2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.88 1998-10-16 08:14:29 adam
8 * Updated record control system.
10 * Revision 1.87 1998/10/15 13:10:33 adam
11 * Fixed bug in Zebra that caused it to stop indexing when empty
14 * Revision 1.86 1998/10/13 20:33:53 adam
15 * Fixed one log message and change use ordinal to be an unsigned char.
17 * Revision 1.85 1998/09/22 10:03:41 adam
18 * Changed result sets to be persistent in the sense that they can
19 * be re-searched if needed.
20 * Fixed memory leak in rsm_or.
22 * Revision 1.84 1998/06/11 15:42:22 adam
23 * Changed the way use attributes are specified in the recordId
26 * Revision 1.83 1998/06/08 14:43:10 adam
27 * Added suport for EXPLAIN Proxy servers - added settings databasePath
28 * and explainDatabase to facilitate this. Increased maximum number
29 * of databases and attributes in one register.
31 * Revision 1.82 1998/05/20 10:12:15 adam
32 * Implemented automatic EXPLAIN database maintenance.
33 * Modified Zebra to work with ASN.1 compiled version of YAZ.
35 * Revision 1.81 1998/03/11 11:19:04 adam
36 * Changed the way sequence numbers are generated.
38 * Revision 1.80 1998/03/05 08:45:11 adam
39 * New result set model and modular ranking system. Moved towards
40 * descent server API. System information stored as "SGML" records.
42 * Revision 1.79 1998/02/17 10:32:52 adam
43 * Fixed bug: binary files weren't opened with flag b on NT.
45 * Revision 1.78 1998/02/10 12:03:05 adam
48 * Revision 1.77 1998/01/12 15:04:08 adam
49 * The test option (-s) only uses read-lock (and not write lock).
51 * Revision 1.76 1997/10/27 14:33:04 adam
52 * Moved towards generic character mapping depending on "structure"
53 * field in abstract syntax file. Fixed a few memory leaks. Fixed
54 * bug with negative integers when doing searches with relational
57 * Revision 1.75 1997/09/17 12:19:12 adam
58 * Zebra version corresponds to YAZ version 1.4.
59 * Changed Zebra server so that it doesn't depend on global common_resource.
61 * Revision 1.74 1997/09/09 13:38:06 adam
62 * Partial port to WIN95/NT.
64 * Revision 1.73 1997/09/04 13:57:20 adam
65 * New file extract/retrieve method tellf (added).
66 * Added O_BINARY for open calls.
68 * Revision 1.72 1997/07/15 16:32:29 adam
69 * Bug fix: Match handler didn't terminate the resulting string!
71 * Revision 1.71 1997/07/15 16:28:41 adam
72 * Bug fix: storeData didn't work with files with multiple records.
73 * Bug fix: fixed memory management with records; not really well
76 * Revision 1.70 1997/07/01 13:00:42 adam
77 * Bug fix in routine searchRecordKey: uninitialized variables.
79 * Revision 1.69 1997/04/29 09:26:03 adam
80 * Bug fix: generic recordId handling didn't work for compressed internal
83 * Revision 1.68 1997/02/12 20:39:45 adam
84 * Implemented options -f <n> that limits the log to the first <n>
86 * Changed some log messages also.
88 * Revision 1.67 1996/11/15 15:02:14 adam
89 * Minor changes regarding logging.
91 * Revision 1.66 1996/11/14 09:52:21 adam
92 * Strings in record keys bound by IT_MAX_WORD.
94 * Revision 1.65 1996/11/14 08:57:56 adam
95 * Reduction of storeKeys area.
97 * Revision 1.64 1996/11/08 11:10:16 adam
98 * Buffers used during file match got bigger.
99 * Compressed ISAM support everywhere.
100 * Bug fixes regarding masking characters in queries.
101 * Redesigned Regexp-2 queries.
103 * Revision 1.63 1996/10/29 14:09:39 adam
104 * Use of cisam system - enabled if setting isamc is 1.
106 * Revision 1.62 1996/10/11 10:57:01 adam
107 * New module recctrl. Used to manage records (extract/retrieval).
108 * Several files have been moved to the recctrl sub directory.
110 * Revision 1.61 1996/06/06 12:08:37 quinn
111 * Added showRecord function
113 * Revision 1.60 1996/06/04 10:18:12 adam
114 * Search/scan uses character mapping module.
116 * Revision 1.59 1996/05/14 15:47:07 adam
117 * Cleanup of various buffer size entities.
119 * Revision 1.58 1996/05/14 06:16:38 adam
120 * Compact use/set bytes used in search service.
122 * Revision 1.57 1996/05/13 14:23:04 adam
123 * Work on compaction of set/use bytes in dictionary.
