2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.31 2000-04-05 10:07:02 adam
7 * Minor zebra compile fix.
9 * Revision 1.30 2000/04/05 09:49:35 adam
10 * On Unix, zebra/z'mbol uses automake.
12 * Revision 1.29 2000/03/20 19:08:36 adam
13 * Added remote record import using Z39.50 extended services and Segment
16 * Revision 1.28 2000/03/15 15:00:30 adam
17 * First work on threaded version.
19 * Revision 1.27 2000/02/24 12:31:17 adam
20 * Added zebra_string_norm.
22 * Revision 1.26 1999/11/30 13:48:03 adam
23 * Improved installation. Updated for inclusion of YAZ header files.
25 * Revision 1.25 1999/11/04 15:00:45 adam
26 * Implemented delete result set(s).
28 * Revision 1.24 1999/10/14 14:33:50 adam
29 * Added truncation 5=106.
31 * Revision 1.23 1999/09/07 11:36:32 adam
34 * Revision 1.22 1999/08/02 10:13:47 adam
35 * Fixed bug regarding zebra_hits.
37 * Revision 1.21 1999/07/14 10:59:26 adam
38 * Changed functions isc_getmethod, isams_getmethod.
39 * Improved fatal error handling (such as missing EXPLAIN schema).
41 * Revision 1.20 1999/07/06 12:28:04 adam
42 * Updated record index structure. Format includes version ID. Compression
43 * algorithm ID is stored for each record block.
45 * Revision 1.19 1999/05/26 07:49:13 adam
48 * Revision 1.18 1999/05/15 14:36:38 adam
49 * Updated dictionary. Implemented "compression" of dictionary.
51 * Revision 1.17 1999/05/12 13:08:06 adam
52 * First version of ISAMS.
54 * Revision 1.16 1999/02/19 10:38:30 adam
55 * Implemented chdir-setting.
57 * Revision 1.15 1999/02/17 12:18:12 adam
58 * Fixed zebra_close so that a NULL pointer is ignored.
60 * Revision 1.14 1999/02/02 14:51:11 adam
61 * Updated WIN32 code specific sections. Changed header.
63 * Revision 1.13 1998/12/16 12:23:30 adam
64 * Added facility for database name mapping using resource mapdb.
66 * Revision 1.12 1998/11/16 10:18:10 adam
67 * Better error reporting for result sets.
69 * Revision 1.11 1998/10/16 08:14:34 adam
70 * Updated record control system.
72 * Revision 1.10 1998/09/22 10:03:42 adam
73 * Changed result sets to be persistent in the sense that they can
74 * be re-searched if needed.
75 * Fixed memory leak in rsm_or.
77 * Revision 1.9 1998/09/02 13:53:17 adam
78 * Extra parameter decode added to search routines to implement
81 * Revision 1.8 1998/08/24 17:29:23 adam
84 * Revision 1.7 1998/06/24 12:16:13 adam
85 * Support for relations on text operands. Open range support in
86 * DFA module (i.e. [-j], [g-]).
88 * Revision 1.6 1998/06/22 11:36:47 adam
89 * Added authentication check facility to zebra.
91 * Revision 1.5 1998/06/13 00:14:08 adam
94 * Revision 1.4 1998/06/12 12:22:12 adam
97 * Revision 1.3 1998/05/27 16:57:44 adam
98 * Zebra returns surrogate diagnostic for single records when
101 * Revision 1.2 1998/05/20 10:12:19 adam
102 * Implemented automatic EXPLAIN database maintenance.
103 * Modified Zebra to work with ASN.1 compiled version of YAZ.
105 * Revision 1.1 1998/03/05 08:45:13 adam
106 * New result set model and modular ranking system. Moved towards
107 * descent server API. System information stored as "SGML" records.
