2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.35 2000-07-07 12:49:20 adam
7 * Optimized resultSetInsert{Rank,Sort}.
9 * Revision 1.34 2000/06/09 13:56:38 ian
10 * Added some logging on Authentication and searches.
12 * Revision 1.33 2000/05/18 12:01:36 adam
13 * System call times(2) used again. More 64-bit fixes.
15 * Revision 1.32 2000/04/19 14:35:59 adam
16 * WIN32 update (this version is known not to work on Windows).
18 * Revision 1.31 2000/04/05 10:07:02 adam
19 * Minor zebra compile fix.
21 * Revision 1.30 2000/04/05 09:49:35 adam
22 * On Unix, zebra/z'mbol uses automake.
24 * Revision 1.29 2000/03/20 19:08:36 adam
25 * Added remote record import using Z39.50 extended services and Segment
28 * Revision 1.28 2000/03/15 15:00:30 adam
29 * First work on threaded version.
31 * Revision 1.27 2000/02/24 12:31:17 adam
32 * Added zebra_string_norm.
34 * Revision 1.26 1999/11/30 13:48:03 adam
35 * Improved installation. Updated for inclusion of YAZ header files.
37 * Revision 1.25 1999/11/04 15:00:45 adam
38 * Implemented delete result set(s).
40 * Revision 1.24 1999/10/14 14:33:50 adam
41 * Added truncation 5=106.
43 * Revision 1.23 1999/09/07 11:36:32 adam
46 * Revision 1.22 1999/08/02 10:13:47 adam
47 * Fixed bug regarding zebra_hits.
49 * Revision 1.21 1999/07/14 10:59:26 adam
50 * Changed functions isc_getmethod, isams_getmethod.
51 * Improved fatal error handling (such as missing EXPLAIN schema).
53 * Revision 1.20 1999/07/06 12:28:04 adam
54 * Updated record index structure. Format includes version ID. Compression
55 * algorithm ID is stored for each record block.
57 * Revision 1.19 1999/05/26 07:49:13 adam
60 * Revision 1.18 1999/05/15 14:36:38 adam
61 * Updated dictionary. Implemented "compression" of dictionary.
63 * Revision 1.17 1999/05/12 13:08:06 adam
64 * First version of ISAMS.
66 * Revision 1.16 1999/02/19 10:38:30 adam
67 * Implemented chdir-setting.
69 * Revision 1.15 1999/02/17 12:18:12 adam
70 * Fixed zebra_close so that a NULL pointer is ignored.
72 * Revision 1.14 1999/02/02 14:51:11 adam
73 * Updated WIN32 code specific sections. Changed header.
75 * Revision 1.13 1998/12/16 12:23:30 adam
76 * Added facility for database name mapping using resource mapdb.
78 * Revision 1.12 1998/11/16 10:18:10 adam
79 * Better error reporting for result sets.
81 * Revision 1.11 1998/10/16 08:14:34 adam
82 * Updated record control system.
84 * Revision 1.10 1998/09/22 10:03:42 adam
85 * Changed result sets to be persistent in the sense that they can
86 * be re-searched if needed.
87 * Fixed memory leak in rsm_or.
89 * Revision 1.9 1998/09/02 13:53:17 adam
90 * Extra parameter decode added to search routines to implement
93 * Revision 1.8 1998/08/24 17:29:23 adam
96 * Revision 1.7 1998/06/24 12:16:13 adam
97 * Support for relations on text operands. Open range support in
98 * DFA module (i.e. [-j], [g-]).
100 * Revision 1.6 1998/06/22 11:36:47 adam
101 * Added authentication check facility to zebra.
103 * Revision 1.5 1998/06/13 00:14:08 adam
106 * Revision 1.4 1998/06/12 12:22:12 adam
109 * Revision 1.3 1998/05/27 16:57:44 adam
110 * Zebra returns surrogate diagnostic for single records when
113 * Revision 1.2 1998/05/20 10:12:19 adam
114 * Implemented automatic EXPLAIN database maintenance.
115 * Modified Zebra to work with ASN.1 compiled version of YAZ.
