2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.30 2000-04-05 09:49:35 adam
7 * On Unix, zebra/z'mbol uses automake.
9 * Revision 1.29 2000/03/20 19:08:36 adam
10 * Added remote record import using Z39.50 extended services and Segment
13 * Revision 1.28 2000/03/15 15:00:30 adam
14 * First work on threaded version.
16 * Revision 1.27 2000/02/24 12:31:17 adam
17 * Added zebra_string_norm.
19 * Revision 1.26 1999/11/30 13:48:03 adam
20 * Improved installation. Updated for inclusion of YAZ header files.
22 * Revision 1.25 1999/11/04 15:00:45 adam
23 * Implemented delete result set(s).
25 * Revision 1.24 1999/10/14 14:33:50 adam
26 * Added truncation 5=106.
28 * Revision 1.23 1999/09/07 11:36:32 adam
31 * Revision 1.22 1999/08/02 10:13:47 adam
32 * Fixed bug regarding zebra_hits.
34 * Revision 1.21 1999/07/14 10:59:26 adam
35 * Changed functions isc_getmethod, isams_getmethod.
36 * Improved fatal error handling (such as missing EXPLAIN schema).
38 * Revision 1.20 1999/07/06 12:28:04 adam
39 * Updated record index structure. Format includes version ID. Compression
40 * algorithm ID is stored for each record block.
42 * Revision 1.19 1999/05/26 07:49:13 adam
45 * Revision 1.18 1999/05/15 14:36:38 adam
46 * Updated dictionary. Implemented "compression" of dictionary.
48 * Revision 1.17 1999/05/12 13:08:06 adam
49 * First version of ISAMS.
51 * Revision 1.16 1999/02/19 10:38:30 adam
52 * Implemented chdir-setting.
54 * Revision 1.15 1999/02/17 12:18:12 adam
55 * Fixed zebra_close so that a NULL pointer is ignored.
57 * Revision 1.14 1999/02/02 14:51:11 adam
58 * Updated WIN32 code specific sections. Changed header.
60 * Revision 1.13 1998/12/16 12:23:30 adam
61 * Added facility for database name mapping using resource mapdb.
63 * Revision 1.12 1998/11/16 10:18:10 adam
64 * Better error reporting for result sets.
66 * Revision 1.11 1998/10/16 08:14:34 adam
67 * Updated record control system.
69 * Revision 1.10 1998/09/22 10:03:42 adam
70 * Changed result sets to be persistent in the sense that they can
71 * be re-searched if needed.
72 * Fixed memory leak in rsm_or.
74 * Revision 1.9 1998/09/02 13:53:17 adam
75 * Extra parameter decode added to search routines to implement
78 * Revision 1.8 1998/08/24 17:29:23 adam
81 * Revision 1.7 1998/06/24 12:16:13 adam
82 * Support for relations on text operands. Open range support in
83 * DFA module (i.e. [-j], [g-]).
85 * Revision 1.6 1998/06/22 11:36:47 adam
86 * Added authentication check facility to zebra.
88 * Revision 1.5 1998/06/13 00:14:08 adam
91 * Revision 1.4 1998/06/12 12:22:12 adam
94 * Revision 1.3 1998/05/27 16:57:44 adam
95 * Zebra returns surrogate diagnostic for single records when
98 * Revision 1.2 1998/05/20 10:12:19 adam
99 * Implemented automatic EXPLAIN database maintenance.
100 * Modified Zebra to work with ASN.1 compiled version of YAZ.
102 * Revision 1.1 1998/03/05 08:45:13 adam
103 * New result set model and modular ranking system. Moved towards
104 * descent server API. System information stored as "SGML" records.