121 #include <yaz/diagbib1.h>
125 static void zebra_chdir (ZebraService zh)
127 const char *dir = res_get (zh->res, "chdir");
130 logf (LOG_DEBUG, "chdir %s", dir);
138 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
139 const char *buf, size_t buf_size,
140 const char *databaseName, int delete_flag,
141 int test_mode, int *sysno,
142 int store_keys, int store_data,
143 const char *match_criteria);
145 static int explain_extract (void *handle, Record rec, data1_node *n);
146 static void extract_index (ZebraHandle zh);
148 static void zebra_register_unlock (ZebraHandle zh);
150 static int zebra_register_lock (ZebraHandle zh)
154 if (!zh->service->active)
162 static void zebra_register_unlock (ZebraHandle zh)
166 ZebraHandle zebra_open (ZebraService zs)
177 zh = (ZebraHandle) xmalloc (sizeof(*zh));
186 zh->admin_databaseName = 0;
188 zebra_mutex_cond_lock (&zs->session_lock);
190 zh->next = zs->sessions;
193 zebra_mutex_cond_unlock (&zs->session_lock);
197 static int zebra_register_activate (ZebraService zh);
198 static int zebra_register_deactivate (ZebraService zh);
200 ZebraService zebra_start (const char *configName)
202 ZebraService zh = xmalloc (sizeof(*zh));
204 yaz_log (LOG_LOG, "zebra_start %s", configName);
206 zh->configName = xstrdup(configName);
210 zebra_mutex_cond_init (&zh->session_lock);
211 zebra_register_activate (zh);
215 static int zebra_register_activate (ZebraService zh)
219 yaz_log (LOG_LOG, "zebra_register_activate");
220 if (!(zh->res = res_open (zh->configName)))
222 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
226 zh->dh = data1_create ();
229 zh->bfs = bfs_create (res_get (zh->res, "register"));
232 data1_destroy(zh->dh);
235 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
236 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
237 zh->registerState = -1; /* trigger open of registers! */
238 zh->registerChange = 0;
239 zh->recTypes = recTypes_init (zh->dh);
240 recTypes_default_handlers (zh->recTypes);
243 zh->zebra_maps = zebra_maps_open (zh->res);
244 zh->rank_classes = NULL;
257 zebraRankInstall (zh, rank1_class);
259 if (!res_get (zh->res, "passwd"))
260 zh->passwd_db = NULL;
263 zh->passwd_db = passwd_db_open ();
265 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
267 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
270 if (!(zh->records = rec_open (zh->bfs, 1, 0)))
272 logf (LOG_WARN, "rec_open");
275 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 40, 1, 0)))
277 logf (LOG_WARN, "dict_open");
280 if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0)))
282 logf (LOG_WARN, "sortIdx_open");
285 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
287 struct ISAMS_M_s isams_m;
288 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, 1,
289 key_isams_m(zh->res, &isams_m))))
291 logf (LOG_WARN, "isams_open");
296 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
298 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, 1,
299 sizeof (struct it_key), zh->res)))
301 logf (LOG_WARN, "is_open");
305 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
307 struct ISAMC_M_s isamc_m;
308 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
309 1, key_isamc_m(zh->res, &isamc_m))))
311 logf (LOG_WARN, "isc_open");
315 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
317 struct ISAMD_M_s isamd_m;
319 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
320 1, key_isamd_m(zh->res, &isamd_m))))
322 logf (LOG_WARN, "isamd_open");
327 zh->zei = zebraExplain_open (zh->records, zh->dh,
328 zh->res, 1, 0 /* rGroup */,
332 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
336 yaz_log (LOG_LOG, "zebra_register_activate ok");
340 void zebra_admin_shutdown (ZebraHandle zh)
342 zebraExplain_flush (zh->service->zei, 1, zh);
345 zebra_mutex_cond_lock (&zh->service->session_lock);
346 zh->service->stop_flag = 1;
347 if (!zh->service->sessions)
348 zebra_register_deactivate(zh->service);
349 zebra_mutex_cond_unlock (&zh->service->session_lock);
