2 * Copyright (C) 1995-2000, Index Data
6 * Revision 1.39 2000-11-29 14:24:01 adam
7 * Script configure uses yaz pthreads options. Added locking for
8 * zebra_register_{lock,unlock}.
10 * Revision 1.38 2000/11/08 13:46:58 adam
11 * Fixed scan: server could break if bad attribute/database was selected.
12 * Work on remote update.
14 * Revision 1.37 2000/10/17 12:37:09 adam
15 * Fixed notification of live-updates. Fixed minor problem with mf_init
16 * where it didn't handle shadow area file names correctly.
18 * Revision 1.36 2000/09/06 08:59:36 adam
19 * Using read-only (for now) for server.
21 * Revision 1.35 2000/07/07 12:49:20 adam
22 * Optimized resultSetInsert{Rank,Sort}.
24 * Revision 1.34 2000/06/09 13:56:38 ian
25 * Added some logging on Authentication and searches.
27 * Revision 1.33 2000/05/18 12:01:36 adam
28 * System call times(2) used again. More 64-bit fixes.
30 * Revision 1.32 2000/04/19 14:35:59 adam
31 * WIN32 update (this version is known not to work on Windows).
33 * Revision 1.31 2000/04/05 10:07:02 adam
34 * Minor zebra compile fix.
36 * Revision 1.30 2000/04/05 09:49:35 adam
37 * On Unix, zebra/z'mbol uses automake.
39 * Revision 1.29 2000/03/20 19:08:36 adam
40 * Added remote record import using Z39.50 extended services and Segment
43 * Revision 1.28 2000/03/15 15:00:30 adam
44 * First work on threaded version.
46 * Revision 1.27 2000/02/24 12:31:17 adam
47 * Added zebra_string_norm.
49 * Revision 1.26 1999/11/30 13:48:03 adam
50 * Improved installation. Updated for inclusion of YAZ header files.
52 * Revision 1.25 1999/11/04 15:00:45 adam
53 * Implemented delete result set(s).
55 * Revision 1.24 1999/10/14 14:33:50 adam
56 * Added truncation 5=106.
58 * Revision 1.23 1999/09/07 11:36:32 adam
61 * Revision 1.22 1999/08/02 10:13:47 adam
62 * Fixed bug regarding zebra_hits.
64 * Revision 1.21 1999/07/14 10:59:26 adam
65 * Changed functions isc_getmethod, isams_getmethod.
66 * Improved fatal error handling (such as missing EXPLAIN schema).
68 * Revision 1.20 1999/07/06 12:28:04 adam
69 * Updated record index structure. Format includes version ID. Compression
70 * algorithm ID is stored for each record block.
72 * Revision 1.19 1999/05/26 07:49:13 adam
75 * Revision 1.18 1999/05/15 14:36:38 adam
76 * Updated dictionary. Implemented "compression" of dictionary.
78 * Revision 1.17 1999/05/12 13:08:06 adam
79 * First version of ISAMS.
81 * Revision 1.16 1999/02/19 10:38:30 adam
82 * Implemented chdir-setting.
84 * Revision 1.15 1999/02/17 12:18:12 adam
85 * Fixed zebra_close so that a NULL pointer is ignored.
87 * Revision 1.14 1999/02/02 14:51:11 adam
88 * Updated WIN32 code specific sections. Changed header.
90 * Revision 1.13 1998/12/16 12:23:30 adam
91 * Added facility for database name mapping using resource mapdb.
93 * Revision 1.12 1998/11/16 10:18:10 adam
94 * Better error reporting for result sets.
96 * Revision 1.11 1998/10/16 08:14:34 adam
97 * Updated record control system.
99 * Revision 1.10 1998/09/22 10:03:42 adam
100 * Changed result sets to be persistent in the sense that they can
101 * be re-searched if needed.
102 * Fixed memory leak in rsm_or.
104 * Revision 1.9 1998/09/02 13:53:17 adam
105 * Extra parameter decode added to search routines to implement
106 * persistent queries.
108 * Revision 1.8 1998/08/24 17:29:23 adam
111 * Revision 1.7 1998/06/24 12:16:13 adam
112 * Support for relations on text operands. Open range support in
113 * DFA module (i.e. [-j], [g-]).
115 * Revision 1.6 1998/06/22 11:36:47 adam
116 * Added authentication check facility to zebra.
118 * Revision 1.5 1998/06/13 00:14:08 adam
121 * Revision 1.4 1998/06/12 12:22:12 adam
124 * Revision 1.3 1998/05/27 16:57:44 adam
125 * Zebra returns surrogate diagnostic for single records when
128 * Revision 1.2 1998/05/20 10:12:19 adam
129 * Implemented automatic EXPLAIN database maintenance.
