2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.92 1999-03-09 16:27:49 adam
8 * More work on SDRKit integration.
10 * Revision 1.91 1999/02/12 13:29:22 adam
11 * Implemented position-flag for registers.
13 * Revision 1.90 1999/02/02 14:50:52 adam
14 * Updated WIN32 code specific sections. Changed header.
16 * Revision 1.89 1998/10/28 10:54:38 adam
19 * Revision 1.88 1998/10/16 08:14:29 adam
20 * Updated record control system.
22 * Revision 1.87 1998/10/15 13:10:33 adam
23 * Fixed bug in Zebra that caused it to stop indexing when empty
26 * Revision 1.86 1998/10/13 20:33:53 adam
27 * Fixed one log message and change use ordinal to be an unsigned char.
29 * Revision 1.85 1998/09/22 10:03:41 adam
30 * Changed result sets to be persistent in the sense that they can
31 * be re-searched if needed.
32 * Fixed memory leak in rsm_or.
34 * Revision 1.84 1998/06/11 15:42:22 adam
35 * Changed the way use attributes are specified in the recordId
38 * Revision 1.83 1998/06/08 14:43:10 adam
39 * Added suport for EXPLAIN Proxy servers - added settings databasePath
40 * and explainDatabase to facilitate this. Increased maximum number
41 * of databases and attributes in one register.
43 * Revision 1.82 1998/05/20 10:12:15 adam
44 * Implemented automatic EXPLAIN database maintenance.
45 * Modified Zebra to work with ASN.1 compiled version of YAZ.
47 * Revision 1.81 1998/03/11 11:19:04 adam
48 * Changed the way sequence numbers are generated.
50 * Revision 1.80 1998/03/05 08:45:11 adam
51 * New result set model and modular ranking system. Moved towards
52 * descent server API. System information stored as "SGML" records.
54 * Revision 1.79 1998/02/17 10:32:52 adam
55 * Fixed bug: binary files weren't opened with flag b on NT.
57 * Revision 1.78 1998/02/10 12:03:05 adam
60 * Revision 1.77 1998/01/12 15:04:08 adam
61 * The test option (-s) only uses read-lock (and not write lock).
63 * Revision 1.76 1997/10/27 14:33:04 adam
64 * Moved towards generic character mapping depending on "structure"
65 * field in abstract syntax file. Fixed a few memory leaks. Fixed
66 * bug with negative integers when doing searches with relational
69 * Revision 1.75 1997/09/17 12:19:12 adam
70 * Zebra version corresponds to YAZ version 1.4.
71 * Changed Zebra server so that it doesn't depend on global common_resource.
73 * Revision 1.74 1997/09/09 13:38:06 adam
74 * Partial port to WIN95/NT.
76 * Revision 1.73 1997/09/04 13:57:20 adam
77 * New file extract/retrieve method tellf (added).
78 * Added O_BINARY for open calls.
80 * Revision 1.72 1997/07/15 16:32:29 adam
81 * Bug fix: Match handler didn't terminate the resulting string!
83 * Revision 1.71 1997/07/15 16:28:41 adam
84 * Bug fix: storeData didn't work with files with multiple records.
85 * Bug fix: fixed memory management with records; not really well
88 * Revision 1.70 1997/07/01 13:00:42 adam
89 * Bug fix in routine searchRecordKey: uninitialized variables.
91 * Revision 1.69 1997/04/29 09:26:03 adam
92 * Bug fix: generic recordId handling didn't work for compressed internal
95 * Revision 1.68 1997/02/12 20:39:45 adam
96 * Implemented options -f <n> that limits the log to the first <n>
98 * Changed some log messages also.
100 * Revision 1.67 1996/11/15 15:02:14 adam
101 * Minor changes regarding logging.
103 * Revision 1.66 1996/11/14 09:52:21 adam
104 * Strings in record keys bound by IT_MAX_WORD.
106 * Revision 1.65 1996/11/14 08:57:56 adam
107 * Reduction of storeKeys area.
109 * Revision 1.64 1996/11/08 11:10:16 adam
110 * Buffers used during file match got bigger.
111 * Compressed ISAM support everywhere.
112 * Bug fixes regarding masking characters in queries.
113 * Redesigned Regexp-2 queries.
115 * Revision 1.63 1996/10/29 14:09:39 adam
116 * Use of cisam system - enabled if setting isamc is 1.
118 * Revision 1.62 1996/10/11 10:57:01 adam
119 * New module recctrl. Used to manage records (extract/retrieval).
120 * Several files have been moved to the recctrl sub directory.
122 * Revision 1.61 1996/06/06 12:08:37 quinn
123 * Added showRecord function
125 * Revision 1.60 1996/06/04 10:18:12 adam
126 * Search/scan uses character mapping module.
128 * Revision 1.59 1996/05/14 15:47:07 adam
129 * Cleanup of various buffer size entities.
