2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.36 2000-09-06 08:59:36 adam
7 * Using read-only (for now) for server.
9 * Revision 1.35 2000/07/07 12:49:20 adam
10 * Optimized resultSetInsert{Rank,Sort}.
12 * Revision 1.34 2000/06/09 13:56:38 ian
13 * Added some logging on Authentication and searches.
15 * Revision 1.33 2000/05/18 12:01:36 adam
16 * System call times(2) used again. More 64-bit fixes.
18 * Revision 1.32 2000/04/19 14:35:59 adam
19 * WIN32 update (this version is known not to work on Windows).
21 * Revision 1.31 2000/04/05 10:07:02 adam
22 * Minor zebra compile fix.
24 * Revision 1.30 2000/04/05 09:49:35 adam
25 * On Unix, zebra/z'mbol uses automake.
27 * Revision 1.29 2000/03/20 19:08:36 adam
28 * Added remote record import using Z39.50 extended services and Segment
31 * Revision 1.28 2000/03/15 15:00:30 adam
32 * First work on threaded version.
34 * Revision 1.27 2000/02/24 12:31:17 adam
35 * Added zebra_string_norm.
37 * Revision 1.26 1999/11/30 13:48:03 adam
38 * Improved installation. Updated for inclusion of YAZ header files.
40 * Revision 1.25 1999/11/04 15:00:45 adam
41 * Implemented delete result set(s).
43 * Revision 1.24 1999/10/14 14:33:50 adam
44 * Added truncation 5=106.
46 * Revision 1.23 1999/09/07 11:36:32 adam
49 * Revision 1.22 1999/08/02 10:13:47 adam
50 * Fixed bug regarding zebra_hits.
52 * Revision 1.21 1999/07/14 10:59:26 adam
53 * Changed functions isc_getmethod, isams_getmethod.
54 * Improved fatal error handling (such as missing EXPLAIN schema).
56 * Revision 1.20 1999/07/06 12:28:04 adam
57 * Updated record index structure. Format includes version ID. Compression
58 * algorithm ID is stored for each record block.
60 * Revision 1.19 1999/05/26 07:49:13 adam
63 * Revision 1.18 1999/05/15 14:36:38 adam
64 * Updated dictionary. Implemented "compression" of dictionary.
66 * Revision 1.17 1999/05/12 13:08:06 adam
67 * First version of ISAMS.
69 * Revision 1.16 1999/02/19 10:38:30 adam
70 * Implemented chdir-setting.
72 * Revision 1.15 1999/02/17 12:18:12 adam
73 * Fixed zebra_close so that a NULL pointer is ignored.
75 * Revision 1.14 1999/02/02 14:51:11 adam
76 * Updated WIN32 code specific sections. Changed header.
78 * Revision 1.13 1998/12/16 12:23:30 adam
79 * Added facility for database name mapping using resource mapdb.
81 * Revision 1.12 1998/11/16 10:18:10 adam
82 * Better error reporting for result sets.
84 * Revision 1.11 1998/10/16 08:14:34 adam
85 * Updated record control system.
87 * Revision 1.10 1998/09/22 10:03:42 adam
88 * Changed result sets to be persistent in the sense that they can
89 * be re-searched if needed.
90 * Fixed memory leak in rsm_or.
92 * Revision 1.9 1998/09/02 13:53:17 adam
93 * Extra parameter decode added to search routines to implement
96 * Revision 1.8 1998/08/24 17:29:23 adam
99 * Revision 1.7 1998/06/24 12:16:13 adam
100 * Support for relations on text operands. Open range support in
101 * DFA module (i.e. [-j], [g-]).
103 * Revision 1.6 1998/06/22 11:36:47 adam
104 * Added authentication check facility to zebra.
106 * Revision 1.5 1998/06/13 00:14:08 adam
109 * Revision 1.4 1998/06/12 12:22:12 adam
112 * Revision 1.3 1998/05/27 16:57:44 adam
113 * Zebra returns surrogate diagnostic for single records when
116 * Revision 1.2 1998/05/20 10:12:19 adam
117 * Implemented automatic EXPLAIN database maintenance.
118 * Modified Zebra to work with ASN.1 compiled version of YAZ.