117 * Revision 1.1 1998/03/05 08:45:13 adam
118 * New result set model and modular ranking system. Moved towards
119 * descent server API. System information stored as "SGML" records.
133 #include <yaz/diagbib1.h>
137 static void zebra_chdir (ZebraService zh)
139 const char *dir = res_get (zh->res, "chdir");
142 logf (LOG_DEBUG, "chdir %s", dir);
150 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
151 const char *buf, size_t buf_size,
152 const char *databaseName, int delete_flag,
153 int test_mode, int *sysno,
154 int store_keys, int store_data,
155 const char *match_criteria);
157 static int explain_extract (void *handle, Record rec, data1_node *n);
158 static void extract_index (ZebraHandle zh);
160 static void zebra_register_unlock (ZebraHandle zh);
162 static int zebra_register_lock (ZebraHandle zh)
166 if (!zh->service->active)
177 static void zebra_register_unlock (ZebraHandle zh)
181 logf (LOG_LOG, "user/system: %ld/%ld",
182 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
183 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
188 ZebraHandle zebra_open (ZebraService zs)
196 zh = (ZebraHandle) xmalloc (sizeof(*zh));
205 zh->admin_databaseName = 0;
207 zebra_mutex_cond_lock (&zs->session_lock);
209 zh->next = zs->sessions;
212 zebra_mutex_cond_unlock (&zs->session_lock);
214 logf(LOG_APP,"CONNECT:");
219 static int zebra_register_activate (ZebraService zh);
220 static int zebra_register_deactivate (ZebraService zh);
222 ZebraService zebra_start (const char *configName)
224 ZebraService zh = xmalloc (sizeof(*zh));
226 yaz_log (LOG_LOG, "zebra_start %s", configName);
228 zh->configName = xstrdup(configName);
232 zebra_mutex_cond_init (&zh->session_lock);
233 zebra_register_activate (zh);
237 static int zebra_register_activate (ZebraService zh)
241 yaz_log (LOG_LOG, "zebra_register_activate");
242 if (!(zh->res = res_open (zh->configName)))
244 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
248 zh->dh = data1_create ();
251 zh->bfs = bfs_create (res_get (zh->res, "register"));
254 data1_destroy(zh->dh);
257 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
258 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
259 zh->registerState = -1; /* trigger open of registers! */
260 zh->registerChange = 0;
261 zh->recTypes = recTypes_init (zh->dh);
262 recTypes_default_handlers (zh->recTypes);
265 zh->zebra_maps = zebra_maps_open (zh->res);
266 zh->rank_classes = NULL;
279 zebraRankInstall (zh, rank1_class);
281 if (!res_get (zh->res, "passwd"))
282 zh->passwd_db = NULL;
285 zh->passwd_db = passwd_db_open ();
287 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
289 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
292 if (!(zh->records = rec_open (zh->bfs, 1, 0)))
294 logf (LOG_WARN, "rec_open");
297 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, 1, 0)))
299 logf (LOG_WARN, "dict_open");
302 if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0)))
304 logf (LOG_WARN, "sortIdx_open");
307 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
309 struct ISAMS_M_s isams_m;
310 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, 1,
311 key_isams_m(zh->res, &isams_m))))
313 logf (LOG_WARN, "isams_open");
318 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
320 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, 1,
321 sizeof (struct it_key), zh->res)))
323 logf (LOG_WARN, "is_open");
327 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
329 struct ISAMC_M_s isamc_m;
330 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
331 1, key_isamc_m(zh->res, &isamc_m))))
333 logf (LOG_WARN, "isc_open");
337 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
339 struct ISAMD_M_s isamd_m;
341 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
342 1, key_isamd_m(zh->res, &isamd_m))))
344 logf (LOG_WARN, "isamd_open");
349 zh->zei = zebraExplain_open (zh->records, zh->dh,
350 zh->res, 1, 0 /* rGroup */,
354 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
358 yaz_log (LOG_LOG, "zebra_register_activate ok");
362 void zebra_admin_shutdown (ZebraHandle zh)