131 * Revision 1.58 1996/05/14 06:16:38 adam
132 * Compact use/set bytes used in search service.
134 * Revision 1.57 1996/05/13 14:23:04 adam
135 * Work on compaction of set/use bytes in dictionary.
137 * Revision 1.56 1996/05/09 09:54:42 adam
138 * Server supports maps from one logical attributes to a list of physical
140 * The extraction process doesn't make space consuming 'any' keys.
142 * Revision 1.55 1996/05/09 07:28:55 quinn
143 * Work towards phrases and multiple registers
145 * Revision 1.54 1996/05/01 13:46:35 adam
146 * First work on multiple records in one file.
147 * New option, -offset, to the "unread" command in the filter module.
149 * Revision 1.53 1996/04/26 12:09:43 adam
150 * Added a few comments.
152 * Revision 1.52 1996/04/25 13:27:57 adam
153 * Function recordExtract modified so that files with no keys (possibly empty)
156 * Revision 1.51 1996/03/19 11:08:42 adam
157 * Bug fix: Log preamble wasn't always turned off after recordExtract.
159 * Revision 1.50 1996/02/12 18:45:36 adam
160 * New fileVerboseFlag in record group control.
162 * Revision 1.49 1996/02/05 12:29:57 adam
163 * Logging reduced a bit.
164 * The remaining running time is estimated during register merge.
166 * Revision 1.48 1996/02/01 20:53:26 adam
167 * The temporary per-record keys are compacted a little, and duplication
168 * of the per-records keys are avoided when they are saved in the record
169 * information buffer.
171 * Revision 1.47 1996/01/17 14:57:48 adam
172 * Prototype changed for reader functions in extract/retrieve. File
173 * is identified by 'void *' instead of 'int.
175 * Revision 1.46 1995/12/15 14:57:16 adam
178 * Revision 1.45 1995/12/15 12:37:41 adam
179 * In addRecordKeyAny: Writes key only when attrSet != -1.
181 * Revision 1.44 1995/12/12 16:00:54 adam
182 * System call sync(2) used after update/commit.
183 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
186 * Revision 1.43 1995/12/11 09:12:46 adam
187 * The rec_get function returns NULL if record doesn't exist - will
188 * happen in the server if the result set records have been deleted since
189 * the creation of the set (i.e. the search).
190 * The server saves a result temporarily if it is 'volatile', i.e. the
191 * set is register dependent.
193 * Revision 1.42 1995/12/07 17:38:46 adam
194 * Work locking mechanisms for concurrent updates/commit.
196 * Revision 1.41 1995/12/06 16:06:42 adam
197 * Better diagnostics. Work on 'real' dictionary deletion.
199 * Revision 1.40 1995/12/05 16:57:40 adam
200 * More work on regular patterns.
202 * Revision 1.39 1995/12/05 13:20:18 adam
203 * Bug fix: file_read sometimes returned early EOF.
205 * Revision 1.38 1995/12/04 17:59:21 adam
206 * More work on regular expression conversion.
208 * Revision 1.37 1995/12/04 14:22:27 adam
209 * Extra arg to recType_byName.
210 * Started work on new regular expression parsed input to
211 * structured records.
213 * Revision 1.36 1995/11/30 08:34:29 adam
214 * Started work on commit facility.
215 * Changed a few malloc/free to xmalloc/xfree.
217 * Revision 1.35 1995/11/28 14:26:21 adam
218 * Bug fix: recordId with constant wasn't right.
219 * Bug fix: recordId dictionary entry wasn't deleted when needed.
221 * Revision 1.34 1995/11/28 09:09:38 adam
222 * Zebra config renamed.
223 * Use setting 'recordId' to identify record now.
224 * Bug fix in recindex.c: rec_release_blocks was invokeded even
225 * though the blocks were already released.
226 * File traversal properly deletes records when needed.
228 * Revision 1.33 1995/11/27 09:56:20 adam
229 * Record info elements better enumerated. Internal store of records.
231 * Revision 1.32 1995/11/25 10:24:05 adam
232 * More record fields - they are enumerated now.
233 * New options: flagStoreData flagStoreKey.
235 * Revision 1.31 1995/11/24 11:31:35 adam
236 * Commands add & del read filenames from stdin if source directory is
238 * Match criteria supports 'constant' strings.
240 * Revision 1.30 1995/11/22 17:19:16 adam
241 * Record management uses the bfile system.
243 * Revision 1.29 1995/11/21 15:01:14 adam
244 * New general match criteria implemented.
245 * New feature: document groups.
247 * Revision 1.28 1995/11/21 09:20:30 adam
248 * Yet more work on record match.
250 * Revision 1.27 1995/11/20 16:59:45 adam
251 * New update method: the 'old' keys are saved for each records.
253 * Revision 1.26 1995/11/20 11:56:24 adam
254 * Work on new traversal.