130 * Modified Zebra to work with ASN.1 compiled version of YAZ.
132 * Revision 1.1 1998/03/05 08:45:13 adam
133 * New result set model and modular ranking system. Moved towards
134 * descent server API. System information stored as "SGML" records.
148 #include <yaz/diagbib1.h>
152 static void zebra_chdir (ZebraService zh)
154 const char *dir = res_get (zh->res, "chdir");
157 logf (LOG_DEBUG, "chdir %s", dir);
165 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
166 const char *buf, size_t buf_size,
167 const char *databaseName, int delete_flag,
168 int test_mode, int *sysno,
169 int store_keys, int store_data,
170 const char *match_criteria);
172 static int explain_extract (void *handle, Record rec, data1_node *n);
173 static void extract_index (ZebraHandle zh);
175 static void zebra_register_unlock (ZebraHandle zh);
177 static int zebra_register_activate (ZebraService zh, int rw);
178 static int zebra_register_deactivate (ZebraService zh);
180 static int zebra_register_lock (ZebraHandle zh, int rw)
186 if (!zh->service->active)
196 zebra_mutex_cond_lock (&zh->service->session_lock);
198 state = zebra_server_lock_get_state(zh->service, &lastChange);
200 zebra_server_lock (zh->service, state);
211 logf (LOG_LOG, "Register in read/write mode");
212 else if (zh->service->registerState == state)
214 if (zh->service->registerChange >= lastChange)
216 logf (LOG_LOG, "Register completely updated since last access");
218 else if (zh->service->registerState == -1)
219 logf (LOG_LOG, "Reading register using state %d pid=%ld", state,
222 logf (LOG_LOG, "Register has changed state from %d to %d",
223 zh->service->registerState, state);
224 zh->service->registerChange = lastChange;
226 zebra_register_deactivate (zh->service);
228 zh->service->registerState = state;
230 zebra_register_activate (zh->service, rw);
235 static void zebra_register_unlock (ZebraHandle zh)
237 if (zh->service->registerState != -1)
238 zebra_server_unlock (zh->service, zh->service->registerState);
239 zebra_mutex_cond_unlock (&zh->service->session_lock);
242 logf (LOG_LOG, "user/system: %ld/%ld",
243 (long) (zh->tms2.tms_utime - zh->tms1.tms_utime),
244 (long) (zh->tms2.tms_stime - zh->tms1.tms_stime));
249 ZebraHandle zebra_open (ZebraService zs)
257 zh = (ZebraHandle) xmalloc (sizeof(*zh));
258 yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh);
267 zh->admin_databaseName = 0;
269 zebra_mutex_cond_lock (&zs->session_lock);
271 zh->next = zs->sessions;
274 zebra_mutex_cond_unlock (&zs->session_lock);
280 ZebraService zebra_start (const char *configName)
282 ZebraService zh = xmalloc (sizeof(*zh));
284 yaz_log (LOG_LOG, "zebra_start %s", configName);
286 zh->configName = xstrdup(configName);
291 zh->registerState = -1;
292 zh->registerChange = 0;
294 if (!(zh->res = res_open (zh->configName)))
296 logf (LOG_WARN, "Failed to read resources `%s'", zh->configName);
300 zebra_server_lock_init (zh);
301 zebra_mutex_cond_init (&zh->session_lock);
305 static int zebra_register_activate (ZebraService zh, int rw)
309 yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)",
313 yaz_log (LOG_LOG, "zebra_register_activate shadow=%s",
314 zh->registerState ? "yes" : "no");
316 zh->dh = data1_create ();
319 zh->bfs = bfs_create (res_get (zh->res, "register"));
322 data1_destroy(zh->dh);
325 bf_lockDir (zh->bfs, res_get (zh->res, "lockDir"));
326 bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL);
327 data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath"));
328 zh->recTypes = recTypes_init (zh->dh);
329 recTypes_default_handlers (zh->recTypes);
332 zh->zebra_maps = zebra_maps_open (zh->res);
333 zh->rank_classes = NULL;
346 zebraRankInstall (zh, rank1_class);
348 if (!res_get (zh->res, "passwd"))
349 zh->passwd_db = NULL;
352 zh->passwd_db = passwd_db_open ();
354 logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed");
356 passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd"));
359 if (!(zh->records = rec_open (zh->bfs, rw, 0)))
361 logf (LOG_WARN, "rec_open");
364 if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0)))
366 logf (LOG_WARN, "dict_open");
369 if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw)))
371 logf (LOG_WARN, "sortIdx_open");
374 if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT))
376 struct ISAMS_M_s isams_m;
