2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.96 1999-05-26 07:49:13 adam
10 * Revision 1.95 1999/05/21 12:00:17 adam
11 * Better diagnostics for extraction process.
13 * Revision 1.94 1999/05/20 12:57:18 adam
14 * Implemented TCL filter. Updated recctrl system.
16 * Revision 1.93 1999/05/15 14:36:38 adam
17 * Updated dictionary. Implemented "compression" of dictionary.
19 * Revision 1.92 1999/03/09 16:27:49 adam
20 * More work on SDRKit integration.
22 * Revision 1.91 1999/02/12 13:29:22 adam
23 * Implemented position-flag for registers.
25 * Revision 1.90 1999/02/02 14:50:52 adam
26 * Updated WIN32 code specific sections. Changed header.
28 * Revision 1.89 1998/10/28 10:54:38 adam
31 * Revision 1.88 1998/10/16 08:14:29 adam
32 * Updated record control system.
34 * Revision 1.87 1998/10/15 13:10:33 adam
35 * Fixed bug in Zebra that caused it to stop indexing when empty
38 * Revision 1.86 1998/10/13 20:33:53 adam
39 * Fixed one log message and change use ordinal to be an unsigned char.
41 * Revision 1.85 1998/09/22 10:03:41 adam
42 * Changed result sets to be persistent in the sense that they can
43 * be re-searched if needed.
44 * Fixed memory leak in rsm_or.
46 * Revision 1.84 1998/06/11 15:42:22 adam
47 * Changed the way use attributes are specified in the recordId
50 * Revision 1.83 1998/06/08 14:43:10 adam
51 * Added suport for EXPLAIN Proxy servers - added settings databasePath
52 * and explainDatabase to facilitate this. Increased maximum number
53 * of databases and attributes in one register.
55 * Revision 1.82 1998/05/20 10:12:15 adam
56 * Implemented automatic EXPLAIN database maintenance.
57 * Modified Zebra to work with ASN.1 compiled version of YAZ.
59 * Revision 1.81 1998/03/11 11:19:04 adam
60 * Changed the way sequence numbers are generated.
62 * Revision 1.80 1998/03/05 08:45:11 adam
63 * New result set model and modular ranking system. Moved towards
64 * descent server API. System information stored as "SGML" records.
66 * Revision 1.79 1998/02/17 10:32:52 adam
67 * Fixed bug: binary files weren't opened with flag b on NT.
69 * Revision 1.78 1998/02/10 12:03:05 adam
72 * Revision 1.77 1998/01/12 15:04:08 adam
73 * The test option (-s) only uses read-lock (and not write lock).
75 * Revision 1.76 1997/10/27 14:33:04 adam
76 * Moved towards generic character mapping depending on "structure"
77 * field in abstract syntax file. Fixed a few memory leaks. Fixed
78 * bug with negative integers when doing searches with relational
81 * Revision 1.75 1997/09/17 12:19:12 adam
82 * Zebra version corresponds to YAZ version 1.4.
83 * Changed Zebra server so that it doesn't depend on global common_resource.
85 * Revision 1.74 1997/09/09 13:38:06 adam
86 * Partial port to WIN95/NT.
88 * Revision 1.73 1997/09/04 13:57:20 adam
89 * New file extract/retrieve method tellf (added).
90 * Added O_BINARY for open calls.
92 * Revision 1.72 1997/07/15 16:32:29 adam
93 * Bug fix: Match handler didn't terminate the resulting string!
95 * Revision 1.71 1997/07/15 16:28:41 adam
96 * Bug fix: storeData didn't work with files with multiple records.
97 * Bug fix: fixed memory management with records; not really well
100 * Revision 1.70 1997/07/01 13:00:42 adam
101 * Bug fix in routine searchRecordKey: uninitialized variables.
103 * Revision 1.69 1997/04/29 09:26:03 adam
104 * Bug fix: generic recordId handling didn't work for compressed internal
107 * Revision 1.68 1997/02/12 20:39:45 adam
108 * Implemented options -f <n> that limits the log to the first <n>
110 * Changed some log messages also.
112 * Revision 1.67 1996/11/15 15:02:14 adam
113 * Minor changes regarding logging.
115 * Revision 1.66 1996/11/14 09:52:21 adam
116 * Strings in record keys bound by IT_MAX_WORD.
118 * Revision 1.65 1996/11/14 08:57:56 adam
119 * Reduction of storeKeys area.
121 * Revision 1.64 1996/11/08 11:10:16 adam
122 * Buffers used during file match got bigger.
123 * Compressed ISAM support everywhere.
124 * Bug fixes regarding masking characters in queries.
125 * Redesigned Regexp-2 queries.
127 * Revision 1.63 1996/10/29 14:09:39 adam
128 * Use of cisam system - enabled if setting isamc is 1.
130 * Revision 1.62 1996/10/11 10:57:01 adam
131 * New module recctrl. Used to manage records (extract/retrieval).
132 * Several files have been moved to the recctrl sub directory.
