2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.43 2000-12-05 12:22:53 adam
7 * Termlist source implemented (so that we can index values of XML/SGML
10 * Revision 1.42 2000/12/05 10:01:44 adam
11 * Fixed bug regarding user-defined attribute sets.
13 * Revision 1.41 2000/12/01 17:59:08 adam
14 * Fixed bug regarding online updates on WIN32.
15 * When zebra.cfg is not available the server will not abort.
17 * Revision 1.40 2000/11/29 15:21:31 adam
18 * Fixed problem with passwd db.
20 * Revision 1.39 2000/11/29 14:24:01 adam
21 * Script configure uses yaz pthreads options. Added locking for
22 * zebra_register_{lock,unlock}.
24 * Revision 1.38 2000/11/08 13:46:58 adam
25 * Fixed scan: server could break if bad attribute/database was selected.
26 * Work on remote update.
28 * Revision 1.37 2000/10/17 12:37:09 adam
29 * Fixed notification of live-updates. Fixed minor problem with mf_init
30 * where it didn't handle shadow area file names correctly.
32 * Revision 1.36 2000/09/06 08:59:36 adam
33 * Using read-only (for now) for server.
35 * Revision 1.35 2000/07/07 12:49:20 adam
36 * Optimized resultSetInsert{Rank,Sort}.
38 * Revision 1.34 2000/06/09 13:56:38 ian
39 * Added some logging on Authentication and searches.
41 * Revision 1.33 2000/05/18 12:01:36 adam
42 * System call times(2) used again. More 64-bit fixes.
44 * Revision 1.32 2000/04/19 14:35:59 adam
45 * WIN32 update (this version is known not to work on Windows).
47 * Revision 1.31 2000/04/05 10:07:02 adam
48 * Minor zebra compile fix.
50 * Revision 1.30 2000/04/05 09:49:35 adam
51 * On Unix, zebra/z'mbol uses automake.
53 * Revision 1.29 2000/03/20 19:08:36 adam
54 * Added remote record import using Z39.50 extended services and Segment
57 * Revision 1.28 2000/03/15 15:00:30 adam
58 * First work on threaded version.
60 * Revision 1.27 2000/02/24 12:31:17 adam
61 * Added zebra_string_norm.
63 * Revision 1.26 1999/11/30 13:48:03 adam
64 * Improved installation. Updated for inclusion of YAZ header files.
66 * Revision 1.25 1999/11/04 15:00:45 adam
67 * Implemented delete result set(s).
69 * Revision 1.24 1999/10/14 14:33:50 adam
70 * Added truncation 5=106.
72 * Revision 1.23 1999/09/07 11:36:32 adam
75 * Revision 1.22 1999/08/02 10:13:47 adam
76 * Fixed bug regarding zebra_hits.
78 * Revision 1.21 1999/07/14 10:59:26 adam
79 * Changed functions isc_getmethod, isams_getmethod.
80 * Improved fatal error handling (such as missing EXPLAIN schema).
82 * Revision 1.20 1999/07/06 12:28:04 adam
83 * Updated record index structure. Format includes version ID. Compression
84 * algorithm ID is stored for each record block.
86 * Revision 1.19 1999/05/26 07:49:13 adam
89 * Revision 1.18 1999/05/15 14:36:38 adam
90 * Updated dictionary. Implemented "compression" of dictionary.
92 * Revision 1.17 1999/05/12 13:08:06 adam
93 * First version of ISAMS.
95 * Revision 1.16 1999/02/19 10:38:30 adam
96 * Implemented chdir-setting.
98 * Revision 1.15 1999/02/17 12:18:12 adam
99 * Fixed zebra_close so that a NULL pointer is ignored.
101 * Revision 1.14 1999/02/02 14:51:11 adam
102 * Updated WIN32 code specific sections. Changed header.
104 * Revision 1.13 1998/12/16 12:23:30 adam
105 * Added facility for database name mapping using resource mapdb.
107 * Revision 1.12 1998/11/16 10:18:10 adam
108 * Better error reporting for result sets.
110 * Revision 1.11 1998/10/16 08:14:34 adam
111 * Updated record control system.
113 * Revision 1.10 1998/09/22 10:03:42 adam
114 * Changed result sets to be persistent in the sense that they can
115 * be re-searched if needed.
116 * Fixed memory leak in rsm_or.
118 * Revision 1.9 1998/09/02 13:53:17 adam
119 * Extra parameter decode added to search routines to implement
120 * persistent queries.