352 void zebra_admin_start (ZebraHandle zh)
354 ZebraService zs = zh->service;
356 zebra_mutex_cond_lock (&zs->session_lock);
358 zebra_register_activate(zs);
359 zebra_mutex_cond_unlock (&zs->session_lock);
362 static int zebra_register_deactivate (ZebraService zh)
367 yaz_log(LOG_LOG, "zebra_register_deactivate");
371 zebraExplain_close (zh->zei, 1);
372 dict_close (zh->dict);
373 sortIdx_close (zh->sortIdx);
375 isams_close (zh->isams);
380 isc_close (zh->isamc);
382 isamd_close (zh->isamd);
384 rec_close (&zh->records);
386 recTypes_destroy (zh->recTypes);
387 zebra_maps_close (zh->zebra_maps);
388 zebraRankDestroy (zh);
389 bfs_destroy (zh->bfs);
390 data1_destroy (zh->dh);
393 passwd_db_close (zh->passwd_db);
399 void zebra_stop(ZebraService zh)
403 yaz_log (LOG_LOG, "zebra_stop");
405 assert (!zh->sessions);
407 zebra_mutex_cond_destroy (&zh->session_lock);
409 zebra_register_deactivate(zh);
410 xfree (zh->configName);
414 void zebra_close (ZebraHandle zh)
416 ZebraService zs = zh->service;
417 struct zebra_session **sp;
420 resultSetDestroy (zh, -1, 0, 0);
427 xfree (zh->admin_databaseName);
428 zebra_mutex_cond_lock (&zs->session_lock);
440 if (!zs->sessions && zs->stop_flag)
441 zebra_register_deactivate(zs);
442 zebra_mutex_cond_unlock (&zs->session_lock);
446 struct map_baseinfo {
452 char **new_basenames;
456 void map_basenames_func (void *vp, const char *name, const char *value)
458 struct map_baseinfo *p = (struct map_baseinfo *) vp;
460 char fromdb[128], todb[8][128];
463 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
464 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
465 todb[5], todb[6], todb[7]);
469 for (i = 0; i<p->num_bases; i++)
470 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
473 for (i = 0; i < no; i++)
475 if (p->new_num_bases == p->new_num_max)
477 p->new_basenames[(p->new_num_bases)++] =
478 nmem_strdup (p->mem, todb[i]);
484 void map_basenames (ZebraHandle zh, ODR stream,
485 int *num_bases, char ***basenames)
487 struct map_baseinfo info;
488 struct map_baseinfo *p = &info;
492 info.num_bases = *num_bases;
493 info.basenames = *basenames;
494 info.new_num_max = 128;
495 info.new_num_bases = 0;
496 info.new_basenames = (char **)
497 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
498 info.mem = stream->mem;
500 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
502 for (i = 0; i<p->num_bases; i++)
503 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
505 p->new_basenames[(p->new_num_bases)++] =
506 nmem_strdup (p->mem, p->basenames[i]);
508 *num_bases = info.new_num_bases;
509 *basenames = info.new_basenames;
510 for (i = 0; i<*num_bases; i++)
511 logf (LOG_LOG, "base %s", (*basenames)[i]);
514 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
515 Z_RPNQuery *query, int num_bases, char **basenames,
519 if (zebra_register_lock (zh))
521 map_basenames (zh, stream, &num_bases, &basenames);
522 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
524 zebra_register_unlock (zh);
527 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
528 const char *setname, Z_RecordComposition *comp,
529 oid_value input_format, int num_recs,
530 ZebraRetrievalRecord *recs)
535 if (zebra_register_lock (zh))
537 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
538 for (i = 0; i<num_recs; i++)
539 pos_array[i] = recs[i].position;
540 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
543 logf (LOG_DEBUG, "zebraPosSetCreate error");
545 zh->errString = nmem_strdup (stream->mem, setname);
549 for (i = 0; i<num_recs; i++)
555 sprintf (num_str, "%d", pos_array[i]);
557 zh->errString = nmem_strdup (stream->mem, num_str);
563 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
564 stream, input_format, comp,
565 &recs[i].format, &recs[i].buf,
568 recs[i].errString = NULL;
571 zebraPosSetDestroy (zh, poset, num_recs);
573 zebra_register_unlock (zh);
577 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
578 oid_value attributeset,
579 int num_bases, char **basenames,
