2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.42 2000-12-05 10:01:44 adam
7 * Fixed bug regarding user-defined attribute sets.
9 * Revision 1.41 2000/12/01 17:59:08 adam
10 * Fixed bug regarding online updates on WIN32.
11 * When zebra.cfg is not available the server will not abort.
13 * Revision 1.40 2000/11/29 15:21:31 adam
14 * Fixed problem with passwd db.
16 * Revision 1.39 2000/11/29 14:24:01 adam
17 * Script configure uses yaz pthreads options. Added locking for
18 * zebra_register_{lock,unlock}.
20 * Revision 1.38 2000/11/08 13:46:58 adam
21 * Fixed scan: server could break if bad attribute/database was selected.
22 * Work on remote update.
24 * Revision 1.37 2000/10/17 12:37:09 adam
25 * Fixed notification of live-updates. Fixed minor problem with mf_init
26 * where it didn't handle shadow area file names correctly.
28 * Revision 1.36 2000/09/06 08:59:36 adam
29 * Using read-only (for now) for server.
31 * Revision 1.35 2000/07/07 12:49:20 adam
32 * Optimized resultSetInsert{Rank,Sort}.
34 * Revision 1.34 2000/06/09 13:56:38 ian
35 * Added some logging on Authentication and searches.
37 * Revision 1.33 2000/05/18 12:01:36 adam
38 * System call times(2) used again. More 64-bit fixes.
40 * Revision 1.32 2000/04/19 14:35:59 adam
41 * WIN32 update (this version is known not to work on Windows).
43 * Revision 1.31 2000/04/05 10:07:02 adam
44 * Minor zebra compile fix.
46 * Revision 1.30 2000/04/05 09:49:35 adam
47 * On Unix, zebra/z'mbol uses automake.
49 * Revision 1.29 2000/03/20 19:08:36 adam
50 * Added remote record import using Z39.50 extended services and Segment
53 * Revision 1.28 2000/03/15 15:00:30 adam
54 * First work on threaded version.
56 * Revision 1.27 2000/02/24 12:31:17 adam
57 * Added zebra_string_norm.
59 * Revision 1.26 1999/11/30 13:48:03 adam
60 * Improved installation. Updated for inclusion of YAZ header files.
62 * Revision 1.25 1999/11/04 15:00:45 adam
63 * Implemented delete result set(s).
65 * Revision 1.24 1999/10/14 14:33:50 adam
66 * Added truncation 5=106.
68 * Revision 1.23 1999/09/07 11:36:32 adam
71 * Revision 1.22 1999/08/02 10:13:47 adam
72 * Fixed bug regarding zebra_hits.
74 * Revision 1.21 1999/07/14 10:59:26 adam
75 * Changed functions isc_getmethod, isams_getmethod.
76 * Improved fatal error handling (such as missing EXPLAIN schema).
78 * Revision 1.20 1999/07/06 12:28:04 adam
79 * Updated record index structure. Format includes version ID. Compression
80 * algorithm ID is stored for each record block.
82 * Revision 1.19 1999/05/26 07:49:13 adam
85 * Revision 1.18 1999/05/15 14:36:38 adam
86 * Updated dictionary. Implemented "compression" of dictionary.
88 * Revision 1.17 1999/05/12 13:08:06 adam
89 * First version of ISAMS.
91 * Revision 1.16 1999/02/19 10:38:30 adam
92 * Implemented chdir-setting.
94 * Revision 1.15 1999/02/17 12:18:12 adam
95 * Fixed zebra_close so that a NULL pointer is ignored.
97 * Revision 1.14 1999/02/02 14:51:11 adam
98 * Updated WIN32 code specific sections. Changed header.
100 * Revision 1.13 1998/12/16 12:23:30 adam
101 * Added facility for database name mapping using resource mapdb.
103 * Revision 1.12 1998/11/16 10:18:10 adam
104 * Better error reporting for result sets.
106 * Revision 1.11 1998/10/16 08:14:34 adam
107 * Updated record control system.
109 * Revision 1.10 1998/09/22 10:03:42 adam
110 * Changed result sets to be persistent in the sense that they can
111 * be re-searched if needed.
112 * Fixed memory leak in rsm_or.
114 * Revision 1.9 1998/09/02 13:53:17 adam
115 * Extra parameter decode added to search routines to implement
116 * persistent queries.
118 * Revision 1.8 1998/08/24 17:29:23 adam
121 * Revision 1.7 1998/06/24 12:16:13 adam
122 * Support for relations on text operands. Open range support in
123 * DFA module (i.e. [-j], [g-]).
