2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.103 2000-05-18 12:01:36 adam
8 * System call times(2) used again. More 64-bit fixes.
10 * Revision 1.102 2000/05/15 15:32:33 adam
11 * Added 64 bit file input.
13 * Revision 1.101 2000/05/15 13:02:39 adam
16 * Revision 1.100 2000/03/20 19:08:36 adam
17 * Added remote record import using Z39.50 extended services and Segment
20 * Revision 1.99 2000/02/24 10:57:02 adam
21 * Sequence number incremented after each incomplete-field.
23 * Revision 1.98 1999/09/07 07:19:21 adam
24 * Work on character mapping. Implemented replace rules.
26 * Revision 1.97 1999/07/06 12:28:04 adam
27 * Updated record index structure. Format includes version ID. Compression
28 * algorithm ID is stored for each record block.
30 * Revision 1.96 1999/05/26 07:49:13 adam
33 * Revision 1.95 1999/05/21 12:00:17 adam
34 * Better diagnostics for extraction process.
36 * Revision 1.94 1999/05/20 12:57:18 adam
37 * Implemented TCL filter. Updated recctrl system.
39 * Revision 1.93 1999/05/15 14:36:38 adam
40 * Updated dictionary. Implemented "compression" of dictionary.
42 * Revision 1.92 1999/03/09 16:27:49 adam
43 * More work on SDRKit integration.
45 * Revision 1.91 1999/02/12 13:29:22 adam
46 * Implemented position-flag for registers.
48 * Revision 1.90 1999/02/02 14:50:52 adam
49 * Updated WIN32 code specific sections. Changed header.
51 * Revision 1.89 1998/10/28 10:54:38 adam
54 * Revision 1.88 1998/10/16 08:14:29 adam
55 * Updated record control system.
57 * Revision 1.87 1998/10/15 13:10:33 adam
58 * Fixed bug in Zebra that caused it to stop indexing when empty
61 * Revision 1.86 1998/10/13 20:33:53 adam
62 * Fixed one log message and change use ordinal to be an unsigned char.
64 * Revision 1.85 1998/09/22 10:03:41 adam
65 * Changed result sets to be persistent in the sense that they can
66 * be re-searched if needed.
67 * Fixed memory leak in rsm_or.
69 * Revision 1.84 1998/06/11 15:42:22 adam
70 * Changed the way use attributes are specified in the recordId
73 * Revision 1.83 1998/06/08 14:43:10 adam
74 * Added suport for EXPLAIN Proxy servers - added settings databasePath
75 * and explainDatabase to facilitate this. Increased maximum number
76 * of databases and attributes in one register.
78 * Revision 1.82 1998/05/20 10:12:15 adam
79 * Implemented automatic EXPLAIN database maintenance.
80 * Modified Zebra to work with ASN.1 compiled version of YAZ.
82 * Revision 1.81 1998/03/11 11:19:04 adam
83 * Changed the way sequence numbers are generated.
85 * Revision 1.80 1998/03/05 08:45:11 adam
86 * New result set model and modular ranking system. Moved towards
87 * descent server API. System information stored as "SGML" records.
89 * Revision 1.79 1998/02/17 10:32:52 adam
90 * Fixed bug: binary files weren't opened with flag b on NT.
92 * Revision 1.78 1998/02/10 12:03:05 adam
95 * Revision 1.77 1998/01/12 15:04:08 adam
96 * The test option (-s) only uses read-lock (and not write lock).
98 * Revision 1.76 1997/10/27 14:33:04 adam
99 * Moved towards generic character mapping depending on "structure"
100 * field in abstract syntax file. Fixed a few memory leaks. Fixed
101 * bug with negative integers when doing searches with relational
104 * Revision 1.75 1997/09/17 12:19:12 adam
105 * Zebra version corresponds to YAZ version 1.4.
106 * Changed Zebra server so that it doesn't depend on global common_resource.
108 * Revision 1.74 1997/09/09 13:38:06 adam
109 * Partial port to WIN95/NT.
111 * Revision 1.73 1997/09/04 13:57:20 adam
112 * New file extract/retrieve method tellf (added).
113 * Added O_BINARY for open calls.
115 * Revision 1.72 1997/07/15 16:32:29 adam
116 * Bug fix: Match handler didn't terminate the resulting string!
118 * Revision 1.71 1997/07/15 16:28:41 adam
119 * Bug fix: storeData didn't work with files with multiple records.
120 * Bug fix: fixed memory management with records; not really well
123 * Revision 1.70 1997/07/01 13:00:42 adam
124 * Bug fix in routine searchRecordKey: uninitialized variables.
126 * Revision 1.69 1997/04/29 09:26:03 adam
127 * Bug fix: generic recordId handling didn't work for compressed internal
130 * Revision 1.68 1997/02/12 20:39:45 adam
131 * Implemented options -f <n> that limits the log to the first <n>
133 * Changed some log messages also.
135 * Revision 1.67 1996/11/15 15:02:14 adam
136 * Minor changes regarding logging.
138 * Revision 1.66 1996/11/14 09:52:21 adam
139 * Strings in record keys bound by IT_MAX_WORD.
