2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.87 1998-10-15 13:10:33 adam
8 * Fixed bug in Zebra that caused it to stop indexing when empty
11 * Revision 1.86 1998/10/13 20:33:53 adam
12 * Fixed one log message and change use ordinal to be an unsigned char.
14 * Revision 1.85 1998/09/22 10:03:41 adam
15 * Changed result sets to be persistent in the sense that they can
16 * be re-searched if needed.
17 * Fixed memory leak in rsm_or.
19 * Revision 1.84 1998/06/11 15:42:22 adam
20 * Changed the way use attributes are specified in the recordId
23 * Revision 1.83 1998/06/08 14:43:10 adam
24 * Added suport for EXPLAIN Proxy servers - added settings databasePath
25 * and explainDatabase to facilitate this. Increased maximum number
26 * of databases and attributes in one register.
28 * Revision 1.82 1998/05/20 10:12:15 adam
29 * Implemented automatic EXPLAIN database maintenance.
30 * Modified Zebra to work with ASN.1 compiled version of YAZ.
32 * Revision 1.81 1998/03/11 11:19:04 adam
33 * Changed the way sequence numbers are generated.
35 * Revision 1.80 1998/03/05 08:45:11 adam
36 * New result set model and modular ranking system. Moved towards
37 * descent server API. System information stored as "SGML" records.
39 * Revision 1.79 1998/02/17 10:32:52 adam
40 * Fixed bug: binary files weren't opened with flag b on NT.
42 * Revision 1.78 1998/02/10 12:03:05 adam
45 * Revision 1.77 1998/01/12 15:04:08 adam
46 * The test option (-s) only uses read-lock (and not write lock).
48 * Revision 1.76 1997/10/27 14:33:04 adam
49 * Moved towards generic character mapping depending on "structure"
50 * field in abstract syntax file. Fixed a few memory leaks. Fixed
51 * bug with negative integers when doing searches with relational
54 * Revision 1.75 1997/09/17 12:19:12 adam
55 * Zebra version corresponds to YAZ version 1.4.
56 * Changed Zebra server so that it doesn't depend on global common_resource.
58 * Revision 1.74 1997/09/09 13:38:06 adam
59 * Partial port to WIN95/NT.
61 * Revision 1.73 1997/09/04 13:57:20 adam
62 * New file extract/retrieve method tellf (added).
63 * Added O_BINARY for open calls.
65 * Revision 1.72 1997/07/15 16:32:29 adam
66 * Bug fix: Match handler didn't terminate the resulting string!
68 * Revision 1.71 1997/07/15 16:28:41 adam
69 * Bug fix: storeData didn't work with files with multiple records.
70 * Bug fix: fixed memory management with records; not really well
73 * Revision 1.70 1997/07/01 13:00:42 adam
74 * Bug fix in routine searchRecordKey: uninitialized variables.
76 * Revision 1.69 1997/04/29 09:26:03 adam
77 * Bug fix: generic recordId handling didn't work for compressed internal
80 * Revision 1.68 1997/02/12 20:39:45 adam
81 * Implemented options -f <n> that limits the log to the first <n>
83 * Changed some log messages also.
85 * Revision 1.67 1996/11/15 15:02:14 adam
86 * Minor changes regarding logging.
88 * Revision 1.66 1996/11/14 09:52:21 adam
89 * Strings in record keys bound by IT_MAX_WORD.
91 * Revision 1.65 1996/11/14 08:57:56 adam
92 * Reduction of storeKeys area.
94 * Revision 1.64 1996/11/08 11:10:16 adam
95 * Buffers used during file match got bigger.
96 * Compressed ISAM support everywhere.
97 * Bug fixes regarding masking characters in queries.
98 * Redesigned Regexp-2 queries.
100 * Revision 1.63 1996/10/29 14:09:39 adam
101 * Use of cisam system - enabled if setting isamc is 1.
103 * Revision 1.62 1996/10/11 10:57:01 adam
104 * New module recctrl. Used to manage records (extract/retrieval).
105 * Several files have been moved to the recctrl sub directory.
107 * Revision 1.61 1996/06/06 12:08:37 quinn
108 * Added showRecord function
110 * Revision 1.60 1996/06/04 10:18:12 adam
111 * Search/scan uses character mapping module.
113 * Revision 1.59 1996/05/14 15:47:07 adam
114 * Cleanup of various buffer size entities.
116 * Revision 1.58 1996/05/14 06:16:38 adam
117 * Compact use/set bytes used in search service.
119 * Revision 1.57 1996/05/13 14:23:04 adam
120 * Work on compaction of set/use bytes in dictionary.