125 * Revision 1.6 1998/06/22 11:36:47 adam
126 * Added authentication check facility to zebra.
128 * Revision 1.5 1998/06/13 00:14:08 adam
131 * Revision 1.4 1998/06/12 12:22:12 adam
134 * Revision 1.3 1998/05/27 16:57:44 adam
135 * Zebra returns surrogate diagnostic for single records when
138 * Revision 1.2 1998/05/20 10:12:19 adam
139 * Implemented automatic EXPLAIN database maintenance.
140 * Modified Zebra to work with ASN.1 compiled version of YAZ.
142 * Revision 1.1 1998/03/05 08:45:13 adam
143 * New result set model and modular ranking system. Moved towards
144 * descent server API. System information stored as "SGML" records.
158 #include <yaz/diagbib1.h>
162 static void zebra_chdir (ZebraService zh)
164 const char *dir = res_get (zh->res, "chdir");
167 logf (LOG_DEBUG, "chdir %s", dir);
175 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
176 const char *buf, size_t buf_size,
177 const char *databaseName, int delete_flag,
178 int test_mode, int *sysno,
179 int store_keys, int store_data,
180 const char *match_criteria);
182 static int explain_extract (void *handle, Record rec, data1_node *n);
183 static void extract_index (ZebraHandle zh);
185 static void zebra_register_unlock (ZebraHandle zh);
187 static int zebra_register_activate (ZebraService zh, int rw);
188 static int zebra_register_deactivate (ZebraService zh);
190 static int zebra_register_lock (ZebraHandle zh, int rw)
196 if (!zh->service->active)
206 zebra_mutex_cond_lock (&zh->service->session_lock);
208 state = zebra_server_lock_get_state(zh->service, &lastChange);
210 zebra_server_lock (zh->service, state);
221 logf (LOG_LOG, "Register in read/write mode");
222 else if (zh->service->registerState == state)
224 logf (LOG_LOG, "registerChange = %ld lastChange = %ld",
225 (long) zh->service->registerChange, (long)lastChange);
226 if (zh->service->registerChange >= lastChange)
230 logf (LOG_LOG, "Register completely updated since last access");
232 else if (zh->service->registerState == -1)
233 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
236 logf (LOG_LOG, "Register has changed state from %d to %d",
237 zh->service->registerState, state);
238 zh->service->registerChange = lastChange;
240 zebra_register_deactivate (zh->service);
242 zh->service->registerState = state;
244 zebra_register_activate (zh->service, rw);
249 static void zebra_register_unlock (ZebraHandle zh)
251 if (zh->service->registerState != -1)
252 zebra_server_unlock (zh->service, zh->service->registerState);
253 zebra_mutex_cond_unlock (&zh->service->session_lock);
256 logf (LOG_LOG, "user/system: %ld/%ld",
257 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
258 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
263 ZebraHandle zebra_open (ZebraService zs)
271 zh = (ZebraHandle) xmalloc (sizeof(*zh));
272 yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh);
281 zh->admin_databaseName = 0;
283 zebra_mutex_cond_lock (&zs->session_lock);
285 zh->next = zs->sessions;
288 zebra_mutex_cond_unlock (&zs->session_lock);
294 ZebraService zebra_start (const char *configName)
296 ZebraService zh = xmalloc (sizeof(*zh));
298 yaz_log (LOG_LOG, "zebra_start %s", configName);
300 zh->configName = xstrdup(configName);
305 zh->registerState = -1;
306 zh->registerChange = 0;
308 if (!(zh->res = res_open (zh->configName)))
310 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
314 zebra_server_lock_init (zh);
315 zebra_mutex_cond_init (&zh->session_lock);
316 if (!res_get (zh->res, "passwd"))
317 zh->passwd_db = NULL;
320 zh->passwd_db = passwd_db_open ();
322 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
324 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
330 static int zebra_register_activate (ZebraService zh, int rw)
334 yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)",
338 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
339 zh->registerState ? "yes" : "no");
341 zh->dh = data1_create ();
344 zh->bfs = bfs_create (res_get (zh->res, "register"));
347 data1_destroy(zh->dh);
350 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
351 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
352 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
353 zh->recTypes = recTypes_init (zh->dh);
354 recTypes_default_handlers (zh->recTypes);
357 zh->zebra_maps = zebra_maps_open (zh->res);
358 zh->rank_classes = NULL;
371 zebraRankInstall (zh, rank1_class);