256 * Revision 1.25 1995/11/16 15:34:54 adam
257 * Uses new record management system in both indexer and server.
259 * Revision 1.24 1995/11/15 19:13:08 adam
260 * Work on record management.
262 * Revision 1.23 1995/10/27 14:00:10 adam
263 * Implemented detection of database availability.
265 * Revision 1.22 1995/10/17 18:02:07 adam
266 * New feature: databases. Implemented as prefix to words in dictionary.
268 * Revision 1.21 1995/10/10 12:24:38 adam
269 * Temporary sort files are compressed.
271 * Revision 1.20 1995/10/06 13:52:05 adam
272 * Bug fixes. Handler may abort further scanning.
274 * Revision 1.19 1995/10/04 12:55:16 adam
275 * Bug fix in ranked search. Use=Any keys inserted.
277 * Revision 1.18 1995/10/04 09:37:08 quinn
280 * Revision 1.17 1995/10/03 14:28:57 adam
281 * Buffered read in extract works.
283 * Revision 1.16 1995/10/03 14:28:45 adam
284 * Work on more effecient read handler in extract.
286 * Revision 1.15 1995/10/02 15:42:53 adam
287 * Extract uses file descriptors instead of FILE pointers.
289 * Revision 1.14 1995/10/02 15:29:13 adam
290 * More logging in file_extract.
292 * Revision 1.13 1995/09/29 14:01:39 adam
295 * Revision 1.12 1995/09/28 14:22:56 adam
296 * Sort uses smaller temporary files.
298 * Revision 1.11 1995/09/28 12:10:31 adam
299 * Bug fixes. Field prefix used in queries.
301 * Revision 1.10 1995/09/28 09:19:41 adam
302 * xfree/xmalloc used everywhere.
303 * Extract/retrieve method seems to work for text records.
305 * Revision 1.9 1995/09/27 12:22:28 adam
306 * More work on extract in record control.
307 * Field name is not in isam keys but in prefix in dictionary words.
309 * Revision 1.8 1995/09/14 07:48:22 adam
310 * Record control management.
312 * Revision 1.7 1995/09/11 13:09:32 adam
313 * More work on relevance feedback.
315 * Revision 1.6 1995/09/08 14:52:27 adam
316 * Minor changes. Dictionary is lower case now.
318 * Revision 1.5 1995/09/06 16:11:16 adam
319 * Option: only one word key per file.
321 * Revision 1.4 1995/09/05 15:28:39 adam
322 * More work on search engine.
324 * Revision 1.3 1995/09/04 12:33:41 adam
325 * Various cleanup. YAZ util used instead.
327 * Revision 1.2 1995/09/04 09:10:34 adam
328 * More work on index add/del/update.
329 * Merge sort implemented.
330 * Initial work on z39 server.
332 * Revision 1.1 1995/09/01 14:06:35 adam
333 * Split of work into more files.
357 #include "zebrasdr.h"
360 static Dict matchDict;
362 static Records records = NULL;
363 static SortIdx sortIdx = NULL;
365 static char **key_buf;
366 static size_t ptr_top;
368 static size_t key_buf_used;
369 static int key_file_no;
371 static int records_inserted = 0;
372 static int records_updated = 0;
373 static int records_deleted = 0;
374 static int records_processed = 0;
376 static ZebraExplainInfo zti = NULL;
378 static void logRecord (int showFlag)
382 if (showFlag || !(records_processed % 1000))
384 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
385 records_processed, records_inserted, records_updated,
390 static int explain_extract (void *handle, Record drec, data1_node *n);
392 int key_open (struct recordGroup *rGroup, int mem)
394 BFiles bfs = rGroup->bfs;
395 int rw = rGroup->flagRw;
396 data1_handle dh = rGroup->dh;
398 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
401 key_buf = xmalloc (mem);
402 ptr_top = mem/sizeof(char*);
408 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
410 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
414 records = rec_open (bfs, rw);
417 dict_close (matchDict);
420 zti = zebraExplain_open (records, dh, common_resource,
421 rw, rGroup, explain_extract);
424 rec_close (&records);
425 dict_close (matchDict);
428 sortIdx = sortIdx_open (bfs, 1);
439 void encode_key_init (struct encode_info *i)
446 char *encode_key_int (int d, char *bp)
455 else if (d <= 4194303)
457 *bp++ = 128 + (d>>16);
458 *bp++ = (d>>8) & 255;
463 *bp++ = 192 + (d>>24);
464 *bp++ = (d>>16) & 255;
465 *bp++ = (d>>8) & 255;
471 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
476 while ((*bp++ = *k++))
478 memcpy (&key, k+1, sizeof(struct it_key));
479 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
480 if (i->sysno != key.sysno)
482 i->sysno = key.sysno;
485 else if (!i->seqno && !key.seqno && i->cmd == *k)
487 bp = encode_key_int (key.seqno - i->seqno, bp);
488 i->seqno = key.seqno;
490 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
492 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
500 static int key_y_len;
502 static int key_y_compare (const void *p1, const void *p2)
506 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