364 zebraExplain_flush (zh->service->zei, 1, zh);
367 zebra_mutex_cond_lock (&zh->service->session_lock);
368 zh->service->stop_flag = 1;
369 if (!zh->service->sessions)
370 zebra_register_deactivate(zh->service);
371 zebra_mutex_cond_unlock (&zh->service->session_lock);
374 void zebra_admin_start (ZebraHandle zh)
376 ZebraService zs = zh->service;
378 zebra_mutex_cond_lock (&zs->session_lock);
380 zebra_register_activate(zs);
381 zebra_mutex_cond_unlock (&zs->session_lock);
384 static int zebra_register_deactivate (ZebraService zh)
389 yaz_log(LOG_LOG, "zebra_register_deactivate");
393 zebraExplain_close (zh->zei, 0);
394 dict_close (zh->dict);
395 sortIdx_close (zh->sortIdx);
397 isams_close (zh->isams);
402 isc_close (zh->isamc);
404 isamd_close (zh->isamd);
406 rec_close (&zh->records);
408 recTypes_destroy (zh->recTypes);
409 zebra_maps_close (zh->zebra_maps);
410 zebraRankDestroy (zh);
411 bfs_destroy (zh->bfs);
412 data1_destroy (zh->dh);
415 passwd_db_close (zh->passwd_db);
421 void zebra_stop(ZebraService zh)
425 yaz_log (LOG_LOG, "zebra_stop");
427 assert (!zh->sessions);
429 zebra_mutex_cond_destroy (&zh->session_lock);
431 zebra_register_deactivate(zh);
432 xfree (zh->configName);
436 void zebra_close (ZebraHandle zh)
438 ZebraService zs = zh->service;
439 struct zebra_session **sp;
442 resultSetDestroy (zh, -1, 0, 0);
449 xfree (zh->admin_databaseName);
450 zebra_mutex_cond_lock (&zs->session_lock);
462 if (!zs->sessions && zs->stop_flag)
463 zebra_register_deactivate(zs);
464 zebra_mutex_cond_unlock (&zs->session_lock);
468 struct map_baseinfo {
474 char **new_basenames;
478 void map_basenames_func (void *vp, const char *name, const char *value)
480 struct map_baseinfo *p = (struct map_baseinfo *) vp;
482 char fromdb[128], todb[8][128];
485 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
486 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
487 todb[5], todb[6], todb[7]);
491 for (i = 0; i<p->num_bases; i++)
492 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
495 for (i = 0; i < no; i++)
497 if (p->new_num_bases == p->new_num_max)
499 p->new_basenames[(p->new_num_bases)++] =
500 nmem_strdup (p->mem, todb[i]);
506 void map_basenames (ZebraHandle zh, ODR stream,
507 int *num_bases, char ***basenames)
509 struct map_baseinfo info;
510 struct map_baseinfo *p = &info;
514 info.num_bases = *num_bases;
515 info.basenames = *basenames;
516 info.new_num_max = 128;
517 info.new_num_bases = 0;
518 info.new_basenames = (char **)
519 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
520 info.mem = stream->mem;
522 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
524 for (i = 0; i<p->num_bases; i++)
525 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
527 p->new_basenames[(p->new_num_bases)++] =
528 nmem_strdup (p->mem, p->basenames[i]);
530 *num_bases = info.new_num_bases;
531 *basenames = info.new_basenames;
532 for (i = 0; i<*num_bases; i++)
533 logf (LOG_LOG, "base %s", (*basenames)[i]);
536 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
537 Z_RPNQuery *query, int num_bases, char **basenames,
541 if (zebra_register_lock (zh))
543 map_basenames (zh, stream, &num_bases, &basenames);
544 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
546 zebra_register_unlock (zh);
548 logf(LOG_APP,"SEARCH:%d:",zh->hits);
551 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
552 const char *setname, Z_RecordComposition *comp,
553 oid_value input_format, int num_recs,
554 ZebraRetrievalRecord *recs)
559 if (zebra_register_lock (zh))
561 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
562 for (i = 0; i<num_recs; i++)
563 pos_array[i] = recs[i].position;
564 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
567 logf (LOG_DEBUG, "zebraPosSetCreate error");
569 zh->errString = nmem_strdup (stream->mem, setname);
573 for (i = 0; i<num_recs; i++)
579 sprintf (num_str, "%d", pos_array[i]);
581 zh->errString = nmem_strdup (stream->mem, num_str);
587 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
588 stream, input_format, comp,