134 * Revision 1.61 1996/06/06 12:08:37 quinn
135 * Added showRecord function
137 * Revision 1.60 1996/06/04 10:18:12 adam
138 * Search/scan uses character mapping module.
140 * Revision 1.59 1996/05/14 15:47:07 adam
141 * Cleanup of various buffer size entities.
143 * Revision 1.58 1996/05/14 06:16:38 adam
144 * Compact use/set bytes used in search service.
146 * Revision 1.57 1996/05/13 14:23:04 adam
147 * Work on compaction of set/use bytes in dictionary.
149 * Revision 1.56 1996/05/09 09:54:42 adam
150 * Server supports maps from one logical attributes to a list of physical
152 * The extraction process doesn't make space consuming 'any' keys.
154 * Revision 1.55 1996/05/09 07:28:55 quinn
155 * Work towards phrases and multiple registers
157 * Revision 1.54 1996/05/01 13:46:35 adam
158 * First work on multiple records in one file.
159 * New option, -offset, to the "unread" command in the filter module.
161 * Revision 1.53 1996/04/26 12:09:43 adam
162 * Added a few comments.
164 * Revision 1.52 1996/04/25 13:27:57 adam
165 * Function recordExtract modified so that files with no keys (possibly empty)
168 * Revision 1.51 1996/03/19 11:08:42 adam
169 * Bug fix: Log preamble wasn't always turned off after recordExtract.
171 * Revision 1.50 1996/02/12 18:45:36 adam
172 * New fileVerboseFlag in record group control.
174 * Revision 1.49 1996/02/05 12:29:57 adam
175 * Logging reduced a bit.
176 * The remaining running time is estimated during register merge.
178 * Revision 1.48 1996/02/01 20:53:26 adam
179 * The temporary per-record keys are compacted a little, and duplication
180 * of the per-records keys are avoided when they are saved in the record
181 * information buffer.
183 * Revision 1.47 1996/01/17 14:57:48 adam
184 * Prototype changed for reader functions in extract/retrieve. File
185 * is identified by 'void *' instead of 'int.
187 * Revision 1.46 1995/12/15 14:57:16 adam
190 * Revision 1.45 1995/12/15 12:37:41 adam
191 * In addRecordKeyAny: Writes key only when attrSet != -1.
193 * Revision 1.44 1995/12/12 16:00:54 adam
194 * System call sync(2) used after update/commit.
195 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
198 * Revision 1.43 1995/12/11 09:12:46 adam
199 * The rec_get function returns NULL if record doesn't exist - will
200 * happen in the server if the result set records have been deleted since
201 * the creation of the set (i.e. the search).
202 * The server saves a result temporarily if it is 'volatile', i.e. the
203 * set is register dependent.
205 * Revision 1.42 1995/12/07 17:38:46 adam
206 * Work locking mechanisms for concurrent updates/commit.
208 * Revision 1.41 1995/12/06 16:06:42 adam
209 * Better diagnostics. Work on 'real' dictionary deletion.
211 * Revision 1.40 1995/12/05 16:57:40 adam
212 * More work on regular patterns.
214 * Revision 1.39 1995/12/05 13:20:18 adam
215 * Bug fix: file_read sometimes returned early EOF.
217 * Revision 1.38 1995/12/04 17:59:21 adam
218 * More work on regular expression conversion.
220 * Revision 1.37 1995/12/04 14:22:27 adam
221 * Extra arg to recType_byName.
222 * Started work on new regular expression parsed input to
223 * structured records.
225 * Revision 1.36 1995/11/30 08:34:29 adam
226 * Started work on commit facility.
227 * Changed a few malloc/free to xmalloc/xfree.
229 * Revision 1.35 1995/11/28 14:26:21 adam
230 * Bug fix: recordId with constant wasn't right.
231 * Bug fix: recordId dictionary entry wasn't deleted when needed.
233 * Revision 1.34 1995/11/28 09:09:38 adam
234 * Zebra config renamed.
235 * Use setting 'recordId' to identify record now.
236 * Bug fix in recindex.c: rec_release_blocks was invokeded even
237 * though the blocks were already released.
238 * File traversal properly deletes records when needed.
240 * Revision 1.33 1995/11/27 09:56:20 adam
241 * Record info elements better enumerated. Internal store of records.
243 * Revision 1.32 1995/11/25 10:24:05 adam
244 * More record fields - they are enumerated now.
245 * New options: flagStoreData flagStoreKey.
247 * Revision 1.31 1995/11/24 11:31:35 adam
248 * Commands add & del read filenames from stdin if source directory is
250 * Match criteria supports 'constant' strings.
252 * Revision 1.30 1995/11/22 17:19:16 adam
253 * Record management uses the bfile system.
255 * Revision 1.29 1995/11/21 15:01:14 adam
256 * New general match criteria implemented.
257 * New feature: document groups.
259 * Revision 1.28 1995/11/21 09:20:30 adam
260 * Yet more work on record match.
