2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.34 2000-06-09 13:56:38 ian
7 * Added some logging on Authentication and searches.
9 * Revision 1.33 2000/05/18 12:01:36 adam
10 * System call times(2) used again. More 64-bit fixes.
12 * Revision 1.32 2000/04/19 14:35:59 adam
13 * WIN32 update (this version is known not to work on Windows).
15 * Revision 1.31 2000/04/05 10:07:02 adam
16 * Minor zebra compile fix.
18 * Revision 1.30 2000/04/05 09:49:35 adam
19 * On Unix, zebra/z'mbol uses automake.
21 * Revision 1.29 2000/03/20 19:08:36 adam
22 * Added remote record import using Z39.50 extended services and Segment
25 * Revision 1.28 2000/03/15 15:00:30 adam
26 * First work on threaded version.
28 * Revision 1.27 2000/02/24 12:31:17 adam
29 * Added zebra_string_norm.
31 * Revision 1.26 1999/11/30 13:48:03 adam
32 * Improved installation. Updated for inclusion of YAZ header files.
34 * Revision 1.25 1999/11/04 15:00:45 adam
35 * Implemented delete result set(s).
37 * Revision 1.24 1999/10/14 14:33:50 adam
38 * Added truncation 5=106.
40 * Revision 1.23 1999/09/07 11:36:32 adam
43 * Revision 1.22 1999/08/02 10:13:47 adam
44 * Fixed bug regarding zebra_hits.
46 * Revision 1.21 1999/07/14 10:59:26 adam
47 * Changed functions isc_getmethod, isams_getmethod.
48 * Improved fatal error handling (such as missing EXPLAIN schema).
50 * Revision 1.20 1999/07/06 12:28:04 adam
51 * Updated record index structure. Format includes version ID. Compression
52 * algorithm ID is stored for each record block.
54 * Revision 1.19 1999/05/26 07:49:13 adam
57 * Revision 1.18 1999/05/15 14:36:38 adam
58 * Updated dictionary. Implemented "compression" of dictionary.
60 * Revision 1.17 1999/05/12 13:08:06 adam
61 * First version of ISAMS.
63 * Revision 1.16 1999/02/19 10:38:30 adam
64 * Implemented chdir-setting.
66 * Revision 1.15 1999/02/17 12:18:12 adam
67 * Fixed zebra_close so that a NULL pointer is ignored.
69 * Revision 1.14 1999/02/02 14:51:11 adam
70 * Updated WIN32 code specific sections. Changed header.
72 * Revision 1.13 1998/12/16 12:23:30 adam
73 * Added facility for database name mapping using resource mapdb.
75 * Revision 1.12 1998/11/16 10:18:10 adam
76 * Better error reporting for result sets.
78 * Revision 1.11 1998/10/16 08:14:34 adam
79 * Updated record control system.
81 * Revision 1.10 1998/09/22 10:03:42 adam
82 * Changed result sets to be persistent in the sense that they can
83 * be re-searched if needed.
84 * Fixed memory leak in rsm_or.
86 * Revision 1.9 1998/09/02 13:53:17 adam
87 * Extra parameter decode added to search routines to implement
90 * Revision 1.8 1998/08/24 17:29:23 adam
93 * Revision 1.7 1998/06/24 12:16:13 adam
94 * Support for relations on text operands. Open range support in
95 * DFA module (i.e. [-j], [g-]).
97 * Revision 1.6 1998/06/22 11:36:47 adam
98 * Added authentication check facility to zebra.
100 * Revision 1.5 1998/06/13 00:14:08 adam
103 * Revision 1.4 1998/06/12 12:22:12 adam
106 * Revision 1.3 1998/05/27 16:57:44 adam
107 * Zebra returns surrogate diagnostic for single records when
110 * Revision 1.2 1998/05/20 10:12:19 adam
111 * Implemented automatic EXPLAIN database maintenance.
112 * Modified Zebra to work with ASN.1 compiled version of YAZ.
114 * Revision 1.1 1998/03/05 08:45:13 adam
115 * New result set model and modular ranking system. Moved towards
116 * descent server API. System information stored as "SGML" records.
