2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.41 2000-12-01 17:59:08 adam
7 * Fixed bug regarding online updates on WIN32.
8 * When zebra.cfg is not available the server will not abort.
10 * Revision 1.40 2000/11/29 15:21:31 adam
11 * Fixed problem with passwd db.
13 * Revision 1.39 2000/11/29 14:24:01 adam
14 * Script configure uses yaz pthreads options. Added locking for
15 * zebra_register_{lock,unlock}.
17 * Revision 1.38 2000/11/08 13:46:58 adam
18 * Fixed scan: server could break if bad attribute/database was selected.
19 * Work on remote update.
21 * Revision 1.37 2000/10/17 12:37:09 adam
22 * Fixed notification of live-updates. Fixed minor problem with mf_init
23 * where it didn't handle shadow area file names correctly.
25 * Revision 1.36 2000/09/06 08:59:36 adam
26 * Using read-only (for now) for server.
28 * Revision 1.35 2000/07/07 12:49:20 adam
29 * Optimized resultSetInsert{Rank,Sort}.
31 * Revision 1.34 2000/06/09 13:56:38 ian
32 * Added some logging on Authentication and searches.
34 * Revision 1.33 2000/05/18 12:01:36 adam
35 * System call times(2) used again. More 64-bit fixes.
37 * Revision 1.32 2000/04/19 14:35:59 adam
38 * WIN32 update (this version is known not to work on Windows).
40 * Revision 1.31 2000/04/05 10:07:02 adam
41 * Minor zebra compile fix.
43 * Revision 1.30 2000/04/05 09:49:35 adam
44 * On Unix, zebra/z'mbol uses automake.
46 * Revision 1.29 2000/03/20 19:08:36 adam
47 * Added remote record import using Z39.50 extended services and Segment
50 * Revision 1.28 2000/03/15 15:00:30 adam
51 * First work on threaded version.
53 * Revision 1.27 2000/02/24 12:31:17 adam
54 * Added zebra_string_norm.
56 * Revision 1.26 1999/11/30 13:48:03 adam
57 * Improved installation. Updated for inclusion of YAZ header files.
59 * Revision 1.25 1999/11/04 15:00:45 adam
60 * Implemented delete result set(s).
62 * Revision 1.24 1999/10/14 14:33:50 adam
63 * Added truncation 5=106.
65 * Revision 1.23 1999/09/07 11:36:32 adam
68 * Revision 1.22 1999/08/02 10:13:47 adam
69 * Fixed bug regarding zebra_hits.
71 * Revision 1.21 1999/07/14 10:59:26 adam
72 * Changed functions isc_getmethod, isams_getmethod.
73 * Improved fatal error handling (such as missing EXPLAIN schema).
75 * Revision 1.20 1999/07/06 12:28:04 adam
76 * Updated record index structure. Format includes version ID. Compression
77 * algorithm ID is stored for each record block.
79 * Revision 1.19 1999/05/26 07:49:13 adam
82 * Revision 1.18 1999/05/15 14:36:38 adam
83 * Updated dictionary. Implemented "compression" of dictionary.
85 * Revision 1.17 1999/05/12 13:08:06 adam
86 * First version of ISAMS.
88 * Revision 1.16 1999/02/19 10:38:30 adam
89 * Implemented chdir-setting.
91 * Revision 1.15 1999/02/17 12:18:12 adam
92 * Fixed zebra_close so that a NULL pointer is ignored.
94 * Revision 1.14 1999/02/02 14:51:11 adam
95 * Updated WIN32 code specific sections. Changed header.
97 * Revision 1.13 1998/12/16 12:23:30 adam
98 * Added facility for database name mapping using resource mapdb.
100 * Revision 1.12 1998/11/16 10:18:10 adam
101 * Better error reporting for result sets.
103 * Revision 1.11 1998/10/16 08:14:34 adam
104 * Updated record control system.
106 * Revision 1.10 1998/09/22 10:03:42 adam
107 * Changed result sets to be persistent in the sense that they can
108 * be re-searched if needed.
109 * Fixed memory leak in rsm_or.
111 * Revision 1.9 1998/09/02 13:53:17 adam
112 * Extra parameter decode added to search routines to implement
113 * persistent queries.
115 * Revision 1.8 1998/08/24 17:29:23 adam
118 * Revision 1.7 1998/06/24 12:16:13 adam
119 * Support for relations on text operands. Open range support in
120 * DFA module (i.e. [-j], [g-]).