507 *(char**) p2 + key_y_len + 1)))
509 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
512 static int key_x_compare (const void *p1, const void *p2)
514 return strcmp (*(char**) p1, *(char**) p2);
518 void key_flush (void)
523 struct encode_info encode_info;
532 logf (LOG_LOG, "sorting section %d", key_file_no);
534 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
535 getFnameTmp (out_fname, key_file_no);
537 if (!(outf = fopen (out_fname, "wb")))
539 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
542 logf (LOG_LOG, "writing section %d", key_file_no);
543 prevcp = cp = key_buf[ptr_top-ptr_i];
545 encode_key_init (&encode_info);
546 encode_key_write (cp, &encode_info, outf);
549 cp = key_buf[ptr_top-ptr_i];
550 if (strcmp (cp, prevcp))
552 encode_key_init (&encode_info);
553 encode_key_write (cp, &encode_info, outf);
557 encode_key_write (cp + strlen(cp), &encode_info, outf);
560 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
561 getFnameTmp (out_fname, key_file_no);
563 if (!(outf = fopen (out_fname, "wb")))
565 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
568 logf (LOG_LOG, "writing section %d", key_file_no);
570 prevcp = key_buf[ptr_top-i];
572 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
574 key_y_len = strlen(prevcp)+1;
576 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
577 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
579 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
580 sizeof(char*), key_y_compare);
581 cp = key_buf[ptr_top-ptr_i];
583 encode_key_init (&encode_info);
584 encode_key_write (cp, &encode_info, outf);
587 cp = key_buf[ptr_top-ptr_i];
588 encode_key_write (cp+key_y_len, &encode_info, outf);
592 prevcp = key_buf[ptr_top-ptr_i];
597 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
600 logf (LOG_LOG, "finished section %d", key_file_no);
605 int key_close (struct recordGroup *rGroup)
607 int rw = rGroup->flagRw;
609 zebraExplain_runNumberIncrement (zti, 1);
610 zebraExplain_close (zti, rw, 0);
613 rec_close (&records);
614 dict_close (matchDict);
615 sortIdx_close (sortIdx);
621 static void wordInit (struct recExtractCtrl *p, RecWord *w)
623 w->zebra_maps = p->zebra_maps;
624 w->seqnos = p->seqno;
625 w->attrSet = VAL_BIB1;
630 static struct sortKey {
635 struct sortKey *next;
638 static struct recKeys {
647 static void addIndexString (RecWord *p, const char *string, int length)
650 unsigned char attrSet;
651 unsigned short attrUse;
654 int *pseqno = &p->seqnos[p->reg_type];
656 if (reckeys.buf_used+1024 > reckeys.buf_max)
660 b = xmalloc (reckeys.buf_max += 128000);
661 if (reckeys.buf_used > 0)
662 memcpy (b, reckeys.buf, reckeys.buf_used);
666 dst = reckeys.buf + reckeys.buf_used;
668 attrSet = p->attrSet;
669 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
672 reckeys.prevAttrSet = attrSet;
673 attrUse = p->attrUse;
674 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
677 reckeys.prevAttrUse = attrUse;
679 diff = 1 + *pseqno - reckeys.prevSeqNo;
680 if (diff >= 1 && diff <= 15)
685 reckeys.prevSeqNo = *pseqno;
691 memcpy (dst, &attrSet, sizeof(attrSet));
692 dst += sizeof(attrSet);
696 memcpy (dst, &attrUse, sizeof(attrUse));
697 dst += sizeof(attrUse);
699 *dst++ = p->reg_type;
700 memcpy (dst, string, length);
706 memcpy (dst, pseqno, sizeof(*pseqno));
707 dst += sizeof(*pseqno);
709 reckeys.buf_used = dst - reckeys.buf;
714 static void addSortString (RecWord *p, const char *string, int length)
718 for (sk = sortKeys; sk; sk = sk->next)
719 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
722 sk = xmalloc (sizeof(*sk));
726 sk->string = xmalloc (length);
728 memcpy (sk->string, string, length);
730 sk->attrSet = p->attrSet;
731 sk->attrUse = p->attrUse;
734 static void addString (RecWord *p, const char *string, int length)
737 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
738 addSortString (p, string, length);
740 addIndexString (p, string, length);
743 static void addIncompleteField (RecWord *p)
745 const char *b = p->string;
746 int remain = p->length;
747 const char **map = 0;
750 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
754 char buf[IT_MAX_WORD+1];
758 while (map && *map && **map == *CHR_SPACE)
760 remain = p->length - (b - p->string);
762 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
769 while (map && *map && **map != *CHR_SPACE)
771 const char *cp = *map;
773 while (i < IT_MAX_WORD && *cp)
775 remain = p->length - (b - p->string);
777 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
783 addString (p, buf, i);
787 static void addCompleteField (RecWord *p)