122 * Revision 1.56 1996/05/09 09:54:42 adam
123 * Server supports maps from one logical attributes to a list of physical
125 * The extraction process doesn't make space consuming 'any' keys.
127 * Revision 1.55 1996/05/09 07:28:55 quinn
128 * Work towards phrases and multiple registers
130 * Revision 1.54 1996/05/01 13:46:35 adam
131 * First work on multiple records in one file.
132 * New option, -offset, to the "unread" command in the filter module.
134 * Revision 1.53 1996/04/26 12:09:43 adam
135 * Added a few comments.
137 * Revision 1.52 1996/04/25 13:27:57 adam
138 * Function recordExtract modified so that files with no keys (possibly empty)
141 * Revision 1.51 1996/03/19 11:08:42 adam
142 * Bug fix: Log preamble wasn't always turned off after recordExtract.
144 * Revision 1.50 1996/02/12 18:45:36 adam
145 * New fileVerboseFlag in record group control.
147 * Revision 1.49 1996/02/05 12:29:57 adam
148 * Logging reduced a bit.
149 * The remaining running time is estimated during register merge.
151 * Revision 1.48 1996/02/01 20:53:26 adam
152 * The temporary per-record keys are compacted a little, and duplication
153 * of the per-records keys are avoided when they are saved in the record
154 * information buffer.
156 * Revision 1.47 1996/01/17 14:57:48 adam
157 * Prototype changed for reader functions in extract/retrieve. File
158 * is identified by 'void *' instead of 'int.
160 * Revision 1.46 1995/12/15 14:57:16 adam
163 * Revision 1.45 1995/12/15 12:37:41 adam
164 * In addRecordKeyAny: Writes key only when attrSet != -1.
166 * Revision 1.44 1995/12/12 16:00:54 adam
167 * System call sync(2) used after update/commit.
168 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
171 * Revision 1.43 1995/12/11 09:12:46 adam
172 * The rec_get function returns NULL if record doesn't exist - will
173 * happen in the server if the result set records have been deleted since
174 * the creation of the set (i.e. the search).
175 * The server saves a result temporarily if it is 'volatile', i.e. the
176 * set is register dependent.
178 * Revision 1.42 1995/12/07 17:38:46 adam
179 * Work locking mechanisms for concurrent updates/commit.
181 * Revision 1.41 1995/12/06 16:06:42 adam
182 * Better diagnostics. Work on 'real' dictionary deletion.
184 * Revision 1.40 1995/12/05 16:57:40 adam
185 * More work on regular patterns.
187 * Revision 1.39 1995/12/05 13:20:18 adam
188 * Bug fix: file_read sometimes returned early EOF.
190 * Revision 1.38 1995/12/04 17:59:21 adam
191 * More work on regular expression conversion.
193 * Revision 1.37 1995/12/04 14:22:27 adam
194 * Extra arg to recType_byName.
195 * Started work on new regular expression parsed input to
196 * structured records.
198 * Revision 1.36 1995/11/30 08:34:29 adam
199 * Started work on commit facility.
200 * Changed a few malloc/free to xmalloc/xfree.
202 * Revision 1.35 1995/11/28 14:26:21 adam
203 * Bug fix: recordId with constant wasn't right.
204 * Bug fix: recordId dictionary entry wasn't deleted when needed.
206 * Revision 1.34 1995/11/28 09:09:38 adam
207 * Zebra config renamed.
208 * Use setting 'recordId' to identify record now.
209 * Bug fix in recindex.c: rec_release_blocks was invokeded even
210 * though the blocks were already released.
211 * File traversal properly deletes records when needed.
213 * Revision 1.33 1995/11/27 09:56:20 adam
214 * Record info elements better enumerated. Internal store of records.
216 * Revision 1.32 1995/11/25 10:24:05 adam
217 * More record fields - they are enumerated now.
218 * New options: flagStoreData flagStoreKey.
220 * Revision 1.31 1995/11/24 11:31:35 adam
221 * Commands add & del read filenames from stdin if source directory is
223 * Match criteria supports 'constant' strings.
225 * Revision 1.30 1995/11/22 17:19:16 adam
226 * Record management uses the bfile system.
228 * Revision 1.29 1995/11/21 15:01:14 adam
229 * New general match criteria implemented.
230 * New feature: document groups.
232 * Revision 1.28 1995/11/21 09:20:30 adam
233 * Yet more work on record match.
235 * Revision 1.27 1995/11/20 16:59:45 adam
236 * New update method: the 'old' keys are saved for each records.
238 * Revision 1.26 1995/11/20 11:56:24 adam
239 * Work on new traversal.
241 * Revision 1.25 1995/11/16 15:34:54 adam
242 * Uses new record management system in both indexer and server.