122 * Revision 1.6 1998/06/22 11:36:47 adam
123 * Added authentication check facility to zebra.
125 * Revision 1.5 1998/06/13 00:14:08 adam
128 * Revision 1.4 1998/06/12 12:22:12 adam
131 * Revision 1.3 1998/05/27 16:57:44 adam
132 * Zebra returns surrogate diagnostic for single records when
135 * Revision 1.2 1998/05/20 10:12:19 adam
136 * Implemented automatic EXPLAIN database maintenance.
137 * Modified Zebra to work with ASN.1 compiled version of YAZ.
139 * Revision 1.1 1998/03/05 08:45:13 adam
140 * New result set model and modular ranking system. Moved towards
141 * descent server API. System information stored as "SGML" records.
155 #include <yaz/diagbib1.h>
159 static void zebra_chdir (ZebraService zh)
161 const char *dir = res_get (zh->res, "chdir");
164 logf (LOG_DEBUG, "chdir %s", dir);
172 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
173 const char *buf, size_t buf_size,
174 const char *databaseName, int delete_flag,
175 int test_mode, int *sysno,
176 int store_keys, int store_data,
177 const char *match_criteria);
179 static int explain_extract (void *handle, Record rec, data1_node *n);
180 static void extract_index (ZebraHandle zh);
182 static void zebra_register_unlock (ZebraHandle zh);
184 static int zebra_register_activate (ZebraService zh, int rw);
185 static int zebra_register_deactivate (ZebraService zh);
187 static int zebra_register_lock (ZebraHandle zh, int rw)
193 if (!zh->service->active)
203 zebra_mutex_cond_lock (&zh->service->session_lock);
205 state = zebra_server_lock_get_state(zh->service, &lastChange);
207 zebra_server_lock (zh->service, state);
218 logf (LOG_LOG, "Register in read/write mode");
219 else if (zh->service->registerState == state)
221 logf (LOG_LOG, "registerChange = %ld lastChange = %ld",
222 (long) zh->service->registerChange, (long)lastChange);
223 if (zh->service->registerChange >= lastChange)
227 logf (LOG_LOG, "Register completely updated since last access");
229 else if (zh->service->registerState == -1)
230 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
233 logf (LOG_LOG, "Register has changed state from %d to %d",
234 zh->service->registerState, state);
235 zh->service->registerChange = lastChange;
237 zebra_register_deactivate (zh->service);
239 zh->service->registerState = state;
241 zebra_register_activate (zh->service, rw);
246 static void zebra_register_unlock (ZebraHandle zh)
248 if (zh->service->registerState != -1)
249 zebra_server_unlock (zh->service, zh->service->registerState);
250 zebra_mutex_cond_unlock (&zh->service->session_lock);
253 logf (LOG_LOG, "user/system: %ld/%ld",
254 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
255 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
260 ZebraHandle zebra_open (ZebraService zs)
268 zh = (ZebraHandle) xmalloc (sizeof(*zh));
269 yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh);
278 zh->admin_databaseName = 0;
280 zebra_mutex_cond_lock (&zs->session_lock);
282 zh->next = zs->sessions;
285 zebra_mutex_cond_unlock (&zs->session_lock);
291 ZebraService zebra_start (const char *configName)
293 ZebraService zh = xmalloc (sizeof(*zh));
295 yaz_log (LOG_LOG, "zebra_start %s", configName);
297 zh->configName = xstrdup(configName);
302 zh->registerState = -1;
303 zh->registerChange = 0;
305 if (!(zh->res = res_open (zh->configName)))
307 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
311 zebra_server_lock_init (zh);
312 zebra_mutex_cond_init (&zh->session_lock);
313 if (!res_get (zh->res, "passwd"))
314 zh->passwd_db = NULL;
317 zh->passwd_db = passwd_db_open ();
319 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
321 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
327 static int zebra_register_activate (ZebraService zh, int rw)
331 yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)",
335 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
336 zh->registerState ? "yes" : "no");
338 zh->dh = data1_create ();
341 zh->bfs = bfs_create (res_get (zh->res, "register"));
344 data1_destroy(zh->dh);
347 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
348 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
349 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
350 zh->recTypes = recTypes_init (zh->dh);
351 recTypes_default_handlers (zh->recTypes);
354 zh->zebra_maps = zebra_maps_open (zh->res);