262 * Revision 1.27 1995/11/20 16:59:45 adam
263 * New update method: the 'old' keys are saved for each records.
265 * Revision 1.26 1995/11/20 11:56:24 adam
266 * Work on new traversal.
268 * Revision 1.25 1995/11/16 15:34:54 adam
269 * Uses new record management system in both indexer and server.
271 * Revision 1.24 1995/11/15 19:13:08 adam
272 * Work on record management.
274 * Revision 1.23 1995/10/27 14:00:10 adam
275 * Implemented detection of database availability.
277 * Revision 1.22 1995/10/17 18:02:07 adam
278 * New feature: databases. Implemented as prefix to words in dictionary.
280 * Revision 1.21 1995/10/10 12:24:38 adam
281 * Temporary sort files are compressed.
283 * Revision 1.20 1995/10/06 13:52:05 adam
284 * Bug fixes. Handler may abort further scanning.
286 * Revision 1.19 1995/10/04 12:55:16 adam
287 * Bug fix in ranked search. Use=Any keys inserted.
289 * Revision 1.18 1995/10/04 09:37:08 quinn
292 * Revision 1.17 1995/10/03 14:28:57 adam
293 * Buffered read in extract works.
295 * Revision 1.16 1995/10/03 14:28:45 adam
296 * Work on more effecient read handler in extract.
298 * Revision 1.15 1995/10/02 15:42:53 adam
299 * Extract uses file descriptors instead of FILE pointers.
301 * Revision 1.14 1995/10/02 15:29:13 adam
302 * More logging in file_extract.
304 * Revision 1.13 1995/09/29 14:01:39 adam
307 * Revision 1.12 1995/09/28 14:22:56 adam
308 * Sort uses smaller temporary files.
310 * Revision 1.11 1995/09/28 12:10:31 adam
311 * Bug fixes. Field prefix used in queries.
313 * Revision 1.10 1995/09/28 09:19:41 adam
314 * xfree/xmalloc used everywhere.
315 * Extract/retrieve method seems to work for text records.
317 * Revision 1.9 1995/09/27 12:22:28 adam
318 * More work on extract in record control.
319 * Field name is not in isam keys but in prefix in dictionary words.
321 * Revision 1.8 1995/09/14 07:48:22 adam
322 * Record control management.
324 * Revision 1.7 1995/09/11 13:09:32 adam
325 * More work on relevance feedback.
327 * Revision 1.6 1995/09/08 14:52:27 adam
328 * Minor changes. Dictionary is lower case now.
330 * Revision 1.5 1995/09/06 16:11:16 adam
331 * Option: only one word key per file.
333 * Revision 1.4 1995/09/05 15:28:39 adam
334 * More work on search engine.
336 * Revision 1.3 1995/09/04 12:33:41 adam
337 * Various cleanup. YAZ util used instead.
339 * Revision 1.2 1995/09/04 09:10:34 adam
340 * More work on index add/del/update.
341 * Merge sort implemented.
342 * Initial work on z39 server.
344 * Revision 1.1 1995/09/01 14:06:35 adam
345 * Split of work into more files.
369 #include "zebrasdr.h"
372 static Dict matchDict;
374 static Records records = NULL;
375 static SortIdx sortIdx = NULL;
377 static char **key_buf;
378 static size_t ptr_top;
380 static size_t key_buf_used;
381 static int key_file_no;
383 static int records_inserted = 0;
384 static int records_updated = 0;
385 static int records_deleted = 0;
386 static int records_processed = 0;
388 static ZebraExplainInfo zti = NULL;
390 static void logRecord (int showFlag)
394 if (showFlag || !(records_processed % 1000))
396 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
397 records_processed, records_inserted, records_updated,
402 static int explain_extract (void *handle, Record drec, data1_node *n);
404 int key_open (struct recordGroup *rGroup, int mem)
406 BFiles bfs = rGroup->bfs;
407 int rw = rGroup->flagRw;
408 data1_handle dh = rGroup->dh;
410 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
413 key_buf = (char **) xmalloc (mem);
414 ptr_top = mem/sizeof(char*);
420 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
422 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
426 records = rec_open (bfs, rw);
429 dict_close (matchDict);
432 zti = zebraExplain_open (records, dh, common_resource,
433 rw, rGroup, explain_extract);
436 rec_close (&records);
437 dict_close (matchDict);
440 sortIdx = sortIdx_open (bfs, 1);
451 void encode_key_init (struct encode_info *i)
458 char *encode_key_int (int d, char *bp)
467 else if (d <= 4194303)
469 *bp++ = 128 + (d>>16);
470 *bp++ = (d>>8) & 255;
475 *bp++ = 192 + (d>>24);
476 *bp++ = (d>>16) & 255;
477 *bp++ = (d>>8) & 255;
483 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
488 while ((*bp++ = *k++))
490 memcpy (&key, k+1, sizeof(struct it_key));
491 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
492 if (i->sysno != key.sysno)
494 i->sysno = key.sysno;
497 else if (!i->seqno && !key.seqno && i->cmd == *k)
499 bp = encode_key_int (key.seqno - i->seqno, bp);
500 i->seqno = key.seqno;
502 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
504 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
512 static int key_y_len;
