmd, struct sortKey **skp)
1059 struct sortKey *sk = *skp;
1060 SortIdx sortIdx = zh->service->sortIdx;
1062 sortIdx_sysno (sortIdx, sysno);
1065 struct sortKey *sk_next = sk->next;
1066 sortIdx_type (sortIdx, sk->attrUse);
1067 sortIdx_add (sortIdx, sk->string, sk->length);
1075 struct encode_info {
1082 void encode_key_init (struct encode_info *i)
1089 char *encode_key_int (int d, char *bp)
1093 else if (d <= 16383)
1095 *bp++ = 64 + (d>>8);
1098 else if (d <= 4194303)
1100 *bp++ = 128 + (d>>16);
1101 *bp++ = (d>>8) & 255;
1106 *bp++ = 192 + (d>>24);
1107 *bp++ = (d>>16) & 255;
1108 *bp++ = (d>>8) & 255;
1114 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1119 while ((*bp++ = *k++))
1121 memcpy (&key, k+1, sizeof(struct it_key));
1122 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1123 if (i->sysno != key.sysno)
1125 i->sysno = key.sysno;
1128 else if (!i->seqno && !key.seqno && i->cmd == *k)
1130 bp = encode_key_int (key.seqno - i->seqno, bp);
1131 i->seqno = key.seqno;
1133 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1135 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1140 static void extract_flushWriteKeys (ZebraHandle zh)
1143 char out_fname[200];
1145 struct encode_info encode_info;
1146 int ptr_i = zh->ptr_i;
1150 if (!zh->key_buf || ptr_i <= 0)
1153 (zh->key_file_no)++;
1154 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1156 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1158 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1160 if (!(outf = fopen (out_fname, "wb")))
1162 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1165 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1166 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1168 encode_key_init (&encode_info);
1169 encode_key_write (cp, &encode_info, outf);
1173 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1174 if (strcmp (cp, prevcp))
1176 encode_key_init (&encode_info);
1177 encode_key_write (cp, &encode_info, outf);
1181 encode_key_write (cp + strlen(cp), &encode_info, outf);
1184 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1185 extract_get_fname_tmp (out_fname, key_file_no);
1187 if (!(outf = fopen (out_fname, "wb")))
1189 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1192 logf (LOG_LOG, "writing section %d", key_file_no);
1194 prevcp = key_buf[ptr_top-i];
1196 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1198 key_y_len = strlen(prevcp)+1;
1200 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1201 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1203 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1204 sizeof(char*), key_y_compare);
1205 cp = key_buf[ptr_top-ptr_i];
1207 encode_key_init (&encode_info);
1208 encode_key_write (cp, &encode_info, outf);
1211 cp = key_buf[ptr_top-ptr_i];
1212 encode_key_write (cp+key_y_len, &encode_info, outf);
1216 prevcp = key_buf[ptr_top-ptr_i];
1221 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1224 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1226 zh->key_buf_used = 0;
1229 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1230 int cmd, struct recKeys *reckeys)
1232 unsigned char attrSet = (unsigned char) -1;
1233 unsigned short attrUse = (unsigned short) -1;
1236 ZebraExplainInfo zei = zh->service->zei;
1240 int mem = 8*1024*1024;
1241 zh->key_buf = (char**) xmalloc (mem);
1242 zh->ptr_top = mem/sizeof(char*);
1244 zh->key_buf_used = 0;
1245 zh->key_file_no = 0;
1247 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1248 while (off < reckeys->buf_used)
1250 const char *src = reckeys->buf + off;
1258 memcpy (&attrSet, src, sizeof(attrSet));
1259 src += sizeof(attrSet);
1263 memcpy (&attrUse, src, sizeof(attrUse));
1264 src += sizeof(attrUse);
1266 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1267 extract_flushWriteKeys (zh);
1269 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1270 (char*)zh->key_buf + zh->key_buf_used;
1272 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1274 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1277 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1280 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1282 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1283 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1286 seqno += ((lead>>2) & 15)-1;