141 * Revision 1.65 1996/11/14 08:57:56 adam
142 * Reduction of storeKeys area.
144 * Revision 1.64 1996/11/08 11:10:16 adam
145 * Buffers used during file match got bigger.
146 * Compressed ISAM support everywhere.
147 * Bug fixes regarding masking characters in queries.
148 * Redesigned Regexp-2 queries.
150 * Revision 1.63 1996/10/29 14:09:39 adam
151 * Use of cisam system - enabled if setting isamc is 1.
153 * Revision 1.62 1996/10/11 10:57:01 adam
154 * New module recctrl. Used to manage records (extract/retrieval).
155 * Several files have been moved to the recctrl sub directory.
157 * Revision 1.61 1996/06/06 12:08:37 quinn
158 * Added showRecord function
160 * Revision 1.60 1996/06/04 10:18:12 adam
161 * Search/scan uses character mapping module.
163 * Revision 1.59 1996/05/14 15:47:07 adam
164 * Cleanup of various buffer size entities.
166 * Revision 1.58 1996/05/14 06:16:38 adam
167 * Compact use/set bytes used in search service.
169 * Revision 1.57 1996/05/13 14:23:04 adam
170 * Work on compaction of set/use bytes in dictionary.
172 * Revision 1.56 1996/05/09 09:54:42 adam
173 * Server supports maps from one logical attributes to a list of physical
175 * The extraction process doesn't make space consuming 'any' keys.
177 * Revision 1.55 1996/05/09 07:28:55 quinn
178 * Work towards phrases and multiple registers
180 * Revision 1.54 1996/05/01 13:46:35 adam
181 * First work on multiple records in one file.
182 * New option, -offset, to the "unread" command in the filter module.
184 * Revision 1.53 1996/04/26 12:09:43 adam
185 * Added a few comments.
187 * Revision 1.52 1996/04/25 13:27:57 adam
188 * Function recordExtract modified so that files with no keys (possibly empty)
191 * Revision 1.51 1996/03/19 11:08:42 adam
192 * Bug fix: Log preamble wasn't always turned off after recordExtract.
194 * Revision 1.50 1996/02/12 18:45:36 adam
195 * New fileVerboseFlag in record group control.
197 * Revision 1.49 1996/02/05 12:29:57 adam
198 * Logging reduced a bit.
199 * The remaining running time is estimated during register merge.
201 * Revision 1.48 1996/02/01 20:53:26 adam
202 * The temporary per-record keys are compacted a little, and duplication
203 * of the per-records keys are avoided when they are saved in the record
204 * information buffer.
206 * Revision 1.47 1996/01/17 14:57:48 adam
207 * Prototype changed for reader functions in extract/retrieve. File
208 * is identified by 'void *' instead of 'int.
210 * Revision 1.46 1995/12/15 14:57:16 adam
213 * Revision 1.45 1995/12/15 12:37:41 adam
214 * In addRecordKeyAny: Writes key only when attrSet != -1.
216 * Revision 1.44 1995/12/12 16:00:54 adam
217 * System call sync(2) used after update/commit.
218 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
221 * Revision 1.43 1995/12/11 09:12:46 adam
222 * The rec_get function returns NULL if record doesn't exist - will
223 * happen in the server if the result set records have been deleted since
224 * the creation of the set (i.e. the search).
225 * The server saves a result temporarily if it is 'volatile', i.e. the
226 * set is register dependent.
228 * Revision 1.42 1995/12/07 17:38:46 adam
229 * Work locking mechanisms for concurrent updates/commit.
231 * Revision 1.41 1995/12/06 16:06:42 adam
232 * Better diagnostics. Work on 'real' dictionary deletion.
234 * Revision 1.40 1995/12/05 16:57:40 adam
235 * More work on regular patterns.
237 * Revision 1.39 1995/12/05 13:20:18 adam
238 * Bug fix: file_read sometimes returned early EOF.
240 * Revision 1.38 1995/12/04 17:59:21 adam
241 * More work on regular expression conversion.
243 * Revision 1.37 1995/12/04 14:22:27 adam
244 * Extra arg to recType_byName.
245 * Started work on new regular expression parsed input to
246 * structured records.
248 * Revision 1.36 1995/11/30 08:34:29 adam
249 * Started work on commit facility.
250 * Changed a few malloc/free to xmalloc/xfree.
252 * Revision 1.35 1995/11/28 14:26:21 adam
253 * Bug fix: recordId with constant wasn't right.
254 * Bug fix: recordId dictionary entry wasn't deleted when needed.
256 * Revision 1.34 1995/11/28 09:09:38 adam
257 * Zebra config renamed.
258 * Use setting 'recordId' to identify record now.
259 * Bug fix in recindex.c: rec_release_blocks was invokeded even
260 * though the blocks were already released.
261 * File traversal properly deletes records when needed.
263 * Revision 1.33 1995/11/27 09:56:20 adam
264 * Record info elements better enumerated. Internal store of records.
266 * Revision 1.32 1995/11/25 10:24:05 adam
267 * More record fields - they are enumerated now.
268 * New options: flagStoreData flagStoreKey.