125 * Revision 1.56 1996/05/09 09:54:42 adam
126 * Server supports maps from one logical attributes to a list of physical
128 * The extraction process doesn't make space consuming 'any' keys.
130 * Revision 1.55 1996/05/09 07:28:55 quinn
131 * Work towards phrases and multiple registers
133 * Revision 1.54 1996/05/01 13:46:35 adam
134 * First work on multiple records in one file.
135 * New option, -offset, to the "unread" command in the filter module.
137 * Revision 1.53 1996/04/26 12:09:43 adam
138 * Added a few comments.
140 * Revision 1.52 1996/04/25 13:27:57 adam
141 * Function recordExtract modified so that files with no keys (possibly empty)
144 * Revision 1.51 1996/03/19 11:08:42 adam
145 * Bug fix: Log preamble wasn't always turned off after recordExtract.
147 * Revision 1.50 1996/02/12 18:45:36 adam
148 * New fileVerboseFlag in record group control.
150 * Revision 1.49 1996/02/05 12:29:57 adam
151 * Logging reduced a bit.
152 * The remaining running time is estimated during register merge.
154 * Revision 1.48 1996/02/01 20:53:26 adam
155 * The temporary per-record keys are compacted a little, and duplication
156 * of the per-records keys are avoided when they are saved in the record
157 * information buffer.
159 * Revision 1.47 1996/01/17 14:57:48 adam
160 * Prototype changed for reader functions in extract/retrieve. File
161 * is identified by 'void *' instead of 'int.
163 * Revision 1.46 1995/12/15 14:57:16 adam
166 * Revision 1.45 1995/12/15 12:37:41 adam
167 * In addRecordKeyAny: Writes key only when attrSet != -1.
169 * Revision 1.44 1995/12/12 16:00:54 adam
170 * System call sync(2) used after update/commit.
171 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
174 * Revision 1.43 1995/12/11 09:12:46 adam
175 * The rec_get function returns NULL if record doesn't exist - will
176 * happen in the server if the result set records have been deleted since
177 * the creation of the set (i.e. the search).
178 * The server saves a result temporarily if it is 'volatile', i.e. the
179 * set is register dependent.
181 * Revision 1.42 1995/12/07 17:38:46 adam
182 * Work locking mechanisms for concurrent updates/commit.
184 * Revision 1.41 1995/12/06 16:06:42 adam
185 * Better diagnostics. Work on 'real' dictionary deletion.
187 * Revision 1.40 1995/12/05 16:57:40 adam
188 * More work on regular patterns.
190 * Revision 1.39 1995/12/05 13:20:18 adam
191 * Bug fix: file_read sometimes returned early EOF.
193 * Revision 1.38 1995/12/04 17:59:21 adam
194 * More work on regular expression conversion.
196 * Revision 1.37 1995/12/04 14:22:27 adam
197 * Extra arg to recType_byName.
198 * Started work on new regular expression parsed input to
199 * structured records.
201 * Revision 1.36 1995/11/30 08:34:29 adam
202 * Started work on commit facility.
203 * Changed a few malloc/free to xmalloc/xfree.
205 * Revision 1.35 1995/11/28 14:26:21 adam
206 * Bug fix: recordId with constant wasn't right.
207 * Bug fix: recordId dictionary entry wasn't deleted when needed.
209 * Revision 1.34 1995/11/28 09:09:38 adam
210 * Zebra config renamed.
211 * Use setting 'recordId' to identify record now.
212 * Bug fix in recindex.c: rec_release_blocks was invokeded even
213 * though the blocks were already released.
214 * File traversal properly deletes records when needed.
216 * Revision 1.33 1995/11/27 09:56:20 adam
217 * Record info elements better enumerated. Internal store of records.
219 * Revision 1.32 1995/11/25 10:24:05 adam
220 * More record fields - they are enumerated now.
221 * New options: flagStoreData flagStoreKey.
223 * Revision 1.31 1995/11/24 11:31:35 adam
224 * Commands add & del read filenames from stdin if source directory is
226 * Match criteria supports 'constant' strings.
228 * Revision 1.30 1995/11/22 17:19:16 adam
229 * Record management uses the bfile system.
231 * Revision 1.29 1995/11/21 15:01:14 adam
232 * New general match criteria implemented.
233 * New feature: document groups.
235 * Revision 1.28 1995/11/21 09:20:30 adam
236 * Yet more work on record match.
238 * Revision 1.27 1995/11/20 16:59:45 adam
239 * New update method: the 'old' keys are saved for each records.
241 * Revision 1.26 1995/11/20 11:56:24 adam
242 * Work on new traversal.