1289 memcpy (&seqno, src, sizeof(seqno));
1290 src += sizeof(seqno);
1294 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1295 (zh->key_buf_used) += sizeof(key);
1296 off = src - reckeys->buf;
1298 assert (off == reckeys->buf_used);
1301 static void extract_index (ZebraHandle zh)
1303 extract_flushWriteKeys (zh);
1304 zebra_index_merge (zh);
1307 static int explain_extract (void *handle, Record rec, data1_node *n)
1309 ZebraHandle zh = (ZebraHandle) handle;
1310 struct recExtractCtrl extractCtrl;
1313 if (zebraExplain_curDatabase (zh->service->zei,
1314 rec->info[recInfo_databaseName]))
1317 if (zebraExplain_newDatabase (zh->service->zei,
1318 rec->info[recInfo_databaseName], 0))
1322 zh->keys.buf_used = 0;
1323 zh->keys.prevAttrUse = -1;
1324 zh->keys.prevAttrSet = -1;
1325 zh->keys.prevSeqNo = 0;
1328 extractCtrl.init = extract_init;
1329 extractCtrl.tokenAdd = extract_token_add;
1330 extractCtrl.schemaAdd = extract_schema_add;
1331 extractCtrl.dh = zh->service->dh;
1332 for (i = 0; i<256; i++)
1333 extractCtrl.seqno[i] = 0;
1334 extractCtrl.zebra_maps = zh->service->zebra_maps;
1335 extractCtrl.flagShowRecords = 0;
1336 extractCtrl.handle = handle;
1338 grs_extract_tree(&extractCtrl, n);
1340 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1342 if (rec->size[recInfo_delKeys])
1344 struct recKeys delkeys;
1345 struct sortKey *sortKeys = 0;
1347 delkeys.buf_used = rec->size[recInfo_delKeys];
1348 delkeys.buf = rec->info[recInfo_delKeys];
1349 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1350 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1352 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1353 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1355 xfree (rec->info[recInfo_delKeys]);
1356 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1357 rec->info[recInfo_delKeys] = zh->keys.buf;
1358 zh->keys.buf = NULL;
1359 zh->keys.buf_max = 0;
1363 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1364 const char *buf, size_t buf_size,
1365 const char *databaseName, int delete_flag,
1366 int test_mode, int *sysno,
1367 int store_keys, int store_data,
1368 const char *match_criteria)
1370 RecordAttr *recordAttr;
1371 struct recExtractCtrl extractCtrl;
1376 const char *fname = "<no file>";
1378 long recordOffset = 0;
1379 struct zebra_fetch_control fc;
1382 fc.record_int_buf = buf;
1383 fc.record_int_len = buf_size;
1384 fc.record_int_pos = 0;
1386 fc.record_offset = 0;
1388 extractCtrl.offset = 0;
1389 extractCtrl.readf = zebra_record_int_read;
1390 extractCtrl.seekf = zebra_record_int_seek;
1391 extractCtrl.tellf = zebra_record_int_tell;
1392 extractCtrl.endf = zebra_record_int_end;
1393 extractCtrl.fh = &fc;
1395 /* announce database */
1396 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1398 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1402 recType_byName (zh->service->recTypes, recordType, subType,
1405 logf (LOG_WARN, "No such record type: %s", recordType);
1409 zh->keys.buf_used = 0;
1410 zh->keys.prevAttrUse = -1;
1411 zh->keys.prevAttrSet = -1;
1412 zh->keys.prevSeqNo = 0;
1415 extractCtrl.subType = subType;
1416 extractCtrl.init = extract_init;
1417 extractCtrl.tokenAdd = extract_token_add;
1418 extractCtrl.schemaAdd = extract_schema_add;
1419 extractCtrl.dh = zh->service->dh;
1420 extractCtrl.handle = zh;
1421 extractCtrl.zebra_maps = zh->service->zebra_maps;
1422 extractCtrl.flagShowRecords = 0;
1423 for (i = 0; i<256; i++)
1425 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1426 extractCtrl.seqno[i] = 1;
1428 extractCtrl.seqno[i] = 0;
1431 r = (*recType->extract)(clientData, &extractCtrl);
1433 if (r == RECCTRL_EXTRACT_EOF)
1435 else if (r == RECCTRL_EXTRACT_ERROR)
1437 /* error occured during extraction ... */
1439 yaz_log (LOG_WARN, "extract error");
1441 if (rGroup->flagRw &&
1442 records_processed < rGroup->fileVerboseLimit)
1444 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1445 fname, (long) recordOffset);
1450 if (zh->keys.buf_used == 0)
1452 /* the extraction process returned no information - the record
1453 is probably empty - unless flagShowRecords is in use */