120 * Revision 1.1 1998/03/05 08:45:13 adam
121 * New result set model and modular ranking system. Moved towards
122 * descent server API. System information stored as "SGML" records.
136 #include <yaz/diagbib1.h>
140 static void zebra_chdir (ZebraService zh)
142 const char *dir = res_get (zh->res, "chdir");
145 logf (LOG_DEBUG, "chdir %s", dir);
153 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
154 const char *buf, size_t buf_size,
155 const char *databaseName, int delete_flag,
156 int test_mode, int *sysno,
157 int store_keys, int store_data,
158 const char *match_criteria);
160 static int explain_extract (void *handle, Record rec, data1_node *n);
161 static void extract_index (ZebraHandle zh);
163 static void zebra_register_unlock (ZebraHandle zh);
165 static int zebra_register_lock (ZebraHandle zh)
169 if (!zh->service->active)
180 static void zebra_register_unlock (ZebraHandle zh)
184 logf (LOG_LOG, "user/system: %ld/%ld",
185 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
186 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
191 ZebraHandle zebra_open (ZebraService zs)
199 zh = (ZebraHandle) xmalloc (sizeof(*zh));
208 zh->admin_databaseName = 0;
210 zebra_mutex_cond_lock (&zs->session_lock);
212 zh->next = zs->sessions;
215 zebra_mutex_cond_unlock (&zs->session_lock);
217 logf(LOG_APP,"CONNECT:");
222 static int zebra_register_activate (ZebraService zh, int rw);
223 static int zebra_register_deactivate (ZebraService zh);
225 ZebraService zebra_start (const char *configName)
227 ZebraService zh = xmalloc (sizeof(*zh));
229 yaz_log (LOG_LOG, "zebra_start %s", configName);
231 zh->configName = xstrdup(configName);
235 zebra_mutex_cond_init (&zh->session_lock);
236 zebra_register_activate (zh, 0);
240 static int zebra_register_activate (ZebraService zh, int rw)
244 yaz_log (LOG_LOG, "zebra_register_activate");
245 if (!(zh->res = res_open (zh->configName)))
247 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
251 zh->dh = data1_create ();
254 zh->bfs = bfs_create (res_get (zh->res, "register"));
257 data1_destroy(zh->dh);
260 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
261 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
262 zh->registerState = -1; /* trigger open of registers! */
263 zh->registerChange = 0;
264 zh->recTypes = recTypes_init (zh->dh);
265 recTypes_default_handlers (zh->recTypes);
268 zh->zebra_maps = zebra_maps_open (zh->res);
269 zh->rank_classes = NULL;
282 zebraRankInstall (zh, rank1_class);
284 if (!res_get (zh->res, "passwd"))
285 zh->passwd_db = NULL;
288 zh->passwd_db = passwd_db_open ();
290 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
292 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
295 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
297 logf (LOG_WARN, "rec_open");
300 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
302 logf (LOG_WARN, "dict_open");
305 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
307 logf (LOG_WARN, "sortIdx_open");
310 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
312 struct ISAMS_M_s isams_m;
313 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
314 key_isams_m(zh->res, &isams_m))))
316 logf (LOG_WARN, "isams_open");
321 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
323 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
324 sizeof (struct it_key), zh->res)))
326 logf (LOG_WARN, "is_open");
330 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
332 struct ISAMC_M_s isamc_m;
333 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
334 rw, key_isamc_m(zh->res, &isamc_m))))
336 logf (LOG_WARN, "isc_open");
340 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
342 struct ISAMD_M_s isamd_m;
344 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
345 rw, key_isamd_m(zh->res, &isamd_m))))
347 logf (LOG_WARN, "isamd_open");
352 zh->zei = zebraExplain_open (zh->records, zh->dh,
353 zh->res, rw, 0 /* rGroup */,
357 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
361 yaz_log (LOG_LOG, "zebra_register_activate ok");
365 void zebra_admin_shutdown (ZebraHandle zh)