244 * Revision 1.25 1995/11/16 15:34:54 adam
245 * Uses new record management system in both indexer and server.
247 * Revision 1.24 1995/11/15 19:13:08 adam
248 * Work on record management.
250 * Revision 1.23 1995/10/27 14:00:10 adam
251 * Implemented detection of database availability.
253 * Revision 1.22 1995/10/17 18:02:07 adam
254 * New feature: databases. Implemented as prefix to words in dictionary.
256 * Revision 1.21 1995/10/10 12:24:38 adam
257 * Temporary sort files are compressed.
259 * Revision 1.20 1995/10/06 13:52:05 adam
260 * Bug fixes. Handler may abort further scanning.
262 * Revision 1.19 1995/10/04 12:55:16 adam
263 * Bug fix in ranked search. Use=Any keys inserted.
265 * Revision 1.18 1995/10/04 09:37:08 quinn
268 * Revision 1.17 1995/10/03 14:28:57 adam
269 * Buffered read in extract works.
271 * Revision 1.16 1995/10/03 14:28:45 adam
272 * Work on more effecient read handler in extract.
274 * Revision 1.15 1995/10/02 15:42:53 adam
275 * Extract uses file descriptors instead of FILE pointers.
277 * Revision 1.14 1995/10/02 15:29:13 adam
278 * More logging in file_extract.
280 * Revision 1.13 1995/09/29 14:01:39 adam
283 * Revision 1.12 1995/09/28 14:22:56 adam
284 * Sort uses smaller temporary files.
286 * Revision 1.11 1995/09/28 12:10:31 adam
287 * Bug fixes. Field prefix used in queries.
289 * Revision 1.10 1995/09/28 09:19:41 adam
290 * xfree/xmalloc used everywhere.
291 * Extract/retrieve method seems to work for text records.
293 * Revision 1.9 1995/09/27 12:22:28 adam
294 * More work on extract in record control.
295 * Field name is not in isam keys but in prefix in dictionary words.
297 * Revision 1.8 1995/09/14 07:48:22 adam
298 * Record control management.
300 * Revision 1.7 1995/09/11 13:09:32 adam
301 * More work on relevance feedback.
303 * Revision 1.6 1995/09/08 14:52:27 adam
304 * Minor changes. Dictionary is lower case now.
306 * Revision 1.5 1995/09/06 16:11:16 adam
307 * Option: only one word key per file.
309 * Revision 1.4 1995/09/05 15:28:39 adam
310 * More work on search engine.
312 * Revision 1.3 1995/09/04 12:33:41 adam
313 * Various cleanup. YAZ util used instead.
315 * Revision 1.2 1995/09/04 09:10:34 adam
316 * More work on index add/del/update.
317 * Merge sort implemented.
318 * Initial work on z39 server.
320 * Revision 1.1 1995/09/01 14:06:35 adam
321 * Split of work into more files.
340 static Dict matchDict;
342 static Records records = NULL;
343 static SortIdx sortIdx = NULL;
345 static char **key_buf;
346 static size_t ptr_top;
348 static size_t key_buf_used;
349 static int key_file_no;
351 static int records_inserted = 0;
352 static int records_updated = 0;
353 static int records_deleted = 0;
354 static int records_processed = 0;
356 static ZebraExplainInfo zti = NULL;
358 static void logRecord (int showFlag)
362 if (showFlag || !(records_processed % 1000))
364 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
365 records_processed, records_inserted, records_updated,
370 static int explain_extract (void *handle, Record drec, data1_node *n);
372 int key_open (struct recordGroup *rGroup, int mem)
374 BFiles bfs = rGroup->bfs;
375 int rw = rGroup->flagRw;
376 data1_handle dh = rGroup->dh;
378 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
381 key_buf = xmalloc (mem);
382 ptr_top = mem/sizeof(char*);
388 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
390 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
394 records = rec_open (bfs, rw);
397 dict_close (matchDict);
400 zti = zebraExplain_open (records, dh, common_resource,
401 rw, rGroup, explain_extract);
404 rec_close (&records);
405 dict_close (matchDict);
408 sortIdx = sortIdx_open (bfs, 1);
418 void encode_key_init (struct encode_info *i)
424 char *encode_key_int (int d, char *bp)
433 else if (d <= 4194303)
435 *bp++ = 128 + (d>>16);
436 *bp++ = (d>>8) & 255;
441 *bp++ = 192 + (d>>24);
442 *bp++ = (d>>16) & 255;
443 *bp++ = (d>>8) & 255;
449 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
454 while ((*bp++ = *k++))
456 memcpy (&key, k+1, sizeof(struct it_key));
457 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
458 if (i->sysno != key.sysno)
460 i->sysno = key.sysno;
463 bp = encode_key_int (key.seqno - i->seqno, bp);
464 i->seqno = key.seqno;
465 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
467 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
475 static int key_y_len;
477 static int key_y_compare (const void *p1, const void *p2)
481 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
