2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.97 1999-07-06 12:28:04 adam
8 * Updated record index structure. Format includes version ID. Compression
9 * algorithm ID is stored for each record block.
11 * Revision 1.96 1999/05/26 07:49:13 adam
14 * Revision 1.95 1999/05/21 12:00:17 adam
15 * Better diagnostics for extraction process.
17 * Revision 1.94 1999/05/20 12:57:18 adam
18 * Implemented TCL filter. Updated recctrl system.
20 * Revision 1.93 1999/05/15 14:36:38 adam
21 * Updated dictionary. Implemented "compression" of dictionary.
23 * Revision 1.92 1999/03/09 16:27:49 adam
24 * More work on SDRKit integration.
26 * Revision 1.91 1999/02/12 13:29:22 adam
27 * Implemented position-flag for registers.
29 * Revision 1.90 1999/02/02 14:50:52 adam
30 * Updated WIN32 code specific sections. Changed header.
32 * Revision 1.89 1998/10/28 10:54:38 adam
35 * Revision 1.88 1998/10/16 08:14:29 adam
36 * Updated record control system.
38 * Revision 1.87 1998/10/15 13:10:33 adam
39 * Fixed bug in Zebra that caused it to stop indexing when empty
42 * Revision 1.86 1998/10/13 20:33:53 adam
43 * Fixed one log message and change use ordinal to be an unsigned char.
45 * Revision 1.85 1998/09/22 10:03:41 adam
46 * Changed result sets to be persistent in the sense that they can
47 * be re-searched if needed.
48 * Fixed memory leak in rsm_or.
50 * Revision 1.84 1998/06/11 15:42:22 adam
51 * Changed the way use attributes are specified in the recordId
54 * Revision 1.83 1998/06/08 14:43:10 adam
55 * Added suport for EXPLAIN Proxy servers - added settings databasePath
56 * and explainDatabase to facilitate this. Increased maximum number
57 * of databases and attributes in one register.
59 * Revision 1.82 1998/05/20 10:12:15 adam
60 * Implemented automatic EXPLAIN database maintenance.
61 * Modified Zebra to work with ASN.1 compiled version of YAZ.
63 * Revision 1.81 1998/03/11 11:19:04 adam
64 * Changed the way sequence numbers are generated.
66 * Revision 1.80 1998/03/05 08:45:11 adam
67 * New result set model and modular ranking system. Moved towards
68 * descent server API. System information stored as "SGML" records.
70 * Revision 1.79 1998/02/17 10:32:52 adam
71 * Fixed bug: binary files weren't opened with flag b on NT.
73 * Revision 1.78 1998/02/10 12:03:05 adam
76 * Revision 1.77 1998/01/12 15:04:08 adam
77 * The test option (-s) only uses read-lock (and not write lock).
79 * Revision 1.76 1997/10/27 14:33:04 adam
80 * Moved towards generic character mapping depending on "structure"
81 * field in abstract syntax file. Fixed a few memory leaks. Fixed
82 * bug with negative integers when doing searches with relational
85 * Revision 1.75 1997/09/17 12:19:12 adam
86 * Zebra version corresponds to YAZ version 1.4.
87 * Changed Zebra server so that it doesn't depend on global common_resource.
89 * Revision 1.74 1997/09/09 13:38:06 adam
90 * Partial port to WIN95/NT.
92 * Revision 1.73 1997/09/04 13:57:20 adam
93 * New file extract/retrieve method tellf (added).
94 * Added O_BINARY for open calls.
96 * Revision 1.72 1997/07/15 16:32:29 adam
97 * Bug fix: Match handler didn't terminate the resulting string!
99 * Revision 1.71 1997/07/15 16:28:41 adam
100 * Bug fix: storeData didn't work with files with multiple records.
101 * Bug fix: fixed memory management with records; not really well
104 * Revision 1.70 1997/07/01 13:00:42 adam
105 * Bug fix in routine searchRecordKey: uninitialized variables.
107 * Revision 1.69 1997/04/29 09:26:03 adam
108 * Bug fix: generic recordId handling didn't work for compressed internal
111 * Revision 1.68 1997/02/12 20:39:45 adam
112 * Implemented options -f <n> that limits the log to the first <n>
114 * Changed some log messages also.
116 * Revision 1.67 1996/11/15 15:02:14 adam
117 * Minor changes regarding logging.
119 * Revision 1.66 1996/11/14 09:52:21 adam
120 * Strings in record keys bound by IT_MAX_WORD.
122 * Revision 1.65 1996/11/14 08:57:56 adam
123 * Reduction of storeKeys area.
125 * Revision 1.64 1996/11/08 11:10:16 adam
126 * Buffers used during file match got bigger.
127 * Compressed ISAM support everywhere.
128 * Bug fixes regarding masking characters in queries.
129 * Redesigned Regexp-2 queries.
131 * Revision 1.63 1996/10/29 14:09:39 adam
132 * Use of cisam system - enabled if setting isamc is 1.
134 * Revision 1.62 1996/10/11 10:57:01 adam
135 * New module recctrl. Used to manage records (extract/retrieval).