130 #include <yaz/diagbib1.h>
134 static void zebra_chdir (ZebraService zh)
136 const char *dir = res_get (zh->res, "chdir");
139 logf (LOG_DEBUG, "chdir %s", dir);
147 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
148 const char *buf, size_t buf_size,
149 const char *databaseName, int delete_flag,
150 int test_mode, int *sysno,
151 int store_keys, int store_data,
152 const char *match_criteria);
154 static int explain_extract (void *handle, Record rec, data1_node *n);
155 static void extract_index (ZebraHandle zh);
157 static void zebra_register_unlock (ZebraHandle zh);
159 static int zebra_register_lock (ZebraHandle zh)
163 if (!zh->service->active)
174 static void zebra_register_unlock (ZebraHandle zh)
178 logf (LOG_LOG, "user/system: %ld/%ld",
179 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
180 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
185 ZebraHandle zebra_open (ZebraService zs)
193 zh = (ZebraHandle) xmalloc (sizeof(*zh));
202 zh->admin_databaseName = 0;
204 zebra_mutex_cond_lock (&zs->session_lock);
206 zh->next = zs->sessions;
209 zebra_mutex_cond_unlock (&zs->session_lock);
211 logf(LOG_APP,"CONNECT:");
216 static int zebra_register_activate (ZebraService zh);
217 static int zebra_register_deactivate (ZebraService zh);
219 ZebraService zebra_start (const char *configName)
221 ZebraService zh = xmalloc (sizeof(*zh));
223 yaz_log (LOG_LOG, "zebra_start %s", configName);
225 zh->configName = xstrdup(configName);
229 zebra_mutex_cond_init (&zh->session_lock);
230 zebra_register_activate (zh);
234 static int zebra_register_activate (ZebraService zh)
238 yaz_log (LOG_LOG, "zebra_register_activate");
239 if (!(zh->res = res_open (zh->configName)))
241 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
245 zh->dh = data1_create ();
248 zh->bfs = bfs_create (res_get (zh->res, "register"));
251 data1_destroy(zh->dh);
254 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
255 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
256 zh->registerState = -1; /* trigger open of registers! */
257 zh->registerChange = 0;
258 zh->recTypes = recTypes_init (zh->dh);
259 recTypes_default_handlers (zh->recTypes);
262 zh->zebra_maps = zebra_maps_open (zh->res);
263 zh->rank_classes = NULL;
276 zebraRankInstall (zh, rank1_class);
278 if (!res_get (zh->res, "passwd"))
279 zh->passwd_db = NULL;
282 zh->passwd_db = passwd_db_open ();
284 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
286 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
289 if (!(zh->records = rec_open (zh->bfs, 1, 0)))
291 logf (LOG_WARN, "rec_open");
294 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, 1, 0)))
296 logf (LOG_WARN, "dict_open");
299 if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0)))
301 logf (LOG_WARN, "sortIdx_open");
304 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
306 struct ISAMS_M_s isams_m;
307 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, 1,
308 key_isams_m(zh->res, &isams_m))))
310 logf (LOG_WARN, "isams_open");
315 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
317 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, 1,
318 sizeof (struct it_key), zh->res)))
320 logf (LOG_WARN, "is_open");
324 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
326 struct ISAMC_M_s isamc_m;
327 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
328 1, key_isamc_m(zh->res, &isamc_m))))
330 logf (LOG_WARN, "isc_open");
334 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
336 struct ISAMD_M_s isamd_m;
338 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
339 1, key_isamd_m(zh->res, &isamd_m))))
341 logf (LOG_WARN, "isamd_open");
346 zh->zei = zebraExplain_open (zh->records, zh->dh,
347 zh->res, 1, 0 /* rGroup */,
351 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
355 yaz_log (LOG_LOG, "zebra_register_activate ok");
359 void zebra_admin_shutdown (ZebraHandle zh)
361 zebraExplain_flush (zh->service->zei, 1, zh);
364 zebra_mutex_cond_lock (&zh->service->session_lock);
365 zh->service->stop_flag = 1;