514 static int key_y_compare (const void *p1, const void *p2)
518 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
519 *(char**) p2 + key_y_len + 1)))
521 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
524 static int key_x_compare (const void *p1, const void *p2)
526 return strcmp (*(char**) p1, *(char**) p2);
530 void key_flush (void)
535 struct encode_info encode_info;
544 logf (LOG_LOG, "sorting section %d", key_file_no);
546 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
547 getFnameTmp (out_fname, key_file_no);
549 if (!(outf = fopen (out_fname, "wb")))
551 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
554 logf (LOG_LOG, "writing section %d", key_file_no);
555 prevcp = cp = key_buf[ptr_top-ptr_i];
557 encode_key_init (&encode_info);
558 encode_key_write (cp, &encode_info, outf);
561 cp = key_buf[ptr_top-ptr_i];
562 if (strcmp (cp, prevcp))
564 encode_key_init (&encode_info);
565 encode_key_write (cp, &encode_info, outf);
569 encode_key_write (cp + strlen(cp), &encode_info, outf);
572 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
573 getFnameTmp (out_fname, key_file_no);
575 if (!(outf = fopen (out_fname, "wb")))
577 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
580 logf (LOG_LOG, "writing section %d", key_file_no);
582 prevcp = key_buf[ptr_top-i];
584 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
586 key_y_len = strlen(prevcp)+1;
588 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
589 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
591 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
592 sizeof(char*), key_y_compare);
593 cp = key_buf[ptr_top-ptr_i];
595 encode_key_init (&encode_info);
596 encode_key_write (cp, &encode_info, outf);
599 cp = key_buf[ptr_top-ptr_i];
600 encode_key_write (cp+key_y_len, &encode_info, outf);
604 prevcp = key_buf[ptr_top-ptr_i];
609 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
612 logf (LOG_LOG, "finished section %d", key_file_no);
617 int key_close (struct recordGroup *rGroup)
619 int rw = rGroup->flagRw;
621 zebraExplain_runNumberIncrement (zti, 1);
622 zebraExplain_close (zti, rw, 0);
625 rec_close (&records);
626 dict_close (matchDict);
627 sortIdx_close (sortIdx);
633 static void wordInit (struct recExtractCtrl *p, RecWord *w)
635 w->zebra_maps = p->zebra_maps;
636 w->seqnos = p->seqno;
637 w->attrSet = VAL_BIB1;
642 static struct sortKey {
647 struct sortKey *next;
650 static struct recKeys {
659 static void addIndexString (RecWord *p, const char *string, int length)
662 unsigned char attrSet;
663 unsigned short attrUse;
666 int *pseqno = &p->seqnos[p->reg_type];
668 if (reckeys.buf_used+1024 > reckeys.buf_max)
672 b = (char *) xmalloc (reckeys.buf_max += 128000);
673 if (reckeys.buf_used > 0)
674 memcpy (b, reckeys.buf, reckeys.buf_used);
678 dst = reckeys.buf + reckeys.buf_used;
680 attrSet = p->attrSet;
681 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
684 reckeys.prevAttrSet = attrSet;
685 attrUse = p->attrUse;
686 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
689 reckeys.prevAttrUse = attrUse;
691 diff = 1 + *pseqno - reckeys.prevSeqNo;
692 if (diff >= 1 && diff <= 15)
697 reckeys.prevSeqNo = *pseqno;
703 memcpy (dst, &attrSet, sizeof(attrSet));
704 dst += sizeof(attrSet);
708 memcpy (dst, &attrUse, sizeof(attrUse));
709 dst += sizeof(attrUse);
711 *dst++ = p->reg_type;
712 memcpy (dst, string, length);
718 memcpy (dst, pseqno, sizeof(*pseqno));
719 dst += sizeof(*pseqno);
721 reckeys.buf_used = dst - reckeys.buf;
726 static void addSortString (RecWord *p, const char *string, int length)
730 for (sk = sortKeys; sk; sk = sk->next)
731 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
734 sk = (struct sortKey *) xmalloc (sizeof(*sk));
738 sk->string = (char *) xmalloc (length);
740 memcpy (sk->string, string, length);
742 sk->attrSet = p->attrSet;
743 sk->attrUse = p->attrUse;
746 static void addString (RecWord *p, const char *string, int length)
749 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
750 addSortString (p, string, length);
752 addIndexString (p, string, length);
755 static void addIncompleteField (RecWord *p)
757 const char *b = p->string;
758 int remain = p->length;
759 const char **map = 0;
762 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
766 char buf[IT_MAX_WORD+1];
770 while (map && *map && **map == *CHR_SPACE)
772 remain = p->length - (b - p->string);
774 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
781 while (map && *map && **map != *CHR_SPACE)
783 const char *cp = *map;
785 while (i < IT_MAX_WORD && *cp)
787 remain = p->length - (b - p->string);
789 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
795 addString (p, buf, i);
799 static void addCompleteField (RecWord *p)