122 * Revision 1.8 1998/08/24 17:29:23 adam
125 * Revision 1.7 1998/06/24 12:16:13 adam
126 * Support for relations on text operands. Open range support in
127 * DFA module (i.e. [-j], [g-]).
129 * Revision 1.6 1998/06/22 11:36:47 adam
130 * Added authentication check facility to zebra.
132 * Revision 1.5 1998/06/13 00:14:08 adam
135 * Revision 1.4 1998/06/12 12:22:12 adam
138 * Revision 1.3 1998/05/27 16:57:44 adam
139 * Zebra returns surrogate diagnostic for single records when
142 * Revision 1.2 1998/05/20 10:12:19 adam
143 * Implemented automatic EXPLAIN database maintenance.
144 * Modified Zebra to work with ASN.1 compiled version of YAZ.
146 * Revision 1.1 1998/03/05 08:45:13 adam
147 * New result set model and modular ranking system. Moved towards
148 * descent server API. System information stored as "SGML" records.
162 #include <yaz/diagbib1.h>
166 static void zebra_chdir (ZebraService zh)
168 const char *dir = res_get (zh->res, "chdir");
171 logf (LOG_DEBUG, "chdir %s", dir);
179 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
180 const char *buf, size_t buf_size,
181 const char *databaseName, int delete_flag,
182 int test_mode, int *sysno,
183 int store_keys, int store_data,
184 const char *match_criteria);
186 static int explain_extract (void *handle, Record rec, data1_node *n);
187 static void extract_index (ZebraHandle zh);
189 static void zebra_register_unlock (ZebraHandle zh);
191 static int zebra_register_activate (ZebraService zh, int rw);
192 static int zebra_register_deactivate (ZebraService zh);
194 static int zebra_register_lock (ZebraHandle zh, int rw)
200 if (!zh->service->active)
210 zebra_mutex_cond_lock (&zh->service->session_lock);
212 state = zebra_server_lock_get_state(zh->service, &lastChange);
214 zebra_server_lock (zh->service, state);
225 logf (LOG_LOG, "Register in read/write mode");
226 else if (zh->service->registerState == state)
228 logf (LOG_DEBUG, "registerChange = %ld lastChange = %ld",
229 (long) zh->service->registerChange, (long)lastChange);
230 if (zh->service->registerChange >= lastChange)
234 logf (LOG_LOG, "Register completely updated since last access");
236 else if (zh->service->registerState == -1)
237 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
240 logf (LOG_LOG, "Register has changed state from %d to %d",
241 zh->service->registerState, state);
242 zh->service->registerChange = lastChange;
244 zebra_register_deactivate (zh->service);
246 zh->service->registerState = state;
248 zebra_register_activate (zh->service, rw);
253 static void zebra_register_unlock (ZebraHandle zh)
255 if (zh->service->registerState != -1)
256 zebra_server_unlock (zh->service, zh->service->registerState);
257 zebra_mutex_cond_unlock (&zh->service->session_lock);
260 logf (LOG_LOG, "user/system: %ld/%ld",
261 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
262 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
267 ZebraHandle zebra_open (ZebraService zs)
275 zh = (ZebraHandle) xmalloc (sizeof(*zh));
276 yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh);
285 zh->admin_databaseName = 0;
287 zebra_mutex_cond_lock (&zs->session_lock);
289 zh->next = zs->sessions;
292 zebra_mutex_cond_unlock (&zs->session_lock);
298 ZebraService zebra_start (const char *configName)
300 ZebraService zh = xmalloc (sizeof(*zh));
302 yaz_log (LOG_LOG, "zebra_start %s", configName);
304 zh->configName = xstrdup(configName);
309 zh->registerState = -1;
310 zh->registerChange = 0;
312 if (!(zh->res = res_open (zh->configName)))
314 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
318 zebra_server_lock_init (zh);
319 zebra_mutex_cond_init (&zh->session_lock);
320 if (!res_get (zh->res, "passwd"))
321 zh->passwd_db = NULL;
324 zh->passwd_db = passwd_db_open ();
326 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
328 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
334 static int zebra_register_activate (ZebraService zh, int rw)
338 yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)",
342 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
343 zh->registerState ? "yes" : "no");
345 zh->dh = data1_create ();
348 zh->bfs = bfs_create (res_get (zh->res, "register"));
351 data1_destroy(zh->dh);
354 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
355 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
356 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
357 zh->recTypes = recTypes_init (zh->dh);