366 if (!zh->service->sessions)
367 zebra_register_deactivate(zh->service);
368 zebra_mutex_cond_unlock (&zh->service->session_lock);
371 void zebra_admin_start (ZebraHandle zh)
373 ZebraService zs = zh->service;
375 zebra_mutex_cond_lock (&zs->session_lock);
377 zebra_register_activate(zs);
378 zebra_mutex_cond_unlock (&zs->session_lock);
381 static int zebra_register_deactivate (ZebraService zh)
386 yaz_log(LOG_LOG, "zebra_register_deactivate");
390 zebraExplain_close (zh->zei, 1);
391 dict_close (zh->dict);
392 sortIdx_close (zh->sortIdx);
394 isams_close (zh->isams);
399 isc_close (zh->isamc);
401 isamd_close (zh->isamd);
403 rec_close (&zh->records);
405 recTypes_destroy (zh->recTypes);
406 zebra_maps_close (zh->zebra_maps);
407 zebraRankDestroy (zh);
408 bfs_destroy (zh->bfs);
409 data1_destroy (zh->dh);
412 passwd_db_close (zh->passwd_db);
418 void zebra_stop(ZebraService zh)
422 yaz_log (LOG_LOG, "zebra_stop");
424 assert (!zh->sessions);
426 zebra_mutex_cond_destroy (&zh->session_lock);
428 zebra_register_deactivate(zh);
429 xfree (zh->configName);
433 void zebra_close (ZebraHandle zh)
435 ZebraService zs = zh->service;
436 struct zebra_session **sp;
439 resultSetDestroy (zh, -1, 0, 0);
446 xfree (zh->admin_databaseName);
447 zebra_mutex_cond_lock (&zs->session_lock);
459 if (!zs->sessions && zs->stop_flag)
460 zebra_register_deactivate(zs);
461 zebra_mutex_cond_unlock (&zs->session_lock);
465 struct map_baseinfo {
471 char **new_basenames;
475 void map_basenames_func (void *vp, const char *name, const char *value)
477 struct map_baseinfo *p = (struct map_baseinfo *) vp;
479 char fromdb[128], todb[8][128];
482 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
483 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
484 todb[5], todb[6], todb[7]);
488 for (i = 0; i<p->num_bases; i++)
489 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
492 for (i = 0; i < no; i++)
494 if (p->new_num_bases == p->new_num_max)
496 p->new_basenames[(p->new_num_bases)++] =
497 nmem_strdup (p->mem, todb[i]);
503 void map_basenames (ZebraHandle zh, ODR stream,
504 int *num_bases, char ***basenames)
506 struct map_baseinfo info;
507 struct map_baseinfo *p = &info;
511 info.num_bases = *num_bases;
512 info.basenames = *basenames;
513 info.new_num_max = 128;
514 info.new_num_bases = 0;
515 info.new_basenames = (char **)
516 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
517 info.mem = stream->mem;
519 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
521 for (i = 0; i<p->num_bases; i++)
522 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
524 p->new_basenames[(p->new_num_bases)++] =
525 nmem_strdup (p->mem, p->basenames[i]);
527 *num_bases = info.new_num_bases;
528 *basenames = info.new_basenames;
529 for (i = 0; i<*num_bases; i++)
530 logf (LOG_LOG, "base %s", (*basenames)[i]);
533 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
534 Z_RPNQuery *query, int num_bases, char **basenames,
538 if (zebra_register_lock (zh))
540 map_basenames (zh, stream, &num_bases, &basenames);
541 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
543 zebra_register_unlock (zh);
545 logf(LOG_APP,"SEARCH:%d:",zh->hits);
548 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
549 const char *setname, Z_RecordComposition *comp,
550 oid_value input_format, int num_recs,
551 ZebraRetrievalRecord *recs)
556 if (zebra_register_lock (zh))
558 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
559 for (i = 0; i<num_recs; i++)
560 pos_array[i] = recs[i].position;
561 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
564 logf (LOG_DEBUG, "zebraPosSetCreate error");
566 zh->errString = nmem_strdup (stream->mem, setname);
570 for (i = 0; i<num_recs; i++)
576 sprintf (num_str, "%d", pos_array[i]);
578 zh->errString = nmem_strdup (stream->mem, num_str);
584 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
585 stream, input_format, comp,
586 &recs[i].format, &recs[i].buf,
589 recs[i].errString = NULL;
592 zebraPosSetDestroy (zh, poset, num_recs);
594 zebra_register_unlock (zh);