367 zebraExplain_flush (zh->service->zei, 1, zh);
370 zebra_mutex_cond_lock (&zh->service->session_lock);
371 zh->service->stop_flag = 1;
372 if (!zh->service->sessions)
373 zebra_register_deactivate(zh->service);
374 zebra_mutex_cond_unlock (&zh->service->session_lock);
377 void zebra_admin_start (ZebraHandle zh)
379 ZebraService zs = zh->service;
381 zebra_mutex_cond_lock (&zs->session_lock);
383 zebra_register_activate(zs, 0);
384 zebra_mutex_cond_unlock (&zs->session_lock);
387 static int zebra_register_deactivate (ZebraService zh)
392 yaz_log(LOG_LOG, "zebra_register_deactivate");
396 zebraExplain_close (zh->zei, 0);
397 dict_close (zh->dict);
398 sortIdx_close (zh->sortIdx);
400 isams_close (zh->isams);
405 isc_close (zh->isamc);
407 isamd_close (zh->isamd);
409 rec_close (&zh->records);
411 recTypes_destroy (zh->recTypes);
412 zebra_maps_close (zh->zebra_maps);
413 zebraRankDestroy (zh);
414 bfs_destroy (zh->bfs);
415 data1_destroy (zh->dh);
418 passwd_db_close (zh->passwd_db);
424 void zebra_stop(ZebraService zh)
428 yaz_log (LOG_LOG, "zebra_stop");
430 assert (!zh->sessions);
432 zebra_mutex_cond_destroy (&zh->session_lock);
434 zebra_register_deactivate(zh);
435 xfree (zh->configName);
439 void zebra_close (ZebraHandle zh)
441 ZebraService zs = zh->service;
442 struct zebra_session **sp;
445 resultSetDestroy (zh, -1, 0, 0);
452 xfree (zh->admin_databaseName);
453 zebra_mutex_cond_lock (&zs->session_lock);
465 if (!zs->sessions && zs->stop_flag)
466 zebra_register_deactivate(zs);
467 zebra_mutex_cond_unlock (&zs->session_lock);
471 struct map_baseinfo {
477 char **new_basenames;
481 void map_basenames_func (void *vp, const char *name, const char *value)
483 struct map_baseinfo *p = (struct map_baseinfo *) vp;
485 char fromdb[128], todb[8][128];
488 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
489 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
490 todb[5], todb[6], todb[7]);
494 for (i = 0; i<p->num_bases; i++)
495 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
498 for (i = 0; i < no; i++)
500 if (p->new_num_bases == p->new_num_max)
502 p->new_basenames[(p->new_num_bases)++] =
503 nmem_strdup (p->mem, todb[i]);
509 void map_basenames (ZebraHandle zh, ODR stream,
510 int *num_bases, char ***basenames)
512 struct map_baseinfo info;
513 struct map_baseinfo *p = &info;
517 info.num_bases = *num_bases;
518 info.basenames = *basenames;
519 info.new_num_max = 128;
520 info.new_num_bases = 0;
521 info.new_basenames = (char **)
522 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
523 info.mem = stream->mem;
525 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
527 for (i = 0; i<p->num_bases; i++)
528 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
530 p->new_basenames[(p->new_num_bases)++] =
531 nmem_strdup (p->mem, p->basenames[i]);
533 *num_bases = info.new_num_bases;
534 *basenames = info.new_basenames;
535 for (i = 0; i<*num_bases; i++)
536 logf (LOG_LOG, "base %s", (*basenames)[i]);
539 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
540 Z_RPNQuery *query, int num_bases, char **basenames,
544 if (zebra_register_lock (zh))
546 map_basenames (zh, stream, &num_bases, &basenames);
547 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
549 zebra_register_unlock (zh);
551 logf(LOG_APP,"SEARCH:%d:",zh->hits);
554 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
555 const char *setname, Z_RecordComposition *comp,
556 oid_value input_format, int num_recs,
557 ZebraRetrievalRecord *recs)
562 if (zebra_register_lock (zh))
564 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
565 for (i = 0; i<num_recs; i++)
566 pos_array[i] = recs[i].position;
567 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
570 logf (LOG_DEBUG, "zebraPosSetCreate error");
572 zh->errString = nmem_strdup (stream->mem, setname);
576 for (i = 0; i<num_recs; i++)
582 sprintf (num_str, "%d", pos_array[i]);
584 zh->errString = nmem_strdup (stream->mem, num_str);
590 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
591 stream, input_format, comp,