244 * Revision 1.24 1995/11/15 19:13:08 adam
245 * Work on record management.
247 * Revision 1.23 1995/10/27 14:00:10 adam
248 * Implemented detection of database availability.
250 * Revision 1.22 1995/10/17 18:02:07 adam
251 * New feature: databases. Implemented as prefix to words in dictionary.
253 * Revision 1.21 1995/10/10 12:24:38 adam
254 * Temporary sort files are compressed.
256 * Revision 1.20 1995/10/06 13:52:05 adam
257 * Bug fixes. Handler may abort further scanning.
259 * Revision 1.19 1995/10/04 12:55:16 adam
260 * Bug fix in ranked search. Use=Any keys inserted.
262 * Revision 1.18 1995/10/04 09:37:08 quinn
265 * Revision 1.17 1995/10/03 14:28:57 adam
266 * Buffered read in extract works.
268 * Revision 1.16 1995/10/03 14:28:45 adam
269 * Work on more effecient read handler in extract.
271 * Revision 1.15 1995/10/02 15:42:53 adam
272 * Extract uses file descriptors instead of FILE pointers.
274 * Revision 1.14 1995/10/02 15:29:13 adam
275 * More logging in file_extract.
277 * Revision 1.13 1995/09/29 14:01:39 adam
280 * Revision 1.12 1995/09/28 14:22:56 adam
281 * Sort uses smaller temporary files.
283 * Revision 1.11 1995/09/28 12:10:31 adam
284 * Bug fixes. Field prefix used in queries.
286 * Revision 1.10 1995/09/28 09:19:41 adam
287 * xfree/xmalloc used everywhere.
288 * Extract/retrieve method seems to work for text records.
290 * Revision 1.9 1995/09/27 12:22:28 adam
291 * More work on extract in record control.
292 * Field name is not in isam keys but in prefix in dictionary words.
294 * Revision 1.8 1995/09/14 07:48:22 adam
295 * Record control management.
297 * Revision 1.7 1995/09/11 13:09:32 adam
298 * More work on relevance feedback.
300 * Revision 1.6 1995/09/08 14:52:27 adam
301 * Minor changes. Dictionary is lower case now.
303 * Revision 1.5 1995/09/06 16:11:16 adam
304 * Option: only one word key per file.
306 * Revision 1.4 1995/09/05 15:28:39 adam
307 * More work on search engine.
309 * Revision 1.3 1995/09/04 12:33:41 adam
310 * Various cleanup. YAZ util used instead.
312 * Revision 1.2 1995/09/04 09:10:34 adam
313 * More work on index add/del/update.
314 * Merge sort implemented.
315 * Initial work on z39 server.
317 * Revision 1.1 1995/09/01 14:06:35 adam
318 * Split of work into more files.
337 static Dict matchDict;
339 static Records records = NULL;
340 static SortIdx sortIdx = NULL;
342 static char **key_buf;
343 static size_t ptr_top;
345 static size_t key_buf_used;
346 static int key_file_no;
348 static int records_inserted = 0;
349 static int records_updated = 0;
350 static int records_deleted = 0;
351 static int records_processed = 0;
353 static ZebraExplainInfo zti = NULL;
355 static void logRecord (int showFlag)
359 if (showFlag || !(records_processed % 1000))
361 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
362 records_processed, records_inserted, records_updated,
367 static int explain_extract (void *handle, Record drec, data1_node *n);
369 int key_open (struct recordGroup *rGroup, int mem)
371 BFiles bfs = rGroup->bfs;
372 int rw = rGroup->flagRw;
373 data1_handle dh = rGroup->dh;
375 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
378 key_buf = xmalloc (mem);
379 ptr_top = mem/sizeof(char*);
385 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
387 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
391 records = rec_open (bfs, rw);
394 dict_close (matchDict);
397 zti = zebraExplain_open (records, dh, common_resource,
398 rw, rGroup, explain_extract);
401 rec_close (&records);
402 dict_close (matchDict);
405 sortIdx = sortIdx_open (bfs, 1);
415 void encode_key_init (struct encode_info *i)
421 char *encode_key_int (int d, char *bp)
430 else if (d <= 4194303)
432 *bp++ = 128 + (d>>16);
433 *bp++ = (d>>8) & 255;
438 *bp++ = 192 + (d>>24);
439 *bp++ = (d>>16) & 255;
440 *bp++ = (d>>8) & 255;
446 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
451 while ((*bp++ = *k++))
453 memcpy (&key, k+1, sizeof(struct it_key));
454 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
455 if (i->sysno != key.sysno)
457 i->sysno = key.sysno;
460 bp = encode_key_int (key.seqno - i->seqno, bp);
461 i->seqno = key.seqno;
462 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
464 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
472 static int key_y_len;
474 static int key_y_compare (const void *p1, const void *p2)
478 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
479 *(char**) p2 + key_y_len + 1)))