589 &recs[i].format, &recs[i].buf,
592 recs[i].errString = NULL;
595 zebraPosSetDestroy (zh, poset, num_recs);
597 zebra_register_unlock (zh);
601 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
602 oid_value attributeset,
603 int num_bases, char **basenames,
604 int *position, int *num_entries, ZebraScanEntry **entries,
607 if (zebra_register_lock (zh))
613 map_basenames (zh, stream, &num_bases, &basenames);
614 rpn_scan (zh, stream, zapt, attributeset,
615 num_bases, basenames, position,
616 num_entries, entries, is_partial);
617 zebra_register_unlock (zh);
620 void zebra_sort (ZebraHandle zh, ODR stream,
621 int num_input_setnames, const char **input_setnames,
622 const char *output_setname, Z_SortKeySpecList *sort_sequence,
625 if (zebra_register_lock (zh))
627 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
628 output_setname, sort_sequence, sort_status);
629 zebra_register_unlock (zh);
632 int zebra_deleleResultSet(ZebraHandle zh, int function,
633 int num_setnames, char **setnames,
637 if (zebra_register_lock (zh))
638 return Z_DeleteStatus_systemProblemAtTarget;
641 case Z_DeleteRequest_list:
642 resultSetDestroy (zh, num_setnames, setnames, statuses);
644 case Z_DeleteRequest_all:
645 resultSetDestroy (zh, -1, 0, statuses);
648 zebra_register_unlock (zh);
649 status = Z_DeleteStatus_success;
650 for (i = 0; i<num_setnames; i++)
651 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
652 status = statuses[i];
656 int zebra_errCode (ZebraHandle zh)
661 const char *zebra_errString (ZebraHandle zh)
663 return diagbib1_str (zh->errCode);
666 char *zebra_errAdd (ZebraHandle zh)
668 return zh->errString;
671 int zebra_hits (ZebraHandle zh)
676 int zebra_auth (ZebraService zh, const char *user, const char *pass)
678 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
680 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
684 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
688 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
690 if (zebra_register_lock (zh))
692 xfree (zh->admin_databaseName);
693 zh->admin_databaseName = xstrdup(database);
694 zebra_register_unlock(zh);
697 void zebra_admin_import_end (ZebraHandle zh)
699 zebraExplain_flush (zh->service->zei, 1, zh);
703 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
707 if (zebra_register_lock (zh))
709 for (i = 0; i<segment->num_segmentRecords; i++)
711 Z_NamePlusRecord *npr = segment->segmentRecords[i];
712 const char *databaseName = npr->databaseName;
715 databaseName = zh->admin_databaseName;
716 printf ("--------------%d--------------------\n", i);
717 if (npr->which == Z_NamePlusRecord_intermediateFragment)
719 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
720 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
722 Odr_oct *oct = fragment->u.notExternallyTagged;
723 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
727 extract_rec_in_mem (zh, "grs.sgml",
735 0 /* match criteria */);
739 zebra_register_unlock(zh);
742 void zebra_admin_create (ZebraHandle zh, const char *database)
744 ZebraService zs = zh->service;
745 if (zebra_register_lock(zh))
750 /* announce database */
751 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
754 zh->errString = "Database already exist";
756 zebra_register_unlock(zh);
759 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
760 const char *input_str, int input_len,
761 char *output_str, int output_len)
764 if (!zh->service->zebra_maps)
766 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
767 input_str, input_len);
770 if (wrbuf_len(wrbuf) >= output_len)
772 if (wrbuf_len(wrbuf))
773 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
774 output_str[wrbuf_len(wrbuf)] = '\0';
775 return wrbuf_len(wrbuf);
778 static void extract_init (struct recExtractCtrl *p, RecWord *w)
780 w->zebra_maps = p->zebra_maps;
781 w->seqnos = p->seqno;
782 w->attrSet = VAL_BIB1;
788 static void extract_add_index_string (RecWord *p, const char *string,
792 unsigned char attrSet;
793 unsigned short attrUse;
796 int *pseqno = &p->seqnos[p->reg_type];