789 const char *b = p->string;
790 char buf[IT_MAX_WORD+1];
791 const char **map = 0;
792 int i = 0, remain = p->length;
795 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
797 while (remain > 0 && i < IT_MAX_WORD)
799 while (map && *map && **map == *CHR_SPACE)
801 remain = p->length - (b - p->string);
803 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
810 if (i && i < IT_MAX_WORD)
811 buf[i++] = *CHR_SPACE;
812 while (map && *map && **map != *CHR_SPACE)
814 const char *cp = *map;
816 if (i >= IT_MAX_WORD)
818 while (i < IT_MAX_WORD && *cp)
820 remain = p->length - (b - p->string);
822 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
830 addString (p, buf, i);
833 static void addRecordKey (RecWord *p)
835 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
836 addCompleteField (p);
838 addIncompleteField(p);
841 static void flushSortKeys (SYSNO sysno, int cmd)
843 struct sortKey *sk = sortKeys;
845 sortIdx_sysno (sortIdx, sysno);
848 struct sortKey *sk_next = sk->next;
849 sortIdx_type (sortIdx, sk->attrUse);
850 sortIdx_add (sortIdx, sk->string, sk->length);
858 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
860 unsigned char attrSet = (unsigned char) -1;
861 unsigned short attrUse = (unsigned short) -1;
865 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
866 while (off < reckeys->buf_used)
868 const char *src = reckeys->buf + off;
876 memcpy (&attrSet, src, sizeof(attrSet));
877 src += sizeof(attrSet);
881 memcpy (&attrUse, src, sizeof(attrUse));
882 src += sizeof(attrUse);
884 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
887 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
889 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
891 ch = zebraExplain_addSU (zti, attrSet, attrUse);
893 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
896 ((char*)key_buf) [key_buf_used++] = *src++;
898 ((char*)key_buf) [key_buf_used++] = '\0';
899 ((char*) key_buf)[key_buf_used++] = cmd;
902 seqno += ((lead>>2) & 15)-1;
905 memcpy (&seqno, src, sizeof(seqno));
906 src += sizeof(seqno);
910 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
911 key_buf_used += sizeof(key);
912 off = src - reckeys->buf;
914 assert (off == reckeys->buf_used);
917 static const char **searchRecordKey (struct recKeys *reckeys,
918 int attrSetS, int attrUseS)
920 static const char *ws[32];
928 for (i = 0; i<32; i++)
931 while (off < reckeys->buf_used)
934 const char *src = reckeys->buf + off;
942 memcpy (&attrSet, src, sizeof(attrSet));
943 src += sizeof(attrSet);
947 memcpy (&attrUse, src, sizeof(attrUse));
948 src += sizeof(attrUse);
954 seqno += ((lead>>2) & 15)-1;
957 memcpy (&seqno, src, sizeof(seqno));
958 src += sizeof(seqno);
960 if (attrUseS == attrUse && attrSetS == attrSet)
967 woff = seqno - startSeq;
968 if (woff >= 0 && woff < 31)
972 off = src - reckeys->buf;
974 assert (off == reckeys->buf_used);
978 struct file_read_info {
979 off_t file_max; /* maximum offset so far */
980 off_t file_offset; /* current offset */
981 off_t file_moffset; /* offset of rec/rec boundary */
988 static struct file_read_info *file_read_start (int fd)
990 struct file_read_info *fi = xmalloc (sizeof(*fi));
994 fi->file_moffset = 0;
1000 static void file_read_stop (struct file_read_info *fi)
1005 static off_t file_seek (void *handle, off_t offset)
1007 struct file_read_info *p = handle;
1008 p->file_offset = offset;
1011 return lseek (p->fd, offset, SEEK_SET);
1014 static off_t file_tell (void *handle)
1016 struct file_read_info *p = handle;
1017 return p->file_offset;
1020 static int file_read (void *handle, char *buf, size_t count)
1022 struct file_read_info *p = handle;
1028 if (r > p->sdrmax - p->file_offset)
1029 r = p->sdrmax - p->file_offset;
1031 memcpy (buf, p->sdrbuf + p->file_offset, r);
1034 r = read (fd, buf, count);
1037 p->file_offset += r;
1038 if (p->file_offset > p->file_max)
1039 p->file_max = p->file_offset;
1044 static void file_begin (void *handle)
1046 struct file_read_info *p = handle;
1048 p->file_offset = p->file_moffset;
1049 if (!p->sdrbuf && p->file_moffset)
1050 lseek (p->fd, p->file_moffset, SEEK_SET);
1054 static void file_end (void *handle, off_t offset)
1056 struct file_read_info *p = handle;
1058 assert (p->file_more == 0);
1060 p->file_moffset = offset;
1063 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1064 const char *fname, const char *spec)
1066 static char dstBuf[2048];
1068 const char *s = spec;
1069 static const char **w;
1073 while (*s == ' ' || *s == '\t')
1079 char attset_str[64], attname_str[64];
1080 data1_attset *attset;
1083 int attSet = 1, attUse = 1;
