2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.86 1998-10-13 20:33:53 adam
8 * Fixed one log message and change use ordinal to be an unsigned char.
10 * Revision 1.85 1998/09/22 10:03:41 adam
11 * Changed result sets to be persistent in the sense that they can
12 * be re-searched if needed.
13 * Fixed memory leak in rsm_or.
15 * Revision 1.84 1998/06/11 15:42:22 adam
16 * Changed the way use attributes are specified in the recordId
19 * Revision 1.83 1998/06/08 14:43:10 adam
20 * Added suport for EXPLAIN Proxy servers - added settings databasePath
21 * and explainDatabase to facilitate this. Increased maximum number
22 * of databases and attributes in one register.
24 * Revision 1.82 1998/05/20 10:12:15 adam
25 * Implemented automatic EXPLAIN database maintenance.
26 * Modified Zebra to work with ASN.1 compiled version of YAZ.
28 * Revision 1.81 1998/03/11 11:19:04 adam
29 * Changed the way sequence numbers are generated.
31 * Revision 1.80 1998/03/05 08:45:11 adam
32 * New result set model and modular ranking system. Moved towards
33 * descent server API. System information stored as "SGML" records.
35 * Revision 1.79 1998/02/17 10:32:52 adam
36 * Fixed bug: binary files weren't opened with flag b on NT.
38 * Revision 1.78 1998/02/10 12:03:05 adam
41 * Revision 1.77 1998/01/12 15:04:08 adam
42 * The test option (-s) only uses read-lock (and not write lock).
44 * Revision 1.76 1997/10/27 14:33:04 adam
45 * Moved towards generic character mapping depending on "structure"
46 * field in abstract syntax file. Fixed a few memory leaks. Fixed
47 * bug with negative integers when doing searches with relational
50 * Revision 1.75 1997/09/17 12:19:12 adam
51 * Zebra version corresponds to YAZ version 1.4.
52 * Changed Zebra server so that it doesn't depend on global common_resource.
54 * Revision 1.74 1997/09/09 13:38:06 adam
55 * Partial port to WIN95/NT.
57 * Revision 1.73 1997/09/04 13:57:20 adam
58 * New file extract/retrieve method tellf (added).
59 * Added O_BINARY for open calls.
61 * Revision 1.72 1997/07/15 16:32:29 adam
62 * Bug fix: Match handler didn't terminate the resulting string!
64 * Revision 1.71 1997/07/15 16:28:41 adam
65 * Bug fix: storeData didn't work with files with multiple records.
66 * Bug fix: fixed memory management with records; not really well
69 * Revision 1.70 1997/07/01 13:00:42 adam
70 * Bug fix in routine searchRecordKey: uninitialized variables.
72 * Revision 1.69 1997/04/29 09:26:03 adam
73 * Bug fix: generic recordId handling didn't work for compressed internal
76 * Revision 1.68 1997/02/12 20:39:45 adam
77 * Implemented options -f <n> that limits the log to the first <n>
79 * Changed some log messages also.
81 * Revision 1.67 1996/11/15 15:02:14 adam
82 * Minor changes regarding logging.
84 * Revision 1.66 1996/11/14 09:52:21 adam
85 * Strings in record keys bound by IT_MAX_WORD.
87 * Revision 1.65 1996/11/14 08:57:56 adam
88 * Reduction of storeKeys area.
90 * Revision 1.64 1996/11/08 11:10:16 adam
91 * Buffers used during file match got bigger.
92 * Compressed ISAM support everywhere.
93 * Bug fixes regarding masking characters in queries.
94 * Redesigned Regexp-2 queries.
96 * Revision 1.63 1996/10/29 14:09:39 adam
97 * Use of cisam system - enabled if setting isamc is 1.
99 * Revision 1.62 1996/10/11 10:57:01 adam
100 * New module recctrl. Used to manage records (extract/retrieval).
101 * Several files have been moved to the recctrl sub directory.
103 * Revision 1.61 1996/06/06 12:08:37 quinn
104 * Added showRecord function
106 * Revision 1.60 1996/06/04 10:18:12 adam
107 * Search/scan uses character mapping module.
109 * Revision 1.59 1996/05/14 15:47:07 adam
110 * Cleanup of various buffer size entities.
112 * Revision 1.58 1996/05/14 06:16:38 adam
113 * Compact use/set bytes used in search service.
115 * Revision 1.57 1996/05/13 14:23:04 adam
116 * Work on compaction of set/use bytes in dictionary.
118 * Revision 1.56 1996/05/09 09:54:42 adam
119 * Server supports maps from one logical attributes to a list of physical
121 * The extraction process doesn't make space consuming 'any' keys.