270 * Revision 1.31 1995/11/24 11:31:35 adam
271 * Commands add & del read filenames from stdin if source directory is
273 * Match criteria supports 'constant' strings.
275 * Revision 1.30 1995/11/22 17:19:16 adam
276 * Record management uses the bfile system.
278 * Revision 1.29 1995/11/21 15:01:14 adam
279 * New general match criteria implemented.
280 * New feature: document groups.
282 * Revision 1.28 1995/11/21 09:20:30 adam
283 * Yet more work on record match.
285 * Revision 1.27 1995/11/20 16:59:45 adam
286 * New update method: the 'old' keys are saved for each records.
288 * Revision 1.26 1995/11/20 11:56:24 adam
289 * Work on new traversal.
291 * Revision 1.25 1995/11/16 15:34:54 adam
292 * Uses new record management system in both indexer and server.
294 * Revision 1.24 1995/11/15 19:13:08 adam
295 * Work on record management.
297 * Revision 1.23 1995/10/27 14:00:10 adam
298 * Implemented detection of database availability.
300 * Revision 1.22 1995/10/17 18:02:07 adam
301 * New feature: databases. Implemented as prefix to words in dictionary.
303 * Revision 1.21 1995/10/10 12:24:38 adam
304 * Temporary sort files are compressed.
306 * Revision 1.20 1995/10/06 13:52:05 adam
307 * Bug fixes. Handler may abort further scanning.
309 * Revision 1.19 1995/10/04 12:55:16 adam
310 * Bug fix in ranked search. Use=Any keys inserted.
312 * Revision 1.18 1995/10/04 09:37:08 quinn
315 * Revision 1.17 1995/10/03 14:28:57 adam
316 * Buffered read in extract works.
318 * Revision 1.16 1995/10/03 14:28:45 adam
319 * Work on more effecient read handler in extract.
321 * Revision 1.15 1995/10/02 15:42:53 adam
322 * Extract uses file descriptors instead of FILE pointers.
324 * Revision 1.14 1995/10/02 15:29:13 adam
325 * More logging in file_extract.
327 * Revision 1.13 1995/09/29 14:01:39 adam
330 * Revision 1.12 1995/09/28 14:22:56 adam
331 * Sort uses smaller temporary files.
333 * Revision 1.11 1995/09/28 12:10:31 adam
334 * Bug fixes. Field prefix used in queries.
336 * Revision 1.10 1995/09/28 09:19:41 adam
337 * xfree/xmalloc used everywhere.
338 * Extract/retrieve method seems to work for text records.
340 * Revision 1.9 1995/09/27 12:22:28 adam
341 * More work on extract in record control.
342 * Field name is not in isam keys but in prefix in dictionary words.
344 * Revision 1.8 1995/09/14 07:48:22 adam
345 * Record control management.
347 * Revision 1.7 1995/09/11 13:09:32 adam
348 * More work on relevance feedback.
350 * Revision 1.6 1995/09/08 14:52:27 adam
351 * Minor changes. Dictionary is lower case now.
353 * Revision 1.5 1995/09/06 16:11:16 adam
354 * Option: only one word key per file.
356 * Revision 1.4 1995/09/05 15:28:39 adam
357 * More work on search engine.
359 * Revision 1.3 1995/09/04 12:33:41 adam
360 * Various cleanup. YAZ util used instead.
362 * Revision 1.2 1995/09/04 09:10:34 adam
363 * More work on index add/del/update.
364 * Merge sort implemented.
365 * Initial work on z39 server.
367 * Revision 1.1 1995/09/01 14:06:35 adam
368 * Split of work into more files.
387 #if _FILE_OFFSET_BITS == 64
388 #define PRINTF_OFF_T "%Ld"
390 #define PRINTF_OFF_T "%ld"
398 #include "zebrasdr.h"
401 static Dict matchDict;
403 static Records records = NULL;
404 static SortIdx sortIdx = NULL;
406 static char **key_buf;
407 static size_t ptr_top;
409 static size_t key_buf_used;
410 static int key_file_no;
412 static int records_inserted = 0;
413 static int records_updated = 0;
414 static int records_deleted = 0;
415 static int records_processed = 0;
417 static ZebraExplainInfo zti = NULL;
419 static void logRecord (int showFlag)
423 if (showFlag || !(records_processed % 1000))
425 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
426 records_processed, records_inserted, records_updated,
431 static int explain_extract (void *handle, Record drec, data1_node *n);
433 int key_open (struct recordGroup *rGroup, int mem)
435 BFiles bfs = rGroup->bfs;
436 int rw = rGroup->flagRw;
437 data1_handle dh = rGroup->dh;
438 char *recordCompression;
439 int record_compression = REC_COMPRESS_NONE;
441 mem = atoi(res_get_def (common_resource, "memMax", "16"))*1024*1024;
444 key_buf = (char **) xmalloc (mem);
445 ptr_top = mem/sizeof(char*);
451 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
453 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
457 recordCompression = res_get_def (common_resource,
458 "recordCompression", "none");
459 if (!strcmp (recordCompression, "none"))
460 record_compression = REC_COMPRESS_NONE;
461 if (!strcmp (recordCompression, "bzip2"))
462 record_compression = REC_COMPRESS_BZIP2;
463 records = rec_open (bfs, rw, record_compression);
466 dict_close (matchDict);
469 zti = zebraExplain_open (records, dh, common_resource,
470 rw, rGroup, explain_extract);
473 rec_close (&records);
474 dict_close (matchDict);
477 sortIdx = sortIdx_open (bfs, 1);
488 void encode_key_init (struct encode_info *i)
495 char *encode_key_int (int d, char *bp)