797 ZebraHandle zh = p->extractCtrl->handle;
798 struct recKeys *keys = &zh->keys;
800 if (keys->buf_used+1024 > keys->buf_max)
804 b = (char *) xmalloc (keys->buf_max += 128000);
805 if (keys->buf_used > 0)
806 memcpy (b, keys->buf, keys->buf_used);
810 dst = keys->buf + keys->buf_used;
812 attrSet = p->attrSet;
813 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
816 keys->prevAttrSet = attrSet;
817 attrUse = p->attrUse;
818 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
821 keys->prevAttrUse = attrUse;
823 diff = 1 + *pseqno - keys->prevSeqNo;
824 if (diff >= 1 && diff <= 15)
829 keys->prevSeqNo = *pseqno;
835 memcpy (dst, &attrSet, sizeof(attrSet));
836 dst += sizeof(attrSet);
840 memcpy (dst, &attrUse, sizeof(attrUse));
841 dst += sizeof(attrUse);
843 *dst++ = p->reg_type;
844 memcpy (dst, string, length);
850 memcpy (dst, pseqno, sizeof(*pseqno));
851 dst += sizeof(*pseqno);
853 keys->buf_used = dst - keys->buf;
858 static void extract_add_sort_string (RecWord *p, const char *string,
862 ZebraHandle zh = p->extractCtrl->handle;
863 struct sortKey *sortKeys = zh->sortKeys;
865 for (sk = sortKeys; sk; sk = sk->next)
866 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
869 sk = (struct sortKey *) xmalloc (sizeof(*sk));
873 sk->string = (char *) xmalloc (length);
875 memcpy (sk->string, string, length);
877 sk->attrSet = p->attrSet;
878 sk->attrUse = p->attrUse;
881 static void extract_add_string (RecWord *p, const char *string, int length)
884 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
885 extract_add_sort_string (p, string, length);
887 extract_add_index_string (p, string, length);
890 static void extract_add_incomplete_field (RecWord *p)
892 const char *b = p->string;
893 int remain = p->length;
894 const char **map = 0;
897 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
901 char buf[IT_MAX_WORD+1];
905 while (map && *map && **map == *CHR_SPACE)
907 remain = p->length - (b - p->string);
909 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
916 while (map && *map && **map != *CHR_SPACE)
918 const char *cp = *map;
920 while (i < IT_MAX_WORD && *cp)
922 remain = p->length - (b - p->string);
924 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
930 extract_add_string (p, buf, i);
932 (p->seqnos[p->reg_type])++; /* to separate this from next one */
935 static void extract_add_complete_field (RecWord *p)
937 const char *b = p->string;
938 char buf[IT_MAX_WORD+1];
939 const char **map = 0;
940 int i = 0, remain = p->length;
943 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
945 while (remain > 0 && i < IT_MAX_WORD)
947 while (map && *map && **map == *CHR_SPACE)
949 remain = p->length - (b - p->string);
951 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
958 if (i && i < IT_MAX_WORD)
959 buf[i++] = *CHR_SPACE;
960 while (map && *map && **map != *CHR_SPACE)
962 const char *cp = *map;
964 if (i >= IT_MAX_WORD)
966 while (i < IT_MAX_WORD && *cp)
968 remain = p->length - (b - p->string);
970 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
978 extract_add_string (p, buf, i);
981 static void extract_token_add (RecWord *p)
984 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
985 p->string, p->length)))
987 p->string = wrbuf_buf(wrbuf);
988 p->length = wrbuf_len(wrbuf);
990 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
991 extract_add_complete_field (p);
993 extract_add_incomplete_field(p);
996 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
998 ZebraHandle zh = (ZebraHandle) (p->handle);
999 zebraExplain_addSchema (zh->service->zei, oid);
1002 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1003 int cmd, struct sortKey **skp)
1005 struct sortKey *sk = *skp;
1006 SortIdx sortIdx = zh->service->sortIdx;
1008 sortIdx_sysno (sortIdx, sysno);
1011 struct sortKey *sk_next = sk->next;
1012 sortIdx_type (sortIdx, sk->attrUse);
1013 sortIdx_add (sortIdx, sk->string, sk->length);
1021 struct encode_info {
1028 void encode_key_init (struct encode_info *i)
1035 char *encode_key_int (int d, char *bp)
1039 else if (d <= 16383)
1041 *bp++ = 64 + (d>>8);