355 zh->rank_classes = NULL;
368 zebraRankInstall (zh, rank1_class);
370 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
372 logf (LOG_WARN, "rec_open");
375 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
377 logf (LOG_WARN, "dict_open");
380 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
382 logf (LOG_WARN, "sortIdx_open");
385 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
387 struct ISAMS_M_s isams_m;
388 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
389 key_isams_m(zh->res, &isams_m))))
391 logf (LOG_WARN, "isams_open");
396 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
398 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
399 sizeof (struct it_key), zh->res)))
401 logf (LOG_WARN, "is_open");
405 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
407 struct ISAMC_M_s isamc_m;
408 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
409 rw, key_isamc_m(zh->res, &isamc_m))))
411 logf (LOG_WARN, "isc_open");
415 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
417 struct ISAMD_M_s isamd_m;
419 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
420 rw, key_isamd_m(zh->res, &isamd_m))))
422 logf (LOG_WARN, "isamd_open");
427 zh->zei = zebraExplain_open (zh->records, zh->dh,
428 zh->res, rw, 0 /* rGroup */,
432 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
436 yaz_log (LOG_LOG, "zebra_register_activate ok");
440 void zebra_admin_shutdown (ZebraHandle zh)
442 zebra_mutex_cond_lock (&zh->service->session_lock);
443 zh->service->stop_flag = 1;
444 if (!zh->service->sessions)
445 zebra_register_deactivate(zh->service);
446 zh->service->active = 0;
447 zebra_mutex_cond_unlock (&zh->service->session_lock);
450 void zebra_admin_start (ZebraHandle zh)
452 ZebraService zs = zh->service;
454 zebra_mutex_cond_lock (&zs->session_lock);
456 zh->service->active = 1;
457 zebra_mutex_cond_unlock (&zs->session_lock);
460 static int zebra_register_deactivate (ZebraService zs)
465 yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)",
469 yaz_log(LOG_LOG, "zebra_register_deactivate");
473 zebraExplain_close (zs->zei, 0);
474 dict_close (zs->dict);
475 sortIdx_close (zs->sortIdx);
477 isams_close (zs->isams);
482 isc_close (zs->isamc);
484 isamd_close (zs->isamd);
486 rec_close (&zs->records);
488 recTypes_destroy (zs->recTypes);
489 zebra_maps_close (zs->zebra_maps);
490 zebraRankDestroy (zs);
491 bfs_destroy (zs->bfs);
492 data1_destroy (zs->dh);
495 passwd_db_close (zs->passwd_db);
500 void zebra_stop(ZebraService zs)
504 yaz_log (LOG_LOG, "zebra_stop");
506 zebra_mutex_cond_lock (&zs->session_lock);
508 zebra_close (zs->sessions);
510 zebra_mutex_cond_unlock (&zs->session_lock);
512 zebra_mutex_cond_destroy (&zs->session_lock);
514 zebra_register_deactivate(zs);
516 xfree (zs->configName);
520 void zebra_close (ZebraHandle zh)
522 ZebraService zs = zh->service;
523 struct zebra_session **sp;
525 yaz_log (LOG_LOG, "zebra_close zh=%p", zh);
528 resultSetDestroy (zh, -1, 0, 0);
535 xfree (zh->admin_databaseName);
536 zebra_mutex_cond_lock (&zs->session_lock);
548 if (!zs->sessions && zs->stop_flag)
549 zebra_register_deactivate(zs);
550 zebra_mutex_cond_unlock (&zs->session_lock);
554 struct map_baseinfo {
560 char **new_basenames;
564 void map_basenames_func (void *vp, const char *name, const char *value)
566 struct map_baseinfo *p = (struct map_baseinfo *) vp;
568 char fromdb[128], todb[8][128];
571 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
572 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
573 todb[5], todb[6], todb[7]);
577 for (i = 0; i<p->num_bases; i++)
578 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
581 for (i = 0; i < no; i++)
583 if (p->new_num_bases == p->new_num_max)
585 p->new_basenames[(p->new_num_bases)++] =
586 nmem_strdup (p->mem, todb[i]);
592 void map_basenames (ZebraHandle zh, ODR stream,
593 int *num_bases, char ***basenames)
595 struct map_baseinfo info;
596 struct map_baseinfo *p = &info;
600 info.num_bases = *num_bases;
601 info.basenames = *basenames;
602 info.new_num_max = 128;
603 info.new_num_bases = 0;
604 info.new_basenames = (char **)
605 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
606 info.mem = stream->mem;
608 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
610 for (i = 0; i<p->num_bases; i++)
611 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