504 else if (d <= 4194303)
506 *bp++ = 128 + (d>>16);
507 *bp++ = (d>>8) & 255;
512 *bp++ = 192 + (d>>24);
513 *bp++ = (d>>16) & 255;
514 *bp++ = (d>>8) & 255;
520 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
525 while ((*bp++ = *k++))
527 memcpy (&key, k+1, sizeof(struct it_key));
528 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
529 if (i->sysno != key.sysno)
531 i->sysno = key.sysno;
534 else if (!i->seqno && !key.seqno && i->cmd == *k)
536 bp = encode_key_int (key.seqno - i->seqno, bp);
537 i->seqno = key.seqno;
539 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
541 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
549 static int key_y_len;
551 static int key_y_compare (const void *p1, const void *p2)
555 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
556 *(char**) p2 + key_y_len + 1)))
558 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
561 static int key_x_compare (const void *p1, const void *p2)
563 return strcmp (*(char**) p1, *(char**) p2);
567 void key_flush (void)
572 struct encode_info encode_info;
581 logf (LOG_LOG, "sorting section %d", key_file_no);
583 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
584 getFnameTmp (common_resource, out_fname, key_file_no);
586 if (!(outf = fopen (out_fname, "wb")))
588 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
591 logf (LOG_LOG, "writing section %d", key_file_no);
592 prevcp = cp = key_buf[ptr_top-ptr_i];
594 encode_key_init (&encode_info);
595 encode_key_write (cp, &encode_info, outf);
598 cp = key_buf[ptr_top-ptr_i];
599 if (strcmp (cp, prevcp))
601 encode_key_init (&encode_info);
602 encode_key_write (cp, &encode_info, outf);
606 encode_key_write (cp + strlen(cp), &encode_info, outf);
609 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
610 getFnameTmp (out_fname, key_file_no);
612 if (!(outf = fopen (out_fname, "wb")))
614 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
617 logf (LOG_LOG, "writing section %d", key_file_no);
619 prevcp = key_buf[ptr_top-i];
621 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
623 key_y_len = strlen(prevcp)+1;
625 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
626 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
628 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
629 sizeof(char*), key_y_compare);
630 cp = key_buf[ptr_top-ptr_i];
632 encode_key_init (&encode_info);
633 encode_key_write (cp, &encode_info, outf);
636 cp = key_buf[ptr_top-ptr_i];
637 encode_key_write (cp+key_y_len, &encode_info, outf);
641 prevcp = key_buf[ptr_top-ptr_i];
646 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
649 logf (LOG_LOG, "finished section %d", key_file_no);
654 int key_close (struct recordGroup *rGroup)
656 int rw = rGroup->flagRw;
658 zebraExplain_runNumberIncrement (zti, 1);
659 zebraExplain_close (zti, rw);
662 rec_close (&records);
663 dict_close (matchDict);
664 sortIdx_close (sortIdx);
670 static void wordInit (struct recExtractCtrl *p, RecWord *w)
672 w->zebra_maps = p->zebra_maps;
673 w->seqnos = p->seqno;
674 w->attrSet = VAL_BIB1;
680 static struct sortKey {
685 struct sortKey *next;
688 static struct recKeys {
697 static void addIndexString (RecWord *p, const char *string, int length)
700 unsigned char attrSet;
701 unsigned short attrUse;
704 int *pseqno = &p->seqnos[p->reg_type];
706 if (reckeys.buf_used+1024 > reckeys.buf_max)
710 b = (char *) xmalloc (reckeys.buf_max += 128000);
711 if (reckeys.buf_used > 0)
712 memcpy (b, reckeys.buf, reckeys.buf_used);
716 dst = reckeys.buf + reckeys.buf_used;
718 attrSet = p->attrSet;
719 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
722 reckeys.prevAttrSet = attrSet;
723 attrUse = p->attrUse;
724 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
727 reckeys.prevAttrUse = attrUse;
729 diff = 1 + *pseqno - reckeys.prevSeqNo;
730 if (diff >= 1 && diff <= 15)
735 reckeys.prevSeqNo = *pseqno;
741 memcpy (dst, &attrSet, sizeof(attrSet));
742 dst += sizeof(attrSet);
746 memcpy (dst, &attrUse, sizeof(attrUse));
747 dst += sizeof(attrUse);
749 *dst++ = p->reg_type;
750 memcpy (dst, string, length);
756 memcpy (dst, pseqno, sizeof(*pseqno));
757 dst += sizeof(*pseqno);
759 reckeys.buf_used = dst - reckeys.buf;
764 static void addSortString (RecWord *p, const char *string, int length)
768 for (sk = sortKeys; sk; sk = sk->next)
769 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
772 sk = (struct sortKey *) xmalloc (sizeof(*sk));
776 sk->string = (char *) xmalloc (length);
778 memcpy (sk->string, string, length);
780 sk->attrSet = p->attrSet;
781 sk->attrUse = p->attrUse;
784 static void addString (RecWord *p, const char *string, int length)
787 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
788 addSortString (p, string, length);
790 addIndexString (p, string, length);
793 static void addIncompleteField (RecWord *p)
795 const char *b = p->string;
796 int remain = p->length;
797 const char **map = 0;
800 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