1044 else if (d <= 4194303)
1046 *bp++ = 128 + (d>>16);
1047 *bp++ = (d>>8) & 255;
1052 *bp++ = 192 + (d>>24);
1053 *bp++ = (d>>16) & 255;
1054 *bp++ = (d>>8) & 255;
1060 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1065 while ((*bp++ = *k++))
1067 memcpy (&key, k+1, sizeof(struct it_key));
1068 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1069 if (i->sysno != key.sysno)
1071 i->sysno = key.sysno;
1074 else if (!i->seqno && !key.seqno && i->cmd == *k)
1076 bp = encode_key_int (key.seqno - i->seqno, bp);
1077 i->seqno = key.seqno;
1079 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1081 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1086 static void extract_flushWriteKeys (ZebraHandle zh)
1089 char out_fname[200];
1091 struct encode_info encode_info;
1092 int ptr_i = zh->ptr_i;
1096 if (!zh->key_buf || ptr_i <= 0)
1099 (zh->key_file_no)++;
1100 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1102 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1104 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1106 if (!(outf = fopen (out_fname, "wb")))
1108 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1111 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1112 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1114 encode_key_init (&encode_info);
1115 encode_key_write (cp, &encode_info, outf);
1119 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1120 if (strcmp (cp, prevcp))
1122 encode_key_init (&encode_info);
1123 encode_key_write (cp, &encode_info, outf);
1127 encode_key_write (cp + strlen(cp), &encode_info, outf);
1130 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1131 extract_get_fname_tmp (out_fname, key_file_no);
1133 if (!(outf = fopen (out_fname, "wb")))
1135 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1138 logf (LOG_LOG, "writing section %d", key_file_no);
1140 prevcp = key_buf[ptr_top-i];
1142 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1144 key_y_len = strlen(prevcp)+1;
1146 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1147 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1149 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1150 sizeof(char*), key_y_compare);
1151 cp = key_buf[ptr_top-ptr_i];
1153 encode_key_init (&encode_info);
1154 encode_key_write (cp, &encode_info, outf);
1157 cp = key_buf[ptr_top-ptr_i];
1158 encode_key_write (cp+key_y_len, &encode_info, outf);
1162 prevcp = key_buf[ptr_top-ptr_i];
1167 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1170 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1172 zh->key_buf_used = 0;
1175 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1176 int cmd, struct recKeys *reckeys)
1178 unsigned char attrSet = (unsigned char) -1;
1179 unsigned short attrUse = (unsigned short) -1;
1182 ZebraExplainInfo zei = zh->service->zei;
1186 int mem = 8*1024*1024;
1187 zh->key_buf = (char**) xmalloc (mem);
1188 zh->ptr_top = mem/sizeof(char*);
1190 zh->key_buf_used = 0;
1191 zh->key_file_no = 0;
1193 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1194 while (off < reckeys->buf_used)
1196 const char *src = reckeys->buf + off;
1204 memcpy (&attrSet, src, sizeof(attrSet));
1205 src += sizeof(attrSet);
1209 memcpy (&attrUse, src, sizeof(attrUse));
1210 src += sizeof(attrUse);
1212 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1213 extract_flushWriteKeys (zh);
1215 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1216 (char*)zh->key_buf + zh->key_buf_used;
1218 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1220 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1223 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1226 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1228 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1229 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1232 seqno += ((lead>>2) & 15)-1;
1235 memcpy (&seqno, src, sizeof(seqno));
1236 src += sizeof(seqno);
1240 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1241 (zh->key_buf_used) += sizeof(key);
1242 off = src - reckeys->buf;