358 recTypes_default_handlers (zh->recTypes);
361 zh->zebra_maps = zebra_maps_open (zh->res);
362 zh->rank_classes = NULL;
375 zebraRankInstall (zh, rank1_class);
377 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
379 logf (LOG_WARN, "rec_open");
382 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
384 logf (LOG_WARN, "dict_open");
387 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
389 logf (LOG_WARN, "sortIdx_open");
392 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
394 struct ISAMS_M_s isams_m;
395 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
396 key_isams_m(zh->res, &isams_m))))
398 logf (LOG_WARN, "isams_open");
403 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
405 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
406 sizeof (struct it_key), zh->res)))
408 logf (LOG_WARN, "is_open");
412 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
414 struct ISAMC_M_s isamc_m;
415 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
416 rw, key_isamc_m(zh->res, &isamc_m))))
418 logf (LOG_WARN, "isc_open");
422 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
424 struct ISAMD_M_s isamd_m;
426 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
427 rw, key_isamd_m(zh->res, &isamd_m))))
429 logf (LOG_WARN, "isamd_open");
434 zh->zei = zebraExplain_open (zh->records, zh->dh,
435 zh->res, rw, 0 /* rGroup */,
439 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
443 yaz_log (LOG_LOG, "zebra_register_activate ok");
447 void zebra_admin_shutdown (ZebraHandle zh)
449 zebra_mutex_cond_lock (&zh->service->session_lock);
450 zh->service->stop_flag = 1;
451 if (!zh->service->sessions)
452 zebra_register_deactivate(zh->service);
453 zh->service->active = 0;
454 zebra_mutex_cond_unlock (&zh->service->session_lock);
457 void zebra_admin_start (ZebraHandle zh)
459 ZebraService zs = zh->service;
461 zebra_mutex_cond_lock (&zs->session_lock);
463 zh->service->active = 1;
464 zebra_mutex_cond_unlock (&zs->session_lock);
467 static int zebra_register_deactivate (ZebraService zs)
472 yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)",
476 yaz_log(LOG_LOG, "zebra_register_deactivate");
480 zebraExplain_close (zs->zei, 0);
481 dict_close (zs->dict);
482 sortIdx_close (zs->sortIdx);
484 isams_close (zs->isams);
489 isc_close (zs->isamc);
491 isamd_close (zs->isamd);
493 rec_close (&zs->records);
495 recTypes_destroy (zs->recTypes);
496 zebra_maps_close (zs->zebra_maps);
497 zebraRankDestroy (zs);
498 bfs_destroy (zs->bfs);
499 data1_destroy (zs->dh);
502 passwd_db_close (zs->passwd_db);
507 void zebra_stop(ZebraService zs)
511 yaz_log (LOG_LOG, "zebra_stop");
513 zebra_mutex_cond_lock (&zs->session_lock);
515 zebra_close (zs->sessions);
517 zebra_mutex_cond_unlock (&zs->session_lock);
519 zebra_mutex_cond_destroy (&zs->session_lock);
521 zebra_register_deactivate(zs);
523 xfree (zs->configName);
527 void zebra_close (ZebraHandle zh)
529 ZebraService zs = zh->service;
530 struct zebra_session **sp;
532 yaz_log (LOG_LOG, "zebra_close zh=%p", zh);
535 resultSetDestroy (zh, -1, 0, 0);
542 xfree (zh->admin_databaseName);
543 zebra_mutex_cond_lock (&zs->session_lock);
555 if (!zs->sessions && zs->stop_flag)
556 zebra_register_deactivate(zs);
557 zebra_mutex_cond_unlock (&zs->session_lock);
561 struct map_baseinfo {
567 char **new_basenames;
571 void map_basenames_func (void *vp, const char *name, const char *value)
573 struct map_baseinfo *p = (struct map_baseinfo *) vp;
575 char fromdb[128], todb[8][128];
578 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
579 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
580 todb[5], todb[6], todb[7]);
584 for (i = 0; i<p->num_bases; i++)
585 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
588 for (i = 0; i < no; i++)
590 if (p->new_num_bases == p->new_num_max)
592 p->new_basenames[(p->new_num_bases)++] =
593 nmem_strdup (p->mem, todb[i]);
599 void map_basenames (ZebraHandle zh, ODR stream,
600 int *num_bases, char ***basenames)
602 struct map_baseinfo info;
603 struct map_baseinfo *p = &info;
607 info.num_bases = *num_bases;
608 info.basenames = *basenames;
609 info.new_num_max = 128;
610 info.new_num_bases = 0;
611 info.new_basenames = (char **)
612 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
613 info.mem = stream->mem;
615 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
617 for (i = 0; i<p->num_bases; i++)