481 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
484 static int key_x_compare (const void *p1, const void *p2)
486 return strcmp (*(char**) p1, *(char**) p2);
490 void key_flush (void)
495 struct encode_info encode_info;
504 logf (LOG_LOG, "sorting section %d", key_file_no);
506 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
507 getFnameTmp (out_fname, key_file_no);
509 if (!(outf = fopen (out_fname, "wb")))
511 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
514 logf (LOG_LOG, "writing section %d", key_file_no);
515 prevcp = cp = key_buf[ptr_top-ptr_i];
517 encode_key_init (&encode_info);
518 encode_key_write (cp, &encode_info, outf);
521 cp = key_buf[ptr_top-ptr_i];
522 if (strcmp (cp, prevcp))
524 encode_key_init (&encode_info);
525 encode_key_write (cp, &encode_info, outf);
529 encode_key_write (cp + strlen(cp), &encode_info, outf);
532 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
533 getFnameTmp (out_fname, key_file_no);
535 if (!(outf = fopen (out_fname, "wb")))
537 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
540 logf (LOG_LOG, "writing section %d", key_file_no);
542 prevcp = key_buf[ptr_top-i];
544 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
546 key_y_len = strlen(prevcp)+1;
548 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
549 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
551 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
552 sizeof(char*), key_y_compare);
553 cp = key_buf[ptr_top-ptr_i];
555 encode_key_init (&encode_info);
556 encode_key_write (cp, &encode_info, outf);
559 cp = key_buf[ptr_top-ptr_i];
560 encode_key_write (cp+key_y_len, &encode_info, outf);
564 prevcp = key_buf[ptr_top-ptr_i];
569 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
572 logf (LOG_LOG, "finished section %d", key_file_no);
577 int key_close (struct recordGroup *rGroup)
579 int rw = rGroup->flagRw;
581 zebraExplain_runNumberIncrement (zti, 1);
582 zebraExplain_close (zti, rw, 0);
585 rec_close (&records);
586 dict_close (matchDict);
587 sortIdx_close (sortIdx);
593 static void wordInit (struct recExtractCtrl *p, RecWord *w)
595 w->zebra_maps = p->zebra_maps;
596 w->seqnos = p->seqno;
597 w->attrSet = VAL_BIB1;
602 static struct sortKey {
607 struct sortKey *next;
610 static struct recKeys {
619 static void addIndexString (RecWord *p, const char *string, int length)
622 unsigned char attrSet;
623 unsigned short attrUse;
626 int *pseqno = &p->seqnos[p->reg_type];
628 if (reckeys.buf_used+1024 > reckeys.buf_max)
632 b = xmalloc (reckeys.buf_max += 128000);
633 if (reckeys.buf_used > 0)
634 memcpy (b, reckeys.buf, reckeys.buf_used);
638 dst = reckeys.buf + reckeys.buf_used;
640 attrSet = p->attrSet;
641 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
644 reckeys.prevAttrSet = attrSet;
645 attrUse = p->attrUse;
646 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
649 reckeys.prevAttrUse = attrUse;
651 diff = 1 + *pseqno - reckeys.prevSeqNo;
652 if (diff >= 1 && diff <= 15)
657 reckeys.prevSeqNo = *pseqno;
663 memcpy (dst, &attrSet, sizeof(attrSet));
664 dst += sizeof(attrSet);
668 memcpy (dst, &attrUse, sizeof(attrUse));
669 dst += sizeof(attrUse);
671 *dst++ = p->reg_type;
672 memcpy (dst, string, length);
678 memcpy (dst, pseqno, sizeof(*pseqno));
679 dst += sizeof(*pseqno);
681 reckeys.buf_used = dst - reckeys.buf;
685 static void addSortString (RecWord *p, const char *string, int length)
689 for (sk = sortKeys; sk; sk = sk->next)
690 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
693 sk = xmalloc (sizeof(*sk));
697 sk->string = xmalloc (length);
699 memcpy (sk->string, string, length);
701 sk->attrSet = p->attrSet;
702 sk->attrUse = p->attrUse;
705 static void addString (RecWord *p, const char *string, int length)
708 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
709 addSortString (p, string, length);
711 addIndexString (p, string, length);
714 static void addIncompleteField (RecWord *p)
716 const char *b = p->string;
717 int remain = p->length;
718 const char **map = 0;
721 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
725 char buf[IT_MAX_WORD+1];
729 while (map && *map && **map == *CHR_SPACE)
731 remain = p->length - (b - p->string);
733 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
740 while (map && *map && **map != *CHR_SPACE)
742 const char *cp = *map;
744 while (i < IT_MAX_WORD && *cp)
746 remain = p->length - (b - p->string);
748 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
