2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.94 1999-05-20 12:57:18 adam
8 * Implemented TCL filter. Updated recctrl system.
10 * Revision 1.93 1999/05/15 14:36:38 adam
11 * Updated dictionary. Implemented "compression" of dictionary.
13 * Revision 1.92 1999/03/09 16:27:49 adam
14 * More work on SDRKit integration.
16 * Revision 1.91 1999/02/12 13:29:22 adam
17 * Implemented position-flag for registers.
19 * Revision 1.90 1999/02/02 14:50:52 adam
20 * Updated WIN32 code specific sections. Changed header.
22 * Revision 1.89 1998/10/28 10:54:38 adam
25 * Revision 1.88 1998/10/16 08:14:29 adam
26 * Updated record control system.
28 * Revision 1.87 1998/10/15 13:10:33 adam
29 * Fixed bug in Zebra that caused it to stop indexing when empty
32 * Revision 1.86 1998/10/13 20:33:53 adam
33 * Fixed one log message and change use ordinal to be an unsigned char.
35 * Revision 1.85 1998/09/22 10:03:41 adam
36 * Changed result sets to be persistent in the sense that they can
37 * be re-searched if needed.
38 * Fixed memory leak in rsm_or.
40 * Revision 1.84 1998/06/11 15:42:22 adam
41 * Changed the way use attributes are specified in the recordId
44 * Revision 1.83 1998/06/08 14:43:10 adam
45 * Added suport for EXPLAIN Proxy servers - added settings databasePath
46 * and explainDatabase to facilitate this. Increased maximum number
47 * of databases and attributes in one register.
49 * Revision 1.82 1998/05/20 10:12:15 adam
50 * Implemented automatic EXPLAIN database maintenance.
51 * Modified Zebra to work with ASN.1 compiled version of YAZ.
53 * Revision 1.81 1998/03/11 11:19:04 adam
54 * Changed the way sequence numbers are generated.
56 * Revision 1.80 1998/03/05 08:45:11 adam
57 * New result set model and modular ranking system. Moved towards
58 * descent server API. System information stored as "SGML" records.
60 * Revision 1.79 1998/02/17 10:32:52 adam
61 * Fixed bug: binary files weren't opened with flag b on NT.
63 * Revision 1.78 1998/02/10 12:03:05 adam
66 * Revision 1.77 1998/01/12 15:04:08 adam
67 * The test option (-s) only uses read-lock (and not write lock).
69 * Revision 1.76 1997/10/27 14:33:04 adam
70 * Moved towards generic character mapping depending on "structure"
71 * field in abstract syntax file. Fixed a few memory leaks. Fixed
72 * bug with negative integers when doing searches with relational
75 * Revision 1.75 1997/09/17 12:19:12 adam
76 * Zebra version corresponds to YAZ version 1.4.
77 * Changed Zebra server so that it doesn't depend on global common_resource.
79 * Revision 1.74 1997/09/09 13:38:06 adam
80 * Partial port to WIN95/NT.
82 * Revision 1.73 1997/09/04 13:57:20 adam
83 * New file extract/retrieve method tellf (added).
84 * Added O_BINARY for open calls.
86 * Revision 1.72 1997/07/15 16:32:29 adam
87 * Bug fix: Match handler didn't terminate the resulting string!
89 * Revision 1.71 1997/07/15 16:28:41 adam
90 * Bug fix: storeData didn't work with files with multiple records.
91 * Bug fix: fixed memory management with records; not really well
94 * Revision 1.70 1997/07/01 13:00:42 adam
95 * Bug fix in routine searchRecordKey: uninitialized variables.
97 * Revision 1.69 1997/04/29 09:26:03 adam
98 * Bug fix: generic recordId handling didn't work for compressed internal
101 * Revision 1.68 1997/02/12 20:39:45 adam
102 * Implemented options -f <n> that limits the log to the first <n>
104 * Changed some log messages also.
106 * Revision 1.67 1996/11/15 15:02:14 adam
107 * Minor changes regarding logging.
109 * Revision 1.66 1996/11/14 09:52:21 adam
110 * Strings in record keys bound by IT_MAX_WORD.
112 * Revision 1.65 1996/11/14 08:57:56 adam
113 * Reduction of storeKeys area.
115 * Revision 1.64 1996/11/08 11:10:16 adam
116 * Buffers used during file match got bigger.
117 * Compressed ISAM support everywhere.
118 * Bug fixes regarding masking characters in queries.
119 * Redesigned Regexp-2 queries.
121 * Revision 1.63 1996/10/29 14:09:39 adam
122 * Use of cisam system - enabled if setting isamc is 1.
124 * Revision 1.62 1996/10/11 10:57:01 adam
125 * New module recctrl. Used to manage records (extract/retrieval).
126 * Several files have been moved to the recctrl sub directory.