377 if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw,
378 key_isams_m(zh->res, &isams_m))))
380 logf (LOG_WARN, "isams_open");
385 else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT))
387 if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw,
388 sizeof (struct it_key), zh->res)))
390 logf (LOG_WARN, "is_open");
394 else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT))
396 struct ISAMC_M_s isamc_m;
397 if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC,
398 rw, key_isamc_m(zh->res, &isamc_m))))
400 logf (LOG_WARN, "isc_open");
404 else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT))
406 struct ISAMD_M_s isamd_m;
408 if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD,
409 rw, key_isamd_m(zh->res, &isamd_m))))
411 logf (LOG_WARN, "isamd_open");
416 zh->zei = zebraExplain_open (zh->records, zh->dh,
417 zh->res, rw, 0 /* rGroup */,
421 logf (LOG_WARN, "Cannot obtain EXPLAIN information");
425 yaz_log (LOG_LOG, "zebra_register_activate ok");
429 void zebra_admin_shutdown (ZebraHandle zh)
431 zebra_mutex_cond_lock (&zh->service->session_lock);
432 zh->service->stop_flag = 1;
433 if (!zh->service->sessions)
434 zebra_register_deactivate(zh->service);
435 zh->service->active = 0;
436 zebra_mutex_cond_unlock (&zh->service->session_lock);
439 void zebra_admin_start (ZebraHandle zh)
441 ZebraService zs = zh->service;
443 zebra_mutex_cond_lock (&zs->session_lock);
445 zh->service->active = 1;
446 zebra_mutex_cond_unlock (&zs->session_lock);
449 static int zebra_register_deactivate (ZebraService zs)
454 yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)",
458 yaz_log(LOG_LOG, "zebra_register_deactivate");
462 zebraExplain_close (zs->zei, 0);
463 dict_close (zs->dict);
464 sortIdx_close (zs->sortIdx);
466 isams_close (zs->isams);
471 isc_close (zs->isamc);
473 isamd_close (zs->isamd);
475 rec_close (&zs->records);
477 recTypes_destroy (zs->recTypes);
478 zebra_maps_close (zs->zebra_maps);
479 zebraRankDestroy (zs);
480 bfs_destroy (zs->bfs);
481 data1_destroy (zs->dh);
484 passwd_db_close (zs->passwd_db);
489 void zebra_stop(ZebraService zs)
493 yaz_log (LOG_LOG, "zebra_stop");
495 assert (!zs->sessions);
497 zebra_mutex_cond_destroy (&zs->session_lock);
499 zebra_register_deactivate(zs);
501 xfree (zs->configName);
505 void zebra_close (ZebraHandle zh)
507 ZebraService zs = zh->service;
508 struct zebra_session **sp;
510 yaz_log (LOG_LOG, "zebra_close zh=%p", zh);
513 resultSetDestroy (zh, -1, 0, 0);
520 xfree (zh->admin_databaseName);
521 zebra_mutex_cond_lock (&zs->session_lock);
533 if (!zs->sessions && zs->stop_flag)
534 zebra_register_deactivate(zs);
535 zebra_mutex_cond_unlock (&zs->session_lock);
539 struct map_baseinfo {
545 char **new_basenames;
549 void map_basenames_func (void *vp, const char *name, const char *value)
551 struct map_baseinfo *p = (struct map_baseinfo *) vp;
553 char fromdb[128], todb[8][128];
556 sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s",
557 fromdb, todb[0], todb[1], todb[2], todb[3], todb[4],
558 todb[5], todb[6], todb[7]);
562 for (i = 0; i<p->num_bases; i++)
563 if (p->basenames[i] && !strcmp (p->basenames[i], fromdb))
566 for (i = 0; i < no; i++)
568 if (p->new_num_bases == p->new_num_max)
570 p->new_basenames[(p->new_num_bases)++] =
571 nmem_strdup (p->mem, todb[i]);
577 void map_basenames (ZebraHandle zh, ODR stream,
578 int *num_bases, char ***basenames)
580 struct map_baseinfo info;
581 struct map_baseinfo *p = &info;
585 info.num_bases = *num_bases;
586 info.basenames = *basenames;
587 info.new_num_max = 128;
588 info.new_num_bases = 0;
589 info.new_basenames = (char **)
590 odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max);
591 info.mem = stream->mem;
593 res_trav (zh->service->res, "mapdb", &info, map_basenames_func);
595 for (i = 0; i<p->num_bases; i++)
596 if (p->basenames[i] && p->new_num_bases < p->new_num_max)
598 p->new_basenames[(p->new_num_bases)++] =
599 nmem_strdup (p->mem, p->basenames[i]);
601 *num_bases = info.new_num_bases;
602 *basenames = info.new_basenames;
603 for (i = 0; i<*num_bases; i++)
604 logf (LOG_LOG, "base %s", (*basenames)[i]);
607 void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode,
608 Z_RPNQuery *query, int num_bases, char **basenames,
612 if (zebra_register_lock (zh, 0))