754 addString (p, buf, i);
758 static void addCompleteField (RecWord *p)
760 const char *b = p->string;
761 char buf[IT_MAX_WORD+1];
762 const char **map = 0;
763 int i = 0, remain = p->length;
766 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
768 while (remain > 0 && i < IT_MAX_WORD)
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 if (i && i < IT_MAX_WORD)
782 buf[i++] = *CHR_SPACE;
783 while (map && *map && **map != *CHR_SPACE)
785 const char *cp = *map;
787 if (i >= IT_MAX_WORD)
789 while (i < IT_MAX_WORD && *cp)
791 remain = p->length - (b - p->string);
793 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
801 addString (p, buf, i);
804 static void addRecordKey (RecWord *p)
806 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
807 addCompleteField (p);
809 addIncompleteField(p);
812 static void flushSortKeys (SYSNO sysno, int cmd)
814 struct sortKey *sk = sortKeys;
816 sortIdx_sysno (sortIdx, sysno);
819 struct sortKey *sk_next = sk->next;
820 sortIdx_type (sortIdx, sk->attrUse);
821 sortIdx_add (sortIdx, sk->string, sk->length);
829 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
831 unsigned char attrSet = (unsigned char) -1;
832 unsigned short attrUse = (unsigned short) -1;
836 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
837 while (off < reckeys->buf_used)
839 const char *src = reckeys->buf + off;
847 memcpy (&attrSet, src, sizeof(attrSet));
848 src += sizeof(attrSet);
852 memcpy (&attrUse, src, sizeof(attrUse));
853 src += sizeof(attrUse);
855 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
858 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
860 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
862 ch = zebraExplain_addSU (zti, attrSet, attrUse);
864 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
867 ((char*)key_buf) [key_buf_used++] = *src++;
869 ((char*)key_buf) [key_buf_used++] = '\0';
870 ((char*) key_buf)[key_buf_used++] = cmd;
873 seqno += ((lead>>2) & 15)-1;
876 memcpy (&seqno, src, sizeof(seqno));
877 src += sizeof(seqno);
881 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
882 key_buf_used += sizeof(key);
883 off = src - reckeys->buf;
885 assert (off == reckeys->buf_used);
888 static const char **searchRecordKey (struct recKeys *reckeys,
889 int attrSetS, int attrUseS)
891 static const char *ws[32];
899 for (i = 0; i<32; i++)
902 while (off < reckeys->buf_used)
905 const char *src = reckeys->buf + off;
913 memcpy (&attrSet, src, sizeof(attrSet));
914 src += sizeof(attrSet);
918 memcpy (&attrUse, src, sizeof(attrUse));
919 src += sizeof(attrUse);
925 seqno += ((lead>>2) & 15)-1;
928 memcpy (&seqno, src, sizeof(seqno));
929 src += sizeof(seqno);
931 if (attrUseS == attrUse && attrSetS == attrSet)
938 woff = seqno - startSeq;
939 if (woff >= 0 && woff < 31)
943 off = src - reckeys->buf;
945 assert (off == reckeys->buf_used);
949 struct file_read_info {
957 static struct file_read_info *file_read_start (int fd)
959 struct file_read_info *fi = xmalloc (sizeof(*fi));
963 fi->file_moffset = 0;
967 static void file_read_stop (struct file_read_info *fi)
972 static off_t file_seek (void *handle, off_t offset)
974 struct file_read_info *p = handle;
975 p->file_offset = offset;
976 return lseek (p->fd, offset, SEEK_SET);
979 static off_t file_tell (void *handle)
981 struct file_read_info *p = handle;
982 return p->file_offset;
985 static int file_read (void *handle, char *buf, size_t count)
987 struct file_read_info *p = handle;
990 r = read (fd, buf, count);
994 if (p->file_offset > p->file_max)
995 p->file_max = p->file_offset;
1000 static void file_begin (void *handle)
1002 struct file_read_info *p = handle;
1004 p->file_offset = p->file_moffset;
1005 if (p->file_moffset)
1006 lseek (p->fd, p->file_moffset, SEEK_SET);
1010 static void file_end (void *handle, off_t offset)
1012 struct file_read_info *p = handle;
1014 assert (p->file_more == 0);
1016 p->file_moffset = offset;
1019 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1020 const char *fname, const char *spec)
1022 static char dstBuf[2048];
1024 const char *s = spec;
1025 static const char **w;
1029 while (*s == ' ' || *s == '\t')
1035 char attset_str[64], attname_str[64];
1036 data1_attset *attset;
1039 int attSet = 1, attUse = 1;
1043 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1045 attset_str[i++] = *s;
1046 attset_str[i] = '\0';
1051 for (i = 0; *s && *s != ')'; s++)
1053 attname_str[i++] = *s;
1054 attname_str[i] = '\0';