128 * Revision 1.61 1996/06/06 12:08:37 quinn
129 * Added showRecord function
131 * Revision 1.60 1996/06/04 10:18:12 adam
132 * Search/scan uses character mapping module.
134 * Revision 1.59 1996/05/14 15:47:07 adam
135 * Cleanup of various buffer size entities.
137 * Revision 1.58 1996/05/14 06:16:38 adam
138 * Compact use/set bytes used in search service.
140 * Revision 1.57 1996/05/13 14:23:04 adam
141 * Work on compaction of set/use bytes in dictionary.
143 * Revision 1.56 1996/05/09 09:54:42 adam
144 * Server supports maps from one logical attributes to a list of physical
146 * The extraction process doesn't make space consuming 'any' keys.
148 * Revision 1.55 1996/05/09 07:28:55 quinn
149 * Work towards phrases and multiple registers
151 * Revision 1.54 1996/05/01 13:46:35 adam
152 * First work on multiple records in one file.
153 * New option, -offset, to the "unread" command in the filter module.
155 * Revision 1.53 1996/04/26 12:09:43 adam
156 * Added a few comments.
158 * Revision 1.52 1996/04/25 13:27:57 adam
159 * Function recordExtract modified so that files with no keys (possibly empty)
162 * Revision 1.51 1996/03/19 11:08:42 adam
163 * Bug fix: Log preamble wasn't always turned off after recordExtract.
165 * Revision 1.50 1996/02/12 18:45:36 adam
166 * New fileVerboseFlag in record group control.
168 * Revision 1.49 1996/02/05 12:29:57 adam
169 * Logging reduced a bit.
170 * The remaining running time is estimated during register merge.
172 * Revision 1.48 1996/02/01 20:53:26 adam
173 * The temporary per-record keys are compacted a little, and duplication
174 * of the per-records keys are avoided when they are saved in the record
175 * information buffer.
177 * Revision 1.47 1996/01/17 14:57:48 adam
178 * Prototype changed for reader functions in extract/retrieve. File
179 * is identified by 'void *' instead of 'int.
181 * Revision 1.46 1995/12/15 14:57:16 adam
184 * Revision 1.45 1995/12/15 12:37:41 adam
185 * In addRecordKeyAny: Writes key only when attrSet != -1.
187 * Revision 1.44 1995/12/12 16:00:54 adam
188 * System call sync(2) used after update/commit.
189 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
192 * Revision 1.43 1995/12/11 09:12:46 adam
193 * The rec_get function returns NULL if record doesn't exist - will
194 * happen in the server if the result set records have been deleted since
195 * the creation of the set (i.e. the search).
196 * The server saves a result temporarily if it is 'volatile', i.e. the
197 * set is register dependent.
199 * Revision 1.42 1995/12/07 17:38:46 adam
200 * Work locking mechanisms for concurrent updates/commit.
202 * Revision 1.41 1995/12/06 16:06:42 adam
203 * Better diagnostics. Work on 'real' dictionary deletion.
205 * Revision 1.40 1995/12/05 16:57:40 adam
206 * More work on regular patterns.
208 * Revision 1.39 1995/12/05 13:20:18 adam
209 * Bug fix: file_read sometimes returned early EOF.
211 * Revision 1.38 1995/12/04 17:59:21 adam
212 * More work on regular expression conversion.
214 * Revision 1.37 1995/12/04 14:22:27 adam
215 * Extra arg to recType_byName.
216 * Started work on new regular expression parsed input to
217 * structured records.
219 * Revision 1.36 1995/11/30 08:34:29 adam
220 * Started work on commit facility.
221 * Changed a few malloc/free to xmalloc/xfree.
223 * Revision 1.35 1995/11/28 14:26:21 adam
224 * Bug fix: recordId with constant wasn't right.
225 * Bug fix: recordId dictionary entry wasn't deleted when needed.
227 * Revision 1.34 1995/11/28 09:09:38 adam
228 * Zebra config renamed.
229 * Use setting 'recordId' to identify record now.
230 * Bug fix in recindex.c: rec_release_blocks was invokeded even
231 * though the blocks were already released.
232 * File traversal properly deletes records when needed.
234 * Revision 1.33 1995/11/27 09:56:20 adam
235 * Record info elements better enumerated. Internal store of records.
237 * Revision 1.32 1995/11/25 10:24:05 adam
238 * More record fields - they are enumerated now.
239 * New options: flagStoreData flagStoreKey.
241 * Revision 1.31 1995/11/24 11:31:35 adam
242 * Commands add & del read filenames from stdin if source directory is
244 * Match criteria supports 'constant' strings.
246 * Revision 1.30 1995/11/22 17:19:16 adam
247 * Record management uses the bfile system.
249 * Revision 1.29 1995/11/21 15:01:14 adam
250 * New general match criteria implemented.
251 * New feature: document groups.
253 * Revision 1.28 1995/11/21 09:20:30 adam
254 * Yet more work on record match.