123 * Revision 1.55 1996/05/09 07:28:55 quinn
124 * Work towards phrases and multiple registers
126 * Revision 1.54 1996/05/01 13:46:35 adam
127 * First work on multiple records in one file.
128 * New option, -offset, to the "unread" command in the filter module.
130 * Revision 1.53 1996/04/26 12:09:43 adam
131 * Added a few comments.
133 * Revision 1.52 1996/04/25 13:27:57 adam
134 * Function recordExtract modified so that files with no keys (possibly empty)
137 * Revision 1.51 1996/03/19 11:08:42 adam
138 * Bug fix: Log preamble wasn't always turned off after recordExtract.
140 * Revision 1.50 1996/02/12 18:45:36 adam
141 * New fileVerboseFlag in record group control.
143 * Revision 1.49 1996/02/05 12:29:57 adam
144 * Logging reduced a bit.
145 * The remaining running time is estimated during register merge.
147 * Revision 1.48 1996/02/01 20:53:26 adam
148 * The temporary per-record keys are compacted a little, and duplication
149 * of the per-records keys are avoided when they are saved in the record
150 * information buffer.
152 * Revision 1.47 1996/01/17 14:57:48 adam
153 * Prototype changed for reader functions in extract/retrieve. File
154 * is identified by 'void *' instead of 'int.
156 * Revision 1.46 1995/12/15 14:57:16 adam
159 * Revision 1.45 1995/12/15 12:37:41 adam
160 * In addRecordKeyAny: Writes key only when attrSet != -1.
162 * Revision 1.44 1995/12/12 16:00:54 adam
163 * System call sync(2) used after update/commit.
164 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
167 * Revision 1.43 1995/12/11 09:12:46 adam
168 * The rec_get function returns NULL if record doesn't exist - will
169 * happen in the server if the result set records have been deleted since
170 * the creation of the set (i.e. the search).
171 * The server saves a result temporarily if it is 'volatile', i.e. the
172 * set is register dependent.
174 * Revision 1.42 1995/12/07 17:38:46 adam
175 * Work locking mechanisms for concurrent updates/commit.
177 * Revision 1.41 1995/12/06 16:06:42 adam
178 * Better diagnostics. Work on 'real' dictionary deletion.
180 * Revision 1.40 1995/12/05 16:57:40 adam
181 * More work on regular patterns.
183 * Revision 1.39 1995/12/05 13:20:18 adam
184 * Bug fix: file_read sometimes returned early EOF.
186 * Revision 1.38 1995/12/04 17:59:21 adam
187 * More work on regular expression conversion.
189 * Revision 1.37 1995/12/04 14:22:27 adam
190 * Extra arg to recType_byName.
191 * Started work on new regular expression parsed input to
192 * structured records.
194 * Revision 1.36 1995/11/30 08:34:29 adam
195 * Started work on commit facility.
196 * Changed a few malloc/free to xmalloc/xfree.
198 * Revision 1.35 1995/11/28 14:26:21 adam
199 * Bug fix: recordId with constant wasn't right.
200 * Bug fix: recordId dictionary entry wasn't deleted when needed.
202 * Revision 1.34 1995/11/28 09:09:38 adam
203 * Zebra config renamed.
204 * Use setting 'recordId' to identify record now.
205 * Bug fix in recindex.c: rec_release_blocks was invokeded even
206 * though the blocks were already released.
207 * File traversal properly deletes records when needed.
209 * Revision 1.33 1995/11/27 09:56:20 adam
210 * Record info elements better enumerated. Internal store of records.
212 * Revision 1.32 1995/11/25 10:24:05 adam
213 * More record fields - they are enumerated now.
214 * New options: flagStoreData flagStoreKey.
216 * Revision 1.31 1995/11/24 11:31:35 adam
217 * Commands add & del read filenames from stdin if source directory is
219 * Match criteria supports 'constant' strings.
221 * Revision 1.30 1995/11/22 17:19:16 adam
222 * Record management uses the bfile system.
224 * Revision 1.29 1995/11/21 15:01:14 adam
225 * New general match criteria implemented.
226 * New feature: document groups.
228 * Revision 1.28 1995/11/21 09:20:30 adam
229 * Yet more work on record match.
231 * Revision 1.27 1995/11/20 16:59:45 adam
232 * New update method: the 'old' keys are saved for each records.
234 * Revision 1.26 1995/11/20 11:56:24 adam
235 * Work on new traversal.
237 * Revision 1.25 1995/11/16 15:34:54 adam
238 * Uses new record management system in both indexer and server.
240 * Revision 1.24 1995/11/15 19:13:08 adam
241 * Work on record management.