801 const char *b = p->string;
802 char buf[IT_MAX_WORD+1];
803 const char **map = 0;
804 int i = 0, remain = p->length;
807 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
809 while (remain > 0 && i < IT_MAX_WORD)
811 while (map && *map && **map == *CHR_SPACE)
813 remain = p->length - (b - p->string);
815 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
822 if (i && i < IT_MAX_WORD)
823 buf[i++] = *CHR_SPACE;
824 while (map && *map && **map != *CHR_SPACE)
826 const char *cp = *map;
828 if (i >= IT_MAX_WORD)
830 while (i < IT_MAX_WORD && *cp)
832 remain = p->length - (b - p->string);
834 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
842 addString (p, buf, i);
845 static void addRecordKey (RecWord *p)
847 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
848 addCompleteField (p);
850 addIncompleteField(p);
853 static void flushSortKeys (SYSNO sysno, int cmd)
855 struct sortKey *sk = sortKeys;
857 sortIdx_sysno (sortIdx, sysno);
860 struct sortKey *sk_next = sk->next;
861 sortIdx_type (sortIdx, sk->attrUse);
862 sortIdx_add (sortIdx, sk->string, sk->length);
870 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
872 unsigned char attrSet = (unsigned char) -1;
873 unsigned short attrUse = (unsigned short) -1;
877 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
878 while (off < reckeys->buf_used)
880 const char *src = reckeys->buf + off;
888 memcpy (&attrSet, src, sizeof(attrSet));
889 src += sizeof(attrSet);
893 memcpy (&attrUse, src, sizeof(attrUse));
894 src += sizeof(attrUse);
896 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
899 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
901 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
903 ch = zebraExplain_addSU (zti, attrSet, attrUse);
905 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
908 ((char*)key_buf) [key_buf_used++] = *src++;
910 ((char*)key_buf) [key_buf_used++] = '\0';
911 ((char*) key_buf)[key_buf_used++] = cmd;
914 seqno += ((lead>>2) & 15)-1;
917 memcpy (&seqno, src, sizeof(seqno));
918 src += sizeof(seqno);
922 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
923 key_buf_used += sizeof(key);
924 off = src - reckeys->buf;
926 assert (off == reckeys->buf_used);
929 static const char **searchRecordKey (struct recKeys *reckeys,
930 int attrSetS, int attrUseS)
932 static const char *ws[32];
940 for (i = 0; i<32; i++)
943 while (off < reckeys->buf_used)
946 const char *src = reckeys->buf + off;
954 memcpy (&attrSet, src, sizeof(attrSet));
955 src += sizeof(attrSet);
959 memcpy (&attrUse, src, sizeof(attrUse));
960 src += sizeof(attrUse);
966 seqno += ((lead>>2) & 15)-1;
969 memcpy (&seqno, src, sizeof(seqno));
970 src += sizeof(seqno);
972 if (attrUseS == attrUse && attrSetS == attrSet)
979 woff = seqno - startSeq;
980 if (woff >= 0 && woff < 31)
984 off = src - reckeys->buf;
986 assert (off == reckeys->buf_used);
990 struct file_read_info {
991 off_t file_max; /* maximum offset so far */
992 off_t file_offset; /* current offset */
993 off_t file_moffset; /* offset of rec/rec boundary */
1000 static struct file_read_info *file_read_start (int fd)
1002 struct file_read_info *fi = (struct file_read_info *)
1003 xmalloc (sizeof(*fi));
1007 fi->file_moffset = 0;
1013 static void file_read_stop (struct file_read_info *fi)
1018 static off_t file_seek (void *handle, off_t offset)
1020 struct file_read_info *p = (struct file_read_info *) handle;
1021 p->file_offset = offset;
1024 return lseek (p->fd, offset, SEEK_SET);
1027 static off_t file_tell (void *handle)
1029 struct file_read_info *p = (struct file_read_info *) handle;
1030 return p->file_offset;
1033 static int file_read (void *handle, char *buf, size_t count)
1035 struct file_read_info *p = (struct file_read_info *) handle;
1041 if (r > p->sdrmax - p->file_offset)
1042 r = p->sdrmax - p->file_offset;
1044 memcpy (buf, p->sdrbuf + p->file_offset, r);
1047 r = read (fd, buf, count);
1050 p->file_offset += r;
1051 if (p->file_offset > p->file_max)
1052 p->file_max = p->file_offset;
1057 static void file_begin (void *handle)
1059 struct file_read_info *p = (struct file_read_info *) handle;
1061 p->file_offset = p->file_moffset;
1062 if (!p->sdrbuf && p->file_moffset)
1063 lseek (p->fd, p->file_moffset, SEEK_SET);
1067 static void file_end (void *handle, off_t offset)
1069 struct file_read_info *p = (struct file_read_info *) handle;
1071 assert (p->file_more == 0);
1073 p->file_moffset = offset;
1076 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1077 const char *fname, const char *spec)
1079 static char dstBuf[2048];
1081 const char *s = spec;
1082 static const char **w;
1086 while (*s == ' ' || *s == '\t')
1092 char attset_str[64], attname_str[64];