614 map_basenames (zh, stream, &num_bases, &basenames);
615 resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname);
617 zebra_register_unlock (zh);
619 logf(LOG_APP,"SEARCH:%d:",zh->hits);
622 void zebra_records_retrieve (ZebraHandle zh, ODR stream,
623 const char *setname, Z_RecordComposition *comp,
624 oid_value input_format, int num_recs,
625 ZebraRetrievalRecord *recs)
630 if (zebra_register_lock (zh, 0))
632 pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
633 for (i = 0; i<num_recs; i++)
634 pos_array[i] = recs[i].position;
635 poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
638 logf (LOG_DEBUG, "zebraPosSetCreate error");
640 zh->errString = nmem_strdup (stream->mem, setname);
644 for (i = 0; i<num_recs; i++)
650 sprintf (num_str, "%d", pos_array[i]);
652 zh->errString = nmem_strdup (stream->mem, num_str);
658 zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
659 stream, input_format, comp,
660 &recs[i].format, &recs[i].buf,
663 recs[i].errString = NULL;
666 zebraPosSetDestroy (zh, poset, num_recs);
668 zebra_register_unlock (zh);
672 void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
673 oid_value attributeset,
674 int num_bases, char **basenames,
675 int *position, int *num_entries, ZebraScanEntry **entries,
678 if (zebra_register_lock (zh, 0))
684 map_basenames (zh, stream, &num_bases, &basenames);
685 rpn_scan (zh, stream, zapt, attributeset,
686 num_bases, basenames, position,
687 num_entries, entries, is_partial);
688 zebra_register_unlock (zh);
691 void zebra_sort (ZebraHandle zh, ODR stream,
692 int num_input_setnames, const char **input_setnames,
693 const char *output_setname, Z_SortKeySpecList *sort_sequence,
696 if (zebra_register_lock (zh, 0))
698 resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
699 output_setname, sort_sequence, sort_status);
700 zebra_register_unlock (zh);
703 int zebra_deleleResultSet(ZebraHandle zh, int function,
704 int num_setnames, char **setnames,
708 if (zebra_register_lock (zh, 0))
709 return Z_DeleteStatus_systemProblemAtTarget;
712 case Z_DeleteRequest_list:
713 resultSetDestroy (zh, num_setnames, setnames, statuses);
715 case Z_DeleteRequest_all:
716 resultSetDestroy (zh, -1, 0, statuses);
719 zebra_register_unlock (zh);
720 status = Z_DeleteStatus_success;
721 for (i = 0; i<num_setnames; i++)
722 if (statuses[i] == Z_DeleteStatus_resultSetDidNotExist)
723 status = statuses[i];
727 int zebra_errCode (ZebraHandle zh)
732 const char *zebra_errString (ZebraHandle zh)
734 return diagbib1_str (zh->errCode);
737 char *zebra_errAdd (ZebraHandle zh)
739 return zh->errString;
742 int zebra_hits (ZebraHandle zh)
747 int zebra_auth (ZebraService zh, const char *user, const char *pass)
749 if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass))
751 logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS");
755 logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS");
759 void zebra_admin_import_begin (ZebraHandle zh, const char *database)
761 if (zebra_register_lock (zh, 1))
763 xfree (zh->admin_databaseName);
764 zh->admin_databaseName = xstrdup(database);
767 void zebra_admin_import_end (ZebraHandle zh)
769 zebraExplain_flush (zh->service->zei, 1, zh);
771 zebra_register_unlock (zh);
774 void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
778 if (zh->service->active < 2)
780 for (i = 0; i<segment->num_segmentRecords; i++)
782 Z_NamePlusRecord *npr = segment->segmentRecords[i];
783 const char *databaseName = npr->databaseName;
786 databaseName = zh->admin_databaseName;
787 printf ("--------------%d--------------------\n", i);
788 if (npr->which == Z_NamePlusRecord_intermediateFragment)
790 Z_FragmentSyntax *fragment = npr->u.intermediateFragment;
791 if (fragment->which == Z_FragmentSyntax_notExternallyTagged)
793 Odr_oct *oct = fragment->u.notExternallyTagged;
794 printf ("%.*s", (oct->len > 100 ? 100 : oct->len) ,
798 extract_rec_in_mem (zh, "grs.sgml",
806 0 /* match criteria */);
812 void zebra_admin_create (ZebraHandle zh, const char *database)
814 ZebraService zs = zh->service;
815 if (zebra_register_lock(zh, 1))
820 /* announce database */
821 if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */))
824 zh->errString = "Database already exist";
826 zebraExplain_flush (zh->service->zei, 1, zh);