373 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
375 logf (LOG_WARN, "rec_open");
378 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
380 logf (LOG_WARN, "dict_open");
383 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
385 logf (LOG_WARN, "sortIdx_open");
388 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
390 struct ISAMS_M_s isams_m;
391 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
392 key_isams_m(zh->res, &isams_m))))
394 logf (LOG_WARN, "isams_open");
399 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
401 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
402 sizeof (struct it_key), zh->res)))
404 logf (LOG_WARN, "is_open");
408 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
410 struct ISAMC_M_s isamc_m;
411 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
412 rw, key_isamc_m(zh->res, &isamc_m))))
414 logf (LOG_WARN, "isc_open");
418 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
420 struct ISAMD_M_s isamd_m;
422 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
423 rw, key_isamd_m(zh->res, &isamd_m))))
425 logf (LOG_WARN, "isamd_open");
430 zh->zei = zebraExplain_open (zh->records, zh->dh,
431 zh->res, rw, 0 /* rGroup */,
435 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
439 yaz_log (LOG_LOG, "zebra_register_activate ok");
443 void zebra_admin_shutdown (ZebraHandle zh)
445 zebra_mutex_cond_lock (&zh->service->session_lock);
446 zh->service->stop_flag = 1;
447 if (!zh->service->sessions)
448 zebra_register_deactivate(zh->service);
449 zh->service->active = 0;
450 zebra_mutex_cond_unlock (&zh->service->session_lock);
453 void zebra_admin_start (ZebraHandle zh)
455 ZebraService zs = zh->service;
457 zebra_mutex_cond_lock (&zs->session_lock);
459 zh->service->active = 1;
460 zebra_mutex_cond_unlock (&zs->session_lock);
463 static int zebra_register_deactivate (ZebraService zs)
468 yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)",
472 yaz_log(LOG_LOG, "zebra_register_deactivate");
476 zebraExplain_close (zs->zei, 0);
477 dict_close (zs->dict);
478 sortIdx_close (zs->sortIdx);
480 isams_close (zs->isams);
485 isc_close (zs->isamc);
487 isamd_close (zs->isamd);
489 rec_close (&zs->records);
491 recTypes_destroy (zs->recTypes);
492 zebra_maps_close (zs->zebra_maps);
493 zebraRankDestroy (zs);
494 bfs_destroy (zs->bfs);
495 data1_destroy (zs->dh);
498 passwd_db_close (zs->passwd_db);
503 void zebra_stop(ZebraService zs)
507 yaz_log (LOG_LOG, "zebra_stop");
509 zebra_mutex_cond_lock (&zs->session_lock);
511 zebra_close (zs->sessions);
513 zebra_mutex_cond_unlock (&zs->session_lock);
515 zebra_mutex_cond_destroy (&zs->session_lock);
517 zebra_register_deactivate(zs);
519 xfree (zs->configName);
523 void zebra_close (ZebraHandle zh)
525 ZebraService zs = zh->service;
526 struct zebra_session **sp;
528 yaz_log (LOG_LOG, "zebra_close zh=%p", zh);
531 resultSetDestroy (zh, -1, 0, 0);
538 xfree (zh->admin_databaseName);
539 zebra_mutex_cond_lock (&zs->session_lock);
551 if (!zs->sessions && zs->stop_flag)
552 zebra_register_deactivate(zs);
553 zebra_mutex_cond_unlock (&zs->session_lock);
557 struct map_baseinfo {
563 char **new_basenames;
567 void map_basenames_func (void *vp, const char *name, const char *value)
569 struct map_baseinfo *p = (struct map_baseinfo *) vp;
571 char fromdb[128], todb[8][128];
574 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
575 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
576 todb[5], todb[6], todb[7]);
580 for (i = 0; i<p->num_bases; i++)
581 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
584 for (i = 0; i < no; i++)
586 if (p->new_num_bases == p->new_num_max)
588 p->new_basenames[(p->new_num_bases)++] =
589 nmem_strdup (p->mem, todb[i]);
595 void map_basenames (ZebraHandle zh, ODR stream,
596 int *num_bases, char ***basenames)
598 struct map_baseinfo info;
599 struct map_baseinfo *p = &info;
603 info.num_bases = *num_bases;
604 info.basenames = *basenames;
605 info.new_num_max = 128;
606 info.new_num_bases = 0;
607 info.new_basenames = (char **)
608 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
609 info.mem = stream->mem;
611 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
613 for (i = 0; i<p->num_bases; i++)
614 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
616 p->new_basenames[(p->new_num_bases)++] =
617 nmem_strdup (p->mem, p->basenames[i]);