243 * Revision 1.23 1995/10/27 14:00:10 adam
244 * Implemented detection of database availability.
246 * Revision 1.22 1995/10/17 18:02:07 adam
247 * New feature: databases. Implemented as prefix to words in dictionary.
249 * Revision 1.21 1995/10/10 12:24:38 adam
250 * Temporary sort files are compressed.
252 * Revision 1.20 1995/10/06 13:52:05 adam
253 * Bug fixes. Handler may abort further scanning.
255 * Revision 1.19 1995/10/04 12:55:16 adam
256 * Bug fix in ranked search. Use=Any keys inserted.
258 * Revision 1.18 1995/10/04 09:37:08 quinn
261 * Revision 1.17 1995/10/03 14:28:57 adam
262 * Buffered read in extract works.
264 * Revision 1.16 1995/10/03 14:28:45 adam
265 * Work on more effecient read handler in extract.
267 * Revision 1.15 1995/10/02 15:42:53 adam
268 * Extract uses file descriptors instead of FILE pointers.
270 * Revision 1.14 1995/10/02 15:29:13 adam
271 * More logging in file_extract.
273 * Revision 1.13 1995/09/29 14:01:39 adam
276 * Revision 1.12 1995/09/28 14:22:56 adam
277 * Sort uses smaller temporary files.
279 * Revision 1.11 1995/09/28 12:10:31 adam
280 * Bug fixes. Field prefix used in queries.
282 * Revision 1.10 1995/09/28 09:19:41 adam
283 * xfree/xmalloc used everywhere.
284 * Extract/retrieve method seems to work for text records.
286 * Revision 1.9 1995/09/27 12:22:28 adam
287 * More work on extract in record control.
288 * Field name is not in isam keys but in prefix in dictionary words.
290 * Revision 1.8 1995/09/14 07:48:22 adam
291 * Record control management.
293 * Revision 1.7 1995/09/11 13:09:32 adam
294 * More work on relevance feedback.
296 * Revision 1.6 1995/09/08 14:52:27 adam
297 * Minor changes. Dictionary is lower case now.
299 * Revision 1.5 1995/09/06 16:11:16 adam
300 * Option: only one word key per file.
302 * Revision 1.4 1995/09/05 15:28:39 adam
303 * More work on search engine.
305 * Revision 1.3 1995/09/04 12:33:41 adam
306 * Various cleanup. YAZ util used instead.
308 * Revision 1.2 1995/09/04 09:10:34 adam
309 * More work on index add/del/update.
310 * Merge sort implemented.
311 * Initial work on z39 server.
313 * Revision 1.1 1995/09/01 14:06:35 adam
314 * Split of work into more files.
333 static Dict matchDict;
335 static Records records = NULL;
336 static SortIdx sortIdx = NULL;
338 static char **key_buf;
339 static size_t ptr_top;
341 static size_t key_buf_used;
342 static int key_file_no;
344 static int records_inserted = 0;
345 static int records_updated = 0;
346 static int records_deleted = 0;
347 static int records_processed = 0;
349 static ZebraExplainInfo zti = NULL;
351 static void logRecord (int showFlag)
355 if (showFlag || !(records_processed % 1000))
357 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
358 records_processed, records_inserted, records_updated,
363 static int explain_extract (void *handle, Record drec, data1_node *n);
365 int key_open (struct recordGroup *rGroup, int mem)
367 BFiles bfs = rGroup->bfs;
368 int rw = rGroup->flagRw;
369 data1_handle dh = rGroup->dh;
371 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
374 key_buf = xmalloc (mem);
375 ptr_top = mem/sizeof(char*);
381 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
383 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
387 records = rec_open (bfs, rw);
390 dict_close (matchDict);
393 zti = zebraExplain_open (records, dh, common_resource,
394 rw, rGroup, explain_extract);
397 rec_close (&records);
398 dict_close (matchDict);
401 sortIdx = sortIdx_open (bfs, 1);
411 void encode_key_init (struct encode_info *i)
417 char *encode_key_int (int d, char *bp)
426 else if (d <= 4194303)
428 *bp++ = 128 + (d>>16);
429 *bp++ = (d>>8) & 255;
434 *bp++ = 192 + (d>>24);
435 *bp++ = (d>>16) & 255;
436 *bp++ = (d>>8) & 255;
442 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
447 while ((*bp++ = *k++))
449 memcpy (&key, k+1, sizeof(struct it_key));
450 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
451 if (i->sysno != key.sysno)
453 i->sysno = key.sysno;
456 bp = encode_key_int (key.seqno - i->seqno, bp);
457 i->seqno = key.seqno;
458 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
460 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
468 static int key_y_len;
470 static int key_y_compare (const void *p1, const void *p2)
474 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
475 *(char**) p2 + key_y_len + 1)))
477 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