592 &recs[i].format, &recs[i].buf,
595 recs[i].errString = NULL;
598 zebraPosSetDestroy (zh, poset, num_recs);
600 zebra_register_unlock (zh);
604 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
605 oid_value attributeset,
606 int num_bases, char **basenames,
607 int *position, int *num_entries, ZebraScanEntry **entries,
610 if (zebra_register_lock (zh))
616 map_basenames (zh, stream, &num_bases, &basenames);
617 rpn_scan (zh, stream, zapt, attributeset,
618 num_bases, basenames, position,
619 num_entries, entries, is_partial);
620 zebra_register_unlock (zh);
623 void zebra_sort (ZebraHandle zh, ODR stream,
624 int num_input_setnames, const char **input_setnames,
625 const char *output_setname, Z_SortKeySpecList *sort_sequence,
628 if (zebra_register_lock (zh))
630 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
631 output_setname, sort_sequence, sort_status);
632 zebra_register_unlock (zh);
635 int zebra_deleleResultSet(ZebraHandle zh, int function,
636 int num_setnames, char **setnames,
640 if (zebra_register_lock (zh))
641 return Z_DeleteStatus_systemProblemAtTarget;
644 case Z_DeleteRequest_list:
645 resultSetDestroy (zh, num_setnames, setnames, statuses);
647 case Z_DeleteRequest_all:
648 resultSetDestroy (zh, -1, 0, statuses);
651 zebra_register_unlock (zh);
652 status = Z_DeleteStatus_success;
653 for (i = 0; i<num_setnames; i++)
654 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
655 status = statuses[i];
659 int zebra_errCode (ZebraHandle zh)
664 const char *zebra_errString (ZebraHandle zh)
666 return diagbib1_str (zh->errCode);
669 char *zebra_errAdd (ZebraHandle zh)
671 return zh->errString;
674 int zebra_hits (ZebraHandle zh)
679 int zebra_auth (ZebraService zh, const char *user, const char *pass)
681 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
683 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
687 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
691 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
693 if (zebra_register_lock (zh))
695 xfree (zh->admin_databaseName);
696 zh->admin_databaseName = xstrdup(database);
697 zebra_register_unlock(zh);
700 void zebra_admin_import_end (ZebraHandle zh)
702 zebraExplain_flush (zh->service->zei, 1, zh);
706 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
710 if (zebra_register_lock (zh))
712 for (i = 0; i<segment->num_segmentRecords; i++)
714 Z_NamePlusRecord *npr = segment->segmentRecords[i];
715 const char *databaseName = npr->databaseName;
718 databaseName = zh->admin_databaseName;
719 printf ("--------------%d--------------------\n", i);
720 if (npr->which == Z_NamePlusRecord_intermediateFragment)
722 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
723 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
725 Odr_oct *oct = fragment->u.notExternallyTagged;
726 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
730 extract_rec_in_mem (zh, "grs.sgml",
738 0 /* match criteria */);
742 zebra_register_unlock(zh);
745 void zebra_admin_create (ZebraHandle zh, const char *database)
747 ZebraService zs = zh->service;
748 if (zebra_register_lock(zh))
753 /* announce database */
754 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
757 zh->errString = "Database already exist";
759 zebra_register_unlock(zh);
762 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
763 const char *input_str, int input_len,
764 char *output_str, int output_len)
767 if (!zh->service->zebra_maps)
769 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
770 input_str, input_len);
773 if (wrbuf_len(wrbuf) >= output_len)
775 if (wrbuf_len(wrbuf))
776 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
777 output_str[wrbuf_len(wrbuf)] = '\0';
778 return wrbuf_len(wrbuf);
781 static void extract_init (struct recExtractCtrl *p, RecWord *w)
783 w->zebra_maps = p->zebra_maps;
784 w->seqnos = p->seqno;
785 w->attrSet = VAL_BIB1;
791 static void extract_add_index_string (RecWord *p, const char *string,
795 unsigned char attrSet;
796 unsigned short attrUse;
799 int *pseqno = &p->seqnos[p->reg_type];