618 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
620 p->new_basenames[(p->new_num_bases)++] =
621 nmem_strdup (p->mem, p->basenames[i]);
623 *num_bases = info.new_num_bases;
624 *basenames = info.new_basenames;
625 for (i = 0; i<*num_bases; i++)
626 logf (LOG_LOG, "base %s", (*basenames)[i]);
629 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
630 Z_RPNQuery *query, int num_bases, char **basenames,
634 if (zebra_register_lock (zh, 0))
636 map_basenames (zh, stream, &num_bases, &basenames);
637 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
639 zebra_register_unlock (zh);
641 logf(LOG_APP,"SEARCH:%d:",zh->hits);
644 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
645 const char *setname, Z_RecordComposition *comp,
646 oid_value input_format, int num_recs,
647 ZebraRetrievalRecord *recs)
652 if (zebra_register_lock (zh, 0))
654 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
655 for (i = 0; i<num_recs; i++)
656 pos_array[i] = recs[i].position;
657 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
660 logf (LOG_DEBUG, "zebraPosSetCreate error");
662 zh->errString = nmem_strdup (stream->mem, setname);
666 for (i = 0; i<num_recs; i++)
672 sprintf (num_str, "%d", pos_array[i]);
674 zh->errString = nmem_strdup (stream->mem, num_str);
680 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
681 stream, input_format, comp,
682 &recs[i].format, &recs[i].buf,
685 recs[i].errString = NULL;
688 zebraPosSetDestroy (zh, poset, num_recs);
690 zebra_register_unlock (zh);
694 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
695 oid_value attributeset,
696 int num_bases, char **basenames,
697 int *position, int *num_entries, ZebraScanEntry **entries,
700 if (zebra_register_lock (zh, 0))
706 map_basenames (zh, stream, &num_bases, &basenames);
707 rpn_scan (zh, stream, zapt, attributeset,
708 num_bases, basenames, position,
709 num_entries, entries, is_partial);
710 zebra_register_unlock (zh);
713 void zebra_sort (ZebraHandle zh, ODR stream,
714 int num_input_setnames, const char **input_setnames,
715 const char *output_setname, Z_SortKeySpecList *sort_sequence,
718 if (zebra_register_lock (zh, 0))
720 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
721 output_setname, sort_sequence, sort_status);
722 zebra_register_unlock (zh);
725 int zebra_deleleResultSet(ZebraHandle zh, int function,
726 int num_setnames, char **setnames,
730 if (zebra_register_lock (zh, 0))
731 return Z_DeleteStatus_systemProblemAtTarget;
734 case Z_DeleteRequest_list:
735 resultSetDestroy (zh, num_setnames, setnames, statuses);
737 case Z_DeleteRequest_all:
738 resultSetDestroy (zh, -1, 0, statuses);
741 zebra_register_unlock (zh);
742 status = Z_DeleteStatus_success;
743 for (i = 0; i<num_setnames; i++)
744 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
745 status = statuses[i];
749 int zebra_errCode (ZebraHandle zh)
754 const char *zebra_errString (ZebraHandle zh)
756 return diagbib1_str (zh->errCode);
759 char *zebra_errAdd (ZebraHandle zh)
761 return zh->errString;
764 int zebra_hits (ZebraHandle zh)
769 int zebra_auth (ZebraService zh, const char *user, const char *pass)
771 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
773 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
777 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
781 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
783 if (zebra_register_lock (zh, 1))
785 xfree (zh->admin_databaseName);
786 zh->admin_databaseName = xstrdup(database);
789 void zebra_admin_import_end (ZebraHandle zh)
791 zebraExplain_flush (zh->service->zei, 1, zh);
793 zebra_register_unlock (zh);
796 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
800 if (zh->service->active < 2)
802 for (i = 0; i<segment->num_segmentRecords; i++)
804 Z_NamePlusRecord *npr = segment->segmentRecords[i];
805 const char *databaseName = npr->databaseName;
808 databaseName = zh->admin_databaseName;
809 printf ("--------------%d--------------------\n", i);
810 if (npr->which == Z_NamePlusRecord_intermediateFragment)
812 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
813 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
815 Odr_oct *oct = fragment->u.notExternallyTagged;
816 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
820 extract_rec_in_mem (zh, "grs.sgml",
828 0 /* match criteria */);
834 void zebra_admin_create (ZebraHandle zh, const char *database)
836 ZebraService zs = zh->service;