482 *(char**) p2 + key_y_len + 1)))
484 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
487 static int key_x_compare (const void *p1, const void *p2)
489 return strcmp (*(char**) p1, *(char**) p2);
493 void key_flush (void)
498 struct encode_info encode_info;
507 logf (LOG_LOG, "sorting section %d", key_file_no);
509 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
510 getFnameTmp (out_fname, key_file_no);
512 if (!(outf = fopen (out_fname, "wb")))
514 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
517 logf (LOG_LOG, "writing section %d", key_file_no);
518 prevcp = cp = key_buf[ptr_top-ptr_i];
520 encode_key_init (&encode_info);
521 encode_key_write (cp, &encode_info, outf);
524 cp = key_buf[ptr_top-ptr_i];
525 if (strcmp (cp, prevcp))
527 encode_key_init (&encode_info);
528 encode_key_write (cp, &encode_info, outf);
532 encode_key_write (cp + strlen(cp), &encode_info, outf);
535 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
536 getFnameTmp (out_fname, key_file_no);
538 if (!(outf = fopen (out_fname, "wb")))
540 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
543 logf (LOG_LOG, "writing section %d", key_file_no);
545 prevcp = key_buf[ptr_top-i];
547 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
549 key_y_len = strlen(prevcp)+1;
551 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
552 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
554 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
555 sizeof(char*), key_y_compare);
556 cp = key_buf[ptr_top-ptr_i];
558 encode_key_init (&encode_info);
559 encode_key_write (cp, &encode_info, outf);
562 cp = key_buf[ptr_top-ptr_i];
563 encode_key_write (cp+key_y_len, &encode_info, outf);
567 prevcp = key_buf[ptr_top-ptr_i];
572 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
575 logf (LOG_LOG, "finished section %d", key_file_no);
580 int key_close (struct recordGroup *rGroup)
582 int rw = rGroup->flagRw;
584 zebraExplain_runNumberIncrement (zti, 1);
585 zebraExplain_close (zti, rw, 0);
588 rec_close (&records);
589 dict_close (matchDict);
590 sortIdx_close (sortIdx);
596 static void wordInit (struct recExtractCtrl *p, RecWord *w)
598 w->zebra_maps = p->zebra_maps;
599 w->seqnos = p->seqno;
600 w->attrSet = VAL_BIB1;
605 static struct sortKey {
610 struct sortKey *next;
613 static struct recKeys {
622 static void addIndexString (RecWord *p, const char *string, int length)
625 unsigned char attrSet;
626 unsigned short attrUse;
629 int *pseqno = &p->seqnos[p->reg_type];
631 if (reckeys.buf_used+1024 > reckeys.buf_max)
635 b = xmalloc (reckeys.buf_max += 128000);
636 if (reckeys.buf_used > 0)
637 memcpy (b, reckeys.buf, reckeys.buf_used);
641 dst = reckeys.buf + reckeys.buf_used;
643 attrSet = p->attrSet;
644 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
647 reckeys.prevAttrSet = attrSet;
648 attrUse = p->attrUse;
649 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
652 reckeys.prevAttrUse = attrUse;
654 diff = 1 + *pseqno - reckeys.prevSeqNo;
655 if (diff >= 1 && diff <= 15)
660 reckeys.prevSeqNo = *pseqno;
666 memcpy (dst, &attrSet, sizeof(attrSet));
667 dst += sizeof(attrSet);
671 memcpy (dst, &attrUse, sizeof(attrUse));
672 dst += sizeof(attrUse);
674 *dst++ = p->reg_type;
675 memcpy (dst, string, length);
681 memcpy (dst, pseqno, sizeof(*pseqno));
682 dst += sizeof(*pseqno);
684 reckeys.buf_used = dst - reckeys.buf;
688 static void addSortString (RecWord *p, const char *string, int length)
692 for (sk = sortKeys; sk; sk = sk->next)
693 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
696 sk = xmalloc (sizeof(*sk));
700 sk->string = xmalloc (length);
702 memcpy (sk->string, string, length);
704 sk->attrSet = p->attrSet;
705 sk->attrUse = p->attrUse;
708 static void addString (RecWord *p, const char *string, int length)
711 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
712 addSortString (p, string, length);
714 addIndexString (p, string, length);
717 static void addIncompleteField (RecWord *p)
719 const char *b = p->string;
720 int remain = p->length;
721 const char **map = 0;
724 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
728 char buf[IT_MAX_WORD+1];
732 while (map && *map && **map == *CHR_SPACE)
734 remain = p->length - (b - p->string);
736 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
743 while (map && *map && **map != *CHR_SPACE)
745 const char *cp = *map;
747 while (i < IT_MAX_WORD && *cp)
749 remain = p->length - (b - p->string);