619 *num_bases = info.new_num_bases;
620 *basenames = info.new_basenames;
621 for (i = 0; i<*num_bases; i++)
622 logf (LOG_LOG, "base %s", (*basenames)[i]);
625 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
626 Z_RPNQuery *query, int num_bases, char **basenames,
630 if (zebra_register_lock (zh, 0))
632 map_basenames (zh, stream, &num_bases, &basenames);
633 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
635 zebra_register_unlock (zh);
637 logf(LOG_APP,"SEARCH:%d:",zh->hits);
640 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
641 const char *setname, Z_RecordComposition *comp,
642 oid_value input_format, int num_recs,
643 ZebraRetrievalRecord *recs)
648 if (zebra_register_lock (zh, 0))
650 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
651 for (i = 0; i<num_recs; i++)
652 pos_array[i] = recs[i].position;
653 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
656 logf (LOG_DEBUG, "zebraPosSetCreate error");
658 zh->errString = nmem_strdup (stream->mem, setname);
662 for (i = 0; i<num_recs; i++)
668 sprintf (num_str, "%d", pos_array[i]);
670 zh->errString = nmem_strdup (stream->mem, num_str);
676 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
677 stream, input_format, comp,
678 &recs[i].format, &recs[i].buf,
681 recs[i].errString = NULL;
684 zebraPosSetDestroy (zh, poset, num_recs);
686 zebra_register_unlock (zh);
690 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
691 oid_value attributeset,
692 int num_bases, char **basenames,
693 int *position, int *num_entries, ZebraScanEntry **entries,
696 if (zebra_register_lock (zh, 0))
702 map_basenames (zh, stream, &num_bases, &basenames);
703 rpn_scan (zh, stream, zapt, attributeset,
704 num_bases, basenames, position,
705 num_entries, entries, is_partial);
706 zebra_register_unlock (zh);
709 void zebra_sort (ZebraHandle zh, ODR stream,
710 int num_input_setnames, const char **input_setnames,
711 const char *output_setname, Z_SortKeySpecList *sort_sequence,
714 if (zebra_register_lock (zh, 0))
716 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
717 output_setname, sort_sequence, sort_status);
718 zebra_register_unlock (zh);
721 int zebra_deleleResultSet(ZebraHandle zh, int function,
722 int num_setnames, char **setnames,
726 if (zebra_register_lock (zh, 0))
727 return Z_DeleteStatus_systemProblemAtTarget;
730 case Z_DeleteRequest_list:
731 resultSetDestroy (zh, num_setnames, setnames, statuses);
733 case Z_DeleteRequest_all:
734 resultSetDestroy (zh, -1, 0, statuses);
737 zebra_register_unlock (zh);
738 status = Z_DeleteStatus_success;
739 for (i = 0; i<num_setnames; i++)
740 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
741 status = statuses[i];
745 int zebra_errCode (ZebraHandle zh)
750 const char *zebra_errString (ZebraHandle zh)
752 return diagbib1_str (zh->errCode);
755 char *zebra_errAdd (ZebraHandle zh)
757 return zh->errString;
760 int zebra_hits (ZebraHandle zh)
765 int zebra_auth (ZebraService zh, const char *user, const char *pass)
767 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
769 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
773 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
777 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
779 if (zebra_register_lock (zh, 1))
781 xfree (zh->admin_databaseName);
782 zh->admin_databaseName = xstrdup(database);
785 void zebra_admin_import_end (ZebraHandle zh)
787 zebraExplain_flush (zh->service->zei, 1, zh);
789 zebra_register_unlock (zh);
792 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
796 if (zh->service->active < 2)
798 for (i = 0; i<segment->num_segmentRecords; i++)
800 Z_NamePlusRecord *npr = segment->segmentRecords[i];
801 const char *databaseName = npr->databaseName;
804 databaseName = zh->admin_databaseName;
805 printf ("--------------%d--------------------\n", i);
806 if (npr->which == Z_NamePlusRecord_intermediateFragment)
808 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
809 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
811 Odr_oct *oct = fragment->u.notExternallyTagged;
812 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
816 extract_rec_in_mem (zh, "grs.sgml",
824 0 /* match criteria */);
830 void zebra_admin_create (ZebraHandle zh, const char *database)
832 ZebraService zs = zh->service;
833 if (zebra_register_lock(zh, 1))
838 /* announce database */
839 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