1057 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1060 attSet = attset->reference;
1061 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1063 attUse = att->value;
1065 attUse = atoi (attname_str);
1067 w = searchRecordKey (reckeys, attSet, attUse);
1072 for (i = 0; i<32; i++)
1077 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1078 spec, rGroup->groupName ? rGroup->groupName : "none");
1083 for (i = 0; i<32; i++)
1084 if (matchFlag[i] && w[i])
1092 dst += strlen(w[i]);
1096 logf (LOG_WARN, "Record didn't contain match"
1097 " fields in (%s,%s)", attset_str, attname_str);
1105 const char *spec_src = NULL;
1106 const char *s1 = ++s;
1107 while (*s1 && *s1 != ' ' && *s1 != '\t')
1113 memcpy (special, s, spec_len);
1114 special[spec_len] = '\0';
1117 if (!strcmp (special, "group"))
1118 spec_src = rGroup->groupName;
1119 else if (!strcmp (special, "database"))
1120 spec_src = rGroup->databaseName;
1121 else if (!strcmp (special, "filename"))
1123 else if (!strcmp (special, "type"))
1124 spec_src = rGroup->recordType;
1129 strcpy (dst, spec_src);
1130 dst += strlen (spec_src);
1133 else if (*s == '\"' || *s == '\'')
1135 int stopMarker = *s++;
1139 while (*s && *s != stopMarker)
1142 tmpString[i++] = *s++;
1146 tmpString[i] = '\0';
1147 strcpy (dst, tmpString);
1148 dst += strlen (tmpString);
1152 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1153 spec, rGroup->groupName ? rGroup->groupName : "none");
1160 logf (LOG_WARN, "No match criteria for record %s in group %s",
1161 fname, rGroup->groupName ? rGroup->groupName : "none");
1168 struct recordLogInfo {
1171 struct recordGroup *rGroup;
1174 static void recordLogPreamble (int level, const char *msg, void *info)
1176 struct recordLogInfo *p = info;
1177 FILE *outf = log_file ();
1179 if (level & LOG_LOG)
1181 fprintf (outf, "File %s, offset %d, type %s\n",
1182 p->rGroup->recordType, p->recordOffset, p->fname);
1183 log_event_start (NULL, NULL);
1186 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1188 zebraExplain_addSchema (zti, oid);
1191 static int recordExtract (SYSNO *sysno, const char *fname,
1192 struct recordGroup *rGroup, int deleteFlag,
1193 struct file_read_info *fi,
1194 RecType recType, char *subType)
1196 RecordAttr *recordAttr;
1201 struct recordLogInfo logInfo;
1202 off_t recordOffset = 0;
1206 struct recExtractCtrl extractCtrl;
1208 /* we are going to read from a file, so prepare the extraction */
1211 reckeys.buf_used = 0;
1212 reckeys.prevAttrUse = -1;
1213 reckeys.prevAttrSet = -1;
1214 reckeys.prevSeqNo = 0;
1216 recordOffset = fi->file_moffset;
1217 extractCtrl.offset = fi->file_moffset;
1218 extractCtrl.readf = file_read;
1219 extractCtrl.seekf = file_seek;
1220 extractCtrl.tellf = file_tell;
1221 extractCtrl.endf = file_end;
1222 extractCtrl.fh = fi;
1223 extractCtrl.subType = subType;
1224 extractCtrl.init = wordInit;
1225 extractCtrl.addWord = addRecordKey;
1226 extractCtrl.addSchema = addSchema;
1227 extractCtrl.dh = rGroup->dh;
1228 for (i = 0; i<256; i++)
1229 extractCtrl.seqno[i] = 0;
1230 extractCtrl.zebra_maps = rGroup->zebra_maps;
1231 extractCtrl.flagShowRecords = !rGroup->flagRw;
1233 if (!rGroup->flagRw)
1234 printf ("File: %s %ld\n", fname, (long) recordOffset);
1236 logInfo.fname = fname;
1237 logInfo.recordOffset = recordOffset;
1238 logInfo.rGroup = rGroup;
1239 log_event_start (recordLogPreamble, &logInfo);
1241 r = (*recType->extract)(&extractCtrl);
1243 log_event_start (NULL, NULL);
1247 /* error occured during extraction ... */
1248 if (rGroup->flagRw &&
1249 records_processed < rGroup->fileVerboseLimit)
1251 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1252 fname, (long) recordOffset, r);
1256 if (reckeys.buf_used == 0)
1258 /* the extraction process returned no information - the record
1259 is probably empty - unless flagShowRecords is in use */
1260 if (!rGroup->flagRw)
1262 logf (LOG_WARN, "No keys generated for file %s", fname);
1263 logf (LOG_WARN, " The file is probably empty");
1268 /* perform match if sysno not known and if match criteria is specified */
1275 if (rGroup->recordId && *rGroup->recordId)
1279 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1283 rinfo = dict_lookup (matchDict, matchStr);
1285 memcpy (sysno, rinfo+1, sizeof(*sysno));
1289 logf (LOG_WARN, "Bad match criteria");
1300 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1301 fname, (long) recordOffset);
1302 logf (LOG_WARN, "cannot delete record above (seems new)");