751 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
757 addString (p, buf, i);
761 static void addCompleteField (RecWord *p)
763 const char *b = p->string;
764 char buf[IT_MAX_WORD+1];
765 const char **map = 0;
766 int i = 0, remain = p->length;
769 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
771 while (remain > 0 && i < IT_MAX_WORD)
773 while (map && *map && **map == *CHR_SPACE)
775 remain = p->length - (b - p->string);
777 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
784 if (i && i < IT_MAX_WORD)
785 buf[i++] = *CHR_SPACE;
786 while (map && *map && **map != *CHR_SPACE)
788 const char *cp = *map;
790 if (i >= IT_MAX_WORD)
792 while (i < IT_MAX_WORD && *cp)
794 remain = p->length - (b - p->string);
796 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
804 addString (p, buf, i);
807 static void addRecordKey (RecWord *p)
809 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
810 addCompleteField (p);
812 addIncompleteField(p);
815 static void flushSortKeys (SYSNO sysno, int cmd)
817 struct sortKey *sk = sortKeys;
819 sortIdx_sysno (sortIdx, sysno);
822 struct sortKey *sk_next = sk->next;
823 sortIdx_type (sortIdx, sk->attrUse);
824 sortIdx_add (sortIdx, sk->string, sk->length);
832 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
834 unsigned char attrSet = (unsigned char) -1;
835 unsigned short attrUse = (unsigned short) -1;
839 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
840 while (off < reckeys->buf_used)
842 const char *src = reckeys->buf + off;
850 memcpy (&attrSet, src, sizeof(attrSet));
851 src += sizeof(attrSet);
855 memcpy (&attrUse, src, sizeof(attrUse));
856 src += sizeof(attrUse);
858 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
861 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
863 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
865 ch = zebraExplain_addSU (zti, attrSet, attrUse);
867 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
870 ((char*)key_buf) [key_buf_used++] = *src++;
872 ((char*)key_buf) [key_buf_used++] = '\0';
873 ((char*) key_buf)[key_buf_used++] = cmd;
876 seqno += ((lead>>2) & 15)-1;
879 memcpy (&seqno, src, sizeof(seqno));
880 src += sizeof(seqno);
884 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
885 key_buf_used += sizeof(key);
886 off = src - reckeys->buf;
888 assert (off == reckeys->buf_used);
891 static const char **searchRecordKey (struct recKeys *reckeys,
892 int attrSetS, int attrUseS)
894 static const char *ws[32];
902 for (i = 0; i<32; i++)
905 while (off < reckeys->buf_used)
908 const char *src = reckeys->buf + off;
916 memcpy (&attrSet, src, sizeof(attrSet));
917 src += sizeof(attrSet);
921 memcpy (&attrUse, src, sizeof(attrUse));
922 src += sizeof(attrUse);
928 seqno += ((lead>>2) & 15)-1;
931 memcpy (&seqno, src, sizeof(seqno));
932 src += sizeof(seqno);
934 if (attrUseS == attrUse && attrSetS == attrSet)
941 woff = seqno - startSeq;
942 if (woff >= 0 && woff < 31)
946 off = src - reckeys->buf;
948 assert (off == reckeys->buf_used);
952 struct file_read_info {
960 static struct file_read_info *file_read_start (int fd)
962 struct file_read_info *fi = xmalloc (sizeof(*fi));
966 fi->file_moffset = 0;
970 static void file_read_stop (struct file_read_info *fi)
975 static off_t file_seek (void *handle, off_t offset)
977 struct file_read_info *p = handle;
978 p->file_offset = offset;
979 return lseek (p->fd, offset, SEEK_SET);
982 static off_t file_tell (void *handle)
984 struct file_read_info *p = handle;
985 return p->file_offset;
988 static int file_read (void *handle, char *buf, size_t count)
990 struct file_read_info *p = handle;
993 r = read (fd, buf, count);
997 if (p->file_offset > p->file_max)
998 p->file_max = p->file_offset;
1003 static void file_begin (void *handle)
1005 struct file_read_info *p = handle;
1007 p->file_offset = p->file_moffset;
1008 if (p->file_moffset)
1009 lseek (p->fd, p->file_moffset, SEEK_SET);
1013 static void file_end (void *handle, off_t offset)
1015 struct file_read_info *p = handle;
1017 assert (p->file_more == 0);
1019 p->file_moffset = offset;
1022 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1023 const char *fname, const char *spec)
1025 static char dstBuf[2048];
1027 const char *s = spec;
1028 static const char **w;
1032 while (*s == ' ' || *s == '\t')
1038 char attset_str[64], attname_str[64];
1039 data1_attset *attset;
1042 int attSet = 1, attUse = 1;
1046 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1048 attset_str[i++] = *s;
1049 attset_str[i] = '\0';