1093 data1_attset *attset;
1096 int attSet = 1, attUse = 1;
1100 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1102 attset_str[i++] = *s;
1103 attset_str[i] = '\0';
1108 for (i = 0; *s && *s != ')'; s++)
1110 attname_str[i++] = *s;
1111 attname_str[i] = '\0';
1114 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1117 attSet = attset->reference;
1118 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1120 attUse = att->value;
1122 attUse = atoi (attname_str);
1124 w = searchRecordKey (reckeys, attSet, attUse);
1129 for (i = 0; i<32; i++)
1134 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1135 spec, rGroup->groupName ? rGroup->groupName : "none");
1140 for (i = 0; i<32; i++)
1141 if (matchFlag[i] && w[i])
1149 dst += strlen(w[i]);
1153 logf (LOG_WARN, "Record didn't contain match"
1154 " fields in (%s,%s)", attset_str, attname_str);
1162 const char *spec_src = NULL;
1163 const char *s1 = ++s;
1164 while (*s1 && *s1 != ' ' && *s1 != '\t')
1170 memcpy (special, s, spec_len);
1171 special[spec_len] = '\0';
1174 if (!strcmp (special, "group"))
1175 spec_src = rGroup->groupName;
1176 else if (!strcmp (special, "database"))
1177 spec_src = rGroup->databaseName;
1178 else if (!strcmp (special, "filename"))
1180 else if (!strcmp (special, "type"))
1181 spec_src = rGroup->recordType;
1186 strcpy (dst, spec_src);
1187 dst += strlen (spec_src);
1190 else if (*s == '\"' || *s == '\'')
1192 int stopMarker = *s++;
1196 while (*s && *s != stopMarker)
1199 tmpString[i++] = *s++;
1203 tmpString[i] = '\0';
1204 strcpy (dst, tmpString);
1205 dst += strlen (tmpString);
1209 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1210 spec, rGroup->groupName ? rGroup->groupName : "none");
1217 logf (LOG_WARN, "No match criteria for record %s in group %s",
1218 fname, rGroup->groupName ? rGroup->groupName : "none");
1225 struct recordLogInfo {
1228 struct recordGroup *rGroup;
1231 static void recordLogPreamble (int level, const char *msg, void *info)
1233 struct recordLogInfo *p = (struct recordLogInfo *) info;
1234 FILE *outf = log_file ();
1236 if (level & LOG_LOG)
1238 fprintf (outf, "File %s, offset %d, type %s\n",
1239 p->rGroup->recordType, p->recordOffset, p->fname);
1240 log_event_start (NULL, NULL);
1243 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1245 zebraExplain_addSchema (zti, oid);
1248 static int recordExtract (SYSNO *sysno, const char *fname,
1249 struct recordGroup *rGroup, int deleteFlag,
1250 struct file_read_info *fi,
1251 RecType recType, char *subType, void *clientData)
1253 RecordAttr *recordAttr;
1258 struct recordLogInfo logInfo;
1259 off_t recordOffset = 0;
1263 struct recExtractCtrl extractCtrl;
1265 /* we are going to read from a file, so prepare the extraction */
1268 reckeys.buf_used = 0;
1269 reckeys.prevAttrUse = -1;
1270 reckeys.prevAttrSet = -1;
1271 reckeys.prevSeqNo = 0;
1273 recordOffset = fi->file_moffset;
1274 extractCtrl.offset = fi->file_moffset;
1275 extractCtrl.readf = file_read;
1276 extractCtrl.seekf = file_seek;
1277 extractCtrl.tellf = file_tell;
1278 extractCtrl.endf = file_end;
1279 extractCtrl.fh = fi;
1280 extractCtrl.subType = subType;
1281 extractCtrl.init = wordInit;
1282 extractCtrl.addWord = addRecordKey;
1283 extractCtrl.addSchema = addSchema;
1284 extractCtrl.dh = rGroup->dh;
1285 for (i = 0; i<256; i++)
1287 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1288 extractCtrl.seqno[i] = 1;
1290 extractCtrl.seqno[i] = 0;
1292 extractCtrl.zebra_maps = rGroup->zebra_maps;
1293 extractCtrl.flagShowRecords = !rGroup->flagRw;
1295 if (!rGroup->flagRw)
1296 printf ("File: %s %ld\n", fname, (long) recordOffset);
1298 logInfo.fname = fname;
1299 logInfo.recordOffset = recordOffset;
1300 logInfo.rGroup = rGroup;
1301 log_event_start (recordLogPreamble, &logInfo);
1303 r = (*recType->extract)(clientData, &extractCtrl);
1305 log_event_start (NULL, NULL);
1307 if (r == RECCTRL_EXTRACT_EOF)
1309 else if (r == RECCTRL_EXTRACT_ERROR)
1311 /* error occured during extraction ... */
1312 if (rGroup->flagRw &&
1313 records_processed < rGroup->fileVerboseLimit)
1315 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1316 fname, (long) recordOffset);
1320 if (reckeys.buf_used == 0)
1322 /* the extraction process returned no information - the record
1323 is probably empty - unless flagShowRecords is in use */
1324 if (!rGroup->flagRw)
1326 logf (LOG_WARN, "No keys generated for file %s", fname);
1327 logf (LOG_WARN, " The file is probably empty");
1332 /* perform match if sysno not known and if match criteria is specified */
1339 if (rGroup->recordId && *rGroup->recordId)
1343 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1347 rinfo = dict_lookup (matchDict, matchStr);