118 #include <yaz/diagbib1.h>
122 static void zebra_chdir (ZebraService zh)
124 const char *dir = res_get (zh->res, "chdir");
127 logf (LOG_DEBUG, "chdir %s", dir);
135 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
136 const char *buf, size_t buf_size,
137 const char *databaseName, int delete_flag,
138 int test_mode, int *sysno,
139 int store_keys, int store_data,
140 const char *match_criteria);
142 static int explain_extract (void *handle, Record rec, data1_node *n);
143 static void extract_index (ZebraHandle zh);
145 static void zebra_register_unlock (ZebraHandle zh);
147 static int zebra_register_lock (ZebraHandle zh)
151 if (!zh->service->active)
159 static void zebra_register_unlock (ZebraHandle zh)
163 ZebraHandle zebra_open (ZebraService zs)
174 zh = (ZebraHandle) xmalloc (sizeof(*zh));
183 zh->admin_databaseName = 0;
185 zebra_mutex_cond_lock (&zs->session_lock);
187 zh->next = zs->sessions;
190 zebra_mutex_cond_unlock (&zs->session_lock);
194 static int zebra_register_activate (ZebraService zh);
195 static int zebra_register_deactivate (ZebraService zh);
197 ZebraService zebra_start (const char *configName)
199 ZebraService zh = xmalloc (sizeof(*zh));
201 yaz_log (LOG_LOG, "zebra_start %s", configName);
203 zh->configName = xstrdup(configName);
207 zebra_mutex_cond_init (&zh->session_lock);
208 zebra_register_activate (zh);
212 static int zebra_register_activate (ZebraService zh)
216 yaz_log (LOG_LOG, "zebra_register_activate");
217 if (!(zh->res = res_open (zh->configName)))
219 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
223 zh->dh = data1_create ();
226 zh->bfs = bfs_create (res_get (zh->res, "register"));
229 data1_destroy(zh->dh);
232 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
233 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
234 zh->registerState = -1; /* trigger open of registers! */
235 zh->registerChange = 0;
236 zh->recTypes = recTypes_init (zh->dh);
237 recTypes_default_handlers (zh->recTypes);
240 zh->zebra_maps = zebra_maps_open (zh->res);
241 zh->rank_classes = NULL;
252 zebraRankInstall (zh, rank1_class);
254 if (!res_get (zh->res, "passwd"))
255 zh->passwd_db = NULL;
258 zh->passwd_db = passwd_db_open ();
260 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
262 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
265 if (!(zh->records = rec_open (zh->bfs, 1, 0)))
267 logf (LOG_WARN, "rec_open");
270 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 40, 1, 0)))
272 logf (LOG_WARN, "dict_open");
275 if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0)))
277 logf (LOG_WARN, "sortIdx_open");
280 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
282 struct ISAMS_M_s isams_m;
283 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, 1,
284 key_isams_m(zh->res, &isams_m))))
286 logf (LOG_WARN, "isams_open");
291 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
293 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, 1,
294 sizeof (struct it_key), zh->res)))
296 logf (LOG_WARN, "is_open");
300 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
302 struct ISAMC_M_s isamc_m;
303 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
304 1, key_isamc_m(zh->res, &isamc_m))))
306 logf (LOG_WARN, "isc_open");
310 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
312 struct ISAMD_M_s isamd_m;
314 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
315 1, key_isamd_m(zh->res, &isamd_m))))
317 logf (LOG_WARN, "isamd_open");
322 zh->zei = zebraExplain_open (zh->records, zh->dh,
323 zh->res, 1, 0 /* rGroup */,
327 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
331 yaz_log (LOG_LOG, "zebra_register_activate ok");
335 void zebra_admin_shutdown (ZebraHandle zh)
337 zebraExplain_flush (zh->service->zei, 1, zh);
340 zebra_mutex_cond_lock (&zh->service->session_lock);
341 zh->service->stop_flag = 1;
342 if (!zh->service->sessions)
343 zebra_register_deactivate(zh->service);
344 zebra_mutex_cond_unlock (&zh->service->session_lock);