1054 for (i = 0; *s && *s != ')'; s++)
1056 attname_str[i++] = *s;
1057 attname_str[i] = '\0';
1060 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1063 attSet = attset->reference;
1064 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1066 attUse = att->value;
1068 attUse = atoi (attname_str);
1070 w = searchRecordKey (reckeys, attSet, attUse);
1075 for (i = 0; i<32; i++)
1080 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1081 spec, rGroup->groupName ? rGroup->groupName : "none");
1086 for (i = 0; i<32; i++)
1087 if (matchFlag[i] && w[i])
1095 dst += strlen(w[i]);
1099 logf (LOG_WARN, "Record didn't contain match"
1100 " fields in (%s,%s)", attset_str, attname_str);
1108 const char *spec_src = NULL;
1109 const char *s1 = ++s;
1110 while (*s1 && *s1 != ' ' && *s1 != '\t')
1116 memcpy (special, s, spec_len);
1117 special[spec_len] = '\0';
1120 if (!strcmp (special, "group"))
1121 spec_src = rGroup->groupName;
1122 else if (!strcmp (special, "database"))
1123 spec_src = rGroup->databaseName;
1124 else if (!strcmp (special, "filename"))
1126 else if (!strcmp (special, "type"))
1127 spec_src = rGroup->recordType;
1132 strcpy (dst, spec_src);
1133 dst += strlen (spec_src);
1136 else if (*s == '\"' || *s == '\'')
1138 int stopMarker = *s++;
1142 while (*s && *s != stopMarker)
1145 tmpString[i++] = *s++;
1149 tmpString[i] = '\0';
1150 strcpy (dst, tmpString);
1151 dst += strlen (tmpString);
1155 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1156 spec, rGroup->groupName ? rGroup->groupName : "none");
1163 logf (LOG_WARN, "No match criteria for record %s in group %s",
1164 fname, rGroup->groupName ? rGroup->groupName : "none");
1171 struct recordLogInfo {
1174 struct recordGroup *rGroup;
1177 static void recordLogPreamble (int level, const char *msg, void *info)
1179 struct recordLogInfo *p = info;
1180 FILE *outf = log_file ();
1182 if (level & LOG_LOG)
1184 fprintf (outf, "File %s, offset %d, type %s\n",
1185 p->rGroup->recordType, p->recordOffset, p->fname);
1186 log_event_start (NULL, NULL);
1189 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1191 zebraExplain_addSchema (zti, oid);
1194 static int recordExtract (SYSNO *sysno, const char *fname,
1195 struct recordGroup *rGroup, int deleteFlag,
1196 struct file_read_info *fi,
1197 RecType recType, char *subType)
1199 RecordAttr *recordAttr;
1204 struct recordLogInfo logInfo;
1205 off_t recordOffset = 0;
1209 struct recExtractCtrl extractCtrl;
1211 /* we are going to read from a file, so prepare the extraction */
1214 reckeys.buf_used = 0;
1215 reckeys.prevAttrUse = -1;
1216 reckeys.prevAttrSet = -1;
1217 reckeys.prevSeqNo = 0;
1219 recordOffset = fi->file_moffset;
1220 extractCtrl.offset = fi->file_moffset;
1221 extractCtrl.readf = file_read;
1222 extractCtrl.seekf = file_seek;
1223 extractCtrl.tellf = file_tell;
1224 extractCtrl.endf = file_end;
1225 extractCtrl.fh = fi;
1226 extractCtrl.subType = subType;
1227 extractCtrl.init = wordInit;
1228 extractCtrl.addWord = addRecordKey;
1229 extractCtrl.addSchema = addSchema;
1230 extractCtrl.dh = rGroup->dh;
1231 for (i = 0; i<256; i++)
1232 extractCtrl.seqno[i] = 0;
1233 extractCtrl.zebra_maps = rGroup->zebra_maps;
1234 extractCtrl.flagShowRecords = !rGroup->flagRw;
1236 if (!rGroup->flagRw)
1237 printf ("File: %s %ld\n", fname, (long) recordOffset);
1239 logInfo.fname = fname;
1240 logInfo.recordOffset = recordOffset;
1241 logInfo.rGroup = rGroup;
1242 log_event_start (recordLogPreamble, &logInfo);
1244 r = (*recType->extract)(&extractCtrl);
1246 log_event_start (NULL, NULL);
1250 /* error occured during extraction ... */
1251 if (rGroup->flagRw &&
1252 records_processed < rGroup->fileVerboseLimit)
1254 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1255 fname, (long) recordOffset, r);
1259 if (reckeys.buf_used == 0)
1261 /* the extraction process returned no information - the record
1262 is probably empty - unless flagShowRecords is in use */
1263 if (!rGroup->flagRw)
1265 logf (LOG_WARN, "No keys generated for file %s", fname);
1266 logf (LOG_WARN, " The file is probably empty");
1271 /* perform match if sysno not known and if match criteria is specified */
1278 if (rGroup->recordId && *rGroup->recordId)
1282 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1286 rinfo = dict_lookup (matchDict, matchStr);
1288 memcpy (sysno, rinfo+1, sizeof(*sysno));
1292 logf (LOG_WARN, "Bad match criteria");
1303 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1304 fname, (long) recordOffset);