598 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
599 oid_value attributeset,
600 int num_bases, char **basenames,
601 int *position, int *num_entries, ZebraScanEntry **entries,
604 if (zebra_register_lock (zh))
610 map_basenames (zh, stream, &num_bases, &basenames);
611 rpn_scan (zh, stream, zapt, attributeset,
612 num_bases, basenames, position,
613 num_entries, entries, is_partial);
614 zebra_register_unlock (zh);
617 void zebra_sort (ZebraHandle zh, ODR stream,
618 int num_input_setnames, const char **input_setnames,
619 const char *output_setname, Z_SortKeySpecList *sort_sequence,
622 if (zebra_register_lock (zh))
624 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
625 output_setname, sort_sequence, sort_status);
626 zebra_register_unlock (zh);
629 int zebra_deleleResultSet(ZebraHandle zh, int function,
630 int num_setnames, char **setnames,
634 if (zebra_register_lock (zh))
635 return Z_DeleteStatus_systemProblemAtTarget;
638 case Z_DeleteRequest_list:
639 resultSetDestroy (zh, num_setnames, setnames, statuses);
641 case Z_DeleteRequest_all:
642 resultSetDestroy (zh, -1, 0, statuses);
645 zebra_register_unlock (zh);
646 status = Z_DeleteStatus_success;
647 for (i = 0; i<num_setnames; i++)
648 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
649 status = statuses[i];
653 int zebra_errCode (ZebraHandle zh)
658 const char *zebra_errString (ZebraHandle zh)
660 return diagbib1_str (zh->errCode);
663 char *zebra_errAdd (ZebraHandle zh)
665 return zh->errString;
668 int zebra_hits (ZebraHandle zh)
673 int zebra_auth (ZebraService zh, const char *user, const char *pass)
675 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
677 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
681 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
685 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
687 if (zebra_register_lock (zh))
689 xfree (zh->admin_databaseName);
690 zh->admin_databaseName = xstrdup(database);
691 zebra_register_unlock(zh);
694 void zebra_admin_import_end (ZebraHandle zh)
696 zebraExplain_flush (zh->service->zei, 1, zh);
700 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
704 if (zebra_register_lock (zh))
706 for (i = 0; i<segment->num_segmentRecords; i++)
708 Z_NamePlusRecord *npr = segment->segmentRecords[i];
709 const char *databaseName = npr->databaseName;
712 databaseName = zh->admin_databaseName;
713 printf ("--------------%d--------------------\n", i);
714 if (npr->which == Z_NamePlusRecord_intermediateFragment)
716 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
717 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
719 Odr_oct *oct = fragment->u.notExternallyTagged;
720 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
724 extract_rec_in_mem (zh, "grs.sgml",
732 0 /* match criteria */);
736 zebra_register_unlock(zh);
739 void zebra_admin_create (ZebraHandle zh, const char *database)
741 ZebraService zs = zh->service;
742 if (zebra_register_lock(zh))
747 /* announce database */
748 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
751 zh->errString = "Database already exist";
753 zebra_register_unlock(zh);
756 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
757 const char *input_str, int input_len,
758 char *output_str, int output_len)
761 if (!zh->service->zebra_maps)
763 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
764 input_str, input_len);
767 if (wrbuf_len(wrbuf) >= output_len)
769 if (wrbuf_len(wrbuf))
770 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
771 output_str[wrbuf_len(wrbuf)] = '\0';
772 return wrbuf_len(wrbuf);
775 static void extract_init (struct recExtractCtrl *p, RecWord *w)
777 w->zebra_maps = p->zebra_maps;
778 w->seqnos = p->seqno;
779 w->attrSet = VAL_BIB1;
785 static void extract_add_index_string (RecWord *p, const char *string,
789 unsigned char attrSet;
790 unsigned short attrUse;
793 int *pseqno = &p->seqnos[p->reg_type];
794 ZebraHandle zh = p->extractCtrl->handle;
795 struct recKeys *keys = &zh->keys;
797 if (keys->buf_used+1024 > keys->buf_max)