347 void zebra_admin_start (ZebraHandle zh)
349 ZebraService zs = zh->service;
351 zebra_mutex_cond_lock (&zs->session_lock);
353 zebra_register_activate(zs);
354 zebra_mutex_cond_unlock (&zs->session_lock);
357 static int zebra_register_deactivate (ZebraService zh)
362 yaz_log(LOG_LOG, "zebra_register_deactivate");
366 zebraExplain_close (zh->zei, 1);
367 dict_close (zh->dict);
368 sortIdx_close (zh->sortIdx);
370 isams_close (zh->isams);
375 isc_close (zh->isamc);
377 isamd_close (zh->isamd);
379 rec_close (&zh->records);
381 recTypes_destroy (zh->recTypes);
382 zebra_maps_close (zh->zebra_maps);
383 zebraRankDestroy (zh);
384 bfs_destroy (zh->bfs);
385 data1_destroy (zh->dh);
388 passwd_db_close (zh->passwd_db);
394 void zebra_stop(ZebraService zh)
398 yaz_log (LOG_LOG, "zebra_stop");
400 assert (!zh->sessions);
402 zebra_mutex_cond_destroy (&zh->session_lock);
404 zebra_register_deactivate(zh);
405 xfree (zh->configName);
409 void zebra_close (ZebraHandle zh)
411 ZebraService zs = zh->service;
412 struct zebra_session **sp;
415 resultSetDestroy (zh, -1, 0, 0);
422 xfree (zh->admin_databaseName);
423 zebra_mutex_cond_lock (&zs->session_lock);
435 if (!zs->sessions && zs->stop_flag)
436 zebra_register_deactivate(zs);
437 zebra_mutex_cond_unlock (&zs->session_lock);
441 struct map_baseinfo {
447 char **new_basenames;
451 void map_basenames_func (void *vp, const char *name, const char *value)
453 struct map_baseinfo *p = (struct map_baseinfo *) vp;
455 char fromdb[128], todb[8][128];
458 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
459 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
460 todb[5], todb[6], todb[7]);
464 for (i = 0; i<p->num_bases; i++)
465 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
468 for (i = 0; i < no; i++)
470 if (p->new_num_bases == p->new_num_max)
472 p->new_basenames[(p->new_num_bases)++] =
473 nmem_strdup (p->mem, todb[i]);
479 void map_basenames (ZebraHandle zh, ODR stream,
480 int *num_bases, char ***basenames)
482 struct map_baseinfo info;
483 struct map_baseinfo *p = &info;
487 info.num_bases = *num_bases;
488 info.basenames = *basenames;
489 info.new_num_max = 128;
490 info.new_num_bases = 0;
491 info.new_basenames = (char **)
492 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
493 info.mem = stream->mem;
495 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
497 for (i = 0; i<p->num_bases; i++)
498 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
500 p->new_basenames[(p->new_num_bases)++] =
501 nmem_strdup (p->mem, p->basenames[i]);
503 *num_bases = info.new_num_bases;
504 *basenames = info.new_basenames;
505 for (i = 0; i<*num_bases; i++)
506 logf (LOG_LOG, "base %s", (*basenames)[i]);
509 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
510 Z_RPNQuery *query, int num_bases, char **basenames,
514 if (zebra_register_lock (zh))
516 map_basenames (zh, stream, &num_bases, &basenames);
517 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
519 zebra_register_unlock (zh);
522 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
523 const char *setname, Z_RecordComposition *comp,
524 oid_value input_format, int num_recs,
525 ZebraRetrievalRecord *recs)
530 if (zebra_register_lock (zh))
532 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
533 for (i = 0; i<num_recs; i++)
534 pos_array[i] = recs[i].position;
535 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
538 logf (LOG_DEBUG, "zebraPosSetCreate error");
540 zh->errString = nmem_strdup (stream->mem, setname);
544 for (i = 0; i<num_recs; i++)
550 sprintf (num_str, "%d", pos_array[i]);
552 zh->errString = nmem_strdup (stream->mem, num_str);
558 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
559 stream, input_format, comp,
560 &recs[i].format, &recs[i].buf,
563 recs[i].errString = NULL;
566 zebraPosSetDestroy (zh, poset, num_recs);
568 zebra_register_unlock (zh);
572 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
573 oid_value attributeset,
574 int num_bases, char **basenames,