828 zebra_register_unlock(zh);
831 int zebra_string_norm (ZebraHandle zh, unsigned reg_id,
832 const char *input_str, int input_len,
833 char *output_str, int output_len)
836 if (!zh->service->zebra_maps)
838 wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "",
839 input_str, input_len);
842 if (wrbuf_len(wrbuf) >= output_len)
844 if (wrbuf_len(wrbuf))
845 memcpy (output_str, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
846 output_str[wrbuf_len(wrbuf)] = '\0';
847 return wrbuf_len(wrbuf);
850 static void extract_init (struct recExtractCtrl *p, RecWord *w)
852 w->zebra_maps = p->zebra_maps;
853 w->seqnos = p->seqno;
854 w->attrSet = VAL_BIB1;
860 static void extract_add_index_string (RecWord *p, const char *string,
864 unsigned char attrSet;
865 unsigned short attrUse;
868 int *pseqno = &p->seqnos[p->reg_type];
869 ZebraHandle zh = p->extractCtrl->handle;
870 struct recKeys *keys = &zh->keys;
872 if (keys->buf_used+1024 > keys->buf_max)
876 b = (char *) xmalloc (keys->buf_max += 128000);
877 if (keys->buf_used > 0)
878 memcpy (b, keys->buf, keys->buf_used);
882 dst = keys->buf + keys->buf_used;
884 attrSet = p->attrSet;
885 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
888 keys->prevAttrSet = attrSet;
889 attrUse = p->attrUse;
890 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
893 keys->prevAttrUse = attrUse;
895 diff = 1 + *pseqno - keys->prevSeqNo;
896 if (diff >= 1 && diff <= 15)
901 keys->prevSeqNo = *pseqno;
907 memcpy (dst, &attrSet, sizeof(attrSet));
908 dst += sizeof(attrSet);
912 memcpy (dst, &attrUse, sizeof(attrUse));
913 dst += sizeof(attrUse);
915 *dst++ = p->reg_type;
916 memcpy (dst, string, length);
922 memcpy (dst, pseqno, sizeof(*pseqno));
923 dst += sizeof(*pseqno);
925 keys->buf_used = dst - keys->buf;
930 static void extract_add_sort_string (RecWord *p, const char *string,
934 ZebraHandle zh = p->extractCtrl->handle;
935 struct sortKey *sortKeys = zh->sortKeys;
937 for (sk = sortKeys; sk; sk = sk->next)
938 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
941 sk = (struct sortKey *) xmalloc (sizeof(*sk));
945 sk->string = (char *) xmalloc (length);
947 memcpy (sk->string, string, length);
949 sk->attrSet = p->attrSet;
950 sk->attrUse = p->attrUse;
953 static void extract_add_string (RecWord *p, const char *string, int length)
956 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
957 extract_add_sort_string (p, string, length);
959 extract_add_index_string (p, string, length);
962 static void extract_add_incomplete_field (RecWord *p)
964 const char *b = p->string;
965 int remain = p->length;
966 const char **map = 0;
969 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
973 char buf[IT_MAX_WORD+1];
977 while (map && *map && **map == *CHR_SPACE)
979 remain = p->length - (b - p->string);
981 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
988 while (map && *map && **map != *CHR_SPACE)
990 const char *cp = *map;
992 while (i < IT_MAX_WORD && *cp)
994 remain = p->length - (b - p->string);
996 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1002 extract_add_string (p, buf, i);
1004 (p->seqnos[p->reg_type])++; /* to separate this from next one */
1007 static void extract_add_complete_field (RecWord *p)
1009 const char *b = p->string;
1010 char buf[IT_MAX_WORD+1];
1011 const char **map = 0;
1012 int i = 0, remain = p->length;
1015 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1017 while (remain > 0 && i < IT_MAX_WORD)
1019 while (map && *map && **map == *CHR_SPACE)
1021 remain = p->length - (b - p->string);
1023 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1030 if (i && i < IT_MAX_WORD)
1031 buf[i++] = *CHR_SPACE;
1032 while (map && *map && **map != *CHR_SPACE)
1034 const char *cp = *map;
1036 if (i >= IT_MAX_WORD)
1038 while (i < IT_MAX_WORD && *cp)
1040 remain = p->length - (b - p->string);
1042 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1050 extract_add_string (p, buf, i);
1053 static void extract_token_add (RecWord *p)
1056 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1057 p->string, p->length)))
1059 p->string = wrbuf_buf(wrbuf);
1060 p->length = wrbuf_len(wrbuf);
1062 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1063 extract_add_complete_field (p);
1065 extract_add_incomplete_field(p);