480 static int key_x_compare (const void *p1, const void *p2)
482 return strcmp (*(char**) p1, *(char**) p2);
486 void key_flush (void)
491 struct encode_info encode_info;
500 logf (LOG_LOG, "sorting section %d", key_file_no);
502 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
503 getFnameTmp (out_fname, key_file_no);
505 if (!(outf = fopen (out_fname, "wb")))
507 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
510 logf (LOG_LOG, "writing section %d", key_file_no);
511 prevcp = cp = key_buf[ptr_top-ptr_i];
513 encode_key_init (&encode_info);
514 encode_key_write (cp, &encode_info, outf);
517 cp = key_buf[ptr_top-ptr_i];
518 if (strcmp (cp, prevcp))
520 encode_key_init (&encode_info);
521 encode_key_write (cp, &encode_info, outf);
525 encode_key_write (cp + strlen(cp), &encode_info, outf);
528 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
529 getFnameTmp (out_fname, key_file_no);
531 if (!(outf = fopen (out_fname, "wb")))
533 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
536 logf (LOG_LOG, "writing section %d", key_file_no);
538 prevcp = key_buf[ptr_top-i];
540 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
542 key_y_len = strlen(prevcp)+1;
544 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
545 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
547 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
548 sizeof(char*), key_y_compare);
549 cp = key_buf[ptr_top-ptr_i];
551 encode_key_init (&encode_info);
552 encode_key_write (cp, &encode_info, outf);
555 cp = key_buf[ptr_top-ptr_i];
556 encode_key_write (cp+key_y_len, &encode_info, outf);
560 prevcp = key_buf[ptr_top-ptr_i];
565 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
568 logf (LOG_LOG, "finished section %d", key_file_no);
573 int key_close (struct recordGroup *rGroup)
575 int rw = rGroup->flagRw;
577 zebraExplain_runNumberIncrement (zti, 1);
578 zebraExplain_close (zti, rw, 0);
581 rec_close (&records);
582 dict_close (matchDict);
583 sortIdx_close (sortIdx);
589 static void wordInit (struct recExtractCtrl *p, RecWord *w)
591 w->zebra_maps = p->zebra_maps;
592 w->seqnos = p->seqno;
593 w->attrSet = VAL_BIB1;
598 static struct sortKey {
603 struct sortKey *next;
606 static struct recKeys {
615 static void addIndexString (RecWord *p, const char *string, int length)
618 unsigned char attrSet;
619 unsigned short attrUse;
622 int *pseqno = &p->seqnos[p->reg_type];
624 if (reckeys.buf_used+1024 > reckeys.buf_max)
628 b = xmalloc (reckeys.buf_max += 128000);
629 if (reckeys.buf_used > 0)
630 memcpy (b, reckeys.buf, reckeys.buf_used);
634 dst = reckeys.buf + reckeys.buf_used;
636 attrSet = p->attrSet;
637 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
640 reckeys.prevAttrSet = attrSet;
641 attrUse = p->attrUse;
642 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
645 reckeys.prevAttrUse = attrUse;
647 diff = 1 + *pseqno - reckeys.prevSeqNo;
648 if (diff >= 1 && diff <= 15)
653 reckeys.prevSeqNo = *pseqno;
659 memcpy (dst, &attrSet, sizeof(attrSet));
660 dst += sizeof(attrSet);
664 memcpy (dst, &attrUse, sizeof(attrUse));
665 dst += sizeof(attrUse);
667 *dst++ = p->reg_type;
668 memcpy (dst, string, length);
674 memcpy (dst, pseqno, sizeof(*pseqno));
675 dst += sizeof(*pseqno);
677 reckeys.buf_used = dst - reckeys.buf;
681 static void addSortString (RecWord *p, const char *string, int length)
685 for (sk = sortKeys; sk; sk = sk->next)
686 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
689 sk = xmalloc (sizeof(*sk));
693 sk->string = xmalloc (length);
695 memcpy (sk->string, string, length);
697 sk->attrSet = p->attrSet;
698 sk->attrUse = p->attrUse;
701 static void addString (RecWord *p, const char *string, int length)
704 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
705 addSortString (p, string, length);
707 addIndexString (p, string, length);
710 static void addIncompleteField (RecWord *p)
712 const char *b = p->string;
713 int remain = p->length;
714 const char **map = 0;
717 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
721 char buf[IT_MAX_WORD+1];
725 while (map && *map && **map == *CHR_SPACE)
727 remain = p->length - (b - p->string);
729 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
736 while (map && *map && **map != *CHR_SPACE)
738 const char *cp = *map;
740 while (i < IT_MAX_WORD && *cp)
742 remain = p->length - (b - p->string);
744 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
750 addString (p, buf, i);