800 ZebraHandle zh = p->extractCtrl->handle;
801 struct recKeys *keys = &zh->keys;
803 if (keys->buf_used+1024 > keys->buf_max)
807 b = (char *) xmalloc (keys->buf_max += 128000);
808 if (keys->buf_used > 0)
809 memcpy (b, keys->buf, keys->buf_used);
813 dst = keys->buf + keys->buf_used;
815 attrSet = p->attrSet;
816 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
819 keys->prevAttrSet = attrSet;
820 attrUse = p->attrUse;
821 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
824 keys->prevAttrUse = attrUse;
826 diff = 1 + *pseqno - keys->prevSeqNo;
827 if (diff >= 1 && diff <= 15)
832 keys->prevSeqNo = *pseqno;
838 memcpy (dst, &attrSet, sizeof(attrSet));
839 dst += sizeof(attrSet);
843 memcpy (dst, &attrUse, sizeof(attrUse));
844 dst += sizeof(attrUse);
846 *dst++ = p->reg_type;
847 memcpy (dst, string, length);
853 memcpy (dst, pseqno, sizeof(*pseqno));
854 dst += sizeof(*pseqno);
856 keys->buf_used = dst - keys->buf;
861 static void extract_add_sort_string (RecWord *p, const char *string,
865 ZebraHandle zh = p->extractCtrl->handle;
866 struct sortKey *sortKeys = zh->sortKeys;
868 for (sk = sortKeys; sk; sk = sk->next)
869 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
872 sk = (struct sortKey *) xmalloc (sizeof(*sk));
876 sk->string = (char *) xmalloc (length);
878 memcpy (sk->string, string, length);
880 sk->attrSet = p->attrSet;
881 sk->attrUse = p->attrUse;
884 static void extract_add_string (RecWord *p, const char *string, int length)
887 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
888 extract_add_sort_string (p, string, length);
890 extract_add_index_string (p, string, length);
893 static void extract_add_incomplete_field (RecWord *p)
895 const char *b = p->string;
896 int remain = p->length;
897 const char **map = 0;
900 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
904 char buf[IT_MAX_WORD+1];
908 while (map && *map && **map == *CHR_SPACE)
910 remain = p->length - (b - p->string);
912 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
919 while (map && *map && **map != *CHR_SPACE)
921 const char *cp = *map;
923 while (i < IT_MAX_WORD && *cp)
925 remain = p->length - (b - p->string);
927 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
933 extract_add_string (p, buf, i);
935 (p->seqnos[p->reg_type])++; /* to separate this from next one */
938 static void extract_add_complete_field (RecWord *p)
940 const char *b = p->string;
941 char buf[IT_MAX_WORD+1];
942 const char **map = 0;
943 int i = 0, remain = p->length;
946 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
948 while (remain > 0 && i < IT_MAX_WORD)
950 while (map && *map && **map == *CHR_SPACE)
952 remain = p->length - (b - p->string);
954 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
961 if (i && i < IT_MAX_WORD)
962 buf[i++] = *CHR_SPACE;
963 while (map && *map && **map != *CHR_SPACE)
965 const char *cp = *map;
967 if (i >= IT_MAX_WORD)
969 while (i < IT_MAX_WORD && *cp)
971 remain = p->length - (b - p->string);
973 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
981 extract_add_string (p, buf, i);
984 static void extract_token_add (RecWord *p)
987 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
988 p->string, p->length)))
990 p->string = wrbuf_buf(wrbuf);
991 p->length = wrbuf_len(wrbuf);
993 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
994 extract_add_complete_field (p);
996 extract_add_incomplete_field(p);
999 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1001 ZebraHandle zh = (ZebraHandle) (p->handle);
1002 zebraExplain_addSchema (zh->service->zei, oid);
1005 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1006 int cmd, struct sortKey **skp)
1008 struct sortKey *sk = *skp;
1009 SortIdx sortIdx = zh->service->sortIdx;
1011 sortIdx_sysno (sortIdx, sysno);
1014 struct sortKey *sk_next = sk->next;
1015 sortIdx_type (sortIdx, sk->attrUse);
1016 sortIdx_add (sortIdx, sk->string, sk->length);
1024 struct encode_info {
1031 void encode_key_init (struct encode_info *i)
1038 char *encode_key_int (int d, char *bp)
1042 else if (d <= 16383)
1044 *bp++ = 64 + (d>>8);