837 if (zebra_register_lock(zh, 1))
842 /* announce database */
843 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
846 zh->errString = "Database already exist";
848 zebraExplain_flush (zh->service->zei, 1, zh);
850 zebra_register_unlock(zh);
853 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
854 const char *input_str, int input_len,
855 char *output_str, int output_len)
858 if (!zh->service->zebra_maps)
860 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
861 input_str, input_len);
864 if (wrbuf_len(wrbuf) >= output_len)
866 if (wrbuf_len(wrbuf))
867 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
868 output_str[wrbuf_len(wrbuf)] = '\0';
869 return wrbuf_len(wrbuf);
872 static void extract_init (struct recExtractCtrl *p, RecWord *w)
874 w->zebra_maps = p->zebra_maps;
875 w->seqnos = p->seqno;
876 w->attrSet = VAL_BIB1;
882 static void extract_add_index_string (RecWord *p, const char *string,
886 unsigned char attrSet;
887 unsigned short attrUse;
890 int *pseqno = &p->seqnos[p->reg_type];
891 ZebraHandle zh = p->extractCtrl->handle;
892 ZebraExplainInfo zei = zh->service->zei;
893 struct recKeys *keys = &zh->keys;
895 if (keys->buf_used+1024 > keys->buf_max)
899 b = (char *) xmalloc (keys->buf_max += 128000);
900 if (keys->buf_used > 0)
901 memcpy (b, keys->buf, keys->buf_used);
905 dst = keys->buf + keys->buf_used;
907 attrSet = p->attrSet;
908 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
911 keys->prevAttrSet = attrSet;
912 attrUse = p->attrUse;
913 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
916 keys->prevAttrUse = attrUse;
918 diff = 1 + *pseqno - keys->prevSeqNo;
919 if (diff >= 1 && diff <= 15)
924 keys->prevSeqNo = *pseqno;
931 int ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
934 ch = zebraExplain_addSU (zei, attrSet, attrUse);
935 yaz_log (LOG_LOG, "addSU set=%d use=%d SU=%d",
936 attrSet, attrUse, ch);
939 memcpy (dst, &ch, sizeof(ch));
945 memcpy (dst, &attrSet, sizeof(attrSet));
946 dst += sizeof(attrSet);
950 memcpy (dst, &attrUse, sizeof(attrUse));
951 dst += sizeof(attrUse);
954 *dst++ = p->reg_type;
955 memcpy (dst, string, length);
961 memcpy (dst, pseqno, sizeof(*pseqno));
962 dst += sizeof(*pseqno);
964 keys->buf_used = dst - keys->buf;
969 static void extract_add_sort_string (RecWord *p, const char *string,
973 ZebraHandle zh = p->extractCtrl->handle;
974 struct sortKey *sortKeys = zh->sortKeys;
976 for (sk = sortKeys; sk; sk = sk->next)
977 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
980 sk = (struct sortKey *) xmalloc (sizeof(*sk));
984 sk->string = (char *) xmalloc (length);
986 memcpy (sk->string, string, length);
988 sk->attrSet = p->attrSet;
989 sk->attrUse = p->attrUse;
992 static void extract_add_string (RecWord *p, const char *string, int length)
995 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
996 extract_add_sort_string (p, string, length);
998 extract_add_index_string (p, string, length);
1001 static void extract_add_incomplete_field (RecWord *p)
1003 const char *b = p->string;
1004 int remain = p->length;
1005 const char **map = 0;
1008 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1012 char buf[IT_MAX_WORD+1];
1016 while (map && *map && **map == *CHR_SPACE)
1018 remain = p->length - (b - p->string);
1020 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1027 while (map && *map && **map != *CHR_SPACE)
1029 const char *cp = *map;
1031 while (i < IT_MAX_WORD && *cp)
1033 remain = p->length - (b - p->string);
1035 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1041 extract_add_string (p, buf, i);
1043 (p->seqnos[p->reg_type])++; /* to separate this from next one */
1046 static void extract_add_complete_field (RecWord *p)
1048 const char *b = p->string;
1049 char buf[IT_MAX_WORD+1];
1050 const char **map = 0;
1051 int i = 0, remain = p->length;
1054 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1056 while (remain > 0 && i < IT_MAX_WORD)
1058 while (map && *map && **map == *CHR_SPACE)
1060 remain = p->length - (b - p->string);
1062 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1069 if (i && i < IT_MAX_WORD)
1070 buf[i++] = *CHR_SPACE;
1071 while (map && *map && **map != *CHR_SPACE)
1073 const char *cp = *map;
1075 if (i >= IT_MAX_WORD)
1077 while (i < IT_MAX_WORD && *cp)
1079 remain = p->length - (b - p->string);
1081 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1089 extract_add_string (p, buf, i);