1305 logf (LOG_WARN, "cannot delete record above (seems new)");
1308 if (records_processed < rGroup->fileVerboseLimit)
1309 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1310 fname, (long) recordOffset);
1311 rec = rec_new (records);
1313 *sysno = rec->sysno;
1315 recordAttr = rec_init_attr (zti, rec);
1319 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1321 flushRecordKeys (*sysno, 1, &reckeys);
1322 flushSortKeys (*sysno, 1);
1328 /* record already exists */
1329 struct recKeys delkeys;
1331 rec = rec_get (records, *sysno);
1334 recordAttr = rec_init_attr (zti, rec);
1336 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1338 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1339 fname, (long) recordOffset);
1344 delkeys.buf_used = rec->size[recInfo_delKeys];
1345 delkeys.buf = rec->info[recInfo_delKeys];
1346 flushSortKeys (*sysno, 0);
1347 flushRecordKeys (*sysno, 0, &delkeys);
1350 /* record going to be deleted */
1351 if (!delkeys.buf_used)
1353 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1354 fname, (long) recordOffset);
1355 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1359 if (records_processed < rGroup->fileVerboseLimit)
1360 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1361 fname, (long) recordOffset);
1364 dict_delete (matchDict, matchStr);
1365 rec_del (records, &rec);
1373 /* record going to be updated */
1374 if (!delkeys.buf_used)
1376 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1377 fname, (long) recordOffset);
1378 logf (LOG_WARN, "cannot update file above, storeKeys false");
1382 if (records_processed < rGroup->fileVerboseLimit)
1383 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1384 fname, (long) recordOffset);
1385 flushRecordKeys (*sysno, 1, &reckeys);
1390 /* update file type */
1391 xfree (rec->info[recInfo_fileType]);
1392 rec->info[recInfo_fileType] =
1393 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1395 /* update filename */
1396 xfree (rec->info[recInfo_filename]);
1397 rec->info[recInfo_filename] =
1398 rec_strdup (fname, &rec->size[recInfo_filename]);
1400 /* update delete keys */
1401 xfree (rec->info[recInfo_delKeys]);
1402 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1405 rec->size[recInfo_delKeys] = reckeys.buf_used;
1406 rec->info[recInfo_delKeys] = reckeys.buf;
1408 reckeys.buf_max = 0;
1410 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1411 rec->size[recInfo_delKeys] = reckeys.buf_used;
1412 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1413 rec->size[recInfo_delKeys]);
1418 rec->info[recInfo_delKeys] = NULL;
1419 rec->size[recInfo_delKeys] = 0;
1422 /* save file size of original record */
1423 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1424 recordAttr->recordSize = fi->file_moffset - recordOffset;
1425 if (!recordAttr->recordSize)
1426 recordAttr->recordSize = fi->file_max - recordOffset;
1427 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1429 /* set run-number for this record */
1430 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1432 /* update store data */
1433 xfree (rec->info[recInfo_storeData]);
1434 if (rGroup->flagStoreData == 1)
1436 rec->size[recInfo_storeData] = recordAttr->recordSize;
1437 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1438 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1440 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1441 (long) recordOffset, fname);
1444 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1445 < recordAttr->recordSize)
1447 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1448 recordAttr->recordSize, fname);
1454 rec->info[recInfo_storeData] = NULL;
1455 rec->size[recInfo_storeData] = 0;
1457 /* update database name */
1458 xfree (rec->info[recInfo_databaseName]);
1459 rec->info[recInfo_databaseName] =
1460 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1463 recordAttr->recordOffset = recordOffset;
1465 /* commit this record */
1466 rec_put (records, &rec);
1471 int fileExtract (SYSNO *sysno, const char *fname,
1472 const struct recordGroup *rGroupP, int deleteFlag)
1480 struct recordGroup rGroupM;
1481 struct recordGroup *rGroup = &rGroupM;
1482 struct file_read_info *fi;
1484 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1486 if (!rGroup->groupName || !*rGroup->groupName)
1489 sprintf (gprefix, "%s.", rGroup->groupName);
1491 logf (LOG_DEBUG, "fileExtract %s", fname);
1493 /* determine file extension */
1495 for (i = strlen(fname); --i >= 0; )
1496 if (fname[i] == '/')
1498 else if (fname[i] == '.')