580 int *position, int *num_entries, ZebraScanEntry **entries,
583 if (zebra_register_lock (zh))
589 map_basenames (zh, stream, &num_bases, &basenames);
590 rpn_scan (zh, stream, zapt, attributeset,
591 num_bases, basenames, position,
592 num_entries, entries, is_partial);
593 zebra_register_unlock (zh);
596 void zebra_sort (ZebraHandle zh, ODR stream,
597 int num_input_setnames, const char **input_setnames,
598 const char *output_setname, Z_SortKeySpecList *sort_sequence,
601 if (zebra_register_lock (zh))
603 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
604 output_setname, sort_sequence, sort_status);
605 zebra_register_unlock (zh);
608 int zebra_deleleResultSet(ZebraHandle zh, int function,
609 int num_setnames, char **setnames,
613 if (zebra_register_lock (zh))
614 return Z_DeleteStatus_systemProblemAtTarget;
617 case Z_DeleteRequest_list:
618 resultSetDestroy (zh, num_setnames, setnames, statuses);
620 case Z_DeleteRequest_all:
621 resultSetDestroy (zh, -1, 0, statuses);
624 zebra_register_unlock (zh);
625 status = Z_DeleteStatus_success;
626 for (i = 0; i<num_setnames; i++)
627 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
628 status = statuses[i];
632 int zebra_errCode (ZebraHandle zh)
637 const char *zebra_errString (ZebraHandle zh)
639 return diagbib1_str (zh->errCode);
642 char *zebra_errAdd (ZebraHandle zh)
644 return zh->errString;
647 int zebra_hits (ZebraHandle zh)
652 int zebra_auth (ZebraService zh, const char *user, const char *pass)
654 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
659 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
661 if (zebra_register_lock (zh))
663 xfree (zh->admin_databaseName);
664 zh->admin_databaseName = xstrdup(database);
665 zebra_register_unlock(zh);
668 void zebra_admin_import_end (ZebraHandle zh)
670 zebraExplain_flush (zh->service->zei, 1, zh);
674 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
678 if (zebra_register_lock (zh))
680 for (i = 0; i<segment->num_segmentRecords; i++)
682 Z_NamePlusRecord *npr = segment->segmentRecords[i];
683 const char *databaseName = npr->databaseName;
686 databaseName = zh->admin_databaseName;
687 printf ("--------------%d--------------------\n", i);
688 if (npr->which == Z_NamePlusRecord_intermediateFragment)
690 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
691 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
693 Odr_oct *oct = fragment->u.notExternallyTagged;
694 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
698 extract_rec_in_mem (zh, "grs.sgml",
706 0 /* match criteria */);
710 zebra_register_unlock(zh);
713 void zebra_admin_create (ZebraHandle zh, const char *database)
715 ZebraService zs = zh->service;
716 if (zebra_register_lock(zh))
721 /* announce database */
722 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
725 zh->errString = "Database already exist";
727 zebra_register_unlock(zh);
730 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
731 const char *input_str, int input_len,
732 char *output_str, int output_len)
735 if (!zh->service->zebra_maps)
737 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
738 input_str, input_len);
741 if (wrbuf_len(wrbuf) >= output_len)
743 if (wrbuf_len(wrbuf))
744 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
745 output_str[wrbuf_len(wrbuf)] = '\0';
746 return wrbuf_len(wrbuf);
749 static void extract_init (struct recExtractCtrl *p, RecWord *w)
751 w->zebra_maps = p->zebra_maps;
752 w->seqnos = p->seqno;
753 w->attrSet = VAL_BIB1;
759 static void extract_add_index_string (RecWord *p, const char *string,
763 unsigned char attrSet;
764 unsigned short attrUse;
767 int *pseqno = &p->seqnos[p->reg_type];
768 ZebraHandle zh = p->extractCtrl->handle;
769 struct recKeys *keys = &zh->keys;
771 if (keys->buf_used+1024 > keys->buf_max)
775 b = (char *) xmalloc (keys->buf_max += 128000);
776 if (keys->buf_used > 0)
777 memcpy (b, keys->buf, keys->buf_used);
781 dst = keys->buf + keys->buf_used;
783 attrSet = p->attrSet;