1305 if (records_processed < rGroup->fileVerboseLimit)
1306 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1307 fname, (long) recordOffset);
1308 rec = rec_new (records);
1310 *sysno = rec->sysno;
1312 recordAttr = rec_init_attr (zti, rec);
1316 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1318 flushRecordKeys (*sysno, 1, &reckeys);
1319 flushSortKeys (*sysno, 1);
1325 /* record already exists */
1326 struct recKeys delkeys;
1328 rec = rec_get (records, *sysno);
1331 recordAttr = rec_init_attr (zti, rec);
1333 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1335 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1336 fname, (long) recordOffset);
1341 delkeys.buf_used = rec->size[recInfo_delKeys];
1342 delkeys.buf = rec->info[recInfo_delKeys];
1343 flushSortKeys (*sysno, 0);
1344 flushRecordKeys (*sysno, 0, &delkeys);
1347 /* record going to be deleted */
1348 if (!delkeys.buf_used)
1350 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1351 fname, (long) recordOffset);
1352 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1356 if (records_processed < rGroup->fileVerboseLimit)
1357 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1358 fname, (long) recordOffset);
1361 dict_delete (matchDict, matchStr);
1362 rec_del (records, &rec);
1370 /* record going to be updated */
1371 if (!delkeys.buf_used)
1373 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1374 fname, (long) recordOffset);
1375 logf (LOG_WARN, "cannot update file above, storeKeys false");
1379 if (records_processed < rGroup->fileVerboseLimit)
1380 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1381 fname, (long) recordOffset);
1382 flushRecordKeys (*sysno, 1, &reckeys);
1387 /* update file type */
1388 xfree (rec->info[recInfo_fileType]);
1389 rec->info[recInfo_fileType] =
1390 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1392 /* update filename */
1393 xfree (rec->info[recInfo_filename]);
1394 rec->info[recInfo_filename] =
1395 rec_strdup (fname, &rec->size[recInfo_filename]);
1397 /* update delete keys */
1398 xfree (rec->info[recInfo_delKeys]);
1399 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1402 rec->size[recInfo_delKeys] = reckeys.buf_used;
1403 rec->info[recInfo_delKeys] = reckeys.buf;
1405 reckeys.buf_max = 0;
1407 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1408 rec->size[recInfo_delKeys] = reckeys.buf_used;
1409 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1410 rec->size[recInfo_delKeys]);
1415 rec->info[recInfo_delKeys] = NULL;
1416 rec->size[recInfo_delKeys] = 0;
1419 /* save file size of original record */
1420 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1421 recordAttr->recordSize = fi->file_moffset - recordOffset;
1422 if (!recordAttr->recordSize)
1423 recordAttr->recordSize = fi->file_max - recordOffset;
1424 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1426 /* set run-number for this record */
1427 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1429 /* update store data */
1430 xfree (rec->info[recInfo_storeData]);
1431 if (rGroup->flagStoreData == 1)
1433 rec->size[recInfo_storeData] = recordAttr->recordSize;
1434 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1435 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1437 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1438 (long) recordOffset, fname);
1441 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1442 < recordAttr->recordSize)
1444 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1445 recordAttr->recordSize, fname);
1451 rec->info[recInfo_storeData] = NULL;
1452 rec->size[recInfo_storeData] = 0;
1454 /* update database name */
1455 xfree (rec->info[recInfo_databaseName]);
1456 rec->info[recInfo_databaseName] =
1457 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1460 recordAttr->recordOffset = recordOffset;
1462 /* commit this record */
1463 rec_put (records, &rec);
1468 int fileExtract (SYSNO *sysno, const char *fname,
1469 const struct recordGroup *rGroupP, int deleteFlag)
1477 struct recordGroup rGroupM;
1478 struct recordGroup *rGroup = &rGroupM;
1479 struct file_read_info *fi;
1481 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1483 if (!rGroup->groupName || !*rGroup->groupName)
1486 sprintf (gprefix, "%s.", rGroup->groupName);
1488 logf (LOG_DEBUG, "fileExtract %s", fname);
1490 /* determine file extension */
1492 for (i = strlen(fname); --i >= 0; )
1493 if (fname[i] == '/')
1495 else if (fname[i] == '.')