801 b = (char *) xmalloc (keys->buf_max += 128000);
802 if (keys->buf_used > 0)
803 memcpy (b, keys->buf, keys->buf_used);
807 dst = keys->buf + keys->buf_used;
809 attrSet = p->attrSet;
810 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
813 keys->prevAttrSet = attrSet;
814 attrUse = p->attrUse;
815 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
818 keys->prevAttrUse = attrUse;
820 diff = 1 + *pseqno - keys->prevSeqNo;
821 if (diff >= 1 && diff <= 15)
826 keys->prevSeqNo = *pseqno;
832 memcpy (dst, &attrSet, sizeof(attrSet));
833 dst += sizeof(attrSet);
837 memcpy (dst, &attrUse, sizeof(attrUse));
838 dst += sizeof(attrUse);
840 *dst++ = p->reg_type;
841 memcpy (dst, string, length);
847 memcpy (dst, pseqno, sizeof(*pseqno));
848 dst += sizeof(*pseqno);
850 keys->buf_used = dst - keys->buf;
855 static void extract_add_sort_string (RecWord *p, const char *string,
859 ZebraHandle zh = p->extractCtrl->handle;
860 struct sortKey *sortKeys = zh->sortKeys;
862 for (sk = sortKeys; sk; sk = sk->next)
863 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
866 sk = (struct sortKey *) xmalloc (sizeof(*sk));
870 sk->string = (char *) xmalloc (length);
872 memcpy (sk->string, string, length);
874 sk->attrSet = p->attrSet;
875 sk->attrUse = p->attrUse;
878 static void extract_add_string (RecWord *p, const char *string, int length)
881 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
882 extract_add_sort_string (p, string, length);
884 extract_add_index_string (p, string, length);
887 static void extract_add_incomplete_field (RecWord *p)
889 const char *b = p->string;
890 int remain = p->length;
891 const char **map = 0;
894 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
898 char buf[IT_MAX_WORD+1];
902 while (map && *map && **map == *CHR_SPACE)
904 remain = p->length - (b - p->string);
906 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
913 while (map && *map && **map != *CHR_SPACE)
915 const char *cp = *map;
917 while (i < IT_MAX_WORD && *cp)
919 remain = p->length - (b - p->string);
921 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
927 extract_add_string (p, buf, i);
929 (p->seqnos[p->reg_type])++; /* to separate this from next one */
932 static void extract_add_complete_field (RecWord *p)
934 const char *b = p->string;
935 char buf[IT_MAX_WORD+1];
936 const char **map = 0;
937 int i = 0, remain = p->length;
940 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
942 while (remain > 0 && i < IT_MAX_WORD)
944 while (map && *map && **map == *CHR_SPACE)
946 remain = p->length - (b - p->string);
948 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
955 if (i && i < IT_MAX_WORD)
956 buf[i++] = *CHR_SPACE;
957 while (map && *map && **map != *CHR_SPACE)
959 const char *cp = *map;
961 if (i >= IT_MAX_WORD)
963 while (i < IT_MAX_WORD && *cp)
965 remain = p->length - (b - p->string);
967 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
975 extract_add_string (p, buf, i);
978 static void extract_token_add (RecWord *p)
981 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
982 p->string, p->length)))
984 p->string = wrbuf_buf(wrbuf);
985 p->length = wrbuf_len(wrbuf);
987 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
988 extract_add_complete_field (p);
990 extract_add_incomplete_field(p);
993 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
995 ZebraHandle zh = (ZebraHandle) (p->handle);
996 zebraExplain_addSchema (zh->service->zei, oid);
999 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1000 int cmd, struct sortKey **skp)
1002 struct sortKey *sk = *skp;
1003 SortIdx sortIdx = zh->service->sortIdx;
1005 sortIdx_sysno (sortIdx, sysno);
1008 struct sortKey *sk_next = sk->next;
1009 sortIdx_type (sortIdx, sk->attrUse);
1010 sortIdx_add (sortIdx, sk->string, sk->length);
1018 struct encode_info {
1025 void encode_key_init (struct encode_info *i)
1032 char *encode_key_int (int d, char *bp)
1036 else if (d <= 16383)
1038 *bp++ = 64 + (d>>8);
1041 else if (d <= 4194303)
1043 *bp++ = 128 + (d>>16);
1044 *bp++ = (d>>8) & 255;