1092 static void extract_token_add (RecWord *p)
1095 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1096 p->string, p->length)))
1098 p->string = wrbuf_buf(wrbuf);
1099 p->length = wrbuf_len(wrbuf);
1101 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1102 extract_add_complete_field (p);
1104 extract_add_incomplete_field(p);
1107 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1109 ZebraHandle zh = (ZebraHandle) (p->handle);
1110 zebraExplain_addSchema (zh->service->zei, oid);
1113 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1114 int cmd, struct sortKey **skp)
1116 struct sortKey *sk = *skp;
1117 SortIdx sortIdx = zh->service->sortIdx;
1119 sortIdx_sysno (sortIdx, sysno);
1122 struct sortKey *sk_next = sk->next;
1123 sortIdx_type (sortIdx, sk->attrUse);
1124 sortIdx_add (sortIdx, sk->string, sk->length);
1132 struct encode_info {
1139 void encode_key_init (struct encode_info *i)
1146 char *encode_key_int (int d, char *bp)
1150 else if (d <= 16383)
1152 *bp++ = 64 + (d>>8);
1155 else if (d <= 4194303)
1157 *bp++ = 128 + (d>>16);
1158 *bp++ = (d>>8) & 255;
1163 *bp++ = 192 + (d>>24);
1164 *bp++ = (d>>16) & 255;
1165 *bp++ = (d>>8) & 255;
1171 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1176 while ((*bp++ = *k++))
1178 memcpy (&key, k+1, sizeof(struct it_key));
1179 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1180 if (i->sysno != key.sysno)
1182 i->sysno = key.sysno;
1185 else if (!i->seqno && !key.seqno && i->cmd == *k)
1187 bp = encode_key_int (key.seqno - i->seqno, bp);
1188 i->seqno = key.seqno;
1190 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1192 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1197 static void extract_flushWriteKeys (ZebraHandle zh)
1200 char out_fname[200];
1202 struct encode_info encode_info;
1203 int ptr_i = zh->ptr_i;
1207 if (!zh->key_buf || ptr_i <= 0)
1210 (zh->key_file_no)++;
1211 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1213 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1215 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1217 if (!(outf = fopen (out_fname, "wb")))
1219 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1222 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1223 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1225 encode_key_init (&encode_info);
1226 encode_key_write (cp, &encode_info, outf);
1230 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1231 if (strcmp (cp, prevcp))
1233 encode_key_init (&encode_info);
1234 encode_key_write (cp, &encode_info, outf);
1238 encode_key_write (cp + strlen(cp), &encode_info, outf);
1241 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1242 extract_get_fname_tmp (out_fname, key_file_no);
1244 if (!(outf = fopen (out_fname, "wb")))
1246 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1249 logf (LOG_LOG, "writing section %d", key_file_no);
1251 prevcp = key_buf[ptr_top-i];
1253 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1255 key_y_len = strlen(prevcp)+1;
1257 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1258 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1260 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1261 sizeof(char*), key_y_compare);
1262 cp = key_buf[ptr_top-ptr_i];
1264 encode_key_init (&encode_info);
1265 encode_key_write (cp, &encode_info, outf);
1268 cp = key_buf[ptr_top-ptr_i];
1269 encode_key_write (cp+key_y_len, &encode_info, outf);
1273 prevcp = key_buf[ptr_top-ptr_i];
1278 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1281 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1283 zh->key_buf_used = 0;
1286 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1287 int cmd, struct recKeys *reckeys)
1291 unsigned char attrSet = (unsigned char) -1;
1292 unsigned short attrUse = (unsigned short) -1;
1297 ZebraExplainInfo zei = zh->service->zei;
1301 int mem = 8*1024*1024;
1302 zh->key_buf = (char**) xmalloc (mem);
1303 zh->ptr_top = mem/sizeof(char*);
1305 zh->key_buf_used = 0;
1306 zh->key_file_no = 0;
1308 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1309 while (off < reckeys->buf_used)
1311 const char *src = reckeys->buf + off;
1320 memcpy (&ch, src, sizeof(ch));
1326 memcpy (&attrSet, src, sizeof(attrSet));
1327 src += sizeof(attrSet);
1331 memcpy (&attrUse, src, sizeof(attrUse));
1332 src += sizeof(attrUse);
1335 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1336 extract_flushWriteKeys (zh);