842 zh->errString = "Database already exist";
844 zebraExplain_flush (zh->service->zei, 1, zh);
846 zebra_register_unlock(zh);
849 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
850 const char *input_str, int input_len,
851 char *output_str, int output_len)
854 if (!zh->service->zebra_maps)
856 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
857 input_str, input_len);
860 if (wrbuf_len(wrbuf) >= output_len)
862 if (wrbuf_len(wrbuf))
863 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
864 output_str[wrbuf_len(wrbuf)] = '\0';
865 return wrbuf_len(wrbuf);
868 static void extract_init (struct recExtractCtrl *p, RecWord *w)
870 w->zebra_maps = p->zebra_maps;
871 w->seqnos = p->seqno;
872 w->attrSet = VAL_BIB1;
878 static void extract_add_index_string (RecWord *p, const char *string,
882 unsigned char attrSet;
883 unsigned short attrUse;
886 int *pseqno = &p->seqnos[p->reg_type];
887 ZebraHandle zh = p->extractCtrl->handle;
888 ZebraExplainInfo zei = zh->service->zei;
889 struct recKeys *keys = &zh->keys;
891 if (keys->buf_used+1024 > keys->buf_max)
895 b = (char *) xmalloc (keys->buf_max += 128000);
896 if (keys->buf_used > 0)
897 memcpy (b, keys->buf, keys->buf_used);
901 dst = keys->buf + keys->buf_used;
903 attrSet = p->attrSet;
904 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
907 keys->prevAttrSet = attrSet;
908 attrUse = p->attrUse;
909 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
912 keys->prevAttrUse = attrUse;
914 diff = 1 + *pseqno - keys->prevSeqNo;
915 if (diff >= 1 && diff <= 15)
920 keys->prevSeqNo = *pseqno;
927 int ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
930 ch = zebraExplain_addSU (zei, attrSet, attrUse);
931 yaz_log (LOG_LOG, "addSU set=%d use=%d SU=%d",
932 attrSet, attrUse, ch);
935 memcpy (dst, &ch, sizeof(ch));
941 memcpy (dst, &attrSet, sizeof(attrSet));
942 dst += sizeof(attrSet);
946 memcpy (dst, &attrUse, sizeof(attrUse));
947 dst += sizeof(attrUse);
950 *dst++ = p->reg_type;
951 memcpy (dst, string, length);
957 memcpy (dst, pseqno, sizeof(*pseqno));
958 dst += sizeof(*pseqno);
960 keys->buf_used = dst - keys->buf;
965 static void extract_add_sort_string (RecWord *p, const char *string,
969 ZebraHandle zh = p->extractCtrl->handle;
970 struct sortKey *sortKeys = zh->sortKeys;
972 for (sk = sortKeys; sk; sk = sk->next)
973 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
976 sk = (struct sortKey *) xmalloc (sizeof(*sk));
980 sk->string = (char *) xmalloc (length);
982 memcpy (sk->string, string, length);
984 sk->attrSet = p->attrSet;
985 sk->attrUse = p->attrUse;
988 static void extract_add_string (RecWord *p, const char *string, int length)
991 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
992 extract_add_sort_string (p, string, length);
994 extract_add_index_string (p, string, length);
997 static void extract_add_incomplete_field (RecWord *p)
999 const char *b = p->string;
1000 int remain = p->length;
1001 const char **map = 0;
1004 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1008 char buf[IT_MAX_WORD+1];
1012 while (map && *map && **map == *CHR_SPACE)
1014 remain = p->length - (b - p->string);
1016 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1023 while (map && *map && **map != *CHR_SPACE)
1025 const char *cp = *map;
1027 while (i < IT_MAX_WORD && *cp)
1029 remain = p->length - (b - p->string);
1031 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1037 extract_add_string (p, buf, i);
1039 (p->seqnos[p->reg_type])++; /* to separate this from next one */
1042 static void extract_add_complete_field (RecWord *p)
1044 const char *b = p->string;
1045 char buf[IT_MAX_WORD+1];
1046 const char **map = 0;
1047 int i = 0, remain = p->length;
1050 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1052 while (remain > 0 && i < IT_MAX_WORD)
1054 while (map && *map && **map == *CHR_SPACE)
1056 remain = p->length - (b - p->string);
1058 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1065 if (i && i < IT_MAX_WORD)
1066 buf[i++] = *CHR_SPACE;
1067 while (map && *map && **map != *CHR_SPACE)
1069 const char *cp = *map;
1071 if (i >= IT_MAX_WORD)
1073 while (i < IT_MAX_WORD && *cp)
1075 remain = p->length - (b - p->string);
1077 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1085 extract_add_string (p, buf, i);
1088 static void extract_token_add (RecWord *p)