1049 *bp++ = 192 + (d>>24);
1050 *bp++ = (d>>16) & 255;
1051 *bp++ = (d>>8) & 255;
1057 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1062 while ((*bp++ = *k++))
1064 memcpy (&key, k+1, sizeof(struct it_key));
1065 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1066 if (i->sysno != key.sysno)
1068 i->sysno = key.sysno;
1071 else if (!i->seqno && !key.seqno && i->cmd == *k)
1073 bp = encode_key_int (key.seqno - i->seqno, bp);
1074 i->seqno = key.seqno;
1076 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1078 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1083 static void extract_flushWriteKeys (ZebraHandle zh)
1086 char out_fname[200];
1088 struct encode_info encode_info;
1089 int ptr_i = zh->ptr_i;
1093 if (!zh->key_buf || ptr_i <= 0)
1096 (zh->key_file_no)++;
1097 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1099 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1101 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1103 if (!(outf = fopen (out_fname, "wb")))
1105 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1108 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1109 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1111 encode_key_init (&encode_info);
1112 encode_key_write (cp, &encode_info, outf);
1116 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1117 if (strcmp (cp, prevcp))
1119 encode_key_init (&encode_info);
1120 encode_key_write (cp, &encode_info, outf);
1124 encode_key_write (cp + strlen(cp), &encode_info, outf);
1127 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1128 extract_get_fname_tmp (out_fname, key_file_no);
1130 if (!(outf = fopen (out_fname, "wb")))
1132 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1135 logf (LOG_LOG, "writing section %d", key_file_no);
1137 prevcp = key_buf[ptr_top-i];
1139 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1141 key_y_len = strlen(prevcp)+1;
1143 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1144 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1146 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1147 sizeof(char*), key_y_compare);
1148 cp = key_buf[ptr_top-ptr_i];
1150 encode_key_init (&encode_info);
1151 encode_key_write (cp, &encode_info, outf);
1154 cp = key_buf[ptr_top-ptr_i];
1155 encode_key_write (cp+key_y_len, &encode_info, outf);
1159 prevcp = key_buf[ptr_top-ptr_i];
1164 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1167 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1169 zh->key_buf_used = 0;
1172 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1173 int cmd, struct recKeys *reckeys)
1175 unsigned char attrSet = (unsigned char) -1;
1176 unsigned short attrUse = (unsigned short) -1;
1179 ZebraExplainInfo zei = zh->service->zei;
1183 int mem = 8*1024*1024;
1184 zh->key_buf = (char**) xmalloc (mem);
1185 zh->ptr_top = mem/sizeof(char*);
1187 zh->key_buf_used = 0;
1188 zh->key_file_no = 0;
1190 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1191 while (off < reckeys->buf_used)
1193 const char *src = reckeys->buf + off;
1201 memcpy (&attrSet, src, sizeof(attrSet));
1202 src += sizeof(attrSet);
1206 memcpy (&attrUse, src, sizeof(attrUse));
1207 src += sizeof(attrUse);
1209 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1210 extract_flushWriteKeys (zh);
1212 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1213 (char*)zh->key_buf + zh->key_buf_used;
1215 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1217 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1220 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1223 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1225 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1226 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1229 seqno += ((lead>>2) & 15)-1;
1232 memcpy (&seqno, src, sizeof(seqno));
1233 src += sizeof(seqno);
1237 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1238 (zh->key_buf_used) += sizeof(key);
1239 off = src - reckeys->buf;
1241 assert (off == reckeys->buf_used);