784 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
787 keys->prevAttrSet = attrSet;
788 attrUse = p->attrUse;
789 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
792 keys->prevAttrUse = attrUse;
794 diff = 1 + *pseqno - keys->prevSeqNo;
795 if (diff >= 1 && diff <= 15)
800 keys->prevSeqNo = *pseqno;
806 memcpy (dst, &attrSet, sizeof(attrSet));
807 dst += sizeof(attrSet);
811 memcpy (dst, &attrUse, sizeof(attrUse));
812 dst += sizeof(attrUse);
814 *dst++ = p->reg_type;
815 memcpy (dst, string, length);
821 memcpy (dst, pseqno, sizeof(*pseqno));
822 dst += sizeof(*pseqno);
824 keys->buf_used = dst - keys->buf;
829 static void extract_add_sort_string (RecWord *p, const char *string,
833 ZebraHandle zh = p->extractCtrl->handle;
834 struct sortKey *sortKeys = zh->sortKeys;
836 for (sk = sortKeys; sk; sk = sk->next)
837 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
840 sk = (struct sortKey *) xmalloc (sizeof(*sk));
844 sk->string = (char *) xmalloc (length);
846 memcpy (sk->string, string, length);
848 sk->attrSet = p->attrSet;
849 sk->attrUse = p->attrUse;
852 static void extract_add_string (RecWord *p, const char *string, int length)
855 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
856 extract_add_sort_string (p, string, length);
858 extract_add_index_string (p, string, length);
861 static void extract_add_incomplete_field (RecWord *p)
863 const char *b = p->string;
864 int remain = p->length;
865 const char **map = 0;
868 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
872 char buf[IT_MAX_WORD+1];
876 while (map && *map && **map == *CHR_SPACE)
878 remain = p->length - (b - p->string);
880 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
887 while (map && *map && **map != *CHR_SPACE)
889 const char *cp = *map;
891 while (i < IT_MAX_WORD && *cp)
893 remain = p->length - (b - p->string);
895 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
901 extract_add_string (p, buf, i);
903 (p->seqnos[p->reg_type])++; /* to separate this from next one */
906 static void extract_add_complete_field (RecWord *p)
908 const char *b = p->string;
909 char buf[IT_MAX_WORD+1];
910 const char **map = 0;
911 int i = 0, remain = p->length;
914 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
916 while (remain > 0 && i < IT_MAX_WORD)
918 while (map && *map && **map == *CHR_SPACE)
920 remain = p->length - (b - p->string);
922 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
929 if (i && i < IT_MAX_WORD)
930 buf[i++] = *CHR_SPACE;
931 while (map && *map && **map != *CHR_SPACE)
933 const char *cp = *map;
935 if (i >= IT_MAX_WORD)
937 while (i < IT_MAX_WORD && *cp)
939 remain = p->length - (b - p->string);
941 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
949 extract_add_string (p, buf, i);
952 static void extract_token_add (RecWord *p)
955 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
956 p->string, p->length)))
958 p->string = wrbuf_buf(wrbuf);
959 p->length = wrbuf_len(wrbuf);
961 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
962 extract_add_complete_field (p);
964 extract_add_incomplete_field(p);
967 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
969 ZebraHandle zh = (ZebraHandle) (p->handle);
970 zebraExplain_addSchema (zh->service->zei, oid);
973 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
974 int cmd, struct sortKey **skp)
976 struct sortKey *sk = *skp;
977 SortIdx sortIdx = zh->service->sortIdx;
979 sortIdx_sysno (sortIdx, sysno);
982 struct sortKey *sk_next = sk->next;
983 sortIdx_type (sortIdx, sk->attrUse);
984 sortIdx_add (sortIdx, sk->string, sk->length);
999 void encode_key_init (struct encode_info *i)
1006 char *encode_key_int (int d, char *bp)
1010 else if (d <= 16383)
1012 *bp++ = 64 + (d>>8);
1015 else if (d <= 4194303)
1017 *bp++ = 128 + (d>>16);
1018 *bp++ = (d>>8) & 255;
1023 *bp++ = 192 + (d>>24);
1024 *bp++ = (d>>16) & 255;
1025 *bp++ = (d>>8) & 255;
1031 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1036 while ((*bp++ = *k++))