1456 logf (LOG_WARN, "No keys generated for record");
1457 logf (LOG_WARN, " The file is probably empty");
1460 /* match criteria */
1467 logf (LOG_LOG, "delete %s %s %ld", recordType,
1468 fname, (long) recordOffset);
1469 logf (LOG_WARN, "cannot delete record above (seems new)");
1472 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1473 (long) recordOffset);
1474 rec = rec_new (zh->service->records);
1476 *sysno = rec->sysno;
1478 recordAttr = rec_init_attr (zh->service->zei, rec);
1483 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1486 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1487 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1491 /* record already exists */
1492 struct recKeys delkeys;
1494 rec = rec_get (zh->service->records, *sysno);
1497 recordAttr = rec_init_attr (zh->service->zei, rec);
1499 if (recordAttr->runNumber ==
1500 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1502 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1503 fname, (long) recordOffset);
1507 delkeys.buf_used = rec->size[recInfo_delKeys];
1508 delkeys.buf = rec->info[recInfo_delKeys];
1509 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1510 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1513 /* record going to be deleted */
1514 if (!delkeys.buf_used)
1516 logf (LOG_LOG, "delete %s %s %ld", recordType,
1517 fname, (long) recordOffset);
1518 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1522 logf (LOG_LOG, "delete %s %s %ld", recordType,
1523 fname, (long) recordOffset);
1526 dict_delete (matchDict, matchStr);
1528 rec_del (zh->service->records, &rec);
1535 /* record going to be updated */
1536 if (!delkeys.buf_used)
1538 logf (LOG_LOG, "update %s %s %ld", recordType,
1539 fname, (long) recordOffset);
1540 logf (LOG_WARN, "cannot update file above, storeKeys false");
1544 logf (LOG_LOG, "update %s %s %ld", recordType,
1545 fname, (long) recordOffset);
1546 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1550 /* update file type */
1551 xfree (rec->info[recInfo_fileType]);
1552 rec->info[recInfo_fileType] =
1553 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1555 /* update filename */
1556 xfree (rec->info[recInfo_filename]);
1557 rec->info[recInfo_filename] =
1558 rec_strdup (fname, &rec->size[recInfo_filename]);
1560 /* update delete keys */
1561 xfree (rec->info[recInfo_delKeys]);
1562 if (zh->keys.buf_used > 0 && store_keys == 1)
1564 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1565 rec->info[recInfo_delKeys] = zh->keys.buf;
1566 zh->keys.buf = NULL;
1567 zh->keys.buf_max = 0;
1571 rec->info[recInfo_delKeys] = NULL;
1572 rec->size[recInfo_delKeys] = 0;
1575 /* save file size of original record */
1576 zebraExplain_recordBytesIncrement (zh->service->zei,
1577 - recordAttr->recordSize);
1579 recordAttr->recordSize = fi->file_moffset - recordOffset;
1580 if (!recordAttr->recordSize)
1581 recordAttr->recordSize = fi->file_max - recordOffset;
1583 recordAttr->recordSize = buf_size;
1585 zebraExplain_recordBytesIncrement (zh->service->zei,
1586 recordAttr->recordSize);
1588 /* set run-number for this record */
1589 recordAttr->runNumber =
1590 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1592 /* update store data */
1593 xfree (rec->info[recInfo_storeData]);
1594 if (store_data == 1)
1596 rec->size[recInfo_storeData] = recordAttr->recordSize;
1597 rec->info[recInfo_storeData] = (char *)
1598 xmalloc (recordAttr->recordSize);
1600 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1602 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1604 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1605 (long) recordOffset, fname);
1608 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1609 < recordAttr->recordSize)
1611 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1612 recordAttr->recordSize, fname);
1619 rec->info[recInfo_storeData] = NULL;
1620 rec->size[recInfo_storeData] = 0;
1622 /* update database name */
1623 xfree (rec->info[recInfo_databaseName]);
1624 rec->info[recInfo_databaseName] =
1625 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1628 recordAttr->recordOffset = recordOffset;
1630 /* commit this record */
1631 rec_put (zh->service->records, &rec);