256 * Revision 1.27 1995/11/20 16:59:45 adam
257 * New update method: the 'old' keys are saved for each records.
259 * Revision 1.26 1995/11/20 11:56:24 adam
260 * Work on new traversal.
262 * Revision 1.25 1995/11/16 15:34:54 adam
263 * Uses new record management system in both indexer and server.
265 * Revision 1.24 1995/11/15 19:13:08 adam
266 * Work on record management.
268 * Revision 1.23 1995/10/27 14:00:10 adam
269 * Implemented detection of database availability.
271 * Revision 1.22 1995/10/17 18:02:07 adam
272 * New feature: databases. Implemented as prefix to words in dictionary.
274 * Revision 1.21 1995/10/10 12:24:38 adam
275 * Temporary sort files are compressed.
277 * Revision 1.20 1995/10/06 13:52:05 adam
278 * Bug fixes. Handler may abort further scanning.
280 * Revision 1.19 1995/10/04 12:55:16 adam
281 * Bug fix in ranked search. Use=Any keys inserted.
283 * Revision 1.18 1995/10/04 09:37:08 quinn
286 * Revision 1.17 1995/10/03 14:28:57 adam
287 * Buffered read in extract works.
289 * Revision 1.16 1995/10/03 14:28:45 adam
290 * Work on more effecient read handler in extract.
292 * Revision 1.15 1995/10/02 15:42:53 adam
293 * Extract uses file descriptors instead of FILE pointers.
295 * Revision 1.14 1995/10/02 15:29:13 adam
296 * More logging in file_extract.
298 * Revision 1.13 1995/09/29 14:01:39 adam
301 * Revision 1.12 1995/09/28 14:22:56 adam
302 * Sort uses smaller temporary files.
304 * Revision 1.11 1995/09/28 12:10:31 adam
305 * Bug fixes. Field prefix used in queries.
307 * Revision 1.10 1995/09/28 09:19:41 adam
308 * xfree/xmalloc used everywhere.
309 * Extract/retrieve method seems to work for text records.
311 * Revision 1.9 1995/09/27 12:22:28 adam
312 * More work on extract in record control.
313 * Field name is not in isam keys but in prefix in dictionary words.
315 * Revision 1.8 1995/09/14 07:48:22 adam
316 * Record control management.
318 * Revision 1.7 1995/09/11 13:09:32 adam
319 * More work on relevance feedback.
321 * Revision 1.6 1995/09/08 14:52:27 adam
322 * Minor changes. Dictionary is lower case now.
324 * Revision 1.5 1995/09/06 16:11:16 adam
325 * Option: only one word key per file.
327 * Revision 1.4 1995/09/05 15:28:39 adam
328 * More work on search engine.
330 * Revision 1.3 1995/09/04 12:33:41 adam
331 * Various cleanup. YAZ util used instead.
333 * Revision 1.2 1995/09/04 09:10:34 adam
334 * More work on index add/del/update.
335 * Merge sort implemented.
336 * Initial work on z39 server.
338 * Revision 1.1 1995/09/01 14:06:35 adam
339 * Split of work into more files.
363 #include "zebrasdr.h"
366 static Dict matchDict;
368 static Records records = NULL;
369 static SortIdx sortIdx = NULL;
371 static char **key_buf;
372 static size_t ptr_top;
374 static size_t key_buf_used;
375 static int key_file_no;
377 static int records_inserted = 0;
378 static int records_updated = 0;
379 static int records_deleted = 0;
380 static int records_processed = 0;
382 static ZebraExplainInfo zti = NULL;
384 static void logRecord (int showFlag)
388 if (showFlag || !(records_processed % 1000))
390 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
391 records_processed, records_inserted, records_updated,
396 static int explain_extract (void *handle, Record drec, data1_node *n);
398 int key_open (struct recordGroup *rGroup, int mem)
400 BFiles bfs = rGroup->bfs;
401 int rw = rGroup->flagRw;
402 data1_handle dh = rGroup->dh;
404 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
407 key_buf = xmalloc (mem);
408 ptr_top = mem/sizeof(char*);
414 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
416 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
420 records = rec_open (bfs, rw);
423 dict_close (matchDict);
426 zti = zebraExplain_open (records, dh, common_resource,
427 rw, rGroup, explain_extract);
430 rec_close (&records);
431 dict_close (matchDict);
434 sortIdx = sortIdx_open (bfs, 1);
445 void encode_key_init (struct encode_info *i)
452 char *encode_key_int (int d, char *bp)
461 else if (d <= 4194303)
463 *bp++ = 128 + (d>>16);
464 *bp++ = (d>>8) & 255;
469 *bp++ = 192 + (d>>24);
470 *bp++ = (d>>16) & 255;
471 *bp++ = (d>>8) & 255;
477 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
482 while ((*bp++ = *k++))
484 memcpy (&key, k+1, sizeof(struct it_key));
485 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
486 if (i->sysno != key.sysno)
488 i->sysno = key.sysno;
491 else if (!i->seqno && !key.seqno && i->cmd == *k)
493 bp = encode_key_int (key.seqno - i->seqno, bp);
494 i->seqno = key.seqno;
496 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
498 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