1068 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1070 ZebraHandle zh = (ZebraHandle) (p->handle);
1071 zebraExplain_addSchema (zh->service->zei, oid);
1074 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1075 int cmd, struct sortKey **skp)
1077 struct sortKey *sk = *skp;
1078 SortIdx sortIdx = zh->service->sortIdx;
1080 sortIdx_sysno (sortIdx, sysno);
1083 struct sortKey *sk_next = sk->next;
1084 sortIdx_type (sortIdx, sk->attrUse);
1085 sortIdx_add (sortIdx, sk->string, sk->length);
1093 struct encode_info {
1100 void encode_key_init (struct encode_info *i)
1107 char *encode_key_int (int d, char *bp)
1111 else if (d <= 16383)
1113 *bp++ = 64 + (d>>8);
1116 else if (d <= 4194303)
1118 *bp++ = 128 + (d>>16);
1119 *bp++ = (d>>8) & 255;
1124 *bp++ = 192 + (d>>24);
1125 *bp++ = (d>>16) & 255;
1126 *bp++ = (d>>8) & 255;
1132 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1137 while ((*bp++ = *k++))
1139 memcpy (&key, k+1, sizeof(struct it_key));
1140 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1141 if (i->sysno != key.sysno)
1143 i->sysno = key.sysno;
1146 else if (!i->seqno && !key.seqno && i->cmd == *k)
1148 bp = encode_key_int (key.seqno - i->seqno, bp);
1149 i->seqno = key.seqno;
1151 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1153 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
1158 static void extract_flushWriteKeys (ZebraHandle zh)
1161 char out_fname[200];
1163 struct encode_info encode_info;
1164 int ptr_i = zh->ptr_i;
1168 if (!zh->key_buf || ptr_i <= 0)
1171 (zh->key_file_no)++;
1172 logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
1174 qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
1176 extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
1178 if (!(outf = fopen (out_fname, "wb")))
1180 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1183 logf (LOG_LOG, "writing section %d", zh->key_file_no);
1184 prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1186 encode_key_init (&encode_info);
1187 encode_key_write (cp, &encode_info, outf);
1191 cp = (zh->key_buf)[zh->ptr_top - ptr_i];
1192 if (strcmp (cp, prevcp))
1194 encode_key_init (&encode_info);
1195 encode_key_write (cp, &encode_info, outf);
1199 encode_key_write (cp + strlen(cp), &encode_info, outf);
1202 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1203 extract_get_fname_tmp (out_fname, key_file_no);
1205 if (!(outf = fopen (out_fname, "wb")))
1207 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1210 logf (LOG_LOG, "writing section %d", key_file_no);
1212 prevcp = key_buf[ptr_top-i];
1214 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1216 key_y_len = strlen(prevcp)+1;
1218 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1219 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1221 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1222 sizeof(char*), key_y_compare);
1223 cp = key_buf[ptr_top-ptr_i];
1225 encode_key_init (&encode_info);
1226 encode_key_write (cp, &encode_info, outf);
1229 cp = key_buf[ptr_top-ptr_i];
1230 encode_key_write (cp+key_y_len, &encode_info, outf);
1234 prevcp = key_buf[ptr_top-ptr_i];
1239 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1242 logf (LOG_LOG, "finished section %d", zh->key_file_no);
1244 zh->key_buf_used = 0;
1247 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1248 int cmd, struct recKeys *reckeys)
1250 unsigned char attrSet = (unsigned char) -1;
1251 unsigned short attrUse = (unsigned short) -1;
1254 ZebraExplainInfo zei = zh->service->zei;
1258 int mem = 8*1024*1024;
1259 zh->key_buf = (char**) xmalloc (mem);
1260 zh->ptr_top = mem/sizeof(char*);
1262 zh->key_buf_used = 0;
1263 zh->key_file_no = 0;
1265 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1266 while (off < reckeys->buf_used)
1268 const char *src = reckeys->buf + off;
1276 memcpy (&attrSet, src, sizeof(attrSet));
1277 src += sizeof(attrSet);
1281 memcpy (&attrUse, src, sizeof(attrUse));
1282 src += sizeof(attrUse);
1284 if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
1285 extract_flushWriteKeys (zh);
1287 (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
1288 (char*)zh->key_buf + zh->key_buf_used;
1290 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1292 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1295 key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used);