1497 strcpy (ext, fname+i+1);
1500 /* determine file type - depending on extension */
1501 if (!rGroup->recordType)
1503 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1504 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1506 sprintf (ext_res, "%srecordType", gprefix);
1507 rGroup->recordType = res_get (common_resource, ext_res);
1510 if (!rGroup->recordType)
1512 if (records_processed < rGroup->fileVerboseLimit)
1513 logf (LOG_LOG, "? %s", fname);
1516 if (!*rGroup->recordType)
1518 if (!(recType = recType_byName (rGroup->recordType, subType)))
1520 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1524 /* determine match criteria */
1525 if (!rGroup->recordId)
1527 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1528 rGroup->recordId = res_get (common_resource, ext_res);
1531 /* determine database name */
1532 if (!rGroup->databaseName)
1534 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1535 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1537 sprintf (ext_res, "%sdatabase", gprefix);
1538 rGroup->databaseName = res_get (common_resource, ext_res);
1541 if (!rGroup->databaseName)
1542 rGroup->databaseName = "Default";
1544 /* determine if explain database */
1546 sprintf (ext_res, "%sexplainDatabase", gprefix);
1547 rGroup->explainDatabase =
1548 atoi (res_get_def (common_resource, ext_res, "0"));
1550 /* announce database */
1551 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1553 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1554 rGroup->explainDatabase))
1558 if (rGroup->flagStoreData == -1)
1561 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1562 if (!(sval = res_get (common_resource, ext_res)))
1564 sprintf (ext_res, "%sstoreData", gprefix);
1565 sval = res_get (common_resource, ext_res);
1568 rGroup->flagStoreData = atoi (sval);
1570 if (rGroup->flagStoreData == -1)
1571 rGroup->flagStoreData = 0;
1573 if (rGroup->flagStoreKeys == -1)
1577 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1578 if (!(sval = res_get (common_resource, ext_res)))
1580 sprintf (ext_res, "%sstoreKeys", gprefix);
1581 sval = res_get (common_resource, ext_res);
1584 rGroup->flagStoreKeys = atoi (sval);
1586 if (rGroup->flagStoreKeys == -1)
1587 rGroup->flagStoreKeys = 0;
1589 if (sysno && deleteFlag)
1593 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1595 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1599 fi = file_read_start (fd);
1603 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1605 } while (r && !sysno && fi->file_more);
1606 file_read_stop (fi);
1612 static int explain_extract (void *handle, Record rec, data1_node *n)
1614 struct recordGroup *rGroup = (struct recordGroup*) handle;
1615 struct recExtractCtrl extractCtrl;
1618 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1620 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1624 reckeys.buf_used = 0;
1625 reckeys.prevAttrUse = -1;
1626 reckeys.prevAttrSet = -1;
1627 reckeys.prevSeqNo = 0;
1629 extractCtrl.init = wordInit;
1630 extractCtrl.addWord = addRecordKey;
1631 extractCtrl.addSchema = addSchema;
1632 extractCtrl.dh = rGroup->dh;
1633 for (i = 0; i<256; i++)
1634 extractCtrl.seqno[i] = 0;
1635 extractCtrl.zebra_maps = rGroup->zebra_maps;
1636 extractCtrl.flagShowRecords = !rGroup->flagRw;
1638 grs_extract_tree(&extractCtrl, n);
1640 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1642 if (rec->size[recInfo_delKeys])
1644 struct recKeys delkeys;
1646 delkeys.buf_used = rec->size[recInfo_delKeys];
1647 delkeys.buf = rec->info[recInfo_delKeys];
1648 flushSortKeys (rec->sysno, 0);
1649 flushRecordKeys (rec->sysno, 0, &delkeys);
1651 flushRecordKeys (rec->sysno, 1, &reckeys);
1652 flushSortKeys (rec->sysno, 1);
1654 xfree (rec->info[recInfo_delKeys]);
1655 rec->size[recInfo_delKeys] = reckeys.buf_used;
1656 rec->info[recInfo_delKeys] = reckeys.buf;
1658 reckeys.buf_max = 0;