1091 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1092 p->string, p->length)))
1094 p->string = wrbuf_buf(wrbuf);
1095 p->length = wrbuf_len(wrbuf);
1097 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1098 extract_add_complete_field (p);
1100 extract_add_incomplete_field(p);
1103 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1105 ZebraHandle zh = (ZebraHandle) (p->handle);
1106 zebraExplain_addSchema (zh->service->zei, oid);
1109 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1110 int cmd, struct sortKey **skp)
1112 struct sortKey *sk = *skp;
1113 SortIdx sortIdx = zh->service->sortIdx;
1115 sortIdx_sysno (sortIdx, sysno);
1118 struct sortKey *sk_next = sk->next;
1119 sortIdx_type (sortIdx, sk->attrUse);
1120 sortIdx_add (sortIdx, sk->string, sk->length);
1128 struct encode_info {
1135 void encode_key_init (struct encode_info *i)
1142 char *encode_key_int (int d, char *bp)
1146 else if (d <= 16383)
1148 *bp++ = 64 + (d>>8);
1151 else if (d <= 4194303)
1153 *bp++ = 128 + (d>>16);
1154 *bp++ = (d>>8) & 255;
1159 *bp++ = 192 + (d>>24);
1160 *bp++ = (d>>16) & 255;
1161 *bp++ = (d>>8) & 255;
1167 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1172 while ((*bp++ = *k++))
1174 memcpy (&key, k+1, sizeof(struct it_key));
1175 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1176 if (i->sysno != key.sysno)
1178 i->sysno = key.sysno;
1181 else if (!i->seqno && !key.seqno && i->cmd == *k)
1183 bp = encode_key_int (key.seqno - i->seqno, bp);
1184 i->seqno = key.seqno;
1186 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1188 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1193 static void extract_flushWriteKeys (ZebraHandle zh)
1196 char out_fname[200];
1198 struct encode_info encode_info;
1199 int ptr_i = zh->ptr_i;
1203 if (!zh->key_buf || ptr_i <= 0)
1206 (zh->key_file_no)++;
1207 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1209 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1211 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1213 if (!(outf = fopen (out_fname, "wb")))
1215 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1218 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1219 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1221 encode_key_init (&encode_info);
1222 encode_key_write (cp, &encode_info, outf);
1226 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1227 if (strcmp (cp, prevcp))
1229 encode_key_init (&encode_info);
1230 encode_key_write (cp, &encode_info, outf);
1234 encode_key_write (cp + strlen(cp), &encode_info, outf);
1237 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1238 extract_get_fname_tmp (out_fname, key_file_no);
1240 if (!(outf = fopen (out_fname, "wb")))
1242 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1245 logf (LOG_LOG, "writing section %d", key_file_no);
1247 prevcp = key_buf[ptr_top-i];
1249 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1251 key_y_len = strlen(prevcp)+1;
1253 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1254 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1256 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1257 sizeof(char*), key_y_compare);
1258 cp = key_buf[ptr_top-ptr_i];
1260 encode_key_init (&encode_info);
1261 encode_key_write (cp, &encode_info, outf);
1264 cp = key_buf[ptr_top-ptr_i];
1265 encode_key_write (cp+key_y_len, &encode_info, outf);
1269 prevcp = key_buf[ptr_top-ptr_i];
1274 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1277 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1279 zh->key_buf_used = 0;
1282 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1283 int cmd, struct recKeys *reckeys)
1287 unsigned char attrSet = (unsigned char) -1;
1288 unsigned short attrUse = (unsigned short) -1;
1293 ZebraExplainInfo zei = zh->service->zei;
1297 int mem = 8*1024*1024;
1298 zh->key_buf = (char**) xmalloc (mem);
1299 zh->ptr_top = mem/sizeof(char*);
1301 zh->key_buf_used = 0;
1302 zh->key_file_no = 0;
1304 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1305 while (off < reckeys->buf_used)
1307 const char *src = reckeys->buf + off;
1316 memcpy (&ch, src, sizeof(ch));
1322 memcpy (&attrSet, src, sizeof(attrSet));
1323 src += sizeof(attrSet);
1327 memcpy (&attrUse, src, sizeof(attrUse));
1328 src += sizeof(attrUse);
1331 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1332 extract_flushWriteKeys (zh);