1244 assert (off == reckeys->buf_used);
1247 static void extract_index (ZebraHandle zh)
1249 extract_flushWriteKeys (zh);
1250 zebra_index_merge (zh);
1253 static int explain_extract (void *handle, Record rec, data1_node *n)
1255 ZebraHandle zh = (ZebraHandle) handle;
1256 struct recExtractCtrl extractCtrl;
1259 if (zebraExplain_curDatabase (zh->service->zei,
1260 rec->info[recInfo_databaseName]))
1263 if (zebraExplain_newDatabase (zh->service->zei,
1264 rec->info[recInfo_databaseName], 0))
1268 zh->keys.buf_used = 0;
1269 zh->keys.prevAttrUse = -1;
1270 zh->keys.prevAttrSet = -1;
1271 zh->keys.prevSeqNo = 0;
1274 extractCtrl.init = extract_init;
1275 extractCtrl.tokenAdd = extract_token_add;
1276 extractCtrl.schemaAdd = extract_schema_add;
1277 extractCtrl.dh = zh->service->dh;
1278 for (i = 0; i<256; i++)
1279 extractCtrl.seqno[i] = 0;
1280 extractCtrl.zebra_maps = zh->service->zebra_maps;
1281 extractCtrl.flagShowRecords = 0;
1282 extractCtrl.handle = handle;
1284 grs_extract_tree(&extractCtrl, n);
1286 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1288 if (rec->size[recInfo_delKeys])
1290 struct recKeys delkeys;
1291 struct sortKey *sortKeys = 0;
1293 delkeys.buf_used = rec->size[recInfo_delKeys];
1294 delkeys.buf = rec->info[recInfo_delKeys];
1295 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1296 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1298 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1299 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1301 xfree (rec->info[recInfo_delKeys]);
1302 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1303 rec->info[recInfo_delKeys] = zh->keys.buf;
1304 zh->keys.buf = NULL;
1305 zh->keys.buf_max = 0;
1309 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1310 const char *buf, size_t buf_size,
1311 const char *databaseName, int delete_flag,
1312 int test_mode, int *sysno,
1313 int store_keys, int store_data,
1314 const char *match_criteria)
1316 RecordAttr *recordAttr;
1317 struct recExtractCtrl extractCtrl;
1322 const char *fname = "<no file>";
1324 long recordOffset = 0;
1325 struct zebra_fetch_control fc;
1328 fc.record_int_buf = buf;
1329 fc.record_int_len = buf_size;
1330 fc.record_int_pos = 0;
1332 fc.record_offset = 0;
1334 extractCtrl.offset = 0;
1335 extractCtrl.readf = zebra_record_int_read;
1336 extractCtrl.seekf = zebra_record_int_seek;
1337 extractCtrl.tellf = zebra_record_int_tell;
1338 extractCtrl.endf = zebra_record_int_end;
1339 extractCtrl.fh = &fc;
1341 /* announce database */
1342 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1344 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1348 recType_byName (zh->service->recTypes, recordType, subType,
1351 logf (LOG_WARN, "No such record type: %s", recordType);
1355 zh->keys.buf_used = 0;
1356 zh->keys.prevAttrUse = -1;
1357 zh->keys.prevAttrSet = -1;
1358 zh->keys.prevSeqNo = 0;
1361 extractCtrl.subType = subType;
1362 extractCtrl.init = extract_init;
1363 extractCtrl.tokenAdd = extract_token_add;
1364 extractCtrl.schemaAdd = extract_schema_add;
1365 extractCtrl.dh = zh->service->dh;
1366 extractCtrl.handle = zh;
1367 extractCtrl.zebra_maps = zh->service->zebra_maps;
1368 extractCtrl.flagShowRecords = 0;
1369 for (i = 0; i<256; i++)
1371 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1372 extractCtrl.seqno[i] = 1;
1374 extractCtrl.seqno[i] = 0;
1377 r = (*recType->extract)(clientData, &extractCtrl);
1379 if (r == RECCTRL_EXTRACT_EOF)
1381 else if (r == RECCTRL_EXTRACT_ERROR)
1383 /* error occured during extraction ... */
1385 yaz_log (LOG_WARN, "extract error");
1387 if (rGroup->flagRw &&
1388 records_processed < rGroup->fileVerboseLimit)
1390 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1391 fname, (long) recordOffset);
1396 if (zh->keys.buf_used == 0)
1398 /* the extraction process returned no information - the record
1399 is probably empty - unless flagShowRecords is in use */
1402 logf (LOG_WARN, "No keys generated for record");
1403 logf (LOG_WARN, " The file is probably empty");
1406 /* match criteria */