1047 else if (d <= 4194303)
1049 *bp++ = 128 + (d>>16);
1050 *bp++ = (d>>8) & 255;
1055 *bp++ = 192 + (d>>24);
1056 *bp++ = (d>>16) & 255;
1057 *bp++ = (d>>8) & 255;
1063 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1068 while ((*bp++ = *k++))
1070 memcpy (&key, k+1, sizeof(struct it_key));
1071 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1072 if (i->sysno != key.sysno)
1074 i->sysno = key.sysno;
1077 else if (!i->seqno && !key.seqno && i->cmd == *k)
1079 bp = encode_key_int (key.seqno - i->seqno, bp);
1080 i->seqno = key.seqno;
1082 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1084 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1089 static void extract_flushWriteKeys (ZebraHandle zh)
1092 char out_fname[200];
1094 struct encode_info encode_info;
1095 int ptr_i = zh->ptr_i;
1099 if (!zh->key_buf || ptr_i <= 0)
1102 (zh->key_file_no)++;
1103 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1105 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1107 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1109 if (!(outf = fopen (out_fname, "wb")))
1111 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1114 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1115 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1117 encode_key_init (&encode_info);
1118 encode_key_write (cp, &encode_info, outf);
1122 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1123 if (strcmp (cp, prevcp))
1125 encode_key_init (&encode_info);
1126 encode_key_write (cp, &encode_info, outf);
1130 encode_key_write (cp + strlen(cp), &encode_info, outf);
1133 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1134 extract_get_fname_tmp (out_fname, key_file_no);
1136 if (!(outf = fopen (out_fname, "wb")))
1138 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1141 logf (LOG_LOG, "writing section %d", key_file_no);
1143 prevcp = key_buf[ptr_top-i];
1145 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1147 key_y_len = strlen(prevcp)+1;
1149 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1150 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1152 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1153 sizeof(char*), key_y_compare);
1154 cp = key_buf[ptr_top-ptr_i];
1156 encode_key_init (&encode_info);
1157 encode_key_write (cp, &encode_info, outf);
1160 cp = key_buf[ptr_top-ptr_i];
1161 encode_key_write (cp+key_y_len, &encode_info, outf);
1165 prevcp = key_buf[ptr_top-ptr_i];
1170 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1173 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1175 zh->key_buf_used = 0;
1178 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1179 int cmd, struct recKeys *reckeys)
1181 unsigned char attrSet = (unsigned char) -1;
1182 unsigned short attrUse = (unsigned short) -1;
1185 ZebraExplainInfo zei = zh->service->zei;
1189 int mem = 8*1024*1024;
1190 zh->key_buf = (char**) xmalloc (mem);
1191 zh->ptr_top = mem/sizeof(char*);
1193 zh->key_buf_used = 0;
1194 zh->key_file_no = 0;
1196 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1197 while (off < reckeys->buf_used)
1199 const char *src = reckeys->buf + off;
1207 memcpy (&attrSet, src, sizeof(attrSet));
1208 src += sizeof(attrSet);
1212 memcpy (&attrUse, src, sizeof(attrUse));
1213 src += sizeof(attrUse);
1215 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1216 extract_flushWriteKeys (zh);
1218 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1219 (char*)zh->key_buf + zh->key_buf_used;
1221 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1223 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1226 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1229 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1231 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1232 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1235 seqno += ((lead>>2) & 15)-1;
1238 memcpy (&seqno, src, sizeof(seqno));
1239 src += sizeof(seqno);
1243 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1244 (zh->key_buf_used) += sizeof(key);
1245 off = src - reckeys->buf;