1334 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1335 (char*)zh->key_buf + zh->key_buf_used;
1338 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1340 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1344 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1347 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1349 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1350 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1353 seqno += ((lead>>2) & 15)-1;
1356 memcpy (&seqno, src, sizeof(seqno));
1357 src += sizeof(seqno);
1361 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1362 (zh->key_buf_used) += sizeof(key);
1363 off = src - reckeys->buf;
1365 assert (off == reckeys->buf_used);
1368 static void extract_index (ZebraHandle zh)
1370 extract_flushWriteKeys (zh);
1371 zebra_index_merge (zh);
1374 static int explain_extract (void *handle, Record rec, data1_node *n)
1376 ZebraHandle zh = (ZebraHandle) handle;
1377 struct recExtractCtrl extractCtrl;
1380 if (zebraExplain_curDatabase (zh->service->zei,
1381 rec->info[recInfo_databaseName]))
1384 if (zebraExplain_newDatabase (zh->service->zei,
1385 rec->info[recInfo_databaseName], 0))
1389 zh->keys.buf_used = 0;
1390 zh->keys.prevAttrUse = -1;
1391 zh->keys.prevAttrSet = -1;
1392 zh->keys.prevSeqNo = 0;
1395 extractCtrl.init = extract_init;
1396 extractCtrl.tokenAdd = extract_token_add;
1397 extractCtrl.schemaAdd = extract_schema_add;
1398 extractCtrl.dh = zh->service->dh;
1399 for (i = 0; i<256; i++)
1400 extractCtrl.seqno[i] = 0;
1401 extractCtrl.zebra_maps = zh->service->zebra_maps;
1402 extractCtrl.flagShowRecords = 0;
1403 extractCtrl.handle = handle;
1405 grs_extract_tree(&extractCtrl, n);
1407 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1409 if (rec->size[recInfo_delKeys])
1411 struct recKeys delkeys;
1412 struct sortKey *sortKeys = 0;
1414 delkeys.buf_used = rec->size[recInfo_delKeys];
1415 delkeys.buf = rec->info[recInfo_delKeys];
1416 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1417 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1419 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1420 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1422 xfree (rec->info[recInfo_delKeys]);
1423 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1424 rec->info[recInfo_delKeys] = zh->keys.buf;
1425 zh->keys.buf = NULL;
1426 zh->keys.buf_max = 0;
1430 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1431 const char *buf, size_t buf_size,
1432 const char *databaseName, int delete_flag,
1433 int test_mode, int *sysno,
1434 int store_keys, int store_data,
1435 const char *match_criteria)
1437 RecordAttr *recordAttr;
1438 struct recExtractCtrl extractCtrl;
1443 const char *fname = "<no file>";
1445 long recordOffset = 0;
1446 struct zebra_fetch_control fc;
1449 fc.record_int_buf = buf;
1450 fc.record_int_len = buf_size;
1451 fc.record_int_pos = 0;
1453 fc.record_offset = 0;
1455 extractCtrl.offset = 0;
1456 extractCtrl.readf = zebra_record_int_read;
1457 extractCtrl.seekf = zebra_record_int_seek;
1458 extractCtrl.tellf = zebra_record_int_tell;
1459 extractCtrl.endf = zebra_record_int_end;
1460 extractCtrl.fh = &fc;
1462 /* announce database */
1463 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1465 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1469 recType_byName (zh->service->recTypes, recordType, subType,
1472 logf (LOG_WARN, "No such record type: %s", recordType);
1476 zh->keys.buf_used = 0;
1477 zh->keys.prevAttrUse = -1;
1478 zh->keys.prevAttrSet = -1;
1479 zh->keys.prevSeqNo = 0;
1482 extractCtrl.subType = subType;
1483 extractCtrl.init = extract_init;
1484 extractCtrl.tokenAdd = extract_token_add;
1485 extractCtrl.schemaAdd = extract_schema_add;
1486 extractCtrl.dh = zh->service->dh;
1487 extractCtrl.handle = zh;
1488 extractCtrl.zebra_maps = zh->service->zebra_maps;
1489 extractCtrl.flagShowRecords = 0;
1490 for (i = 0; i<256; i++)
1492 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1493 extractCtrl.seqno[i] = 1;
1495 extractCtrl.seqno[i] = 0;
1498 r = (*recType->extract)(clientData, &extractCtrl);
1500 if (r == RECCTRL_EXTRACT_EOF)
1502 else if (r == RECCTRL_EXTRACT_ERROR)
1504 /* error occured during extraction ... */
1506 yaz_log (LOG_WARN, "extract error");
1508 if (rGroup->flagRw &&
1509 records_processed < rGroup->fileVerboseLimit)
1511 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1512 fname, (long) recordOffset);