506 static int key_y_len;
508 static int key_y_compare (const void *p1, const void *p2)
512 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
513 *(char**) p2 + key_y_len + 1)))
515 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
518 static int key_x_compare (const void *p1, const void *p2)
520 return strcmp (*(char**) p1, *(char**) p2);
524 void key_flush (void)
529 struct encode_info encode_info;
538 logf (LOG_LOG, "sorting section %d", key_file_no);
540 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
541 getFnameTmp (out_fname, key_file_no);
543 if (!(outf = fopen (out_fname, "wb")))
545 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
548 logf (LOG_LOG, "writing section %d", key_file_no);
549 prevcp = 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 if (strcmp (cp, prevcp))
558 encode_key_init (&encode_info);
559 encode_key_write (cp, &encode_info, outf);
563 encode_key_write (cp + strlen(cp), &encode_info, outf);
566 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
567 getFnameTmp (out_fname, key_file_no);
569 if (!(outf = fopen (out_fname, "wb")))
571 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
574 logf (LOG_LOG, "writing section %d", key_file_no);
576 prevcp = key_buf[ptr_top-i];
578 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
580 key_y_len = strlen(prevcp)+1;
582 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
583 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
585 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
586 sizeof(char*), key_y_compare);
587 cp = key_buf[ptr_top-ptr_i];
589 encode_key_init (&encode_info);
590 encode_key_write (cp, &encode_info, outf);
593 cp = key_buf[ptr_top-ptr_i];
594 encode_key_write (cp+key_y_len, &encode_info, outf);
598 prevcp = key_buf[ptr_top-ptr_i];
603 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
606 logf (LOG_LOG, "finished section %d", key_file_no);
611 int key_close (struct recordGroup *rGroup)
613 int rw = rGroup->flagRw;
615 zebraExplain_runNumberIncrement (zti, 1);
616 zebraExplain_close (zti, rw, 0);
619 rec_close (&records);
620 dict_close (matchDict);
621 sortIdx_close (sortIdx);
627 static void wordInit (struct recExtractCtrl *p, RecWord *w)
629 w->zebra_maps = p->zebra_maps;
630 w->seqnos = p->seqno;
631 w->attrSet = VAL_BIB1;
636 static struct sortKey {
641 struct sortKey *next;
644 static struct recKeys {
653 static void addIndexString (RecWord *p, const char *string, int length)
656 unsigned char attrSet;
657 unsigned short attrUse;
660 int *pseqno = &p->seqnos[p->reg_type];
662 if (reckeys.buf_used+1024 > reckeys.buf_max)
666 b = xmalloc (reckeys.buf_max += 128000);
667 if (reckeys.buf_used > 0)
668 memcpy (b, reckeys.buf, reckeys.buf_used);
672 dst = reckeys.buf + reckeys.buf_used;
674 attrSet = p->attrSet;
675 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
678 reckeys.prevAttrSet = attrSet;
679 attrUse = p->attrUse;
680 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
683 reckeys.prevAttrUse = attrUse;
685 diff = 1 + *pseqno - reckeys.prevSeqNo;
686 if (diff >= 1 && diff <= 15)
691 reckeys.prevSeqNo = *pseqno;
697 memcpy (dst, &attrSet, sizeof(attrSet));
698 dst += sizeof(attrSet);
702 memcpy (dst, &attrUse, sizeof(attrUse));
703 dst += sizeof(attrUse);
705 *dst++ = p->reg_type;
706 memcpy (dst, string, length);
712 memcpy (dst, pseqno, sizeof(*pseqno));
713 dst += sizeof(*pseqno);
715 reckeys.buf_used = dst - reckeys.buf;
720 static void addSortString (RecWord *p, const char *string, int length)
724 for (sk = sortKeys; sk; sk = sk->next)
725 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
728 sk = xmalloc (sizeof(*sk));
732 sk->string = xmalloc (length);
734 memcpy (sk->string, string, length);
736 sk->attrSet = p->attrSet;
737 sk->attrUse = p->attrUse;
740 static void addString (RecWord *p, const char *string, int length)
743 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
744 addSortString (p, string, length);
746 addIndexString (p, string, length);
749 static void addIncompleteField (RecWord *p)
751 const char *b = p->string;
752 int remain = p->length;
753 const char **map = 0;
756 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
760 char buf[IT_MAX_WORD+1];
764 while (map && *map && **map == *CHR_SPACE)
766 remain = p->length - (b - p->string);
768 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
775 while (map && *map && **map != *CHR_SPACE)
777 const char *cp = *map;
779 while (i < IT_MAX_WORD && *cp)
781 remain = p->length - (b - p->string);
783 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
789 addString (p, buf, i);