575 int *position, int *num_entries, ZebraScanEntry **entries,
578 if (zebra_register_lock (zh))
584 map_basenames (zh, stream, &num_bases, &basenames);
585 rpn_scan (zh, stream, zapt, attributeset,
586 num_bases, basenames, position,
587 num_entries, entries, is_partial);
588 zebra_register_unlock (zh);
591 void zebra_sort (ZebraHandle zh, ODR stream,
592 int num_input_setnames, const char **input_setnames,
593 const char *output_setname, Z_SortKeySpecList *sort_sequence,
596 if (zebra_register_lock (zh))
598 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
599 output_setname, sort_sequence, sort_status);
600 zebra_register_unlock (zh);
603 int zebra_deleleResultSet(ZebraHandle zh, int function,
604 int num_setnames, char **setnames,
608 if (zebra_register_lock (zh))
609 return Z_DeleteStatus_systemProblemAtTarget;
612 case Z_DeleteRequest_list:
613 resultSetDestroy (zh, num_setnames, setnames, statuses);
615 case Z_DeleteRequest_all:
616 resultSetDestroy (zh, -1, 0, statuses);
619 zebra_register_unlock (zh);
620 status = Z_DeleteStatus_success;
621 for (i = 0; i<num_setnames; i++)
622 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
623 status = statuses[i];
627 int zebra_errCode (ZebraHandle zh)
632 const char *zebra_errString (ZebraHandle zh)
634 return diagbib1_str (zh->errCode);
637 char *zebra_errAdd (ZebraHandle zh)
639 return zh->errString;
642 int zebra_hits (ZebraHandle zh)
647 int zebra_auth (ZebraService zh, const char *user, const char *pass)
649 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
654 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
656 if (zebra_register_lock (zh))
658 xfree (zh->admin_databaseName);
659 zh->admin_databaseName = xstrdup(database);
660 zebra_register_unlock(zh);
663 void zebra_admin_import_end (ZebraHandle zh)
665 zebraExplain_flush (zh->service->zei, 1, zh);
669 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
673 if (zebra_register_lock (zh))
675 for (i = 0; i<segment->num_segmentRecords; i++)
677 Z_NamePlusRecord *npr = segment->segmentRecords[i];
678 const char *databaseName = npr->databaseName;
681 databaseName = zh->admin_databaseName;
682 printf ("--------------%d--------------------\n", i);
683 if (npr->which == Z_NamePlusRecord_intermediateFragment)
685 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
686 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
688 Odr_oct *oct = fragment->u.notExternallyTagged;
689 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
693 extract_rec_in_mem (zh, "grs.sgml",
701 0 /* match criteria */);
705 zebra_register_unlock(zh);
708 void zebra_admin_create (ZebraHandle zh, const char *database)
710 ZebraService zs = zh->service;
711 if (zebra_register_lock(zh))
716 /* announce database */
717 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
720 zh->errString = "Database already exist";
722 zebra_register_unlock(zh);
725 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
726 const char *input_str, int input_len,
727 char *output_str, int output_len)
730 if (!zh->service->zebra_maps)
732 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
733 input_str, input_len);
736 if (wrbuf_len(wrbuf) >= output_len)
738 if (wrbuf_len(wrbuf))
739 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
740 output_str[wrbuf_len(wrbuf)] = '\0';
741 return wrbuf_len(wrbuf);
744 static void extract_init (struct recExtractCtrl *p, RecWord *w)
746 w->zebra_maps = p->zebra_maps;
747 w->seqnos = p->seqno;
748 w->attrSet = VAL_BIB1;
754 static void extract_add_index_string (RecWord *p, const char *string,
758 unsigned char attrSet;
759 unsigned short attrUse;
762 int *pseqno = &p->seqnos[p->reg_type];
763 ZebraHandle zh = p->extractCtrl->handle;
764 struct recKeys *keys = &zh->keys;
766 if (keys->buf_used+1024 > keys->buf_max)
770 b = (char *) xmalloc (keys->buf_max += 128000);
771 if (keys->buf_used > 0)
772 memcpy (b, keys->buf, keys->buf_used);
776 dst = keys->buf + keys->buf_used;
778 attrSet = p->attrSet;