1517 if (zh->keys.buf_used == 0)
1519 /* the extraction process returned no information - the record
1520 is probably empty - unless flagShowRecords is in use */
1523 logf (LOG_WARN, "No keys generated for record");
1524 logf (LOG_WARN, " The file is probably empty");
1527 /* match criteria */
1534 logf (LOG_LOG, "delete %s %s %ld", recordType,
1535 fname, (long) recordOffset);
1536 logf (LOG_WARN, "cannot delete record above (seems new)");
1539 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1540 (long) recordOffset);
1541 rec = rec_new (zh->service->records);
1543 *sysno = rec->sysno;
1545 recordAttr = rec_init_attr (zh->service->zei, rec);
1550 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1553 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1554 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1558 /* record already exists */
1559 struct recKeys delkeys;
1561 rec = rec_get (zh->service->records, *sysno);
1564 recordAttr = rec_init_attr (zh->service->zei, rec);
1566 if (recordAttr->runNumber ==
1567 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1569 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1570 fname, (long) recordOffset);
1574 delkeys.buf_used = rec->size[recInfo_delKeys];
1575 delkeys.buf = rec->info[recInfo_delKeys];
1576 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1577 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1580 /* record going to be deleted */
1581 if (!delkeys.buf_used)
1583 logf (LOG_LOG, "delete %s %s %ld", recordType,
1584 fname, (long) recordOffset);
1585 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1589 logf (LOG_LOG, "delete %s %s %ld", recordType,
1590 fname, (long) recordOffset);
1593 dict_delete (matchDict, matchStr);
1595 rec_del (zh->service->records, &rec);
1602 /* record going to be updated */
1603 if (!delkeys.buf_used)
1605 logf (LOG_LOG, "update %s %s %ld", recordType,
1606 fname, (long) recordOffset);
1607 logf (LOG_WARN, "cannot update file above, storeKeys false");
1611 logf (LOG_LOG, "update %s %s %ld", recordType,
1612 fname, (long) recordOffset);
1613 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1617 /* update file type */
1618 xfree (rec->info[recInfo_fileType]);
1619 rec->info[recInfo_fileType] =
1620 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1622 /* update filename */
1623 xfree (rec->info[recInfo_filename]);
1624 rec->info[recInfo_filename] =
1625 rec_strdup (fname, &rec->size[recInfo_filename]);
1627 /* update delete keys */
1628 xfree (rec->info[recInfo_delKeys]);
1629 if (zh->keys.buf_used > 0 && store_keys == 1)
1631 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1632 rec->info[recInfo_delKeys] = zh->keys.buf;
1633 zh->keys.buf = NULL;
1634 zh->keys.buf_max = 0;
1638 rec->info[recInfo_delKeys] = NULL;
1639 rec->size[recInfo_delKeys] = 0;
1642 /* save file size of original record */
1643 zebraExplain_recordBytesIncrement (zh->service->zei,
1644 - recordAttr->recordSize);
1646 recordAttr->recordSize = fi->file_moffset - recordOffset;
1647 if (!recordAttr->recordSize)
1648 recordAttr->recordSize = fi->file_max - recordOffset;
1650 recordAttr->recordSize = buf_size;
1652 zebraExplain_recordBytesIncrement (zh->service->zei,
1653 recordAttr->recordSize);
1655 /* set run-number for this record */
1656 recordAttr->runNumber =
1657 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1659 /* update store data */
1660 xfree (rec->info[recInfo_storeData]);
1661 if (store_data == 1)
1663 rec->size[recInfo_storeData] = recordAttr->recordSize;
1664 rec->info[recInfo_storeData] = (char *)
1665 xmalloc (recordAttr->recordSize);
1667 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1669 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1671 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1672 (long) recordOffset, fname);
1675 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1676 < recordAttr->recordSize)
1678 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1679 recordAttr->recordSize, fname);
1686 rec->info[recInfo_storeData] = NULL;
1687 rec->size[recInfo_storeData] = 0;
1689 /* update database name */
1690 xfree (rec->info[recInfo_databaseName]);
1691 rec->info[recInfo_databaseName] =
1692 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1695 recordAttr->recordOffset = recordOffset;
1697 /* commit this record */
1698 rec_put (zh->service->records, &rec);