804 char buf[IT_MAX_WORD+1];
808 while (map && *map && **map == *CHR_SPACE)
810 remain = p->length - (b - p->string);
812 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
819 while (map && *map && **map != *CHR_SPACE)
821 const char *cp = *map;
823 while (i < IT_MAX_WORD && *cp)
825 remain = p->length - (b - p->string);
827 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
833 addString (p, buf, i);
835 (p->seqnos[p->reg_type])++; /* to separate this from next one */
838 static void addCompleteField (RecWord *p)
840 const char *b = p->string;
841 char buf[IT_MAX_WORD+1];
842 const char **map = 0;
843 int i = 0, remain = p->length;
846 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
848 while (remain > 0 && i < IT_MAX_WORD)
850 while (map && *map && **map == *CHR_SPACE)
852 remain = p->length - (b - p->string);
854 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
861 if (i && i < IT_MAX_WORD)
862 buf[i++] = *CHR_SPACE;
863 while (map && *map && **map != *CHR_SPACE)
865 const char *cp = *map;
867 if (i >= IT_MAX_WORD)
869 while (i < IT_MAX_WORD && *cp)
871 remain = p->length - (b - p->string);
873 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
881 addString (p, buf, i);
884 static void addRecordKey (RecWord *p)
887 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
888 p->string, p->length)))
890 p->string = wrbuf_buf(wrbuf);
891 p->length = wrbuf_len(wrbuf);
893 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
894 addCompleteField (p);
896 addIncompleteField(p);
899 static void flushSortKeys (SYSNO sysno, int cmd)
901 struct sortKey *sk = sortKeys;
903 sortIdx_sysno (sortIdx, sysno);
906 struct sortKey *sk_next = sk->next;
907 sortIdx_type (sortIdx, sk->attrUse);
908 sortIdx_add (sortIdx, sk->string, sk->length);
916 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
918 unsigned char attrSet = (unsigned char) -1;
919 unsigned short attrUse = (unsigned short) -1;
923 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
924 while (off < reckeys->buf_used)
926 const char *src = reckeys->buf + off;
934 memcpy (&attrSet, src, sizeof(attrSet));
935 src += sizeof(attrSet);
939 memcpy (&attrUse, src, sizeof(attrUse));
940 src += sizeof(attrUse);
942 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
945 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
947 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
949 ch = zebraExplain_addSU (zti, attrSet, attrUse);
951 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
954 ((char*)key_buf) [key_buf_used++] = *src++;
956 ((char*)key_buf) [key_buf_used++] = '\0';
957 ((char*) key_buf)[key_buf_used++] = cmd;
960 seqno += ((lead>>2) & 15)-1;
963 memcpy (&seqno, src, sizeof(seqno));
964 src += sizeof(seqno);
968 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
969 key_buf_used += sizeof(key);
970 off = src - reckeys->buf;
972 assert (off == reckeys->buf_used);
975 static const char **searchRecordKey (struct recKeys *reckeys,
976 int attrSetS, int attrUseS)
978 static const char *ws[32];
986 for (i = 0; i<32; i++)
989 while (off < reckeys->buf_used)
992 const char *src = reckeys->buf + off;
1000 memcpy (&attrSet, src, sizeof(attrSet));
1001 src += sizeof(attrSet);
1005 memcpy (&attrUse, src, sizeof(attrUse));
1006 src += sizeof(attrUse);
1012 seqno += ((lead>>2) & 15)-1;
1015 memcpy (&seqno, src, sizeof(seqno));
1016 src += sizeof(seqno);
1018 if (attrUseS == attrUse && attrSetS == attrSet)
1025 woff = seqno - startSeq;
1026 if (woff >= 0 && woff < 31)
1030 off = src - reckeys->buf;
1032 assert (off == reckeys->buf_used);
1036 struct file_read_info {
1037 off_t file_max; /* maximum offset so far */
1038 off_t file_offset; /* current offset */
1039 off_t file_moffset; /* offset of rec/rec boundary */
1046 static struct file_read_info *file_read_start (int fd)
1048 struct file_read_info *fi = (struct file_read_info *)
1049 xmalloc (sizeof(*fi));
1053 fi->file_moffset = 0;
1059 static void file_read_stop (struct file_read_info *fi)
1064 static off_t file_seek (void *handle, off_t offset)
1066 struct file_read_info *p = (struct file_read_info *) handle;
1067 p->file_offset = offset;
1070 return lseek (p->fd, offset, SEEK_SET);
1073 static off_t file_tell (void *handle)
1075 struct file_read_info *p = (struct file_read_info *) handle;
1076 return p->file_offset;
1079 static int file_read (void *handle, char *buf, size_t count)
1081 struct file_read_info *p = (struct file_read_info *) handle;
1087 if (r > p->sdrmax - p->file_offset)
1088 r = p->sdrmax - p->file_offset;
1090 memcpy (buf, p->sdrbuf + p->file_offset, r);
1093 r = read (fd, buf, count);
1096 p->file_offset += r;
1097 if (p->file_offset > p->file_max)
1098 p->file_max = p->file_offset;
1103 static void file_begin (void *handle)
1105 struct file_read_info *p = (struct file_read_info *) handle;
1107 p->file_offset = p->file_moffset;
1108 if (!p->sdrbuf && p->file_moffset)
1109 lseek (p->fd, p->file_moffset, SEEK_SET);