1247 assert (off == reckeys->buf_used);
1250 static void extract_index (ZebraHandle zh)
1252 extract_flushWriteKeys (zh);
1253 zebra_index_merge (zh);
1256 static int explain_extract (void *handle, Record rec, data1_node *n)
1258 ZebraHandle zh = (ZebraHandle) handle;
1259 struct recExtractCtrl extractCtrl;
1262 if (zebraExplain_curDatabase (zh->service->zei,
1263 rec->info[recInfo_databaseName]))
1266 if (zebraExplain_newDatabase (zh->service->zei,
1267 rec->info[recInfo_databaseName], 0))
1271 zh->keys.buf_used = 0;
1272 zh->keys.prevAttrUse = -1;
1273 zh->keys.prevAttrSet = -1;
1274 zh->keys.prevSeqNo = 0;
1277 extractCtrl.init = extract_init;
1278 extractCtrl.tokenAdd = extract_token_add;
1279 extractCtrl.schemaAdd = extract_schema_add;
1280 extractCtrl.dh = zh->service->dh;
1281 for (i = 0; i<256; i++)
1282 extractCtrl.seqno[i] = 0;
1283 extractCtrl.zebra_maps = zh->service->zebra_maps;
1284 extractCtrl.flagShowRecords = 0;
1285 extractCtrl.handle = handle;
1287 grs_extract_tree(&extractCtrl, n);
1289 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1291 if (rec->size[recInfo_delKeys])
1293 struct recKeys delkeys;
1294 struct sortKey *sortKeys = 0;
1296 delkeys.buf_used = rec->size[recInfo_delKeys];
1297 delkeys.buf = rec->info[recInfo_delKeys];
1298 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1299 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1301 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1302 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1304 xfree (rec->info[recInfo_delKeys]);
1305 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1306 rec->info[recInfo_delKeys] = zh->keys.buf;
1307 zh->keys.buf = NULL;
1308 zh->keys.buf_max = 0;
1312 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1313 const char *buf, size_t buf_size,
1314 const char *databaseName, int delete_flag,
1315 int test_mode, int *sysno,
1316 int store_keys, int store_data,
1317 const char *match_criteria)
1319 RecordAttr *recordAttr;
1320 struct recExtractCtrl extractCtrl;
1325 const char *fname = "<no file>";
1327 long recordOffset = 0;
1328 struct zebra_fetch_control fc;
1331 fc.record_int_buf = buf;
1332 fc.record_int_len = buf_size;
1333 fc.record_int_pos = 0;
1335 fc.record_offset = 0;
1337 extractCtrl.offset = 0;
1338 extractCtrl.readf = zebra_record_int_read;
1339 extractCtrl.seekf = zebra_record_int_seek;
1340 extractCtrl.tellf = zebra_record_int_tell;
1341 extractCtrl.endf = zebra_record_int_end;
1342 extractCtrl.fh = &fc;
1344 /* announce database */
1345 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1347 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1351 recType_byName (zh->service->recTypes, recordType, subType,
1354 logf (LOG_WARN, "No such record type: %s", recordType);
1358 zh->keys.buf_used = 0;
1359 zh->keys.prevAttrUse = -1;
1360 zh->keys.prevAttrSet = -1;
1361 zh->keys.prevSeqNo = 0;
1364 extractCtrl.subType = subType;
1365 extractCtrl.init = extract_init;
1366 extractCtrl.tokenAdd = extract_token_add;
1367 extractCtrl.schemaAdd = extract_schema_add;
1368 extractCtrl.dh = zh->service->dh;
1369 extractCtrl.handle = zh;
1370 extractCtrl.zebra_maps = zh->service->zebra_maps;
1371 extractCtrl.flagShowRecords = 0;
1372 for (i = 0; i<256; i++)
1374 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1375 extractCtrl.seqno[i] = 1;
1377 extractCtrl.seqno[i] = 0;
1380 r = (*recType->extract)(clientData, &extractCtrl);
1382 if (r == RECCTRL_EXTRACT_EOF)
1384 else if (r == RECCTRL_EXTRACT_ERROR)
1386 /* error occured during extraction ... */
1388 yaz_log (LOG_WARN, "extract error");
1390 if (rGroup->flagRw &&
1391 records_processed < rGroup->fileVerboseLimit)
1393 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1394 fname, (long) recordOffset);
1399 if (zh->keys.buf_used == 0)
1401 /* the extraction process returned no information - the record
1402 is probably empty - unless flagShowRecords is in use */
1405 logf (LOG_WARN, "No keys generated for record");
1406 logf (LOG_WARN, " The file is probably empty");
1409 /* match criteria */