779 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
782 keys->prevAttrSet = attrSet;
783 attrUse = p->attrUse;
784 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
787 keys->prevAttrUse = attrUse;
789 diff = 1 + *pseqno - keys->prevSeqNo;
790 if (diff >= 1 && diff <= 15)
795 keys->prevSeqNo = *pseqno;
801 memcpy (dst, &attrSet, sizeof(attrSet));
802 dst += sizeof(attrSet);
806 memcpy (dst, &attrUse, sizeof(attrUse));
807 dst += sizeof(attrUse);
809 *dst++ = p->reg_type;
810 memcpy (dst, string, length);
816 memcpy (dst, pseqno, sizeof(*pseqno));
817 dst += sizeof(*pseqno);
819 keys->buf_used = dst - keys->buf;
824 static void extract_add_sort_string (RecWord *p, const char *string,
828 ZebraHandle zh = p->extractCtrl->handle;
829 struct sortKey *sortKeys = zh->sortKeys;
831 for (sk = sortKeys; sk; sk = sk->next)
832 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
835 sk = (struct sortKey *) xmalloc (sizeof(*sk));
839 sk->string = (char *) xmalloc (length);
841 memcpy (sk->string, string, length);
843 sk->attrSet = p->attrSet;
844 sk->attrUse = p->attrUse;
847 static void extract_add_string (RecWord *p, const char *string, int length)
850 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
851 extract_add_sort_string (p, string, length);
853 extract_add_index_string (p, string, length);
856 static void extract_add_incomplete_field (RecWord *p)
858 const char *b = p->string;
859 int remain = p->length;
860 const char **map = 0;
863 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
867 char buf[IT_MAX_WORD+1];
871 while (map && *map && **map == *CHR_SPACE)
873 remain = p->length - (b - p->string);
875 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
882 while (map && *map && **map != *CHR_SPACE)
884 const char *cp = *map;
886 while (i < IT_MAX_WORD && *cp)
888 remain = p->length - (b - p->string);
890 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
896 extract_add_string (p, buf, i);
898 (p->seqnos[p->reg_type])++; /* to separate this from next one */
901 static void extract_add_complete_field (RecWord *p)
903 const char *b = p->string;
904 char buf[IT_MAX_WORD+1];
905 const char **map = 0;
906 int i = 0, remain = p->length;
909 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
911 while (remain > 0 && i < IT_MAX_WORD)
913 while (map && *map && **map == *CHR_SPACE)
915 remain = p->length - (b - p->string);
917 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
924 if (i && i < IT_MAX_WORD)
925 buf[i++] = *CHR_SPACE;
926 while (map && *map && **map != *CHR_SPACE)
928 const char *cp = *map;
930 if (i >= IT_MAX_WORD)
932 while (i < IT_MAX_WORD && *cp)
934 remain = p->length - (b - p->string);
936 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
944 extract_add_string (p, buf, i);
947 static void extract_token_add (RecWord *p)
950 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
951 p->string, p->length)))
953 p->string = wrbuf_buf(wrbuf);
954 p->length = wrbuf_len(wrbuf);
956 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
957 extract_add_complete_field (p);
959 extract_add_incomplete_field(p);
962 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
964 ZebraHandle zh = (ZebraHandle) (p->handle);
965 zebraExplain_addSchema (zh->service->zei, oid);
968 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
969 int cmd, struct sortKey **skp)
971 struct sortKey *sk = *skp;
972 SortIdx sortIdx = zh->service->sortIdx;
974 sortIdx_sysno (sortIdx, sysno);
977 struct sortKey *sk_next = sk->next;
978 sortIdx_type (sortIdx, sk->attrUse);
979 sortIdx_add (sortIdx, sk->string, sk->length);
994 void encode_key_init (struct encode_info *i)
1001 char *encode_key_int (int d, char *bp)
1005 else if (d <= 16383)
1007 *bp++ = 64 + (d>>8);
1010 else if (d <= 4194303)
1012 *bp++ = 128 + (d>>16);
1013 *bp++ = (d>>8) & 255;
1018 *bp++ = 192 + (d>>24);
1019 *bp++ = (d>>16) & 255;
1020 *bp++ = (d>>8) & 255;
1026 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1031 while ((*bp++ = *k++))