1113 static void file_end (void *handle, off_t offset)
1115 struct file_read_info *p = (struct file_read_info *) handle;
1117 assert (p->file_more == 0);
1119 p->file_moffset = offset;
1122 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1123 const char *fname, const char *spec)
1125 static char dstBuf[2048];
1127 const char *s = spec;
1128 static const char **w;
1132 while (*s == ' ' || *s == '\t')
1138 char attset_str[64], attname_str[64];
1139 data1_attset *attset;
1142 int attSet = 1, attUse = 1;
1146 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1148 attset_str[i++] = *s;
1149 attset_str[i] = '\0';
1154 for (i = 0; *s && *s != ')'; s++)
1156 attname_str[i++] = *s;
1157 attname_str[i] = '\0';
1160 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1163 attSet = attset->reference;
1164 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1166 attUse = att->value;
1168 attUse = atoi (attname_str);
1170 w = searchRecordKey (reckeys, attSet, attUse);
1175 for (i = 0; i<32; i++)
1180 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1181 spec, rGroup->groupName ? rGroup->groupName : "none");
1186 for (i = 0; i<32; i++)
1187 if (matchFlag[i] && w[i])
1195 dst += strlen(w[i]);
1199 logf (LOG_WARN, "Record didn't contain match"
1200 " fields in (%s,%s)", attset_str, attname_str);
1208 const char *spec_src = NULL;
1209 const char *s1 = ++s;
1210 while (*s1 && *s1 != ' ' && *s1 != '\t')
1216 memcpy (special, s, spec_len);
1217 special[spec_len] = '\0';
1220 if (!strcmp (special, "group"))
1221 spec_src = rGroup->groupName;
1222 else if (!strcmp (special, "database"))
1223 spec_src = rGroup->databaseName;
1224 else if (!strcmp (special, "filename"))
1226 else if (!strcmp (special, "type"))
1227 spec_src = rGroup->recordType;
1232 strcpy (dst, spec_src);
1233 dst += strlen (spec_src);
1236 else if (*s == '\"' || *s == '\'')
1238 int stopMarker = *s++;
1242 while (*s && *s != stopMarker)
1245 tmpString[i++] = *s++;
1249 tmpString[i] = '\0';
1250 strcpy (dst, tmpString);
1251 dst += strlen (tmpString);
1255 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1256 spec, rGroup->groupName ? rGroup->groupName : "none");
1263 logf (LOG_WARN, "No match criteria for record %s in group %s",
1264 fname, rGroup->groupName ? rGroup->groupName : "none");
1271 struct recordLogInfo {
1274 struct recordGroup *rGroup;
1277 static void recordLogPreamble (int level, const char *msg, void *info)
1279 struct recordLogInfo *p = (struct recordLogInfo *) info;
1280 FILE *outf = log_file ();
1282 if (level & LOG_LOG)
1284 fprintf (outf, "File %s, offset %d, type %s\n",
1285 p->rGroup->recordType, p->recordOffset, p->fname);
1286 log_event_start (NULL, NULL);
1289 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1291 zebraExplain_addSchema (zti, oid);
1294 static int recordExtract (SYSNO *sysno, const char *fname,
1295 struct recordGroup *rGroup, int deleteFlag,
1296 struct file_read_info *fi,
1297 RecType recType, char *subType, void *clientData)
1299 RecordAttr *recordAttr;
1304 struct recordLogInfo logInfo;
1305 off_t recordOffset = 0;
1309 struct recExtractCtrl extractCtrl;
1311 /* we are going to read from a file, so prepare the extraction */
1314 reckeys.buf_used = 0;
1315 reckeys.prevAttrUse = -1;
1316 reckeys.prevAttrSet = -1;
1317 reckeys.prevSeqNo = 0;
1319 recordOffset = fi->file_moffset;
1320 extractCtrl.offset = fi->file_moffset;
1321 extractCtrl.readf = file_read;
1322 extractCtrl.seekf = file_seek;
1323 extractCtrl.tellf = file_tell;
1324 extractCtrl.endf = file_end;
1325 extractCtrl.fh = fi;
1326 extractCtrl.subType = subType;
1327 extractCtrl.init = wordInit;
1328 extractCtrl.tokenAdd = addRecordKey;
1329 extractCtrl.schemaAdd = addSchema;
1330 extractCtrl.dh = rGroup->dh;
1331 for (i = 0; i<256; i++)
1333 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1334 extractCtrl.seqno[i] = 1;
1336 extractCtrl.seqno[i] = 0;
1338 extractCtrl.zebra_maps = rGroup->zebra_maps;
1339 extractCtrl.flagShowRecords = !rGroup->flagRw;
1341 if (!rGroup->flagRw)
1342 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
1344 logInfo.fname = fname;
1345 logInfo.recordOffset = recordOffset;
1346 logInfo.rGroup = rGroup;
1347 log_event_start (recordLogPreamble, &logInfo);
1349 r = (*recType->extract)(clientData, &extractCtrl);
1351 log_event_start (NULL, NULL);
1353 if (r == RECCTRL_EXTRACT_EOF)
1355 else if (r == RECCTRL_EXTRACT_ERROR)
1357 /* error occured during extraction ... */
1358 if (rGroup->flagRw &&
1359 records_processed < rGroup->fileVerboseLimit)
1361 logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType,
1362 fname, recordOffset);
1366 if (reckeys.buf_used == 0)
1368 /* the extraction process returned no information - the record
1369 is probably empty - unless flagShowRecords is in use */
1370 if (!rGroup->flagRw)
1373 logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType,
1374 fname, recordOffset);
1379 /* perform match if sysno not known and if match criteria is specified */