754 static void addCompleteField (RecWord *p)
756 const char *b = p->string;
757 char buf[IT_MAX_WORD+1];
758 const char **map = 0;
759 int i = 0, remain = p->length;
762 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
764 while (remain > 0 && i < IT_MAX_WORD)
766 while (map && *map && **map == *CHR_SPACE)
768 remain = p->length - (b - p->string);
770 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
777 if (i && i < IT_MAX_WORD)
778 buf[i++] = *CHR_SPACE;
779 while (map && *map && **map != *CHR_SPACE)
781 const char *cp = *map;
783 if (i >= IT_MAX_WORD)
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,
797 addString (p, buf, i);
800 static void addRecordKey (RecWord *p)
802 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
803 addCompleteField (p);
805 addIncompleteField(p);
808 static void flushSortKeys (SYSNO sysno, int cmd)
810 struct sortKey *sk = sortKeys;
812 sortIdx_sysno (sortIdx, sysno);
815 struct sortKey *sk_next = sk->next;
816 sortIdx_type (sortIdx, sk->attrUse);
817 sortIdx_add (sortIdx, sk->string, sk->length);
825 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
827 unsigned char attrSet = (unsigned char) -1;
828 unsigned short attrUse = (unsigned short) -1;
832 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
833 while (off < reckeys->buf_used)
835 const char *src = reckeys->buf + off;
843 memcpy (&attrSet, src, sizeof(attrSet));
844 src += sizeof(attrSet);
848 memcpy (&attrUse, src, sizeof(attrUse));
849 src += sizeof(attrUse);
851 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
854 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
856 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
858 ch = zebraExplain_addSU (zti, attrSet, attrUse);
860 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
863 ((char*)key_buf) [key_buf_used++] = *src++;
865 ((char*)key_buf) [key_buf_used++] = '\0';
866 ((char*) key_buf)[key_buf_used++] = cmd;
869 seqno += ((lead>>2) & 15)-1;
872 memcpy (&seqno, src, sizeof(seqno));
873 src += sizeof(seqno);
877 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
878 key_buf_used += sizeof(key);
879 off = src - reckeys->buf;
881 assert (off == reckeys->buf_used);
884 static const char **searchRecordKey (struct recKeys *reckeys,
885 int attrSetS, int attrUseS)
887 static const char *ws[32];
895 for (i = 0; i<32; i++)
898 while (off < reckeys->buf_used)
901 const char *src = reckeys->buf + off;
909 memcpy (&attrSet, src, sizeof(attrSet));
910 src += sizeof(attrSet);
914 memcpy (&attrUse, src, sizeof(attrUse));
915 src += sizeof(attrUse);
921 seqno += ((lead>>2) & 15)-1;
924 memcpy (&seqno, src, sizeof(seqno));
925 src += sizeof(seqno);
927 if (attrUseS == attrUse && attrSetS == attrSet)
934 woff = seqno - startSeq;
935 if (woff >= 0 && woff < 31)
939 off = src - reckeys->buf;
941 assert (off == reckeys->buf_used);
945 struct file_read_info {
953 static struct file_read_info *file_read_start (int fd)
955 struct file_read_info *fi = xmalloc (sizeof(*fi));
959 fi->file_moffset = 0;
963 static void file_read_stop (struct file_read_info *fi)
968 static off_t file_seek (void *handle, off_t offset)
970 struct file_read_info *p = handle;
971 p->file_offset = offset;
972 return lseek (p->fd, offset, SEEK_SET);
975 static off_t file_tell (void *handle)
977 struct file_read_info *p = handle;
978 return p->file_offset;
981 static int file_read (void *handle, char *buf, size_t count)
983 struct file_read_info *p = handle;
986 r = read (fd, buf, count);
990 if (p->file_offset > p->file_max)
991 p->file_max = p->file_offset;
996 static void file_begin (void *handle)
998 struct file_read_info *p = handle;
1000 p->file_offset = p->file_moffset;
1001 if (p->file_moffset)
1002 lseek (p->fd, p->file_moffset, SEEK_SET);
1006 static void file_end (void *handle, off_t offset)
1008 struct file_read_info *p = handle;
1010 assert (p->file_more == 0);
1012 p->file_moffset = offset;
1015 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1016 const char *fname, const char *spec)
1018 static char dstBuf[2048];
1020 const char *s = spec;
1021 static const char **w;
1025 while (*s == ' ' || *s == '\t')
1031 char attset_str[64], attname_str[64];
1032 data1_attset *attset;
1035 int attSet = 1, attUse = 1;
1039 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1041 attset_str[i++] = *s;
1042 attset_str[i] = '\0';
1047 for (i = 0; *s && *s != ')'; s++)
1049 attname_str[i++] = *s;
1050 attname_str[i] = '\0';