1033 memcpy (&key, k+1, sizeof(struct it_key));
1034 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1035 if (i->sysno != key.sysno)
1037 i->sysno = key.sysno;
1040 else if (!i->seqno && !key.seqno && i->cmd == *k)
1042 bp = encode_key_int (key.seqno - i->seqno, bp);
1043 i->seqno = key.seqno;
1045 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1047 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1052 static void extract_flushWriteKeys (ZebraHandle zh)
1055 char out_fname[200];
1057 struct encode_info encode_info;
1058 int ptr_i = zh->ptr_i;
1062 if (!zh->key_buf || ptr_i <= 0)
1065 (zh->key_file_no)++;
1066 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1068 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1070 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1072 if (!(outf = fopen (out_fname, "wb")))
1074 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1077 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1078 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1080 encode_key_init (&encode_info);
1081 encode_key_write (cp, &encode_info, outf);
1085 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1086 if (strcmp (cp, prevcp))
1088 encode_key_init (&encode_info);
1089 encode_key_write (cp, &encode_info, outf);
1093 encode_key_write (cp + strlen(cp), &encode_info, outf);
1096 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1097 extract_get_fname_tmp (out_fname, key_file_no);
1099 if (!(outf = fopen (out_fname, "wb")))
1101 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1104 logf (LOG_LOG, "writing section %d", key_file_no);
1106 prevcp = key_buf[ptr_top-i];
1108 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1110 key_y_len = strlen(prevcp)+1;
1112 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1113 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1115 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1116 sizeof(char*), key_y_compare);
1117 cp = key_buf[ptr_top-ptr_i];
1119 encode_key_init (&encode_info);
1120 encode_key_write (cp, &encode_info, outf);
1123 cp = key_buf[ptr_top-ptr_i];
1124 encode_key_write (cp+key_y_len, &encode_info, outf);
1128 prevcp = key_buf[ptr_top-ptr_i];
1133 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1136 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1138 zh->key_buf_used = 0;
1141 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1142 int cmd, struct recKeys *reckeys)
1144 unsigned char attrSet = (unsigned char) -1;
1145 unsigned short attrUse = (unsigned short) -1;
1148 ZebraExplainInfo zei = zh->service->zei;
1152 int mem = 8*1024*1024;
1153 zh->key_buf = (char**) xmalloc (mem);
1154 zh->ptr_top = mem/sizeof(char*);
1156 zh->key_buf_used = 0;
1157 zh->key_file_no = 0;
1159 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1160 while (off < reckeys->buf_used)
1162 const char *src = reckeys->buf + off;
1170 memcpy (&attrSet, src, sizeof(attrSet));
1171 src += sizeof(attrSet);
1175 memcpy (&attrUse, src, sizeof(attrUse));
1176 src += sizeof(attrUse);
1178 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1179 extract_flushWriteKeys (zh);
1181 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1182 (char*)zh->key_buf + zh->key_buf_used;
1184 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1186 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1189 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1192 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1194 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1195 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1198 seqno += ((lead>>2) & 15)-1;
1201 memcpy (&seqno, src, sizeof(seqno));
1202 src += sizeof(seqno);
1206 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1207 (zh->key_buf_used) += sizeof(key);
1208 off = src - reckeys->buf;
1210 assert (off == reckeys->buf_used);
1213 static void extract_index (ZebraHandle zh)
1215 extract_flushWriteKeys (zh);
1216 zebra_index_merge (zh);