1298 ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
1300 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
1301 ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
1304 seqno += ((lead>>2) & 15)-1;
1307 memcpy (&seqno, src, sizeof(seqno));
1308 src += sizeof(seqno);
1312 memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
1313 (zh->key_buf_used) += sizeof(key);
1314 off = src - reckeys->buf;
1316 assert (off == reckeys->buf_used);
1319 static void extract_index (ZebraHandle zh)
1321 extract_flushWriteKeys (zh);
1322 zebra_index_merge (zh);
1325 static int explain_extract (void *handle, Record rec, data1_node *n)
1327 ZebraHandle zh = (ZebraHandle) handle;
1328 struct recExtractCtrl extractCtrl;
1331 if (zebraExplain_curDatabase (zh->service->zei,
1332 rec->info[recInfo_databaseName]))
1335 if (zebraExplain_newDatabase (zh->service->zei,
1336 rec->info[recInfo_databaseName], 0))
1340 zh->keys.buf_used = 0;
1341 zh->keys.prevAttrUse = -1;
1342 zh->keys.prevAttrSet = -1;
1343 zh->keys.prevSeqNo = 0;
1346 extractCtrl.init = extract_init;
1347 extractCtrl.tokenAdd = extract_token_add;
1348 extractCtrl.schemaAdd = extract_schema_add;
1349 extractCtrl.dh = zh->service->dh;
1350 for (i = 0; i<256; i++)
1351 extractCtrl.seqno[i] = 0;
1352 extractCtrl.zebra_maps = zh->service->zebra_maps;
1353 extractCtrl.flagShowRecords = 0;
1354 extractCtrl.handle = handle;
1356 grs_extract_tree(&extractCtrl, n);
1358 logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
1360 if (rec->size[recInfo_delKeys])
1362 struct recKeys delkeys;
1363 struct sortKey *sortKeys = 0;
1365 delkeys.buf_used = rec->size[recInfo_delKeys];
1366 delkeys.buf = rec->info[recInfo_delKeys];
1367 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1368 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1370 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
1371 extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
1373 xfree (rec->info[recInfo_delKeys]);
1374 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1375 rec->info[recInfo_delKeys] = zh->keys.buf;
1376 zh->keys.buf = NULL;
1377 zh->keys.buf_max = 0;
1381 static int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
1382 const char *buf, size_t buf_size,
1383 const char *databaseName, int delete_flag,
1384 int test_mode, int *sysno,
1385 int store_keys, int store_data,
1386 const char *match_criteria)
1388 RecordAttr *recordAttr;
1389 struct recExtractCtrl extractCtrl;
1394 const char *fname = "<no file>";
1396 long recordOffset = 0;
1397 struct zebra_fetch_control fc;
1400 fc.record_int_buf = buf;
1401 fc.record_int_len = buf_size;
1402 fc.record_int_pos = 0;
1404 fc.record_offset = 0;
1406 extractCtrl.offset = 0;
1407 extractCtrl.readf = zebra_record_int_read;
1408 extractCtrl.seekf = zebra_record_int_seek;
1409 extractCtrl.tellf = zebra_record_int_tell;
1410 extractCtrl.endf = zebra_record_int_end;
1411 extractCtrl.fh = &fc;
1413 /* announce database */
1414 if (zebraExplain_curDatabase (zh->service->zei, databaseName))
1416 if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
1420 recType_byName (zh->service->recTypes, recordType, subType,
1423 logf (LOG_WARN, "No such record type: %s", recordType);
1427 zh->keys.buf_used = 0;
1428 zh->keys.prevAttrUse = -1;
1429 zh->keys.prevAttrSet = -1;
1430 zh->keys.prevSeqNo = 0;
1433 extractCtrl.subType = subType;
1434 extractCtrl.init = extract_init;
1435 extractCtrl.tokenAdd = extract_token_add;
1436 extractCtrl.schemaAdd = extract_schema_add;
1437 extractCtrl.dh = zh->service->dh;
1438 extractCtrl.handle = zh;
1439 extractCtrl.zebra_maps = zh->service->zebra_maps;
1440 extractCtrl.flagShowRecords = 0;
1441 for (i = 0; i<256; i++)
1443 if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
1444 extractCtrl.seqno[i] = 1;
1446 extractCtrl.seqno[i] = 0;
1449 r = (*recType->extract)(clientData, &extractCtrl);
1451 if (r == RECCTRL_EXTRACT_EOF)
1453 else if (r == RECCTRL_EXTRACT_ERROR)
1455 /* error occured during extraction ... */
1457 yaz_log (LOG_WARN, "extract error");
1459 if (rGroup->flagRw &&
1460 records_processed < rGroup->fileVerboseLimit)
1462 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1463 fname, (long) recordOffset);
1468 if (zh->keys.buf_used == 0)
1470 /* the extraction process returned no information - the record