793 static void addCompleteField (RecWord *p)
795 const char *b = p->string;
796 char buf[IT_MAX_WORD+1];
797 const char **map = 0;
798 int i = 0, remain = p->length;
801 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
803 while (remain > 0 && i < IT_MAX_WORD)
805 while (map && *map && **map == *CHR_SPACE)
807 remain = p->length - (b - p->string);
809 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
816 if (i && i < IT_MAX_WORD)
817 buf[i++] = *CHR_SPACE;
818 while (map && *map && **map != *CHR_SPACE)
820 const char *cp = *map;
822 if (i >= IT_MAX_WORD)
824 while (i < IT_MAX_WORD && *cp)
826 remain = p->length - (b - p->string);
828 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
836 addString (p, buf, i);
839 static void addRecordKey (RecWord *p)
841 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
842 addCompleteField (p);
844 addIncompleteField(p);
847 static void flushSortKeys (SYSNO sysno, int cmd)
849 struct sortKey *sk = sortKeys;
851 sortIdx_sysno (sortIdx, sysno);
854 struct sortKey *sk_next = sk->next;
855 sortIdx_type (sortIdx, sk->attrUse);
856 sortIdx_add (sortIdx, sk->string, sk->length);
864 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
866 unsigned char attrSet = (unsigned char) -1;
867 unsigned short attrUse = (unsigned short) -1;
871 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
872 while (off < reckeys->buf_used)
874 const char *src = reckeys->buf + off;
882 memcpy (&attrSet, src, sizeof(attrSet));
883 src += sizeof(attrSet);
887 memcpy (&attrUse, src, sizeof(attrUse));
888 src += sizeof(attrUse);
890 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
893 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
895 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
897 ch = zebraExplain_addSU (zti, attrSet, attrUse);
899 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
902 ((char*)key_buf) [key_buf_used++] = *src++;
904 ((char*)key_buf) [key_buf_used++] = '\0';
905 ((char*) key_buf)[key_buf_used++] = cmd;
908 seqno += ((lead>>2) & 15)-1;
911 memcpy (&seqno, src, sizeof(seqno));
912 src += sizeof(seqno);
916 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
917 key_buf_used += sizeof(key);
918 off = src - reckeys->buf;
920 assert (off == reckeys->buf_used);
923 static const char **searchRecordKey (struct recKeys *reckeys,
924 int attrSetS, int attrUseS)
926 static const char *ws[32];
934 for (i = 0; i<32; i++)
937 while (off < reckeys->buf_used)
940 const char *src = reckeys->buf + off;
948 memcpy (&attrSet, src, sizeof(attrSet));
949 src += sizeof(attrSet);
953 memcpy (&attrUse, src, sizeof(attrUse));
954 src += sizeof(attrUse);
960 seqno += ((lead>>2) & 15)-1;
963 memcpy (&seqno, src, sizeof(seqno));
964 src += sizeof(seqno);
966 if (attrUseS == attrUse && attrSetS == attrSet)
973 woff = seqno - startSeq;
974 if (woff >= 0 && woff < 31)
978 off = src - reckeys->buf;
980 assert (off == reckeys->buf_used);
984 struct file_read_info {
985 off_t file_max; /* maximum offset so far */
986 off_t file_offset; /* current offset */
987 off_t file_moffset; /* offset of rec/rec boundary */
994 static struct file_read_info *file_read_start (int fd)
996 struct file_read_info *fi = xmalloc (sizeof(*fi));
1000 fi->file_moffset = 0;
1006 static void file_read_stop (struct file_read_info *fi)
1011 static off_t file_seek (void *handle, off_t offset)
1013 struct file_read_info *p = handle;
1014 p->file_offset = offset;
1017 return lseek (p->fd, offset, SEEK_SET);
1020 static off_t file_tell (void *handle)
1022 struct file_read_info *p = handle;
1023 return p->file_offset;
1026 static int file_read (void *handle, char *buf, size_t count)
1028 struct file_read_info *p = handle;
1034 if (r > p->sdrmax - p->file_offset)
1035 r = p->sdrmax - p->file_offset;
1037 memcpy (buf, p->sdrbuf + p->file_offset, r);
1040 r = read (fd, buf, count);
1043 p->file_offset += r;
1044 if (p->file_offset > p->file_max)
1045 p->file_max = p->file_offset;
1050 static void file_begin (void *handle)
1052 struct file_read_info *p = handle;
1054 p->file_offset = p->file_moffset;
1055 if (!p->sdrbuf && p->file_moffset)
1056 lseek (p->fd, p->file_moffset, SEEK_SET);
1060 static void file_end (void *handle, off_t offset)
1062 struct file_read_info *p = handle;
1064 assert (p->file_more == 0);
1066 p->file_moffset = offset;
1069 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1070 const char *fname, const char *spec)
1072 static char dstBuf[2048];
1074 const char *s = spec;
1075 static const char **w;
1079 while (*s == ' ' || *s == '\t')
1085 char attset_str[64], attname_str[64];
1086 data1_attset *attset;
1089 int attSet = 1, attUse = 1;