1219 static int explain_extract (void *handle, Record rec, data1_node *n)
1221 ZebraHandle zh = (ZebraHandle) handle;
1222 struct recExtractCtrl extractCtrl;
1225 if (zebraExplain_curDatabase (zh->service->zei,
1226 rec->info[recInfo_databaseName]))
1229 if (zebraExplain_newDatabase (zh->service->zei,
1230 rec->info[recInfo_databaseName], 0))
1234 zh->keys.buf_used = 0;
1235 zh->keys.prevAttrUse = -1;
1236 zh->keys.prevAttrSet = -1;
1237 zh->keys.prevSeqNo = 0;
1240 extractCtrl.init = extract_init;
1241 extractCtrl.tokenAdd = extract_token_add;
1242 extractCtrl.schemaAdd = extract_schema_add;
1243 extractCtrl.dh = zh->service->dh;
1244 for (i = 0; i<256; i++)
1245 extractCtrl.seqno[i] = 0;
1246 extractCtrl.zebra_maps = zh->service->zebra_maps;
1247 extractCtrl.flagShowRecords = 0;
1248 extractCtrl.handle = handle;
1250 grs_extract_tree(&extractCtrl, n);
1252 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1254 if (rec->size[recInfo_delKeys])
1256 struct recKeys delkeys;
1257 struct sortKey *sortKeys = 0;
1259 delkeys.buf_used = rec->size[recInfo_delKeys];
1260 delkeys.buf = rec->info[recInfo_delKeys];
1261 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1262 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1264 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1265 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1267 xfree (rec->info[recInfo_delKeys]);
1268 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1269 rec->info[recInfo_delKeys] = zh->keys.buf;
1270 zh->keys.buf = NULL;
1271 zh->keys.buf_max = 0;
1275 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1276 const char *buf, size_t buf_size,
1277 const char *databaseName, int delete_flag,
1278 int test_mode, int *sysno,
1279 int store_keys, int store_data,
1280 const char *match_criteria)
1282 RecordAttr *recordAttr;
1283 struct recExtractCtrl extractCtrl;
1288 const char *fname = "<no file>";
1290 long recordOffset = 0;
1291 struct zebra_fetch_control fc;
1294 fc.record_int_buf = buf;
1295 fc.record_int_len = buf_size;
1296 fc.record_int_pos = 0;
1298 fc.record_offset = 0;
1300 extractCtrl.offset = 0;
1301 extractCtrl.readf = zebra_record_int_read;
1302 extractCtrl.seekf = zebra_record_int_seek;
1303 extractCtrl.tellf = zebra_record_int_tell;
1304 extractCtrl.endf = zebra_record_int_end;
1305 extractCtrl.fh = &fc;
1307 /* announce database */
1308 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1310 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1314 recType_byName (zh->service->recTypes, recordType, subType,
1317 logf (LOG_WARN, "No such record type: %s", recordType);
1321 zh->keys.buf_used = 0;
1322 zh->keys.prevAttrUse = -1;
1323 zh->keys.prevAttrSet = -1;
1324 zh->keys.prevSeqNo = 0;
1327 extractCtrl.subType = subType;
1328 extractCtrl.init = extract_init;
1329 extractCtrl.tokenAdd = extract_token_add;
1330 extractCtrl.schemaAdd = extract_schema_add;
1331 extractCtrl.dh = zh->service->dh;
1332 extractCtrl.handle = zh;
1333 extractCtrl.zebra_maps = zh->service->zebra_maps;
1334 extractCtrl.flagShowRecords = 0;
1335 for (i = 0; i<256; i++)
1337 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1338 extractCtrl.seqno[i] = 1;
1340 extractCtrl.seqno[i] = 0;
1343 r = (*recType->extract)(clientData, &extractCtrl);
1345 if (r == RECCTRL_EXTRACT_EOF)
1347 else if (r == RECCTRL_EXTRACT_ERROR)
1349 /* error occured during extraction ... */
1351 yaz_log (LOG_WARN, "extract error");
1353 if (rGroup->flagRw &&
1354 records_processed < rGroup->fileVerboseLimit)
1356 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1357 fname, (long) recordOffset);
1362 if (zh->keys.buf_used == 0)
1364 /* the extraction process returned no information - the record
1365 is probably empty - unless flagShowRecords is in use */
1368 logf (LOG_WARN, "No keys generated for record");
1369 logf (LOG_WARN, " The file is probably empty");
1372 /* match criteria */
1379 logf (LOG_LOG, "delete %s %s %ld", recordType,
1380 fname, (long) recordOffset);