1471 is probably empty - unless flagShowRecords is in use */
1474 logf (LOG_WARN, "No keys generated for record");
1475 logf (LOG_WARN, " The file is probably empty");
1478 /* match criteria */
1485 logf (LOG_LOG, "delete %s %s %ld", recordType,
1486 fname, (long) recordOffset);
1487 logf (LOG_WARN, "cannot delete record above (seems new)");
1490 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
1491 (long) recordOffset);
1492 rec = rec_new (zh->service->records);
1494 *sysno = rec->sysno;
1496 recordAttr = rec_init_attr (zh->service->zei, rec);
1501 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1504 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1505 extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
1509 /* record already exists */
1510 struct recKeys delkeys;
1512 rec = rec_get (zh->service->records, *sysno);
1515 recordAttr = rec_init_attr (zh->service->zei, rec);
1517 if (recordAttr->runNumber ==
1518 zebraExplain_runNumberIncrement (zh->service->zei, 0))
1520 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1521 fname, (long) recordOffset);
1525 delkeys.buf_used = rec->size[recInfo_delKeys];
1526 delkeys.buf = rec->info[recInfo_delKeys];
1527 extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
1528 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1531 /* record going to be deleted */
1532 if (!delkeys.buf_used)
1534 logf (LOG_LOG, "delete %s %s %ld", recordType,
1535 fname, (long) recordOffset);
1536 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1540 logf (LOG_LOG, "delete %s %s %ld", recordType,
1541 fname, (long) recordOffset);
1544 dict_delete (matchDict, matchStr);
1546 rec_del (zh->service->records, &rec);
1553 /* record going to be updated */
1554 if (!delkeys.buf_used)
1556 logf (LOG_LOG, "update %s %s %ld", recordType,
1557 fname, (long) recordOffset);
1558 logf (LOG_WARN, "cannot update file above, storeKeys false");
1562 logf (LOG_LOG, "update %s %s %ld", recordType,
1563 fname, (long) recordOffset);
1564 extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
1568 /* update file type */
1569 xfree (rec->info[recInfo_fileType]);
1570 rec->info[recInfo_fileType] =
1571 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1573 /* update filename */
1574 xfree (rec->info[recInfo_filename]);
1575 rec->info[recInfo_filename] =
1576 rec_strdup (fname, &rec->size[recInfo_filename]);
1578 /* update delete keys */
1579 xfree (rec->info[recInfo_delKeys]);
1580 if (zh->keys.buf_used > 0 && store_keys == 1)
1582 rec->size[recInfo_delKeys] = zh->keys.buf_used;
1583 rec->info[recInfo_delKeys] = zh->keys.buf;
1584 zh->keys.buf = NULL;
1585 zh->keys.buf_max = 0;
1589 rec->info[recInfo_delKeys] = NULL;
1590 rec->size[recInfo_delKeys] = 0;
1593 /* save file size of original record */
1594 zebraExplain_recordBytesIncrement (zh->service->zei,
1595 - recordAttr->recordSize);
1597 recordAttr->recordSize = fi->file_moffset - recordOffset;
1598 if (!recordAttr->recordSize)
1599 recordAttr->recordSize = fi->file_max - recordOffset;
1601 recordAttr->recordSize = buf_size;
1603 zebraExplain_recordBytesIncrement (zh->service->zei,
1604 recordAttr->recordSize);
1606 /* set run-number for this record */
1607 recordAttr->runNumber =
1608 zebraExplain_runNumberIncrement (zh->service->zei, 0);
1610 /* update store data */
1611 xfree (rec->info[recInfo_storeData]);
1612 if (store_data == 1)
1614 rec->size[recInfo_storeData] = recordAttr->recordSize;
1615 rec->info[recInfo_storeData] = (char *)
1616 xmalloc (recordAttr->recordSize);
1618 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1620 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1622 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1623 (long) recordOffset, fname);
1626 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1627 < recordAttr->recordSize)
1629 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1630 recordAttr->recordSize, fname);
1637 rec->info[recInfo_storeData] = NULL;
1638 rec->size[recInfo_storeData] = 0;
1640 /* update database name */
1641 xfree (rec->info[recInfo_databaseName]);
1642 rec->info[recInfo_databaseName] =
1643 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1646 recordAttr->recordOffset = recordOffset;
1648 /* commit this record */
1649 rec_put (zh->service->records, &rec);