1093 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1095 attset_str[i++] = *s;
1096 attset_str[i] = '\0';
1101 for (i = 0; *s && *s != ')'; s++)
1103 attname_str[i++] = *s;
1104 attname_str[i] = '\0';
1107 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1110 attSet = attset->reference;
1111 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1113 attUse = att->value;
1115 attUse = atoi (attname_str);
1117 w = searchRecordKey (reckeys, attSet, attUse);
1122 for (i = 0; i<32; i++)
1127 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1128 spec, rGroup->groupName ? rGroup->groupName : "none");
1133 for (i = 0; i<32; i++)
1134 if (matchFlag[i] && w[i])
1142 dst += strlen(w[i]);
1146 logf (LOG_WARN, "Record didn't contain match"
1147 " fields in (%s,%s)", attset_str, attname_str);
1155 const char *spec_src = NULL;
1156 const char *s1 = ++s;
1157 while (*s1 && *s1 != ' ' && *s1 != '\t')
1163 memcpy (special, s, spec_len);
1164 special[spec_len] = '\0';
1167 if (!strcmp (special, "group"))
1168 spec_src = rGroup->groupName;
1169 else if (!strcmp (special, "database"))
1170 spec_src = rGroup->databaseName;
1171 else if (!strcmp (special, "filename"))
1173 else if (!strcmp (special, "type"))
1174 spec_src = rGroup->recordType;
1179 strcpy (dst, spec_src);
1180 dst += strlen (spec_src);
1183 else if (*s == '\"' || *s == '\'')
1185 int stopMarker = *s++;
1189 while (*s && *s != stopMarker)
1192 tmpString[i++] = *s++;
1196 tmpString[i] = '\0';
1197 strcpy (dst, tmpString);
1198 dst += strlen (tmpString);
1202 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1203 spec, rGroup->groupName ? rGroup->groupName : "none");
1210 logf (LOG_WARN, "No match criteria for record %s in group %s",
1211 fname, rGroup->groupName ? rGroup->groupName : "none");
1218 struct recordLogInfo {
1221 struct recordGroup *rGroup;
1224 static void recordLogPreamble (int level, const char *msg, void *info)
1226 struct recordLogInfo *p = info;
1227 FILE *outf = log_file ();
1229 if (level & LOG_LOG)
1231 fprintf (outf, "File %s, offset %d, type %s\n",
1232 p->rGroup->recordType, p->recordOffset, p->fname);
1233 log_event_start (NULL, NULL);
1236 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1238 zebraExplain_addSchema (zti, oid);
1241 static int recordExtract (SYSNO *sysno, const char *fname,
1242 struct recordGroup *rGroup, int deleteFlag,
1243 struct file_read_info *fi,
1244 RecType recType, char *subType, void *clientData)
1246 RecordAttr *recordAttr;
1251 struct recordLogInfo logInfo;
1252 off_t recordOffset = 0;
1256 struct recExtractCtrl extractCtrl;
1258 /* we are going to read from a file, so prepare the extraction */
1261 reckeys.buf_used = 0;
1262 reckeys.prevAttrUse = -1;
1263 reckeys.prevAttrSet = -1;
1264 reckeys.prevSeqNo = 0;
1266 recordOffset = fi->file_moffset;
1267 extractCtrl.offset = fi->file_moffset;
1268 extractCtrl.readf = file_read;
1269 extractCtrl.seekf = file_seek;
1270 extractCtrl.tellf = file_tell;
1271 extractCtrl.endf = file_end;
1272 extractCtrl.fh = fi;
1273 extractCtrl.subType = subType;
1274 extractCtrl.init = wordInit;
1275 extractCtrl.addWord = addRecordKey;
1276 extractCtrl.addSchema = addSchema;
1277 extractCtrl.dh = rGroup->dh;
1278 for (i = 0; i<256; i++)
1280 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1281 extractCtrl.seqno[i] = 1;
1283 extractCtrl.seqno[i] = 0;
1285 extractCtrl.zebra_maps = rGroup->zebra_maps;
1286 extractCtrl.flagShowRecords = !rGroup->flagRw;
1288 if (!rGroup->flagRw)
1289 printf ("File: %s %ld\n", fname, (long) recordOffset);
1291 logInfo.fname = fname;
1292 logInfo.recordOffset = recordOffset;
1293 logInfo.rGroup = rGroup;
1294 log_event_start (recordLogPreamble, &logInfo);
1296 r = (*recType->extract)(clientData, &extractCtrl);
1298 log_event_start (NULL, NULL);
1302 /* error occured during extraction ... */
1303 if (rGroup->flagRw &&
1304 records_processed < rGroup->fileVerboseLimit)
1306 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1307 fname, (long) recordOffset, r);
1311 if (reckeys.buf_used == 0)
1313 /* the extraction process returned no information - the record
1314 is probably empty - unless flagShowRecords is in use */
1315 if (!rGroup->flagRw)
1317 logf (LOG_WARN, "No keys generated for file %s", fname);
1318 logf (LOG_WARN, " The file is probably empty");
1323 /* perform match if sysno not known and if match criteria is specified */
1330 if (rGroup->recordId && *rGroup->recordId)
1334 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1338 rinfo = dict_lookup (matchDict, matchStr);
1340 memcpy (sysno, rinfo+1, sizeof(*sysno));
1344 logf (LOG_WARN, "Bad match criteria");