1244 static void extract_index (ZebraHandle zh)
1246 extract_flushWriteKeys (zh);
1247 zebra_index_merge (zh);
1250 static int explain_extract (void *handle, Record rec, data1_node *n)
1252 ZebraHandle zh = (ZebraHandle) handle;
1253 struct recExtractCtrl extractCtrl;
1256 if (zebraExplain_curDatabase (zh->service->zei,
1257 rec->info[recInfo_databaseName]))
1260 if (zebraExplain_newDatabase (zh->service->zei,
1261 rec->info[recInfo_databaseName], 0))
1265 zh->keys.buf_used = 0;
1266 zh->keys.prevAttrUse = -1;
1267 zh->keys.prevAttrSet = -1;
1268 zh->keys.prevSeqNo = 0;
1271 extractCtrl.init = extract_init;
1272 extractCtrl.tokenAdd = extract_token_add;
1273 extractCtrl.schemaAdd = extract_schema_add;
1274 extractCtrl.dh = zh->service->dh;
1275 for (i = 0; i<256; i++)
1276 extractCtrl.seqno[i] = 0;
1277 extractCtrl.zebra_maps = zh->service->zebra_maps;
1278 extractCtrl.flagShowRecords = 0;
1279 extractCtrl.handle = handle;
1281 grs_extract_tree(&extractCtrl, n);
1283 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1285 if (rec->size[recInfo_delKeys])
1287 struct recKeys delkeys;
1288 struct sortKey *sortKeys = 0;
1290 delkeys.buf_used = rec->size[recInfo_delKeys];
1291 delkeys.buf = rec->info[recInfo_delKeys];
1292 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1293 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1295 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1296 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1298 xfree (rec->info[recInfo_delKeys]);
1299 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1300 rec->info[recInfo_delKeys] = zh->keys.buf;
1301 zh->keys.buf = NULL;
1302 zh->keys.buf_max = 0;
1306 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1307 const char *buf, size_t buf_size,
1308 const char *databaseName, int delete_flag,
1309 int test_mode, int *sysno,
1310 int store_keys, int store_data,
1311 const char *match_criteria)
1313 RecordAttr *recordAttr;
1314 struct recExtractCtrl extractCtrl;
1319 const char *fname = "<no file>";
1321 long recordOffset = 0;
1322 struct zebra_fetch_control fc;
1325 fc.record_int_buf = buf;
1326 fc.record_int_len = buf_size;
1327 fc.record_int_pos = 0;
1329 fc.record_offset = 0;
1331 extractCtrl.offset = 0;
1332 extractCtrl.readf = zebra_record_int_read;
1333 extractCtrl.seekf = zebra_record_int_seek;
1334 extractCtrl.tellf = zebra_record_int_tell;
1335 extractCtrl.endf = zebra_record_int_end;
1336 extractCtrl.fh = &fc;
1338 /* announce database */
1339 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1341 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1345 recType_byName (zh->service->recTypes, recordType, subType,
1348 logf (LOG_WARN, "No such record type: %s", recordType);
1352 zh->keys.buf_used = 0;
1353 zh->keys.prevAttrUse = -1;
1354 zh->keys.prevAttrSet = -1;
1355 zh->keys.prevSeqNo = 0;
1358 extractCtrl.subType = subType;
1359 extractCtrl.init = extract_init;
1360 extractCtrl.tokenAdd = extract_token_add;
1361 extractCtrl.schemaAdd = extract_schema_add;
1362 extractCtrl.dh = zh->service->dh;
1363 extractCtrl.handle = zh;
1364 extractCtrl.zebra_maps = zh->service->zebra_maps;
1365 extractCtrl.flagShowRecords = 0;
1366 for (i = 0; i<256; i++)
1368 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1369 extractCtrl.seqno[i] = 1;
1371 extractCtrl.seqno[i] = 0;
1374 r = (*recType->extract)(clientData, &extractCtrl);
1376 if (r == RECCTRL_EXTRACT_EOF)
1378 else if (r == RECCTRL_EXTRACT_ERROR)
1380 /* error occured during extraction ... */
1382 yaz_log (LOG_WARN, "extract error");
1384 if (rGroup->flagRw &&
1385 records_processed < rGroup->fileVerboseLimit)
1387 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1388 fname, (long) recordOffset);
1393 if (zh->keys.buf_used == 0)
1395 /* the extraction process returned no information - the record
1396 is probably empty - unless flagShowRecords is in use */
1399 logf (LOG_WARN, "No keys generated for record");
1400 logf (LOG_WARN, " The file is probably empty");
1403 /* match criteria */
1410 logf (LOG_LOG, "delete %s %s %ld", recordType,