1500 strcpy (ext, fname+i+1);
1503 /* determine file type - depending on extension */
1504 if (!rGroup->recordType)
1506 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1507 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1509 sprintf (ext_res, "%srecordType", gprefix);
1510 rGroup->recordType = res_get (common_resource, ext_res);
1513 if (!rGroup->recordType)
1515 if (records_processed < rGroup->fileVerboseLimit)
1516 logf (LOG_LOG, "? %s", fname);
1519 if (!*rGroup->recordType)
1522 recType_byName (rGroup->recTypes, rGroup->recordType, subType)))
1524 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1528 /* determine match criteria */
1529 if (!rGroup->recordId)
1531 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1532 rGroup->recordId = res_get (common_resource, ext_res);
1535 /* determine database name */
1536 if (!rGroup->databaseName)
1538 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1539 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1541 sprintf (ext_res, "%sdatabase", gprefix);
1542 rGroup->databaseName = res_get (common_resource, ext_res);
1545 if (!rGroup->databaseName)
1546 rGroup->databaseName = "Default";
1548 /* determine if explain database */
1550 sprintf (ext_res, "%sexplainDatabase", gprefix);
1551 rGroup->explainDatabase =
1552 atoi (res_get_def (common_resource, ext_res, "0"));
1554 /* announce database */
1555 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1557 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1558 rGroup->explainDatabase))
1562 if (rGroup->flagStoreData == -1)
1565 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1566 if (!(sval = res_get (common_resource, ext_res)))
1568 sprintf (ext_res, "%sstoreData", gprefix);
1569 sval = res_get (common_resource, ext_res);
1572 rGroup->flagStoreData = atoi (sval);
1574 if (rGroup->flagStoreData == -1)
1575 rGroup->flagStoreData = 0;
1577 if (rGroup->flagStoreKeys == -1)
1581 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1582 if (!(sval = res_get (common_resource, ext_res)))
1584 sprintf (ext_res, "%sstoreKeys", gprefix);
1585 sval = res_get (common_resource, ext_res);
1588 rGroup->flagStoreKeys = atoi (sval);
1590 if (rGroup->flagStoreKeys == -1)
1591 rGroup->flagStoreKeys = 0;
1593 if (sysno && deleteFlag)
1597 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1599 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1603 fi = file_read_start (fd);
1607 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1609 } while (r && !sysno && fi->file_more);
1610 file_read_stop (fi);
1616 static int explain_extract (void *handle, Record rec, data1_node *n)
1618 struct recordGroup *rGroup = (struct recordGroup*) handle;
1619 struct recExtractCtrl extractCtrl;
1622 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1624 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1628 reckeys.buf_used = 0;
1629 reckeys.prevAttrUse = -1;
1630 reckeys.prevAttrSet = -1;
1631 reckeys.prevSeqNo = 0;
1633 extractCtrl.init = wordInit;
1634 extractCtrl.addWord = addRecordKey;
1635 extractCtrl.addSchema = addSchema;
1636 extractCtrl.dh = rGroup->dh;
1637 for (i = 0; i<256; i++)
1638 extractCtrl.seqno[i] = 0;
1639 extractCtrl.zebra_maps = rGroup->zebra_maps;
1640 extractCtrl.flagShowRecords = !rGroup->flagRw;
1642 grs_extract_tree(&extractCtrl, n);
1644 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1646 if (rec->size[recInfo_delKeys])
1648 struct recKeys delkeys;
1650 delkeys.buf_used = rec->size[recInfo_delKeys];
1651 delkeys.buf = rec->info[recInfo_delKeys];
1652 flushSortKeys (rec->sysno, 0);
1653 flushRecordKeys (rec->sysno, 0, &delkeys);
1655 flushRecordKeys (rec->sysno, 1, &reckeys);
1656 flushSortKeys (rec->sysno, 1);
1658 xfree (rec->info[recInfo_delKeys]);
1659 rec->size[recInfo_delKeys] = reckeys.buf_used;
1660 rec->info[recInfo_delKeys] = reckeys.buf;
1662 reckeys.buf_max = 0;