1038 memcpy (&key, k+1, sizeof(struct it_key));
1039 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1040 if (i->sysno != key.sysno)
1042 i->sysno = key.sysno;
1045 else if (!i->seqno && !key.seqno && i->cmd == *k)
1047 bp = encode_key_int (key.seqno - i->seqno, bp);
1048 i->seqno = key.seqno;
1050 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1052 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1057 static void extract_flushWriteKeys (ZebraHandle zh)
1060 char out_fname[200];
1062 struct encode_info encode_info;
1063 int ptr_i = zh->ptr_i;
1067 if (!zh->key_buf || ptr_i <= 0)
1070 (zh->key_file_no)++;
1071 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1073 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1075 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1077 if (!(outf = fopen (out_fname, "wb")))
1079 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1082 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1083 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1085 encode_key_init (&encode_info);
1086 encode_key_write (cp, &encode_info, outf);
1090 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1091 if (strcmp (cp, prevcp))
1093 encode_key_init (&encode_info);
1094 encode_key_write (cp, &encode_info, outf);
1098 encode_key_write (cp + strlen(cp), &encode_info, outf);
1101 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1102 extract_get_fname_tmp (out_fname, key_file_no);
1104 if (!(outf = fopen (out_fname, "wb")))
1106 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1109 logf (LOG_LOG, "writing section %d", key_file_no);
1111 prevcp = key_buf[ptr_top-i];
1113 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1115 key_y_len = strlen(prevcp)+1;
1117 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1118 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1120 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1121 sizeof(char*), key_y_compare);
1122 cp = key_buf[ptr_top-ptr_i];
1124 encode_key_init (&encode_info);
1125 encode_key_write (cp, &encode_info, outf);
1128 cp = key_buf[ptr_top-ptr_i];
1129 encode_key_write (cp+key_y_len, &encode_info, outf);
1133 prevcp = key_buf[ptr_top-ptr_i];
1138 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1141 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1143 zh->key_buf_used = 0;
1146 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1147 int cmd, struct recKeys *reckeys)
1149 unsigned char attrSet = (unsigned char) -1;
1150 unsigned short attrUse = (unsigned short) -1;
1153 ZebraExplainInfo zei = zh->service->zei;
1157 int mem = 8*1024*1024;
1158 zh->key_buf = (char**) xmalloc (mem);
1159 zh->ptr_top = mem/sizeof(char*);
1161 zh->key_buf_used = 0;
1162 zh->key_file_no = 0;
1164 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1165 while (off < reckeys->buf_used)
1167 const char *src = reckeys->buf + off;
1175 memcpy (&attrSet, src, sizeof(attrSet));
1176 src += sizeof(attrSet);
1180 memcpy (&attrUse, src, sizeof(attrUse));
1181 src += sizeof(attrUse);
1183 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1184 extract_flushWriteKeys (zh);
1186 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1187 (char*)zh->key_buf + zh->key_buf_used;
1189 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1191 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1194 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1197 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1199 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1200 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1203 seqno += ((lead>>2) & 15)-1;
1206 memcpy (&seqno, src, sizeof(seqno));
1207 src += sizeof(seqno);
1211 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1212 (zh->key_buf_used) += sizeof(key);
1213 off = src - reckeys->buf;
1215 assert (off == reckeys->buf_used);
1218 static void extract_index (ZebraHandle zh)
1220 extract_flushWriteKeys (zh);
1221 zebra_index_merge (zh);