1411 fname, (long) recordOffset);
1412 logf (LOG_WARN, "cannot delete record above (seems new)");
1415 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1416 (long) recordOffset);
1417 rec = rec_new (zh->service->records);
1419 *sysno = rec->sysno;
1421 recordAttr = rec_init_attr (zh->service->zei, rec);
1426 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1429 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1430 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1434 /* record already exists */
1435 struct recKeys delkeys;
1437 rec = rec_get (zh->service->records, *sysno);
1440 recordAttr = rec_init_attr (zh->service->zei, rec);
1442 if (recordAttr->runNumber ==
1443 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1445 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1446 fname, (long) recordOffset);
1450 delkeys.buf_used = rec->size[recInfo_delKeys];
1451 delkeys.buf = rec->info[recInfo_delKeys];
1452 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1453 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1456 /* record going to be deleted */
1457 if (!delkeys.buf_used)
1459 logf (LOG_LOG, "delete %s %s %ld", recordType,
1460 fname, (long) recordOffset);
1461 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1465 logf (LOG_LOG, "delete %s %s %ld", recordType,
1466 fname, (long) recordOffset);
1469 dict_delete (matchDict, matchStr);
1471 rec_del (zh->service->records, &rec);
1478 /* record going to be updated */
1479 if (!delkeys.buf_used)
1481 logf (LOG_LOG, "update %s %s %ld", recordType,
1482 fname, (long) recordOffset);
1483 logf (LOG_WARN, "cannot update file above, storeKeys false");
1487 logf (LOG_LOG, "update %s %s %ld", recordType,
1488 fname, (long) recordOffset);
1489 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1493 /* update file type */
1494 xfree (rec->info[recInfo_fileType]);
1495 rec->info[recInfo_fileType] =
1496 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1498 /* update filename */
1499 xfree (rec->info[recInfo_filename]);
1500 rec->info[recInfo_filename] =
1501 rec_strdup (fname, &rec->size[recInfo_filename]);
1503 /* update delete keys */
1504 xfree (rec->info[recInfo_delKeys]);
1505 if (zh->keys.buf_used > 0 && store_keys == 1)
1507 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1508 rec->info[recInfo_delKeys] = zh->keys.buf;
1509 zh->keys.buf = NULL;
1510 zh->keys.buf_max = 0;
1514 rec->info[recInfo_delKeys] = NULL;
1515 rec->size[recInfo_delKeys] = 0;
1518 /* save file size of original record */
1519 zebraExplain_recordBytesIncrement (zh->service->zei,
1520 - recordAttr->recordSize);
1522 recordAttr->recordSize = fi->file_moffset - recordOffset;
1523 if (!recordAttr->recordSize)
1524 recordAttr->recordSize = fi->file_max - recordOffset;
1526 recordAttr->recordSize = buf_size;
1528 zebraExplain_recordBytesIncrement (zh->service->zei,
1529 recordAttr->recordSize);
1531 /* set run-number for this record */
1532 recordAttr->runNumber =
1533 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1535 /* update store data */
1536 xfree (rec->info[recInfo_storeData]);
1537 if (store_data == 1)
1539 rec->size[recInfo_storeData] = recordAttr->recordSize;
1540 rec->info[recInfo_storeData] = (char *)
1541 xmalloc (recordAttr->recordSize);
1543 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1545 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1547 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1548 (long) recordOffset, fname);
1551 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1552 < recordAttr->recordSize)
1554 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1555 recordAttr->recordSize, fname);
1562 rec->info[recInfo_storeData] = NULL;
1563 rec->size[recInfo_storeData] = 0;
1565 /* update database name */
1566 xfree (rec->info[recInfo_databaseName]);
1567 rec->info[recInfo_databaseName] =
1568 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1571 recordAttr->recordOffset = recordOffset;
1573 /* commit this record */
1574 rec_put (zh->service->records, &rec);