1087 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1089 attset_str[i++] = *s;
1090 attset_str[i] = '\0';
1095 for (i = 0; *s && *s != ')'; s++)
1097 attname_str[i++] = *s;
1098 attname_str[i] = '\0';
1101 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1104 attSet = attset->reference;
1105 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1107 attUse = att->value;
1109 attUse = atoi (attname_str);
1111 w = searchRecordKey (reckeys, attSet, attUse);
1116 for (i = 0; i<32; i++)
1121 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1122 spec, rGroup->groupName ? rGroup->groupName : "none");
1127 for (i = 0; i<32; i++)
1128 if (matchFlag[i] && w[i])
1136 dst += strlen(w[i]);
1140 logf (LOG_WARN, "Record didn't contain match"
1141 " fields in (%s,%s)", attset_str, attname_str);
1149 const char *spec_src = NULL;
1150 const char *s1 = ++s;
1151 while (*s1 && *s1 != ' ' && *s1 != '\t')
1157 memcpy (special, s, spec_len);
1158 special[spec_len] = '\0';
1161 if (!strcmp (special, "group"))
1162 spec_src = rGroup->groupName;
1163 else if (!strcmp (special, "database"))
1164 spec_src = rGroup->databaseName;
1165 else if (!strcmp (special, "filename"))
1167 else if (!strcmp (special, "type"))
1168 spec_src = rGroup->recordType;
1173 strcpy (dst, spec_src);
1174 dst += strlen (spec_src);
1177 else if (*s == '\"' || *s == '\'')
1179 int stopMarker = *s++;
1183 while (*s && *s != stopMarker)
1186 tmpString[i++] = *s++;
1190 tmpString[i] = '\0';
1191 strcpy (dst, tmpString);
1192 dst += strlen (tmpString);
1196 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1197 spec, rGroup->groupName ? rGroup->groupName : "none");
1204 logf (LOG_WARN, "No match criteria for record %s in group %s",
1205 fname, rGroup->groupName ? rGroup->groupName : "none");
1212 struct recordLogInfo {
1215 struct recordGroup *rGroup;
1218 static void recordLogPreamble (int level, const char *msg, void *info)
1220 struct recordLogInfo *p = info;
1221 FILE *outf = log_file ();
1223 if (level & LOG_LOG)
1225 fprintf (outf, "File %s, offset %d, type %s\n",
1226 p->rGroup->recordType, p->recordOffset, p->fname);
1227 log_event_start (NULL, NULL);
1230 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1232 zebraExplain_addSchema (zti, oid);
1235 static int recordExtract (SYSNO *sysno, const char *fname,
1236 struct recordGroup *rGroup, int deleteFlag,
1237 struct file_read_info *fi,
1238 RecType recType, char *subType)
1240 RecordAttr *recordAttr;
1245 struct recordLogInfo logInfo;
1246 off_t recordOffset = 0;
1250 struct recExtractCtrl extractCtrl;
1252 /* we are going to read from a file, so prepare the extraction */
1255 reckeys.buf_used = 0;
1256 reckeys.prevAttrUse = -1;
1257 reckeys.prevAttrSet = -1;
1258 reckeys.prevSeqNo = 0;
1260 recordOffset = fi->file_moffset;
1261 extractCtrl.offset = fi->file_moffset;
1262 extractCtrl.readf = file_read;
1263 extractCtrl.seekf = file_seek;
1264 extractCtrl.tellf = file_tell;
1265 extractCtrl.endf = file_end;
1266 extractCtrl.fh = fi;
1267 extractCtrl.subType = subType;
1268 extractCtrl.init = wordInit;
1269 extractCtrl.addWord = addRecordKey;
1270 extractCtrl.addSchema = addSchema;
1271 extractCtrl.dh = rGroup->dh;
1272 for (i = 0; i<256; i++)
1274 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1275 extractCtrl.seqno[i] = 1;
1277 extractCtrl.seqno[i] = 0;
1279 extractCtrl.zebra_maps = rGroup->zebra_maps;
1280 extractCtrl.flagShowRecords = !rGroup->flagRw;
1282 if (!rGroup->flagRw)
1283 printf ("File: %s %ld\n", fname, (long) recordOffset);
1285 logInfo.fname = fname;
1286 logInfo.recordOffset = recordOffset;
1287 logInfo.rGroup = rGroup;
1288 log_event_start (recordLogPreamble, &logInfo);
1290 r = (*recType->extract)(&extractCtrl);
1292 log_event_start (NULL, NULL);
1296 /* error occured during extraction ... */
1297 if (rGroup->flagRw &&
1298 records_processed < rGroup->fileVerboseLimit)
1300 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1301 fname, (long) recordOffset, r);
1305 if (reckeys.buf_used == 0)
1307 /* the extraction process returned no information - the record
1308 is probably empty - unless flagShowRecords is in use */
1309 if (!rGroup->flagRw)
1311 logf (LOG_WARN, "No keys generated for file %s", fname);
1312 logf (LOG_WARN, " The file is probably empty");
1317 /* perform match if sysno not known and if match criteria is specified */
1324 if (rGroup->recordId && *rGroup->recordId)
1328 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1332 rinfo = dict_lookup (matchDict, matchStr);
1334 memcpy (sysno, rinfo+1, sizeof(*sysno));
1338 logf (LOG_WARN, "Bad match criteria");