1386 if (rGroup->recordId && *rGroup->recordId)
1390 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1394 rinfo = dict_lookup (matchDict, matchStr);
1396 memcpy (sysno, rinfo+1, sizeof(*sysno));
1400 logf (LOG_WARN, "Bad match criteria");
1411 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType,
1412 fname, recordOffset);
1413 logf (LOG_WARN, "cannot delete record above (seems new)");
1416 if (records_processed < rGroup->fileVerboseLimit)
1417 logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
1418 fname, recordOffset);
1419 rec = rec_new (records);
1421 *sysno = rec->sysno;
1423 recordAttr = rec_init_attr (zti, rec);
1427 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1429 flushRecordKeys (*sysno, 1, &reckeys);
1430 flushSortKeys (*sysno, 1);
1436 /* record already exists */
1437 struct recKeys delkeys;
1439 rec = rec_get (records, *sysno);
1442 recordAttr = rec_init_attr (zti, rec);
1444 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1446 logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType,
1447 fname, recordOffset);
1452 delkeys.buf_used = rec->size[recInfo_delKeys];
1453 delkeys.buf = rec->info[recInfo_delKeys];
1454 flushSortKeys (*sysno, 0);
1455 flushRecordKeys (*sysno, 0, &delkeys);
1458 /* record going to be deleted */
1459 if (!delkeys.buf_used)
1461 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
1462 rGroup->recordType, fname, recordOffset);
1463 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1467 if (records_processed < rGroup->fileVerboseLimit)
1468 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
1469 rGroup->recordType, fname, recordOffset);
1472 dict_delete (matchDict, matchStr);
1473 rec_del (records, &rec);
1481 /* record going to be updated */
1482 if (!delkeys.buf_used)
1484 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
1485 rGroup->recordType, fname, recordOffset);
1486 logf (LOG_WARN, "cannot update file above, storeKeys false");
1490 if (records_processed < rGroup->fileVerboseLimit)
1491 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
1492 rGroup->recordType, fname, recordOffset);
1493 flushRecordKeys (*sysno, 1, &reckeys);
1498 /* update file type */
1499 xfree (rec->info[recInfo_fileType]);
1500 rec->info[recInfo_fileType] =
1501 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1503 /* update filename */
1504 xfree (rec->info[recInfo_filename]);
1505 rec->info[recInfo_filename] =
1506 rec_strdup (fname, &rec->size[recInfo_filename]);
1508 /* update delete keys */
1509 xfree (rec->info[recInfo_delKeys]);
1510 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1513 rec->size[recInfo_delKeys] = reckeys.buf_used;
1514 rec->info[recInfo_delKeys] = reckeys.buf;
1516 reckeys.buf_max = 0;
1518 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1519 rec->size[recInfo_delKeys] = reckeys.buf_used;
1520 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1521 rec->size[recInfo_delKeys]);
1526 rec->info[recInfo_delKeys] = NULL;
1527 rec->size[recInfo_delKeys] = 0;
1530 /* save file size of original record */
1531 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1532 recordAttr->recordSize = fi->file_moffset - recordOffset;
1533 if (!recordAttr->recordSize)
1534 recordAttr->recordSize = fi->file_max - recordOffset;
1535 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1537 /* set run-number for this record */
1538 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1540 /* update store data */
1541 xfree (rec->info[recInfo_storeData]);
1542 if (rGroup->flagStoreData == 1)
1544 rec->size[recInfo_storeData] = recordAttr->recordSize;
1545 rec->info[recInfo_storeData] = (char *)
1546 xmalloc (recordAttr->recordSize);
1547 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1549 logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
1550 recordOffset, fname);
1553 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1554 < recordAttr->recordSize)
1556 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1557 recordAttr->recordSize, fname);
1563 rec->info[recInfo_storeData] = NULL;
1564 rec->size[recInfo_storeData] = 0;
1566 /* update database name */
1567 xfree (rec->info[recInfo_databaseName]);
1568 rec->info[recInfo_databaseName] =
1569 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1572 recordAttr->recordOffset = recordOffset;
1574 /* commit this record */
1575 rec_put (records, &rec);
1580 int fileExtract (SYSNO *sysno, const char *fname,
1581 const struct recordGroup *rGroupP, int deleteFlag)
1589 struct recordGroup rGroupM;
1590 struct recordGroup *rGroup = &rGroupM;
1591 struct file_read_info *fi;
1594 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1596 if (!rGroup->groupName || !*rGroup->groupName)
1599 sprintf (gprefix, "%s.", rGroup->groupName);
1601 logf (LOG_DEBUG, "fileExtract %s", fname);
1603 /* determine file extension */
1605 for (i = strlen(fname); --i >= 0; )
1606 if (fname[i] == '/')
1608 else if (fname[i] == '.')