613 p->new_basenames[(p->new_num_bases)++] =
614 nmem_strdup (p->mem, p->basenames[i]);
616 *num_bases = info.new_num_bases;
617 *basenames = info.new_basenames;
618 for (i = 0; i<*num_bases; i++)
619 logf (LOG_LOG, "base %s", (*basenames)[i]);
622 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
623 Z_RPNQuery *query, int num_bases, char **basenames,
627 if (zebra_register_lock (zh, 0))
629 map_basenames (zh, stream, &num_bases, &basenames);
630 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
632 zebra_register_unlock (zh);
634 logf(LOG_APP,"SEARCH:%d:",zh->hits);
637 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
638 const char *setname, Z_RecordComposition *comp,
639 oid_value input_format, int num_recs,
640 ZebraRetrievalRecord *recs)
645 if (zebra_register_lock (zh, 0))
647 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
648 for (i = 0; i<num_recs; i++)
649 pos_array[i] = recs[i].position;
650 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
653 logf (LOG_DEBUG, "zebraPosSetCreate error");
655 zh->errString = nmem_strdup (stream->mem, setname);
659 for (i = 0; i<num_recs; i++)
665 sprintf (num_str, "%d", pos_array[i]);
667 zh->errString = nmem_strdup (stream->mem, num_str);
673 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
674 stream, input_format, comp,
675 &recs[i].format, &recs[i].buf,
678 recs[i].errString = NULL;
681 zebraPosSetDestroy (zh, poset, num_recs);
683 zebra_register_unlock (zh);
687 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
688 oid_value attributeset,
689 int num_bases, char **basenames,
690 int *position, int *num_entries, ZebraScanEntry **entries,
693 if (zebra_register_lock (zh, 0))
699 map_basenames (zh, stream, &num_bases, &basenames);
700 rpn_scan (zh, stream, zapt, attributeset,
701 num_bases, basenames, position,
702 num_entries, entries, is_partial);
703 zebra_register_unlock (zh);
706 void zebra_sort (ZebraHandle zh, ODR stream,
707 int num_input_setnames, const char **input_setnames,
708 const char *output_setname, Z_SortKeySpecList *sort_sequence,
711 if (zebra_register_lock (zh, 0))
713 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
714 output_setname, sort_sequence, sort_status);
715 zebra_register_unlock (zh);
718 int zebra_deleleResultSet(ZebraHandle zh, int function,
719 int num_setnames, char **setnames,
723 if (zebra_register_lock (zh, 0))
724 return Z_DeleteStatus_systemProblemAtTarget;
727 case Z_DeleteRequest_list:
728 resultSetDestroy (zh, num_setnames, setnames, statuses);
730 case Z_DeleteRequest_all:
731 resultSetDestroy (zh, -1, 0, statuses);
734 zebra_register_unlock (zh);
735 status = Z_DeleteStatus_success;
736 for (i = 0; i<num_setnames; i++)
737 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
738 status = statuses[i];
742 int zebra_errCode (ZebraHandle zh)
747 const char *zebra_errString (ZebraHandle zh)
749 return diagbib1_str (zh->errCode);
752 char *zebra_errAdd (ZebraHandle zh)
754 return zh->errString;
757 int zebra_hits (ZebraHandle zh)
762 int zebra_auth (ZebraService zh, const char *user, const char *pass)
764 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
766 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
770 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
774 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
776 if (zebra_register_lock (zh, 1))
778 xfree (zh->admin_databaseName);
779 zh->admin_databaseName = xstrdup(database);
782 void zebra_admin_import_end (ZebraHandle zh)
784 zebraExplain_flush (zh->service->zei, 1, zh);
786 zebra_register_unlock (zh);
789 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
793 if (zh->service->active < 2)
795 for (i = 0; i<segment->num_segmentRecords; i++)
797 Z_NamePlusRecord *npr = segment->segmentRecords[i];
798 const char *databaseName = npr->databaseName;
801 databaseName = zh->admin_databaseName;
802 printf ("--------------%d--------------------\n", i);
803 if (npr->which == Z_NamePlusRecord_intermediateFragment)
805 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
806 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
808 Odr_oct *oct = fragment->u.notExternallyTagged;
809 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
813 extract_rec_in_mem (zh, "grs.sgml",
821 0 /* match criteria */);
827 void zebra_admin_create (ZebraHandle zh, const char *database)