1349 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1350 fname, (long) recordOffset);
1351 logf (LOG_WARN, "cannot delete record above (seems new)");
1354 if (records_processed < rGroup->fileVerboseLimit)
1355 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1356 fname, (long) recordOffset);
1357 rec = rec_new (records);
1359 *sysno = rec->sysno;
1361 recordAttr = rec_init_attr (zti, rec);
1365 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1367 flushRecordKeys (*sysno, 1, &reckeys);
1368 flushSortKeys (*sysno, 1);
1374 /* record already exists */
1375 struct recKeys delkeys;
1377 rec = rec_get (records, *sysno);
1380 recordAttr = rec_init_attr (zti, rec);
1382 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1384 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1385 fname, (long) recordOffset);
1390 delkeys.buf_used = rec->size[recInfo_delKeys];
1391 delkeys.buf = rec->info[recInfo_delKeys];
1392 flushSortKeys (*sysno, 0);
1393 flushRecordKeys (*sysno, 0, &delkeys);
1396 /* record going to be deleted */
1397 if (!delkeys.buf_used)
1399 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1400 fname, (long) recordOffset);
1401 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1405 if (records_processed < rGroup->fileVerboseLimit)
1406 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1407 fname, (long) recordOffset);
1410 dict_delete (matchDict, matchStr);
1411 rec_del (records, &rec);
1419 /* record going to be updated */
1420 if (!delkeys.buf_used)
1422 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1423 fname, (long) recordOffset);
1424 logf (LOG_WARN, "cannot update file above, storeKeys false");
1428 if (records_processed < rGroup->fileVerboseLimit)
1429 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1430 fname, (long) recordOffset);
1431 flushRecordKeys (*sysno, 1, &reckeys);
1436 /* update file type */
1437 xfree (rec->info[recInfo_fileType]);
1438 rec->info[recInfo_fileType] =
1439 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1441 /* update filename */
1442 xfree (rec->info[recInfo_filename]);
1443 rec->info[recInfo_filename] =
1444 rec_strdup (fname, &rec->size[recInfo_filename]);
1446 /* update delete keys */
1447 xfree (rec->info[recInfo_delKeys]);
1448 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1451 rec->size[recInfo_delKeys] = reckeys.buf_used;
1452 rec->info[recInfo_delKeys] = reckeys.buf;
1454 reckeys.buf_max = 0;
1456 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1457 rec->size[recInfo_delKeys] = reckeys.buf_used;
1458 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1459 rec->size[recInfo_delKeys]);
1464 rec->info[recInfo_delKeys] = NULL;
1465 rec->size[recInfo_delKeys] = 0;
1468 /* save file size of original record */
1469 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1470 recordAttr->recordSize = fi->file_moffset - recordOffset;
1471 if (!recordAttr->recordSize)
1472 recordAttr->recordSize = fi->file_max - recordOffset;
1473 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1475 /* set run-number for this record */
1476 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1478 /* update store data */
1479 xfree (rec->info[recInfo_storeData]);
1480 if (rGroup->flagStoreData == 1)
1482 rec->size[recInfo_storeData] = recordAttr->recordSize;
1483 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1484 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1486 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1487 (long) recordOffset, fname);
1490 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1491 < recordAttr->recordSize)
1493 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1494 recordAttr->recordSize, fname);
1500 rec->info[recInfo_storeData] = NULL;
1501 rec->size[recInfo_storeData] = 0;
1503 /* update database name */
1504 xfree (rec->info[recInfo_databaseName]);
1505 rec->info[recInfo_databaseName] =
1506 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1509 recordAttr->recordOffset = recordOffset;
1511 /* commit this record */
1512 rec_put (records, &rec);
1517 int fileExtract (SYSNO *sysno, const char *fname,
1518 const struct recordGroup *rGroupP, int deleteFlag)
1526 struct recordGroup rGroupM;
1527 struct recordGroup *rGroup = &rGroupM;
1528 struct file_read_info *fi;
1530 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1532 if (!rGroup->groupName || !*rGroup->groupName)
1535 sprintf (gprefix, "%s.", rGroup->groupName);
1537 logf (LOG_DEBUG, "fileExtract %s", fname);
1539 /* determine file extension */
1541 for (i = strlen(fname); --i >= 0; )
1542 if (fname[i] == '/')
1544 else if (fname[i] == '.')