1349 memcpy (sysno, rinfo+1, sizeof(*sysno));
1353 logf (LOG_WARN, "Bad match criteria");
1364 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1365 fname, (long) recordOffset);
1366 logf (LOG_WARN, "cannot delete record above (seems new)");
1369 if (records_processed < rGroup->fileVerboseLimit)
1370 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1371 fname, (long) recordOffset);
1372 rec = rec_new (records);
1374 *sysno = rec->sysno;
1376 recordAttr = rec_init_attr (zti, rec);
1380 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1382 flushRecordKeys (*sysno, 1, &reckeys);
1383 flushSortKeys (*sysno, 1);
1389 /* record already exists */
1390 struct recKeys delkeys;
1392 rec = rec_get (records, *sysno);
1395 recordAttr = rec_init_attr (zti, rec);
1397 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1399 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1400 fname, (long) recordOffset);
1405 delkeys.buf_used = rec->size[recInfo_delKeys];
1406 delkeys.buf = rec->info[recInfo_delKeys];
1407 flushSortKeys (*sysno, 0);
1408 flushRecordKeys (*sysno, 0, &delkeys);
1411 /* record going to be deleted */
1412 if (!delkeys.buf_used)
1414 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1415 fname, (long) recordOffset);
1416 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1420 if (records_processed < rGroup->fileVerboseLimit)
1421 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1422 fname, (long) recordOffset);
1425 dict_delete (matchDict, matchStr);
1426 rec_del (records, &rec);
1434 /* record going to be updated */
1435 if (!delkeys.buf_used)
1437 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1438 fname, (long) recordOffset);
1439 logf (LOG_WARN, "cannot update file above, storeKeys false");
1443 if (records_processed < rGroup->fileVerboseLimit)
1444 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1445 fname, (long) recordOffset);
1446 flushRecordKeys (*sysno, 1, &reckeys);
1451 /* update file type */
1452 xfree (rec->info[recInfo_fileType]);
1453 rec->info[recInfo_fileType] =
1454 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1456 /* update filename */
1457 xfree (rec->info[recInfo_filename]);
1458 rec->info[recInfo_filename] =
1459 rec_strdup (fname, &rec->size[recInfo_filename]);
1461 /* update delete keys */
1462 xfree (rec->info[recInfo_delKeys]);
1463 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1466 rec->size[recInfo_delKeys] = reckeys.buf_used;
1467 rec->info[recInfo_delKeys] = reckeys.buf;
1469 reckeys.buf_max = 0;
1471 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1472 rec->size[recInfo_delKeys] = reckeys.buf_used;
1473 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1474 rec->size[recInfo_delKeys]);
1479 rec->info[recInfo_delKeys] = NULL;
1480 rec->size[recInfo_delKeys] = 0;
1483 /* save file size of original record */
1484 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1485 recordAttr->recordSize = fi->file_moffset - recordOffset;
1486 if (!recordAttr->recordSize)
1487 recordAttr->recordSize = fi->file_max - recordOffset;
1488 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1490 /* set run-number for this record */
1491 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1493 /* update store data */
1494 xfree (rec->info[recInfo_storeData]);
1495 if (rGroup->flagStoreData == 1)
1497 rec->size[recInfo_storeData] = recordAttr->recordSize;
1498 rec->info[recInfo_storeData] = (char *)
1499 xmalloc (recordAttr->recordSize);
1500 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1502 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1503 (long) recordOffset, fname);
1506 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1507 < recordAttr->recordSize)
1509 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1510 recordAttr->recordSize, fname);
1516 rec->info[recInfo_storeData] = NULL;
1517 rec->size[recInfo_storeData] = 0;
1519 /* update database name */
1520 xfree (rec->info[recInfo_databaseName]);
1521 rec->info[recInfo_databaseName] =
1522 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1525 recordAttr->recordOffset = recordOffset;
1527 /* commit this record */
1528 rec_put (records, &rec);
1533 int fileExtract (SYSNO *sysno, const char *fname,
1534 const struct recordGroup *rGroupP, int deleteFlag)
1542 struct recordGroup rGroupM;
1543 struct recordGroup *rGroup = &rGroupM;
1544 struct file_read_info *fi;
1547 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1549 if (!rGroup->groupName || !*rGroup->groupName)
1552 sprintf (gprefix, "%s.", rGroup->groupName);
1554 logf (LOG_DEBUG, "fileExtract %s", fname);
1556 /* determine file extension */
1558 for (i = strlen(fname); --i >= 0; )
1559 if (fname[i] == '/')
1561 else if (fname[i] == '.')