829 ZebraService zs = zh->service;
830 if (zebra_register_lock(zh, 1))
835 /* announce database */
836 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
839 zh->errString = "Database already exist";
841 zebraExplain_flush (zh->service->zei, 1, zh);
843 zebra_register_unlock(zh);
846 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
847 const char *input_str, int input_len,
848 char *output_str, int output_len)
851 if (!zh->service->zebra_maps)
853 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
854 input_str, input_len);
857 if (wrbuf_len(wrbuf) >= output_len)
859 if (wrbuf_len(wrbuf))
860 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
861 output_str[wrbuf_len(wrbuf)] = '\0';
862 return wrbuf_len(wrbuf);
865 static void extract_init (struct recExtractCtrl *p, RecWord *w)
867 w->zebra_maps = p->zebra_maps;
868 w->seqnos = p->seqno;
869 w->attrSet = VAL_BIB1;
875 static void extract_add_index_string (RecWord *p, const char *string,
879 unsigned char attrSet;
880 unsigned short attrUse;
883 int *pseqno = &p->seqnos[p->reg_type];
884 ZebraHandle zh = p->extractCtrl->handle;
885 struct recKeys *keys = &zh->keys;
887 if (keys->buf_used+1024 > keys->buf_max)
891 b = (char *) xmalloc (keys->buf_max += 128000);
892 if (keys->buf_used > 0)
893 memcpy (b, keys->buf, keys->buf_used);
897 dst = keys->buf + keys->buf_used;
899 attrSet = p->attrSet;
900 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
903 keys->prevAttrSet = attrSet;
904 attrUse = p->attrUse;
905 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
908 keys->prevAttrUse = attrUse;
910 diff = 1 + *pseqno - keys->prevSeqNo;
911 if (diff >= 1 && diff <= 15)
916 keys->prevSeqNo = *pseqno;
922 memcpy (dst, &attrSet, sizeof(attrSet));
923 dst += sizeof(attrSet);
927 memcpy (dst, &attrUse, sizeof(attrUse));
928 dst += sizeof(attrUse);
930 *dst++ = p->reg_type;
931 memcpy (dst, string, length);
937 memcpy (dst, pseqno, sizeof(*pseqno));
938 dst += sizeof(*pseqno);
940 keys->buf_used = dst - keys->buf;
945 static void extract_add_sort_string (RecWord *p, const char *string,
949 ZebraHandle zh = p->extractCtrl->handle;
950 struct sortKey *sortKeys = zh->sortKeys;
952 for (sk = sortKeys; sk; sk = sk->next)
953 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
956 sk = (struct sortKey *) xmalloc (sizeof(*sk));
960 sk->string = (char *) xmalloc (length);
962 memcpy (sk->string, string, length);
964 sk->attrSet = p->attrSet;
965 sk->attrUse = p->attrUse;
968 static void extract_add_string (RecWord *p, const char *string, int length)
971 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
972 extract_add_sort_string (p, string, length);
974 extract_add_index_string (p, string, length);
977 static void extract_add_incomplete_field (RecWord *p)
979 const char *b = p->string;
980 int remain = p->length;
981 const char **map = 0;
984 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
988 char buf[IT_MAX_WORD+1];
992 while (map && *map && **map == *CHR_SPACE)
994 remain = p->length - (b - p->string);
996 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1003 while (map && *map && **map != *CHR_SPACE)
1005 const char *cp = *map;
1007 while (i < IT_MAX_WORD && *cp)
1009 remain = p->length - (b - p->string);
1011 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1017 extract_add_string (p, buf, i);
1019 (p->seqnos[p->reg_type])++; /* to separate this from next one */
1022 static void extract_add_complete_field (RecWord *p)
1024 const char *b = p->string;
1025 char buf[IT_MAX_WORD+1];
1026 const char **map = 0;
1027 int i = 0, remain = p->length;
1030 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1032 while (remain > 0 && i < IT_MAX_WORD)
1034 while (map && *map && **map == *CHR_SPACE)
1036 remain = p->length - (b - p->string);
1038 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1045 if (i && i < IT_MAX_WORD)
1046 buf[i++] = *CHR_SPACE;
1047 while (map && *map && **map != *CHR_SPACE)
1049 const char *cp = *map;
1051 if (i >= IT_MAX_WORD)
1053 while (i < IT_MAX_WORD && *cp)
1055 remain = p->length - (b - p->string);
1057 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1065 extract_add_string (p, buf, i);
1068 static void extract_token_add (RecWord *p)
1071 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1072 p->string, p->length)))
1074 p->string = wrbuf_buf(wrbuf);