1224 static int explain_extract (void *handle, Record rec, data1_node *n)
1226 ZebraHandle zh = (ZebraHandle) handle;
1227 struct recExtractCtrl extractCtrl;
1230 if (zebraExplain_curDatabase (zh->service->zei,
1231 rec->info[recInfo_databaseName]))
1234 if (zebraExplain_newDatabase (zh->service->zei,
1235 rec->info[recInfo_databaseName], 0))
1239 zh->keys.buf_used = 0;
1240 zh->keys.prevAttrUse = -1;
1241 zh->keys.prevAttrSet = -1;
1242 zh->keys.prevSeqNo = 0;
1245 extractCtrl.init = extract_init;
1246 extractCtrl.tokenAdd = extract_token_add;
1247 extractCtrl.schemaAdd = extract_schema_add;
1248 extractCtrl.dh = zh->service->dh;
1249 for (i = 0; i<256; i++)
1250 extractCtrl.seqno[i] = 0;
1251 extractCtrl.zebra_maps = zh->service->zebra_maps;
1252 extractCtrl.flagShowRecords = 0;
1253 extractCtrl.handle = handle;
1255 grs_extract_tree(&extractCtrl, n);
1257 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1259 if (rec->size[recInfo_delKeys])
1261 struct recKeys delkeys;
1262 struct sortKey *sortKeys = 0;
1264 delkeys.buf_used = rec->size[recInfo_delKeys];
1265 delkeys.buf = rec->info[recInfo_delKeys];
1266 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1267 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1269 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1270 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1272 xfree (rec->info[recInfo_delKeys]);
1273 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1274 rec->info[recInfo_delKeys] = zh->keys.buf;
1275 zh->keys.buf = NULL;
1276 zh->keys.buf_max = 0;
1280 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1281 const char *buf, size_t buf_size,
1282 const char *databaseName, int delete_flag,
1283 int test_mode, int *sysno,
1284 int store_keys, int store_data,
1285 const char *match_criteria)
1287 RecordAttr *recordAttr;
1288 struct recExtractCtrl extractCtrl;
1293 const char *fname = "<no file>";
1295 long recordOffset = 0;
1296 struct zebra_fetch_control fc;
1299 fc.record_int_buf = buf;
1300 fc.record_int_len = buf_size;
1301 fc.record_int_pos = 0;
1303 fc.record_offset = 0;
1305 extractCtrl.offset = 0;
1306 extractCtrl.readf = zebra_record_int_read;
1307 extractCtrl.seekf = zebra_record_int_seek;
1308 extractCtrl.tellf = zebra_record_int_tell;
1309 extractCtrl.endf = zebra_record_int_end;
1310 extractCtrl.fh = &fc;
1312 /* announce database */
1313 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1315 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1319 recType_byName (zh->service->recTypes, recordType, subType,
1322 logf (LOG_WARN, "No such record type: %s", recordType);
1326 zh->keys.buf_used = 0;
1327 zh->keys.prevAttrUse = -1;
1328 zh->keys.prevAttrSet = -1;
1329 zh->keys.prevSeqNo = 0;
1332 extractCtrl.subType = subType;
1333 extractCtrl.init = extract_init;
1334 extractCtrl.tokenAdd = extract_token_add;
1335 extractCtrl.schemaAdd = extract_schema_add;
1336 extractCtrl.dh = zh->service->dh;
1337 extractCtrl.handle = zh;
1338 extractCtrl.zebra_maps = zh->service->zebra_maps;
1339 extractCtrl.flagShowRecords = 0;
1340 for (i = 0; i<256; i++)
1342 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1343 extractCtrl.seqno[i] = 1;
1345 extractCtrl.seqno[i] = 0;
1348 r = (*recType->extract)(clientData, &extractCtrl);
1350 if (r == RECCTRL_EXTRACT_EOF)
1352 else if (r == RECCTRL_EXTRACT_ERROR)
1354 /* error occured during extraction ... */
1356 yaz_log (LOG_WARN, "extract error");
1358 if (rGroup->flagRw &&
1359 records_processed < rGroup->fileVerboseLimit)
1361 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1362 fname, (long) recordOffset);
1367 if (zh->keys.buf_used == 0)
1369 /* the extraction process returned no information - the record
1370 is probably empty - unless flagShowRecords is in use */
1373 logf (LOG_WARN, "No keys generated for record");
1374 logf (LOG_WARN, " The file is probably empty");
1377 /* match criteria */
1384 logf (LOG_LOG, "delete %s %s %ld", recordType,
1385 fname, (long) recordOffset);