1413 logf (LOG_LOG, "delete %s %s %ld", recordType,
1414 fname, (long) recordOffset);
1415 logf (LOG_WARN, "cannot delete record above (seems new)");
1418 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1419 (long) recordOffset);
1420 rec = rec_new (zh->service->records);
1422 *sysno = rec->sysno;
1424 recordAttr = rec_init_attr (zh->service->zei, rec);
1429 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1432 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1433 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1437 /* record already exists */
1438 struct recKeys delkeys;
1440 rec = rec_get (zh->service->records, *sysno);
1443 recordAttr = rec_init_attr (zh->service->zei, rec);
1445 if (recordAttr->runNumber ==
1446 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1448 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1449 fname, (long) recordOffset);
1453 delkeys.buf_used = rec->size[recInfo_delKeys];
1454 delkeys.buf = rec->info[recInfo_delKeys];
1455 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1456 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1459 /* record going to be deleted */
1460 if (!delkeys.buf_used)
1462 logf (LOG_LOG, "delete %s %s %ld", recordType,
1463 fname, (long) recordOffset);
1464 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1468 logf (LOG_LOG, "delete %s %s %ld", recordType,
1469 fname, (long) recordOffset);
1472 dict_delete (matchDict, matchStr);
1474 rec_del (zh->service->records, &rec);
1481 /* record going to be updated */
1482 if (!delkeys.buf_used)
1484 logf (LOG_LOG, "update %s %s %ld", recordType,
1485 fname, (long) recordOffset);
1486 logf (LOG_WARN, "cannot update file above, storeKeys false");
1490 logf (LOG_LOG, "update %s %s %ld", recordType,
1491 fname, (long) recordOffset);
1492 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1496 /* update file type */
1497 xfree (rec->info[recInfo_fileType]);
1498 rec->info[recInfo_fileType] =
1499 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1501 /* update filename */
1502 xfree (rec->info[recInfo_filename]);
1503 rec->info[recInfo_filename] =
1504 rec_strdup (fname, &rec->size[recInfo_filename]);
1506 /* update delete keys */
1507 xfree (rec->info[recInfo_delKeys]);
1508 if (zh->keys.buf_used > 0 && store_keys == 1)
1510 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1511 rec->info[recInfo_delKeys] = zh->keys.buf;
1512 zh->keys.buf = NULL;
1513 zh->keys.buf_max = 0;
1517 rec->info[recInfo_delKeys] = NULL;
1518 rec->size[recInfo_delKeys] = 0;
1521 /* save file size of original record */
1522 zebraExplain_recordBytesIncrement (zh->service->zei,
1523 - recordAttr->recordSize);
1525 recordAttr->recordSize = fi->file_moffset - recordOffset;
1526 if (!recordAttr->recordSize)
1527 recordAttr->recordSize = fi->file_max - recordOffset;
1529 recordAttr->recordSize = buf_size;
1531 zebraExplain_recordBytesIncrement (zh->service->zei,
1532 recordAttr->recordSize);
1534 /* set run-number for this record */
1535 recordAttr->runNumber =
1536 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1538 /* update store data */
1539 xfree (rec->info[recInfo_storeData]);
1540 if (store_data == 1)
1542 rec->size[recInfo_storeData] = recordAttr->recordSize;
1543 rec->info[recInfo_storeData] = (char *)
1544 xmalloc (recordAttr->recordSize);
1546 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1548 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1550 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1551 (long) recordOffset, fname);
1554 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1555 < recordAttr->recordSize)
1557 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1558 recordAttr->recordSize, fname);
1565 rec->info[recInfo_storeData] = NULL;
1566 rec->size[recInfo_storeData] = 0;
1568 /* update database name */
1569 xfree (rec->info[recInfo_databaseName]);
1570 rec->info[recInfo_databaseName] =
1571 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1574 recordAttr->recordOffset = recordOffset;
1576 /* commit this record */
1577 rec_put (zh->service->records, &rec);