1075 p->length = wrbuf_len(wrbuf);
1077 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1078 extract_add_complete_field (p);
1080 extract_add_incomplete_field(p);
1083 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1085 ZebraHandle zh = (ZebraHandle) (p->handle);
1086 zebraExplain_addSchema (zh->service->zei, oid);
1089 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1090 int cmd, struct sortKey **skp)
1092 struct sortKey *sk = *skp;
1093 SortIdx sortIdx = zh->service->sortIdx;
1095 sortIdx_sysno (sortIdx, sysno);
1098 struct sortKey *sk_next = sk->next;
1099 sortIdx_type (sortIdx, sk->attrUse);
1100 sortIdx_add (sortIdx, sk->string, sk->length);
1108 struct encode_info {
1115 void encode_key_init (struct encode_info *i)
1122 char *encode_key_int (int d, char *bp)
1126 else if (d <= 16383)
1128 *bp++ = 64 + (d>>8);
1131 else if (d <= 4194303)
1133 *bp++ = 128 + (d>>16);
1134 *bp++ = (d>>8) & 255;
1139 *bp++ = 192 + (d>>24);
1140 *bp++ = (d>>16) & 255;
1141 *bp++ = (d>>8) & 255;
1147 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1152 while ((*bp++ = *k++))
1154 memcpy (&key, k+1, sizeof(struct it_key));
1155 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1156 if (i->sysno != key.sysno)
1158 i->sysno = key.sysno;
1161 else if (!i->seqno && !key.seqno && i->cmd == *k)
1163 bp = encode_key_int (key.seqno - i->seqno, bp);
1164 i->seqno = key.seqno;
1166 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1168 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1173 static void extract_flushWriteKeys (ZebraHandle zh)
1176 char out_fname[200];
1178 struct encode_info encode_info;
1179 int ptr_i = zh->ptr_i;
1183 if (!zh->key_buf || ptr_i <= 0)
1186 (zh->key_file_no)++;
1187 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1189 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1191 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1193 if (!(outf = fopen (out_fname, "wb")))
1195 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1198 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1199 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1201 encode_key_init (&encode_info);
1202 encode_key_write (cp, &encode_info, outf);
1206 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1207 if (strcmp (cp, prevcp))
1209 encode_key_init (&encode_info);
1210 encode_key_write (cp, &encode_info, outf);
1214 encode_key_write (cp + strlen(cp), &encode_info, outf);
1217 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1218 extract_get_fname_tmp (out_fname, key_file_no);
1220 if (!(outf = fopen (out_fname, "wb")))
1222 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1225 logf (LOG_LOG, "writing section %d", key_file_no);
1227 prevcp = key_buf[ptr_top-i];
1229 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1231 key_y_len = strlen(prevcp)+1;
1233 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1234 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1236 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1237 sizeof(char*), key_y_compare);
1238 cp = key_buf[ptr_top-ptr_i];
1240 encode_key_init (&encode_info);
1241 encode_key_write (cp, &encode_info, outf);
1244 cp = key_buf[ptr_top-ptr_i];
1245 encode_key_write (cp+key_y_len, &encode_info, outf);
1249 prevcp = key_buf[ptr_top-ptr_i];
1254 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1257 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1259 zh->key_buf_used = 0;
1262 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1263 int cmd, struct recKeys *reckeys)
1265 unsigned char attrSet = (unsigned char) -1;
1266 unsigned short attrUse = (unsigned short) -1;
1269 ZebraExplainInfo zei = zh->service->zei;
1273 int mem = 8*1024*1024;
1274 zh->key_buf = (char**) xmalloc (mem);
1275 zh->ptr_top = mem/sizeof(char*);
1277 zh->key_buf_used = 0;
1278 zh->key_file_no = 0;
1280 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1281 while (off < reckeys->buf_used)
1283 const char *src = reckeys->buf + off;
1291 memcpy (&attrSet, src, sizeof(attrSet));
1292 src += sizeof(attrSet);
1296 memcpy (&attrUse, src, sizeof(attrUse));
1297 src += sizeof(attrUse);
1299 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1300 extract_flushWriteKeys (zh);
1302 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1303 (char*)zh->key_buf + zh->key_buf_used;