1053 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1056 attSet = attset->reference;
1057 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1059 attUse = att->value;
1061 attUse = atoi (attname_str);
1063 w = searchRecordKey (reckeys, attSet, attUse);
1068 for (i = 0; i<32; i++)
1073 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1074 spec, rGroup->groupName ? rGroup->groupName : "none");
1079 for (i = 0; i<32; i++)
1080 if (matchFlag[i] && w[i])
1088 dst += strlen(w[i]);
1092 logf (LOG_WARN, "Record didn't contain match"
1093 " fields in (%s,%s)", attset_str, attname_str);
1101 const char *spec_src = NULL;
1102 const char *s1 = ++s;
1103 while (*s1 && *s1 != ' ' && *s1 != '\t')
1109 memcpy (special, s, spec_len);
1110 special[spec_len] = '\0';
1113 if (!strcmp (special, "group"))
1114 spec_src = rGroup->groupName;
1115 else if (!strcmp (special, "database"))
1116 spec_src = rGroup->databaseName;
1117 else if (!strcmp (special, "filename"))
1119 else if (!strcmp (special, "type"))
1120 spec_src = rGroup->recordType;
1125 strcpy (dst, spec_src);
1126 dst += strlen (spec_src);
1129 else if (*s == '\"' || *s == '\'')
1131 int stopMarker = *s++;
1135 while (*s && *s != stopMarker)
1138 tmpString[i++] = *s++;
1142 tmpString[i] = '\0';
1143 strcpy (dst, tmpString);
1144 dst += strlen (tmpString);
1148 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1149 spec, rGroup->groupName ? rGroup->groupName : "none");
1156 logf (LOG_WARN, "No match criteria for record %s in group %s",
1157 fname, rGroup->groupName ? rGroup->groupName : "none");
1164 struct recordLogInfo {
1167 struct recordGroup *rGroup;
1170 static void recordLogPreamble (int level, const char *msg, void *info)
1172 struct recordLogInfo *p = info;
1173 FILE *outf = log_file ();
1175 if (level & LOG_LOG)
1177 fprintf (outf, "File %s, offset %d, type %s\n",
1178 p->rGroup->recordType, p->recordOffset, p->fname);
1179 log_event_start (NULL, NULL);
1182 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1184 zebraExplain_addSchema (zti, oid);
1187 static int recordExtract (SYSNO *sysno, const char *fname,
1188 struct recordGroup *rGroup, int deleteFlag,
1189 struct file_read_info *fi,
1190 RecType recType, char *subType)
1192 RecordAttr *recordAttr;
1197 struct recordLogInfo logInfo;
1198 off_t recordOffset = 0;
1202 struct recExtractCtrl extractCtrl;
1204 /* we are going to read from a file, so prepare the extraction */
1207 reckeys.buf_used = 0;
1208 reckeys.prevAttrUse = -1;
1209 reckeys.prevAttrSet = -1;
1210 reckeys.prevSeqNo = 0;
1212 recordOffset = fi->file_moffset;
1213 extractCtrl.offset = fi->file_moffset;
1214 extractCtrl.readf = file_read;
1215 extractCtrl.seekf = file_seek;
1216 extractCtrl.tellf = file_tell;
1217 extractCtrl.endf = file_end;
1218 extractCtrl.fh = fi;
1219 extractCtrl.subType = subType;
1220 extractCtrl.init = wordInit;
1221 extractCtrl.addWord = addRecordKey;
1222 extractCtrl.addSchema = addSchema;
1223 extractCtrl.dh = rGroup->dh;
1224 for (i = 0; i<256; i++)
1225 extractCtrl.seqno[i] = 0;
1226 extractCtrl.zebra_maps = rGroup->zebra_maps;
1227 extractCtrl.flagShowRecords = !rGroup->flagRw;
1229 if (!rGroup->flagRw)
1230 printf ("File: %s %ld\n", fname, (long) recordOffset);
1232 logInfo.fname = fname;
1233 logInfo.recordOffset = recordOffset;
1234 logInfo.rGroup = rGroup;
1235 log_event_start (recordLogPreamble, &logInfo);
1237 r = (*recType->extract)(&extractCtrl);
1239 log_event_start (NULL, NULL);
1243 /* error occured during extraction ... */
1244 if (rGroup->flagRw &&
1245 records_processed < rGroup->fileVerboseLimit)
1247 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1248 fname, (long) recordOffset, r);
1252 if (reckeys.buf_used == 0)
1254 /* the extraction process returned no information - the record
1255 is probably empty - unless flagShowRecords is in use */
1256 if (!rGroup->flagRw)
1258 logf (LOG_WARN, "No keys generated for file %s", fname);
1259 logf (LOG_WARN, " The file is probably empty");
1264 /* perform match if sysno not known and if match criteria is specified */
1271 if (rGroup->recordId && *rGroup->recordId)
1275 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1279 rinfo = dict_lookup (matchDict, matchStr);
1281 memcpy (sysno, rinfo+1, sizeof(*sysno));
1285 logf (LOG_WARN, "Bad match criteria");
1296 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1297 fname, (long) recordOffset);
1298 logf (LOG_WARN, "cannot delete record above (seems new)");