1355 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1356 fname, (long) recordOffset);
1357 logf (LOG_WARN, "cannot delete record above (seems new)");
1360 if (records_processed < rGroup->fileVerboseLimit)
1361 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1362 fname, (long) recordOffset);
1363 rec = rec_new (records);
1365 *sysno = rec->sysno;
1367 recordAttr = rec_init_attr (zti, rec);
1371 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1373 flushRecordKeys (*sysno, 1, &reckeys);
1374 flushSortKeys (*sysno, 1);
1380 /* record already exists */
1381 struct recKeys delkeys;
1383 rec = rec_get (records, *sysno);
1386 recordAttr = rec_init_attr (zti, rec);
1388 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1390 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1391 fname, (long) recordOffset);
1396 delkeys.buf_used = rec->size[recInfo_delKeys];
1397 delkeys.buf = rec->info[recInfo_delKeys];
1398 flushSortKeys (*sysno, 0);
1399 flushRecordKeys (*sysno, 0, &delkeys);
1402 /* record going to be deleted */
1403 if (!delkeys.buf_used)
1405 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1406 fname, (long) recordOffset);
1407 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1411 if (records_processed < rGroup->fileVerboseLimit)
1412 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1413 fname, (long) recordOffset);
1416 dict_delete (matchDict, matchStr);
1417 rec_del (records, &rec);
1425 /* record going to be updated */
1426 if (!delkeys.buf_used)
1428 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1429 fname, (long) recordOffset);
1430 logf (LOG_WARN, "cannot update file above, storeKeys false");
1434 if (records_processed < rGroup->fileVerboseLimit)
1435 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1436 fname, (long) recordOffset);
1437 flushRecordKeys (*sysno, 1, &reckeys);
1442 /* update file type */
1443 xfree (rec->info[recInfo_fileType]);
1444 rec->info[recInfo_fileType] =
1445 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1447 /* update filename */
1448 xfree (rec->info[recInfo_filename]);
1449 rec->info[recInfo_filename] =
1450 rec_strdup (fname, &rec->size[recInfo_filename]);
1452 /* update delete keys */
1453 xfree (rec->info[recInfo_delKeys]);
1454 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1457 rec->size[recInfo_delKeys] = reckeys.buf_used;
1458 rec->info[recInfo_delKeys] = reckeys.buf;
1460 reckeys.buf_max = 0;
1462 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1463 rec->size[recInfo_delKeys] = reckeys.buf_used;
1464 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1465 rec->size[recInfo_delKeys]);
1470 rec->info[recInfo_delKeys] = NULL;
1471 rec->size[recInfo_delKeys] = 0;
1474 /* save file size of original record */
1475 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1476 recordAttr->recordSize = fi->file_moffset - recordOffset;
1477 if (!recordAttr->recordSize)
1478 recordAttr->recordSize = fi->file_max - recordOffset;
1479 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1481 /* set run-number for this record */
1482 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1484 /* update store data */
1485 xfree (rec->info[recInfo_storeData]);
1486 if (rGroup->flagStoreData == 1)
1488 rec->size[recInfo_storeData] = recordAttr->recordSize;
1489 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1490 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1492 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1493 (long) recordOffset, fname);
1496 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1497 < recordAttr->recordSize)
1499 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1500 recordAttr->recordSize, fname);
1506 rec->info[recInfo_storeData] = NULL;
1507 rec->size[recInfo_storeData] = 0;
1509 /* update database name */
1510 xfree (rec->info[recInfo_databaseName]);
1511 rec->info[recInfo_databaseName] =
1512 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1515 recordAttr->recordOffset = recordOffset;
1517 /* commit this record */
1518 rec_put (records, &rec);
1523 int fileExtract (SYSNO *sysno, const char *fname,
1524 const struct recordGroup *rGroupP, int deleteFlag)
1532 struct recordGroup rGroupM;
1533 struct recordGroup *rGroup = &rGroupM;
1534 struct file_read_info *fi;
1537 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1539 if (!rGroup->groupName || !*rGroup->groupName)
1542 sprintf (gprefix, "%s.", rGroup->groupName);
1544 logf (LOG_DEBUG, "fileExtract %s", fname);
1546 /* determine file extension */
1548 for (i = strlen(fname); --i >= 0; )
1549 if (fname[i] == '/')
1551 else if (fname[i] == '.')