1305 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1307 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1310 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1313 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1315 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1316 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1319 seqno += ((lead>>2) & 15)-1;
1322 memcpy (&seqno, src, sizeof(seqno));
1323 src += sizeof(seqno);
1327 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1328 (zh->key_buf_used) += sizeof(key);
1329 off = src - reckeys->buf;
1331 assert (off == reckeys->buf_used);
1334 static void extract_index (ZebraHandle zh)
1336 extract_flushWriteKeys (zh);
1337 zebra_index_merge (zh);
1340 static int explain_extract (void *handle, Record rec, data1_node *n)
1342 ZebraHandle zh = (ZebraHandle) handle;
1343 struct recExtractCtrl extractCtrl;
1346 if (zebraExplain_curDatabase (zh->service->zei,
1347 rec->info[recInfo_databaseName]))
1350 if (zebraExplain_newDatabase (zh->service->zei,
1351 rec->info[recInfo_databaseName], 0))
1355 zh->keys.buf_used = 0;
1356 zh->keys.prevAttrUse = -1;
1357 zh->keys.prevAttrSet = -1;
1358 zh->keys.prevSeqNo = 0;
1361 extractCtrl.init = extract_init;
1362 extractCtrl.tokenAdd = extract_token_add;
1363 extractCtrl.schemaAdd = extract_schema_add;
1364 extractCtrl.dh = zh->service->dh;
1365 for (i = 0; i<256; i++)
1366 extractCtrl.seqno[i] = 0;
1367 extractCtrl.zebra_maps = zh->service->zebra_maps;
1368 extractCtrl.flagShowRecords = 0;
1369 extractCtrl.handle = handle;
1371 grs_extract_tree(&extractCtrl, n);
1373 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1375 if (rec->size[recInfo_delKeys])
1377 struct recKeys delkeys;
1378 struct sortKey *sortKeys = 0;
1380 delkeys.buf_used = rec->size[recInfo_delKeys];
1381 delkeys.buf = rec->info[recInfo_delKeys];
1382 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1383 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1385 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1386 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1388 xfree (rec->info[recInfo_delKeys]);
1389 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1390 rec->info[recInfo_delKeys] = zh->keys.buf;
1391 zh->keys.buf = NULL;
1392 zh->keys.buf_max = 0;
1396 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1397 const char *buf, size_t buf_size,
1398 const char *databaseName, int delete_flag,
1399 int test_mode, int *sysno,
1400 int store_keys, int store_data,
1401 const char *match_criteria)
1403 RecordAttr *recordAttr;
1404 struct recExtractCtrl extractCtrl;
1409 const char *fname = "<no file>";
1411 long recordOffset = 0;
1412 struct zebra_fetch_control fc;
1415 fc.record_int_buf = buf;
1416 fc.record_int_len = buf_size;
1417 fc.record_int_pos = 0;
1419 fc.record_offset = 0;
1421 extractCtrl.offset = 0;
1422 extractCtrl.readf = zebra_record_int_read;
1423 extractCtrl.seekf = zebra_record_int_seek;
1424 extractCtrl.tellf = zebra_record_int_tell;
1425 extractCtrl.endf = zebra_record_int_end;
1426 extractCtrl.fh = &fc;
1428 /* announce database */
1429 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1431 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1435 recType_byName (zh->service->recTypes, recordType, subType,
1438 logf (LOG_WARN, "No such record type: %s", recordType);
1442 zh->keys.buf_used = 0;
1443 zh->keys.prevAttrUse = -1;
1444 zh->keys.prevAttrSet = -1;
1445 zh->keys.prevSeqNo = 0;
1448 extractCtrl.subType = subType;
1449 extractCtrl.init = extract_init;
1450 extractCtrl.tokenAdd = extract_token_add;
1451 extractCtrl.schemaAdd = extract_schema_add;
1452 extractCtrl.dh = zh->service->dh;
1453 extractCtrl.handle = zh;
1454 extractCtrl.zebra_maps = zh->service->zebra_maps;
1455 extractCtrl.flagShowRecords = 0;
1456 for (i = 0; i<256; i++)
1458 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1459 extractCtrl.seqno[i] = 1;
1461 extractCtrl.seqno[i] = 0;
1464 r = (*recType->extract)(clientData, &extractCtrl);
1466 if (r == RECCTRL_EXTRACT_EOF)
1468 else if (r == RECCTRL_EXTRACT_ERROR)
1470 /* error occured during extraction ... */
1472 yaz_log (LOG_WARN, "extract error");
1474 if (rGroup->flagRw &&
1475 records_processed < rGroup->fileVerboseLimit)
1477 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1478 fname, (long) recordOffset);
1483 if (zh->keys.buf_used == 0)