1546 strcpy (ext, fname+i+1);
1549 /* determine file type - depending on extension */
1550 if (!rGroup->recordType)
1552 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1553 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1555 sprintf (ext_res, "%srecordType", gprefix);
1556 rGroup->recordType = res_get (common_resource, ext_res);
1559 if (!rGroup->recordType)
1561 if (records_processed < rGroup->fileVerboseLimit)
1562 logf (LOG_LOG, "? %s", fname);
1565 if (!*rGroup->recordType)
1568 recType_byName (rGroup->recTypes, rGroup->recordType, subType)))
1570 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1574 /* determine match criteria */
1575 if (!rGroup->recordId)
1577 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1578 rGroup->recordId = res_get (common_resource, ext_res);
1581 /* determine database name */
1582 if (!rGroup->databaseName)
1584 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1585 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1587 sprintf (ext_res, "%sdatabase", gprefix);
1588 rGroup->databaseName = res_get (common_resource, ext_res);
1591 if (!rGroup->databaseName)
1592 rGroup->databaseName = "Default";
1594 /* determine if explain database */
1596 sprintf (ext_res, "%sexplainDatabase", gprefix);
1597 rGroup->explainDatabase =
1598 atoi (res_get_def (common_resource, ext_res, "0"));
1600 /* announce database */
1601 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1603 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1604 rGroup->explainDatabase))
1608 if (rGroup->flagStoreData == -1)
1611 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1612 if (!(sval = res_get (common_resource, ext_res)))
1614 sprintf (ext_res, "%sstoreData", gprefix);
1615 sval = res_get (common_resource, ext_res);
1618 rGroup->flagStoreData = atoi (sval);
1620 if (rGroup->flagStoreData == -1)
1621 rGroup->flagStoreData = 0;
1623 if (rGroup->flagStoreKeys == -1)
1627 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1628 if (!(sval = res_get (common_resource, ext_res)))
1630 sprintf (ext_res, "%sstoreKeys", gprefix);
1631 sval = res_get (common_resource, ext_res);
1634 rGroup->flagStoreKeys = atoi (sval);
1636 if (rGroup->flagStoreKeys == -1)
1637 rGroup->flagStoreKeys = 0;
1643 char xname[128], *xp;
1645 strncpy (xname, fname, 127);
1646 if (!(xp = strchr (xname, '.')))
1649 if (strcmp (xp+1, "sdr.bits"))
1652 h = zebraSdr_open (xname);
1655 logf (LOG_WARN, "sdr open %s", xname);
1664 segmentno = zebraSdr_segment (h, 0);
1665 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1671 r = zebraSdr_read (h, &buf);
1676 fi = file_read_start (0);
1682 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1684 } while (r && !sysno && fi->file_more);
1685 file_read_stop (fi);
1692 if (sysno && deleteFlag)
1696 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1698 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1702 fi = file_read_start (fd);
1706 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1708 } while (r && !sysno && fi->file_more);
1709 file_read_stop (fi);
1715 static int explain_extract (void *handle, Record rec, data1_node *n)
1717 struct recordGroup *rGroup = (struct recordGroup*) handle;
1718 struct recExtractCtrl extractCtrl;
1721 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1723 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1727 reckeys.buf_used = 0;
1728 reckeys.prevAttrUse = -1;
1729 reckeys.prevAttrSet = -1;
1730 reckeys.prevSeqNo = 0;
1732 extractCtrl.init = wordInit;
1733 extractCtrl.addWord = addRecordKey;
1734 extractCtrl.addSchema = addSchema;
1735 extractCtrl.dh = rGroup->dh;
1736 for (i = 0; i<256; i++)
1737 extractCtrl.seqno[i] = 0;
1738 extractCtrl.zebra_maps = rGroup->zebra_maps;
1739 extractCtrl.flagShowRecords = !rGroup->flagRw;
1741 grs_extract_tree(&extractCtrl, n);
1743 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1745 if (rec->size[recInfo_delKeys])
1747 struct recKeys delkeys;
1749 delkeys.buf_used = rec->size[recInfo_delKeys];
1750 delkeys.buf = rec->info[recInfo_delKeys];
1751 flushSortKeys (rec->sysno, 0);
1752 flushRecordKeys (rec->sysno, 0, &delkeys);
1754 flushRecordKeys (rec->sysno, 1, &reckeys);
1755 flushSortKeys (rec->sysno, 1);
1757 xfree (rec->info[recInfo_delKeys]);
1758 rec->size[recInfo_delKeys] = reckeys.buf_used;
1759 rec->info[recInfo_delKeys] = reckeys.buf;
1761 reckeys.buf_max = 0;