1338 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1339 (char*)zh->key_buf + zh->key_buf_used;
1342 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1344 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1348 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1351 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1353 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1354 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1357 seqno += ((lead>>2) & 15)-1;
1360 memcpy (&seqno, src, sizeof(seqno));
1361 src += sizeof(seqno);
1365 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1366 (zh->key_buf_used) += sizeof(key);
1367 off = src - reckeys->buf;
1369 assert (off == reckeys->buf_used);
1372 static void extract_index (ZebraHandle zh)
1374 extract_flushWriteKeys (zh);
1375 zebra_index_merge (zh);
1378 static int explain_extract (void *handle, Record rec, data1_node *n)
1380 ZebraHandle zh = (ZebraHandle) handle;
1381 struct recExtractCtrl extractCtrl;
1384 if (zebraExplain_curDatabase (zh->service->zei,
1385 rec->info[recInfo_databaseName]))
1388 if (zebraExplain_newDatabase (zh->service->zei,
1389 rec->info[recInfo_databaseName], 0))
1393 zh->keys.buf_used = 0;
1394 zh->keys.prevAttrUse = -1;
1395 zh->keys.prevAttrSet = -1;
1396 zh->keys.prevSeqNo = 0;
1399 extractCtrl.init = extract_init;
1400 extractCtrl.tokenAdd = extract_token_add;
1401 extractCtrl.schemaAdd = extract_schema_add;
1402 extractCtrl.dh = zh->service->dh;
1403 for (i = 0; i<256; i++)
1404 extractCtrl.seqno[i] = 0;
1405 extractCtrl.zebra_maps = zh->service->zebra_maps;
1406 extractCtrl.flagShowRecords = 0;
1407 extractCtrl.handle = handle;
1409 grs_extract_tree(&extractCtrl, n);
1411 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1413 if (rec->size[recInfo_delKeys])
1415 struct recKeys delkeys;
1416 struct sortKey *sortKeys = 0;
1418 delkeys.buf_used = rec->size[recInfo_delKeys];
1419 delkeys.buf = rec->info[recInfo_delKeys];
1420 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1421 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1423 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1424 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1426 xfree (rec->info[recInfo_delKeys]);
1427 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1428 rec->info[recInfo_delKeys] = zh->keys.buf;
1429 zh->keys.buf = NULL;
1430 zh->keys.buf_max = 0;
1434 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1435 const char *buf, size_t buf_size,
1436 const char *databaseName, int delete_flag,
1437 int test_mode, int *sysno,
1438 int store_keys, int store_data,
1439 const char *match_criteria)
1441 RecordAttr *recordAttr;
1442 struct recExtractCtrl extractCtrl;
1447 const char *fname = "<no file>";
1449 long recordOffset = 0;
1450 struct zebra_fetch_control fc;
1453 fc.record_int_buf = buf;
1454 fc.record_int_len = buf_size;
1455 fc.record_int_pos = 0;
1457 fc.record_offset = 0;
1459 extractCtrl.offset = 0;
1460 extractCtrl.readf = zebra_record_int_read;
1461 extractCtrl.seekf = zebra_record_int_seek;
1462 extractCtrl.tellf = zebra_record_int_tell;
1463 extractCtrl.endf = zebra_record_int_end;
1464 extractCtrl.fh = &fc;
1466 /* announce database */
1467 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1469 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1473 recType_byName (zh->service->recTypes, recordType, subType,
1476 logf (LOG_WARN, "No such record type: %s", recordType);
1480 zh->keys.buf_used = 0;
1481 zh->keys.prevAttrUse = -1;
1482 zh->keys.prevAttrSet = -1;
1483 zh->keys.prevSeqNo = 0;
1486 extractCtrl.subType = subType;
1487 extractCtrl.init = extract_init;
1488 extractCtrl.tokenAdd = extract_token_add;
1489 extractCtrl.schemaAdd = extract_schema_add;
1490 extractCtrl.dh = zh->service->dh;
1491 extractCtrl.handle = zh;
1492 extractCtrl.zebra_maps = zh->service->zebra_maps;
1493 extractCtrl.flagShowRecords = 0;
1494 for (i = 0; i<256; i++)
1496 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1497 extractCtrl.seqno[i] = 1;
1499 extractCtrl.seqno[i] = 0;
1502 r = (*recType->extract)(clientData, &extractCtrl);
1504 if (r == RECCTRL_EXTRACT_EOF)
1506 else if (r == RECCTRL_EXTRACT_ERROR)
1508 /* error occured during extraction ... */
1510 yaz_log (LOG_WARN, "extract error");
1512 if (rGroup->flagRw &&
1513 records_processed < rGroup->fileVerboseLimit)
1515 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1516 fname, (long) recordOffset);