1485 /* the extraction process returned no information - the record
1486 is probably empty - unless flagShowRecords is in use */
1489 logf (LOG_WARN, "No keys generated for record");
1490 logf (LOG_WARN, " The file is probably empty");
1493 /* match criteria */
1500 logf (LOG_LOG, "delete %s %s %ld", recordType,
1501 fname, (long) recordOffset);
1502 logf (LOG_WARN, "cannot delete record above (seems new)");
1505 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1506 (long) recordOffset);
1507 rec = rec_new (zh->service->records);
1509 *sysno = rec->sysno;
1511 recordAttr = rec_init_attr (zh->service->zei, rec);
1516 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1519 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1520 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1524 /* record already exists */
1525 struct recKeys delkeys;
1527 rec = rec_get (zh->service->records, *sysno);
1530 recordAttr = rec_init_attr (zh->service->zei, rec);
1532 if (recordAttr->runNumber ==
1533 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1535 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1536 fname, (long) recordOffset);
1540 delkeys.buf_used = rec->size[recInfo_delKeys];
1541 delkeys.buf = rec->info[recInfo_delKeys];
1542 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1543 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1546 /* record going to be deleted */
1547 if (!delkeys.buf_used)
1549 logf (LOG_LOG, "delete %s %s %ld", recordType,
1550 fname, (long) recordOffset);
1551 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1555 logf (LOG_LOG, "delete %s %s %ld", recordType,
1556 fname, (long) recordOffset);
1559 dict_delete (matchDict, matchStr);
1561 rec_del (zh->service->records, &rec);
1568 /* record going to be updated */
1569 if (!delkeys.buf_used)
1571 logf (LOG_LOG, "update %s %s %ld", recordType,
1572 fname, (long) recordOffset);
1573 logf (LOG_WARN, "cannot update file above, storeKeys false");
1577 logf (LOG_LOG, "update %s %s %ld", recordType,
1578 fname, (long) recordOffset);
1579 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1583 /* update file type */
1584 xfree (rec->info[recInfo_fileType]);
1585 rec->info[recInfo_fileType] =
1586 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1588 /* update filename */
1589 xfree (rec->info[recInfo_filename]);
1590 rec->info[recInfo_filename] =
1591 rec_strdup (fname, &rec->size[recInfo_filename]);
1593 /* update delete keys */
1594 xfree (rec->info[recInfo_delKeys]);
1595 if (zh->keys.buf_used > 0 && store_keys == 1)
1597 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1598 rec->info[recInfo_delKeys] = zh->keys.buf;
1599 zh->keys.buf = NULL;
1600 zh->keys.buf_max = 0;
1604 rec->info[recInfo_delKeys] = NULL;
1605 rec->size[recInfo_delKeys] = 0;
1608 /* save file size of original record */
1609 zebraExplain_recordBytesIncrement (zh->service->zei,
1610 - recordAttr->recordSize);
1612 recordAttr->recordSize = fi->file_moffset - recordOffset;
1613 if (!recordAttr->recordSize)
1614 recordAttr->recordSize = fi->file_max - recordOffset;
1616 recordAttr->recordSize = buf_size;
1618 zebraExplain_recordBytesIncrement (zh->service->zei,
1619 recordAttr->recordSize);
1621 /* set run-number for this record */
1622 recordAttr->runNumber =
1623 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1625 /* update store data */
1626 xfree (rec->info[recInfo_storeData]);
1627 if (store_data == 1)
1629 rec->size[recInfo_storeData] = recordAttr->recordSize;
1630 rec->info[recInfo_storeData] = (char *)
1631 xmalloc (recordAttr->recordSize);
1633 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1635 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1637 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1638 (long) recordOffset, fname);
1641 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1642 < recordAttr->recordSize)
1644 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1645 recordAttr->recordSize, fname);
1652 rec->info[recInfo_storeData] = NULL;
1653 rec->size[recInfo_storeData] = 0;
1655 /* update database name */
1656 xfree (rec->info[recInfo_databaseName]);
1657 rec->info[recInfo_databaseName] =
1658 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1661 recordAttr->recordOffset = recordOffset;
1663 /* commit this record */
1664 rec_put (zh->service->records, &rec);