1563 strcpy (ext, fname+i+1);
1566 /* determine file type - depending on extension */
1567 if (!rGroup->recordType)
1569 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1570 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1572 sprintf (ext_res, "%srecordType", gprefix);
1573 rGroup->recordType = res_get (common_resource, ext_res);
1576 if (!rGroup->recordType)
1578 if (records_processed < rGroup->fileVerboseLimit)
1579 logf (LOG_LOG, "? %s", fname);
1582 if (!*rGroup->recordType)
1585 recType_byName (rGroup->recTypes, rGroup->recordType, subType,
1588 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1592 /* determine match criteria */
1593 if (!rGroup->recordId)
1595 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1596 rGroup->recordId = res_get (common_resource, ext_res);
1599 /* determine database name */
1600 if (!rGroup->databaseName)
1602 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1603 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1605 sprintf (ext_res, "%sdatabase", gprefix);
1606 rGroup->databaseName = res_get (common_resource, ext_res);
1609 if (!rGroup->databaseName)
1610 rGroup->databaseName = "Default";
1612 /* determine if explain database */
1614 sprintf (ext_res, "%sexplainDatabase", gprefix);
1615 rGroup->explainDatabase =
1616 atoi (res_get_def (common_resource, ext_res, "0"));
1618 /* announce database */
1619 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1621 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1622 rGroup->explainDatabase))
1626 if (rGroup->flagStoreData == -1)
1629 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1630 if (!(sval = res_get (common_resource, ext_res)))
1632 sprintf (ext_res, "%sstoreData", gprefix);
1633 sval = res_get (common_resource, ext_res);
1636 rGroup->flagStoreData = atoi (sval);
1638 if (rGroup->flagStoreData == -1)
1639 rGroup->flagStoreData = 0;
1641 if (rGroup->flagStoreKeys == -1)
1645 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1646 if (!(sval = res_get (common_resource, ext_res)))
1648 sprintf (ext_res, "%sstoreKeys", gprefix);
1649 sval = res_get (common_resource, ext_res);
1652 rGroup->flagStoreKeys = atoi (sval);
1654 if (rGroup->flagStoreKeys == -1)
1655 rGroup->flagStoreKeys = 0;
1661 char xname[128], *xp;
1663 strncpy (xname, fname, 127);
1664 if (!(xp = strchr (xname, '.')))
1667 if (strcmp (xp+1, "sdr.bits"))
1670 h = zebraSdr_open (xname);
1673 logf (LOG_WARN, "sdr open %s", xname);
1682 segmentno = zebraSdr_segment (h, 0);
1683 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1689 r = zebraSdr_read (h, &buf);
1694 fi = file_read_start (0);
1700 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1702 } while (r && !sysno && fi->file_more);
1703 file_read_stop (fi);
1710 if (sysno && deleteFlag)
1714 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1716 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1720 fi = file_read_start (fd);
1724 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1725 recType, subType, clientData);
1726 } while (r && !sysno && fi->file_more);
1727 file_read_stop (fi);
1733 static int explain_extract (void *handle, Record rec, data1_node *n)
1735 struct recordGroup *rGroup = (struct recordGroup*) handle;
1736 struct recExtractCtrl extractCtrl;
1739 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1741 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1745 reckeys.buf_used = 0;
1746 reckeys.prevAttrUse = -1;
1747 reckeys.prevAttrSet = -1;
1748 reckeys.prevSeqNo = 0;
1750 extractCtrl.init = wordInit;
1751 extractCtrl.addWord = addRecordKey;
1752 extractCtrl.addSchema = addSchema;
1753 extractCtrl.dh = rGroup->dh;
1754 for (i = 0; i<256; i++)
1755 extractCtrl.seqno[i] = 0;
1756 extractCtrl.zebra_maps = rGroup->zebra_maps;
1757 extractCtrl.flagShowRecords = !rGroup->flagRw;
1759 grs_extract_tree(&extractCtrl, n);
1761 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1763 if (rec->size[recInfo_delKeys])
1765 struct recKeys delkeys;
1767 delkeys.buf_used = rec->size[recInfo_delKeys];
1768 delkeys.buf = rec->info[recInfo_delKeys];
1769 flushSortKeys (rec->sysno, 0);
1770 flushRecordKeys (rec->sysno, 0, &delkeys);
1772 flushRecordKeys (rec->sysno, 1, &reckeys);
1773 flushSortKeys (rec->sysno, 1);
1775 xfree (rec->info[recInfo_delKeys]);
1776 rec->size[recInfo_delKeys] = reckeys.buf_used;
1777 rec->info[recInfo_delKeys] = reckeys.buf;
1779 reckeys.buf_max = 0;