1521 if (zh->keys.buf_used == 0)
1523 /* the extraction process returned no information - the record
1524 is probably empty - unless flagShowRecords is in use */
1527 logf (LOG_WARN, "No keys generated for record");
1528 logf (LOG_WARN, " The file is probably empty");
1531 /* match criteria */
1538 logf (LOG_LOG, "delete %s %s %ld", recordType,
1539 fname, (long) recordOffset);
1540 logf (LOG_WARN, "cannot delete record above (seems new)");
1543 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1544 (long) recordOffset);
1545 rec = rec_new (zh->service->records);
1547 *sysno = rec->sysno;
1549 recordAttr = rec_init_attr (zh->service->zei, rec);
1554 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1557 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1558 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1562 /* record already exists */
1563 struct recKeys delkeys;
1565 rec = rec_get (zh->service->records, *sysno);
1568 recordAttr = rec_init_attr (zh->service->zei, rec);
1570 if (recordAttr->runNumber ==
1571 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1573 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1574 fname, (long) recordOffset);
1578 delkeys.buf_used = rec->size[recInfo_delKeys];
1579 delkeys.buf = rec->info[recInfo_delKeys];
1580 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1581 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1584 /* record going to be deleted */
1585 if (!delkeys.buf_used)
1587 logf (LOG_LOG, "delete %s %s %ld", recordType,
1588 fname, (long) recordOffset);
1589 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1593 logf (LOG_LOG, "delete %s %s %ld", recordType,
1594 fname, (long) recordOffset);
1597 dict_delete (matchDict, matchStr);
1599 rec_del (zh->service->records, &rec);
1606 /* record going to be updated */
1607 if (!delkeys.buf_used)
1609 logf (LOG_LOG, "update %s %s %ld", recordType,
1610 fname, (long) recordOffset);
1611 logf (LOG_WARN, "cannot update file above, storeKeys false");
1615 logf (LOG_LOG, "update %s %s %ld", recordType,
1616 fname, (long) recordOffset);
1617 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1621 /* update file type */
1622 xfree (rec->info[recInfo_fileType]);
1623 rec->info[recInfo_fileType] =
1624 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1626 /* update filename */
1627 xfree (rec->info[recInfo_filename]);
1628 rec->info[recInfo_filename] =
1629 rec_strdup (fname, &rec->size[recInfo_filename]);
1631 /* update delete keys */
1632 xfree (rec->info[recInfo_delKeys]);
1633 if (zh->keys.buf_used > 0 && store_keys == 1)
1635 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1636 rec->info[recInfo_delKeys] = zh->keys.buf;
1637 zh->keys.buf = NULL;
1638 zh->keys.buf_max = 0;
1642 rec->info[recInfo_delKeys] = NULL;
1643 rec->size[recInfo_delKeys] = 0;
1646 /* save file size of original record */
1647 zebraExplain_recordBytesIncrement (zh->service->zei,
1648 - recordAttr->recordSize);
1650 recordAttr->recordSize = fi->file_moffset - recordOffset;
1651 if (!recordAttr->recordSize)
1652 recordAttr->recordSize = fi->file_max - recordOffset;
1654 recordAttr->recordSize = buf_size;
1656 zebraExplain_recordBytesIncrement (zh->service->zei,
1657 recordAttr->recordSize);
1659 /* set run-number for this record */
1660 recordAttr->runNumber =
1661 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1663 /* update store data */
1664 xfree (rec->info[recInfo_storeData]);
1665 if (store_data == 1)
1667 rec->size[recInfo_storeData] = recordAttr->recordSize;
1668 rec->info[recInfo_storeData] = (char *)
1669 xmalloc (recordAttr->recordSize);
1671 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1673 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1675 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1676 (long) recordOffset, fname);
1679 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1680 < recordAttr->recordSize)
1682 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1683 recordAttr->recordSize, fname);
1690 rec->info[recInfo_storeData] = NULL;
1691 rec->size[recInfo_storeData] = 0;
1693 /* update database name */
1694 xfree (rec->info[recInfo_databaseName]);
1695 rec->info[recInfo_databaseName] =
1696 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1699 recordAttr->recordOffset = recordOffset;
1701 /* commit this record */
1702 rec_put (zh->service->records, &rec);