1553 strcpy (ext, fname+i+1);
1556 /* determine file type - depending on extension */
1557 if (!rGroup->recordType)
1559 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1560 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1562 sprintf (ext_res, "%srecordType", gprefix);
1563 rGroup->recordType = res_get (common_resource, ext_res);
1566 if (!rGroup->recordType)
1568 if (records_processed < rGroup->fileVerboseLimit)
1569 logf (LOG_LOG, "? %s", fname);
1572 if (!*rGroup->recordType)
1575 recType_byName (rGroup->recTypes, rGroup->recordType, subType,
1578 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1582 /* determine match criteria */
1583 if (!rGroup->recordId)
1585 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1586 rGroup->recordId = res_get (common_resource, ext_res);
1589 /* determine database name */
1590 if (!rGroup->databaseName)
1592 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1593 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1595 sprintf (ext_res, "%sdatabase", gprefix);
1596 rGroup->databaseName = res_get (common_resource, ext_res);
1599 if (!rGroup->databaseName)
1600 rGroup->databaseName = "Default";
1602 /* determine if explain database */
1604 sprintf (ext_res, "%sexplainDatabase", gprefix);
1605 rGroup->explainDatabase =
1606 atoi (res_get_def (common_resource, ext_res, "0"));
1608 /* announce database */
1609 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1611 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1612 rGroup->explainDatabase))
1616 if (rGroup->flagStoreData == -1)
1619 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1620 if (!(sval = res_get (common_resource, ext_res)))
1622 sprintf (ext_res, "%sstoreData", gprefix);
1623 sval = res_get (common_resource, ext_res);
1626 rGroup->flagStoreData = atoi (sval);
1628 if (rGroup->flagStoreData == -1)
1629 rGroup->flagStoreData = 0;
1631 if (rGroup->flagStoreKeys == -1)
1635 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1636 if (!(sval = res_get (common_resource, ext_res)))
1638 sprintf (ext_res, "%sstoreKeys", gprefix);
1639 sval = res_get (common_resource, ext_res);
1642 rGroup->flagStoreKeys = atoi (sval);
1644 if (rGroup->flagStoreKeys == -1)
1645 rGroup->flagStoreKeys = 0;
1651 char xname[128], *xp;
1653 strncpy (xname, fname, 127);
1654 if (!(xp = strchr (xname, '.')))
1657 if (strcmp (xp+1, "sdr.bits"))
1660 h = zebraSdr_open (xname);
1663 logf (LOG_WARN, "sdr open %s", xname);
1672 segmentno = zebraSdr_segment (h, 0);
1673 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1679 r = zebraSdr_read (h, &buf);
1684 fi = file_read_start (0);
1690 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1692 } while (r && !sysno && fi->file_more);
1693 file_read_stop (fi);
1700 if (sysno && deleteFlag)
1704 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1706 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1710 fi = file_read_start (fd);
1714 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1715 recType, subType, clientData);
1716 } while (r && !sysno && fi->file_more);
1717 file_read_stop (fi);
1723 static int explain_extract (void *handle, Record rec, data1_node *n)
1725 struct recordGroup *rGroup = (struct recordGroup*) handle;
1726 struct recExtractCtrl extractCtrl;
1729 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1731 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1735 reckeys.buf_used = 0;
1736 reckeys.prevAttrUse = -1;
1737 reckeys.prevAttrSet = -1;
1738 reckeys.prevSeqNo = 0;
1740 extractCtrl.init = wordInit;
1741 extractCtrl.addWord = addRecordKey;
1742 extractCtrl.addSchema = addSchema;
1743 extractCtrl.dh = rGroup->dh;
1744 for (i = 0; i<256; i++)
1745 extractCtrl.seqno[i] = 0;
1746 extractCtrl.zebra_maps = rGroup->zebra_maps;
1747 extractCtrl.flagShowRecords = !rGroup->flagRw;
1749 grs_extract_tree(&extractCtrl, n);
1751 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1753 if (rec->size[recInfo_delKeys])
1755 struct recKeys delkeys;
1757 delkeys.buf_used = rec->size[recInfo_delKeys];
1758 delkeys.buf = rec->info[recInfo_delKeys];
1759 flushSortKeys (rec->sysno, 0);
1760 flushRecordKeys (rec->sysno, 0, &delkeys);
1762 flushRecordKeys (rec->sysno, 1, &reckeys);
1763 flushSortKeys (rec->sysno, 1);
1765 xfree (rec->info[recInfo_delKeys]);
1766 rec->size[recInfo_delKeys] = reckeys.buf_used;
1767 rec->info[recInfo_delKeys] = reckeys.buf;
1769 reckeys.buf_max = 0;