1610 strcpy (ext, fname+i+1);
1613 /* determine file type - depending on extension */
1614 if (!rGroup->recordType)
1616 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1617 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1619 sprintf (ext_res, "%srecordType", gprefix);
1620 rGroup->recordType = res_get (common_resource, ext_res);
1623 if (!rGroup->recordType)
1625 if (records_processed < rGroup->fileVerboseLimit)
1626 logf (LOG_LOG, "? %s", fname);
1629 if (!*rGroup->recordType)
1632 recType_byName (rGroup->recTypes, rGroup->recordType, subType,
1635 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1639 /* determine match criteria */
1640 if (!rGroup->recordId)
1642 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1643 rGroup->recordId = res_get (common_resource, ext_res);
1646 /* determine database name */
1647 if (!rGroup->databaseName)
1649 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1650 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1652 sprintf (ext_res, "%sdatabase", gprefix);
1653 rGroup->databaseName = res_get (common_resource, ext_res);
1656 if (!rGroup->databaseName)
1657 rGroup->databaseName = "Default";
1659 /* determine if explain database */
1661 sprintf (ext_res, "%sexplainDatabase", gprefix);
1662 rGroup->explainDatabase =
1663 atoi (res_get_def (common_resource, ext_res, "0"));
1665 /* announce database */
1666 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1668 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1669 rGroup->explainDatabase))
1673 if (rGroup->flagStoreData == -1)
1676 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1677 if (!(sval = res_get (common_resource, ext_res)))
1679 sprintf (ext_res, "%sstoreData", gprefix);
1680 sval = res_get (common_resource, ext_res);
1683 rGroup->flagStoreData = atoi (sval);
1685 if (rGroup->flagStoreData == -1)
1686 rGroup->flagStoreData = 0;
1688 if (rGroup->flagStoreKeys == -1)
1692 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1693 if (!(sval = res_get (common_resource, ext_res)))
1695 sprintf (ext_res, "%sstoreKeys", gprefix);
1696 sval = res_get (common_resource, ext_res);
1699 rGroup->flagStoreKeys = atoi (sval);
1701 if (rGroup->flagStoreKeys == -1)
1702 rGroup->flagStoreKeys = 0;
1704 if (sysno && deleteFlag)
1708 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1710 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1714 fi = file_read_start (fd);
1718 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1719 recType, subType, clientData);
1720 } while (r && !sysno && fi->file_more);
1721 file_read_stop (fi);
1727 static int explain_extract (void *handle, Record rec, data1_node *n)
1729 struct recordGroup *rGroup = (struct recordGroup*) handle;
1730 struct recExtractCtrl extractCtrl;
1733 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1735 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1739 reckeys.buf_used = 0;
1740 reckeys.prevAttrUse = -1;
1741 reckeys.prevAttrSet = -1;
1742 reckeys.prevSeqNo = 0;
1744 extractCtrl.init = wordInit;
1745 extractCtrl.tokenAdd = addRecordKey;
1746 extractCtrl.schemaAdd = addSchema;
1747 extractCtrl.dh = rGroup->dh;
1748 for (i = 0; i<256; i++)
1749 extractCtrl.seqno[i] = 0;
1750 extractCtrl.zebra_maps = rGroup->zebra_maps;
1751 extractCtrl.flagShowRecords = !rGroup->flagRw;
1753 grs_extract_tree(&extractCtrl, n);
1755 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1757 if (rec->size[recInfo_delKeys])
1759 struct recKeys delkeys;
1761 delkeys.buf_used = rec->size[recInfo_delKeys];
1762 delkeys.buf = rec->info[recInfo_delKeys];
1763 flushSortKeys (rec->sysno, 0);
1764 flushRecordKeys (rec->sysno, 0, &delkeys);
1766 flushRecordKeys (rec->sysno, 1, &reckeys);
1767 flushSortKeys (rec->sysno, 1);
1769 xfree (rec->info[recInfo_delKeys]);
1770 rec->size[recInfo_delKeys] = reckeys.buf_used;
1771 rec->info[recInfo_delKeys] = reckeys.buf;
1773 reckeys.buf_max = 0;