1301 if (records_processed < rGroup->fileVerboseLimit)
1302 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1303 fname, (long) recordOffset);
1304 rec = rec_new (records);
1306 *sysno = rec->sysno;
1308 recordAttr = rec_init_attr (zti, rec);
1312 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1314 flushRecordKeys (*sysno, 1, &reckeys);
1315 flushSortKeys (*sysno, 1);
1321 /* record already exists */
1322 struct recKeys delkeys;
1324 rec = rec_get (records, *sysno);
1327 recordAttr = rec_init_attr (zti, rec);
1329 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1331 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1332 fname, (long) recordOffset);
1337 delkeys.buf_used = rec->size[recInfo_delKeys];
1338 delkeys.buf = rec->info[recInfo_delKeys];
1339 flushSortKeys (*sysno, 0);
1340 flushRecordKeys (*sysno, 0, &delkeys);
1343 /* record going to be deleted */
1344 if (!delkeys.buf_used)
1346 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1347 fname, (long) recordOffset);
1348 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1352 if (records_processed < rGroup->fileVerboseLimit)
1353 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1354 fname, (long) recordOffset);
1357 dict_delete (matchDict, matchStr);
1358 rec_del (records, &rec);
1366 /* record going to be updated */
1367 if (!delkeys.buf_used)
1369 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1370 fname, (long) recordOffset);
1371 logf (LOG_WARN, "cannot update file above, storeKeys false");
1375 if (records_processed < rGroup->fileVerboseLimit)
1376 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1377 fname, (long) recordOffset);
1378 flushRecordKeys (*sysno, 1, &reckeys);
1383 /* update file type */
1384 xfree (rec->info[recInfo_fileType]);
1385 rec->info[recInfo_fileType] =
1386 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1388 /* update filename */
1389 xfree (rec->info[recInfo_filename]);
1390 rec->info[recInfo_filename] =
1391 rec_strdup (fname, &rec->size[recInfo_filename]);
1393 /* update delete keys */
1394 xfree (rec->info[recInfo_delKeys]);
1395 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1398 rec->size[recInfo_delKeys] = reckeys.buf_used;
1399 rec->info[recInfo_delKeys] = reckeys.buf;
1401 reckeys.buf_max = 0;
1403 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1404 rec->size[recInfo_delKeys] = reckeys.buf_used;
1405 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1406 rec->size[recInfo_delKeys]);
1411 rec->info[recInfo_delKeys] = NULL;
1412 rec->size[recInfo_delKeys] = 0;
1415 /* save file size of original record */
1416 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1417 recordAttr->recordSize = fi->file_moffset - recordOffset;
1418 if (!recordAttr->recordSize)
1419 recordAttr->recordSize = fi->file_max - recordOffset;
1420 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1422 /* set run-number for this record */
1423 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1425 /* update store data */
1426 xfree (rec->info[recInfo_storeData]);
1427 if (rGroup->flagStoreData == 1)
1429 rec->size[recInfo_storeData] = recordAttr->recordSize;
1430 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1431 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1433 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1434 (long) recordOffset, fname);
1437 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1438 < recordAttr->recordSize)
1440 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1441 recordAttr->recordSize, fname);
1447 rec->info[recInfo_storeData] = NULL;
1448 rec->size[recInfo_storeData] = 0;
1450 /* update database name */
1451 xfree (rec->info[recInfo_databaseName]);
1452 rec->info[recInfo_databaseName] =
1453 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1456 recordAttr->recordOffset = recordOffset;
1458 /* commit this record */
1459 rec_put (records, &rec);
1464 int fileExtract (SYSNO *sysno, const char *fname,
1465 const struct recordGroup *rGroupP, int deleteFlag)
1473 struct recordGroup rGroupM;
1474 struct recordGroup *rGroup = &rGroupM;
1475 struct file_read_info *fi;
1477 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1479 if (!rGroup->groupName || !*rGroup->groupName)
1482 sprintf (gprefix, "%s.", rGroup->groupName);
1484 logf (LOG_DEBUG, "fileExtract %s", fname);
1486 /* determine file extension */
1488 for (i = strlen(fname); --i >= 0; )
1489 if (fname[i] == '/')
1491 else if (fname[i] == '.')