1416 logf (LOG_LOG, "delete %s %s %ld", recordType,
1417 fname, (long) recordOffset);
1418 logf (LOG_WARN, "cannot delete record above (seems new)");
1421 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1422 (long) recordOffset);
1423 rec = rec_new (zh->service->records);
1425 *sysno = rec->sysno;
1427 recordAttr = rec_init_attr (zh->service->zei, rec);
1432 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1435 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1436 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1440 /* record already exists */
1441 struct recKeys delkeys;
1443 rec = rec_get (zh->service->records, *sysno);
1446 recordAttr = rec_init_attr (zh->service->zei, rec);
1448 if (recordAttr->runNumber ==
1449 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1451 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1452 fname, (long) recordOffset);
1456 delkeys.buf_used = rec->size[recInfo_delKeys];
1457 delkeys.buf = rec->info[recInfo_delKeys];
1458 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1459 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1462 /* record going to be deleted */
1463 if (!delkeys.buf_used)
1465 logf (LOG_LOG, "delete %s %s %ld", recordType,
1466 fname, (long) recordOffset);
1467 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1471 logf (LOG_LOG, "delete %s %s %ld", recordType,
1472 fname, (long) recordOffset);
1475 dict_delete (matchDict, matchStr);
1477 rec_del (zh->service->records, &rec);
1484 /* record going to be updated */
1485 if (!delkeys.buf_used)
1487 logf (LOG_LOG, "update %s %s %ld", recordType,
1488 fname, (long) recordOffset);
1489 logf (LOG_WARN, "cannot update file above, storeKeys false");
1493 logf (LOG_LOG, "update %s %s %ld", recordType,
1494 fname, (long) recordOffset);
1495 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1499 /* update file type */
1500 xfree (rec->info[recInfo_fileType]);
1501 rec->info[recInfo_fileType] =
1502 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1504 /* update filename */
1505 xfree (rec->info[recInfo_filename]);
1506 rec->info[recInfo_filename] =
1507 rec_strdup (fname, &rec->size[recInfo_filename]);
1509 /* update delete keys */
1510 xfree (rec->info[recInfo_delKeys]);
1511 if (zh->keys.buf_used > 0 && store_keys == 1)
1513 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1514 rec->info[recInfo_delKeys] = zh->keys.buf;
1515 zh->keys.buf = NULL;
1516 zh->keys.buf_max = 0;
1520 rec->info[recInfo_delKeys] = NULL;
1521 rec->size[recInfo_delKeys] = 0;
1524 /* save file size of original record */
1525 zebraExplain_recordBytesIncrement (zh->service->zei,
1526 - recordAttr->recordSize);
1528 recordAttr->recordSize = fi->file_moffset - recordOffset;
1529 if (!recordAttr->recordSize)
1530 recordAttr->recordSize = fi->file_max - recordOffset;
1532 recordAttr->recordSize = buf_size;
1534 zebraExplain_recordBytesIncrement (zh->service->zei,
1535 recordAttr->recordSize);
1537 /* set run-number for this record */
1538 recordAttr->runNumber =
1539 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1541 /* update store data */
1542 xfree (rec->info[recInfo_storeData]);
1543 if (store_data == 1)
1545 rec->size[recInfo_storeData] = recordAttr->recordSize;
1546 rec->info[recInfo_storeData] = (char *)
1547 xmalloc (recordAttr->recordSize);
1549 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1551 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1553 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1554 (long) recordOffset, fname);
1557 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1558 < recordAttr->recordSize)
1560 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1561 recordAttr->recordSize, fname);
1568 rec->info[recInfo_storeData] = NULL;
1569 rec->size[recInfo_storeData] = 0;
1571 /* update database name */
1572 xfree (rec->info[recInfo_databaseName]);
1573 rec->info[recInfo_databaseName] =
1574 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1577 recordAttr->recordOffset = recordOffset;
1579 /* commit this record */
1580 rec_put (zh->service->records, &rec);