1381 logf (LOG_WARN, "cannot delete record above (seems new)");
1384 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1385 (long) recordOffset);
1386 rec = rec_new (zh->service->records);
1388 *sysno = rec->sysno;
1390 recordAttr = rec_init_attr (zh->service->zei, rec);
1395 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1398 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1399 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1403 /* record already exists */
1404 struct recKeys delkeys;
1406 rec = rec_get (zh->service->records, *sysno);
1409 recordAttr = rec_init_attr (zh->service->zei, rec);
1411 if (recordAttr->runNumber ==
1412 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1414 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1415 fname, (long) recordOffset);
1419 delkeys.buf_used = rec->size[recInfo_delKeys];
1420 delkeys.buf = rec->info[recInfo_delKeys];
1421 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1422 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1425 /* record going to be deleted */
1426 if (!delkeys.buf_used)
1428 logf (LOG_LOG, "delete %s %s %ld", recordType,
1429 fname, (long) recordOffset);
1430 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1434 logf (LOG_LOG, "delete %s %s %ld", recordType,
1435 fname, (long) recordOffset);
1438 dict_delete (matchDict, matchStr);
1440 rec_del (zh->service->records, &rec);
1447 /* record going to be updated */
1448 if (!delkeys.buf_used)
1450 logf (LOG_LOG, "update %s %s %ld", recordType,
1451 fname, (long) recordOffset);
1452 logf (LOG_WARN, "cannot update file above, storeKeys false");
1456 logf (LOG_LOG, "update %s %s %ld", recordType,
1457 fname, (long) recordOffset);
1458 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1462 /* update file type */
1463 xfree (rec->info[recInfo_fileType]);
1464 rec->info[recInfo_fileType] =
1465 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1467 /* update filename */
1468 xfree (rec->info[recInfo_filename]);
1469 rec->info[recInfo_filename] =
1470 rec_strdup (fname, &rec->size[recInfo_filename]);
1472 /* update delete keys */
1473 xfree (rec->info[recInfo_delKeys]);
1474 if (zh->keys.buf_used > 0 && store_keys == 1)
1476 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1477 rec->info[recInfo_delKeys] = zh->keys.buf;
1478 zh->keys.buf = NULL;
1479 zh->keys.buf_max = 0;
1483 rec->info[recInfo_delKeys] = NULL;
1484 rec->size[recInfo_delKeys] = 0;
1487 /* save file size of original record */
1488 zebraExplain_recordBytesIncrement (zh->service->zei,
1489 - recordAttr->recordSize);
1491 recordAttr->recordSize = fi->file_moffset - recordOffset;
1492 if (!recordAttr->recordSize)
1493 recordAttr->recordSize = fi->file_max - recordOffset;
1495 recordAttr->recordSize = buf_size;
1497 zebraExplain_recordBytesIncrement (zh->service->zei,
1498 recordAttr->recordSize);
1500 /* set run-number for this record */
1501 recordAttr->runNumber =
1502 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1504 /* update store data */
1505 xfree (rec->info[recInfo_storeData]);
1506 if (store_data == 1)
1508 rec->size[recInfo_storeData] = recordAttr->recordSize;
1509 rec->info[recInfo_storeData] = (char *)
1510 xmalloc (recordAttr->recordSize);
1512 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1514 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1516 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1517 (long) recordOffset, fname);
1520 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1521 < recordAttr->recordSize)
1523 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1524 recordAttr->recordSize, fname);
1531 rec->info[recInfo_storeData] = NULL;
1532 rec->size[recInfo_storeData] = 0;
1534 /* update database name */
1535 xfree (rec->info[recInfo_databaseName]);
1536 rec->info[recInfo_databaseName] =
1537 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1540 recordAttr->recordOffset = recordOffset;
1542 /* commit this record */
1543 rec_put (zh->service->records, &rec);