1493 strcpy (ext, fname+i+1);
1496 /* determine file type - depending on extension */
1497 if (!rGroup->recordType)
1499 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1500 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1502 sprintf (ext_res, "%srecordType", gprefix);
1503 rGroup->recordType = res_get (common_resource, ext_res);
1506 if (!rGroup->recordType)
1508 if (records_processed < rGroup->fileVerboseLimit)
1509 logf (LOG_LOG, "? %s", fname);
1512 if (!*rGroup->recordType)
1514 if (!(recType = recType_byName (rGroup->recordType, subType)))
1516 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1520 /* determine match criteria */
1521 if (!rGroup->recordId)
1523 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1524 rGroup->recordId = res_get (common_resource, ext_res);
1527 /* determine database name */
1528 if (!rGroup->databaseName)
1530 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1531 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1533 sprintf (ext_res, "%sdatabase", gprefix);
1534 rGroup->databaseName = res_get (common_resource, ext_res);
1537 if (!rGroup->databaseName)
1538 rGroup->databaseName = "Default";
1540 /* determine if explain database */
1542 sprintf (ext_res, "%sexplainDatabase", gprefix);
1543 rGroup->explainDatabase =
1544 atoi (res_get_def (common_resource, ext_res, "0"));
1546 /* announce database */
1547 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1549 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1550 rGroup->explainDatabase))
1554 if (rGroup->flagStoreData == -1)
1557 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1558 if (!(sval = res_get (common_resource, ext_res)))
1560 sprintf (ext_res, "%sstoreData", gprefix);
1561 sval = res_get (common_resource, ext_res);
1564 rGroup->flagStoreData = atoi (sval);
1566 if (rGroup->flagStoreData == -1)
1567 rGroup->flagStoreData = 0;
1569 if (rGroup->flagStoreKeys == -1)
1573 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1574 if (!(sval = res_get (common_resource, ext_res)))
1576 sprintf (ext_res, "%sstoreKeys", gprefix);
1577 sval = res_get (common_resource, ext_res);
1580 rGroup->flagStoreKeys = atoi (sval);
1582 if (rGroup->flagStoreKeys == -1)
1583 rGroup->flagStoreKeys = 0;
1585 if (sysno && deleteFlag)
1589 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1591 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1595 fi = file_read_start (fd);
1599 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1601 } while (r && !sysno && fi->file_more);
1602 file_read_stop (fi);
1608 static int explain_extract (void *handle, Record rec, data1_node *n)
1610 struct recordGroup *rGroup = (struct recordGroup*) handle;
1611 struct recExtractCtrl extractCtrl;
1614 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1616 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1620 reckeys.buf_used = 0;
1621 reckeys.prevAttrUse = -1;
1622 reckeys.prevAttrSet = -1;
1623 reckeys.prevSeqNo = 0;
1625 extractCtrl.init = wordInit;
1626 extractCtrl.addWord = addRecordKey;
1627 extractCtrl.addSchema = addSchema;
1628 extractCtrl.dh = rGroup->dh;
1629 for (i = 0; i<256; i++)
1630 extractCtrl.seqno[i] = 0;
1631 extractCtrl.zebra_maps = rGroup->zebra_maps;
1632 extractCtrl.flagShowRecords = !rGroup->flagRw;
1634 grs_extract_tree(&extractCtrl, n);
1636 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1638 if (rec->size[recInfo_delKeys])
1640 struct recKeys delkeys;
1642 delkeys.buf_used = rec->size[recInfo_delKeys];
1643 delkeys.buf = rec->info[recInfo_delKeys];
1644 flushSortKeys (rec->sysno, 0);
1645 flushRecordKeys (rec->sysno, 0, &delkeys);
1647 flushRecordKeys (rec->sysno, 1, &reckeys);
1648 flushSortKeys (rec->sysno, 1);
1650 xfree (rec->info[recInfo_delKeys]);
1651 rec->size[recInfo_delKeys] = reckeys.buf_used;
1652 rec->info[recInfo_delKeys] = reckeys.buf;
1654 reckeys.buf_max = 0;