1386 logf (LOG_WARN, "cannot delete record above (seems new)");
1389 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1390 (long) recordOffset);
1391 rec = rec_new (zh->service->records);
1393 *sysno = rec->sysno;
1395 recordAttr = rec_init_attr (zh->service->zei, rec);
1400 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1403 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1404 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1408 /* record already exists */
1409 struct recKeys delkeys;
1411 rec = rec_get (zh->service->records, *sysno);
1414 recordAttr = rec_init_attr (zh->service->zei, rec);
1416 if (recordAttr->runNumber ==
1417 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1419 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1420 fname, (long) recordOffset);
1424 delkeys.buf_used = rec->size[recInfo_delKeys];
1425 delkeys.buf = rec->info[recInfo_delKeys];
1426 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1427 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1430 /* record going to be deleted */
1431 if (!delkeys.buf_used)
1433 logf (LOG_LOG, "delete %s %s %ld", recordType,
1434 fname, (long) recordOffset);
1435 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1439 logf (LOG_LOG, "delete %s %s %ld", recordType,
1440 fname, (long) recordOffset);
1443 dict_delete (matchDict, matchStr);
1445 rec_del (zh->service->records, &rec);
1452 /* record going to be updated */
1453 if (!delkeys.buf_used)
1455 logf (LOG_LOG, "update %s %s %ld", recordType,
1456 fname, (long) recordOffset);
1457 logf (LOG_WARN, "cannot update file above, storeKeys false");
1461 logf (LOG_LOG, "update %s %s %ld", recordType,
1462 fname, (long) recordOffset);
1463 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1467 /* update file type */
1468 xfree (rec->info[recInfo_fileType]);
1469 rec->info[recInfo_fileType] =
1470 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1472 /* update filename */
1473 xfree (rec->info[recInfo_filename]);
1474 rec->info[recInfo_filename] =
1475 rec_strdup (fname, &rec->size[recInfo_filename]);
1477 /* update delete keys */
1478 xfree (rec->info[recInfo_delKeys]);
1479 if (zh->keys.buf_used > 0 && store_keys == 1)
1481 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1482 rec->info[recInfo_delKeys] = zh->keys.buf;
1483 zh->keys.buf = NULL;
1484 zh->keys.buf_max = 0;
1488 rec->info[recInfo_delKeys] = NULL;
1489 rec->size[recInfo_delKeys] = 0;
1492 /* save file size of original record */
1493 zebraExplain_recordBytesIncrement (zh->service->zei,
1494 - recordAttr->recordSize);
1496 recordAttr->recordSize = fi->file_moffset - recordOffset;
1497 if (!recordAttr->recordSize)
1498 recordAttr->recordSize = fi->file_max - recordOffset;
1500 recordAttr->recordSize = buf_size;
1502 zebraExplain_recordBytesIncrement (zh->service->zei,
1503 recordAttr->recordSize);
1505 /* set run-number for this record */
1506 recordAttr->runNumber =
1507 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1509 /* update store data */
1510 xfree (rec->info[recInfo_storeData]);
1511 if (store_data == 1)
1513 rec->size[recInfo_storeData] = recordAttr->recordSize;
1514 rec->info[recInfo_storeData] = (char *)
1515 xmalloc (recordAttr->recordSize);
1517 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1519 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1521 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1522 (long) recordOffset, fname);
1525 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1526 < recordAttr->recordSize)
1528 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1529 recordAttr->recordSize, fname);
1536 rec->info[recInfo_storeData] = NULL;
1537 rec->size[recInfo_storeData] = 0;
1539 /* update database name */
1540 xfree (rec->info[recInfo_databaseName]);
1541 rec->info[recInfo_databaseName] =
1542 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1545 recordAttr->recordOffset = recordOffset;
1547 /* commit this record */
1548 rec_put (zh->service->records, &rec);