136 * Several files have been moved to the recctrl sub directory.
138 * Revision 1.61 1996/06/06 12:08:37 quinn
139 * Added showRecord function
141 * Revision 1.60 1996/06/04 10:18:12 adam
142 * Search/scan uses character mapping module.
144 * Revision 1.59 1996/05/14 15:47:07 adam
145 * Cleanup of various buffer size entities.
147 * Revision 1.58 1996/05/14 06:16:38 adam
148 * Compact use/set bytes used in search service.
150 * Revision 1.57 1996/05/13 14:23:04 adam
151 * Work on compaction of set/use bytes in dictionary.
153 * Revision 1.56 1996/05/09 09:54:42 adam
154 * Server supports maps from one logical attributes to a list of physical
156 * The extraction process doesn't make space consuming 'any' keys.
158 * Revision 1.55 1996/05/09 07:28:55 quinn
159 * Work towards phrases and multiple registers
161 * Revision 1.54 1996/05/01 13:46:35 adam
162 * First work on multiple records in one file.
163 * New option, -offset, to the "unread" command in the filter module.
165 * Revision 1.53 1996/04/26 12:09:43 adam
166 * Added a few comments.
168 * Revision 1.52 1996/04/25 13:27:57 adam
169 * Function recordExtract modified so that files with no keys (possibly empty)
172 * Revision 1.51 1996/03/19 11:08:42 adam
173 * Bug fix: Log preamble wasn't always turned off after recordExtract.
175 * Revision 1.50 1996/02/12 18:45:36 adam
176 * New fileVerboseFlag in record group control.
178 * Revision 1.49 1996/02/05 12:29:57 adam
179 * Logging reduced a bit.
180 * The remaining running time is estimated during register merge.
182 * Revision 1.48 1996/02/01 20:53:26 adam
183 * The temporary per-record keys are compacted a little, and duplication
184 * of the per-records keys are avoided when they are saved in the record
185 * information buffer.
187 * Revision 1.47 1996/01/17 14:57:48 adam
188 * Prototype changed for reader functions in extract/retrieve. File
189 * is identified by 'void *' instead of 'int.
191 * Revision 1.46 1995/12/15 14:57:16 adam
194 * Revision 1.45 1995/12/15 12:37:41 adam
195 * In addRecordKeyAny: Writes key only when attrSet != -1.
197 * Revision 1.44 1995/12/12 16:00:54 adam
198 * System call sync(2) used after update/commit.
199 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
202 * Revision 1.43 1995/12/11 09:12:46 adam
203 * The rec_get function returns NULL if record doesn't exist - will
204 * happen in the server if the result set records have been deleted since
205 * the creation of the set (i.e. the search).
206 * The server saves a result temporarily if it is 'volatile', i.e. the
207 * set is register dependent.
209 * Revision 1.42 1995/12/07 17:38:46 adam
210 * Work locking mechanisms for concurrent updates/commit.
212 * Revision 1.41 1995/12/06 16:06:42 adam
213 * Better diagnostics. Work on 'real' dictionary deletion.
215 * Revision 1.40 1995/12/05 16:57:40 adam
216 * More work on regular patterns.
218 * Revision 1.39 1995/12/05 13:20:18 adam
219 * Bug fix: file_read sometimes returned early EOF.
221 * Revision 1.38 1995/12/04 17:59:21 adam
222 * More work on regular expression conversion.
224 * Revision 1.37 1995/12/04 14:22:27 adam
225 * Extra arg to recType_byName.
226 * Started work on new regular expression parsed input to
227 * structured records.
229 * Revision 1.36 1995/11/30 08:34:29 adam
230 * Started work on commit facility.
231 * Changed a few malloc/free to xmalloc/xfree.
233 * Revision 1.35 1995/11/28 14:26:21 adam
234 * Bug fix: recordId with constant wasn't right.
235 * Bug fix: recordId dictionary entry wasn't deleted when needed.
237 * Revision 1.34 1995/11/28 09:09:38 adam
238 * Zebra config renamed.
239 * Use setting 'recordId' to identify record now.
240 * Bug fix in recindex.c: rec_release_blocks was invokeded even
241 * though the blocks were already released.
242 * File traversal properly deletes records when needed.
244 * Revision 1.33 1995/11/27 09:56:20 adam
245 * Record info elements better enumerated. Internal store of records.
247 * Revision 1.32 1995/11/25 10:24:05 adam
248 * More record fields - they are enumerated now.
249 * New options: flagStoreData flagStoreKey.
251 * Revision 1.31 1995/11/24 11:31:35 adam
252 * Commands add & del read filenames from stdin if source directory is
254 * Match criteria supports 'constant' strings.
256 * Revision 1.30 1995/11/22 17:19:16 adam
257 * Record management uses the bfile system.
259 * Revision 1.29 1995/11/21 15:01:14 adam
260 * New general match criteria implemented.
261 * New feature: document groups.
263 * Revision 1.28 1995/11/21 09:20:30 adam
264 * Yet more work on record match.
266 * Revision 1.27 1995/11/20 16:59:45 adam
267 * New update method: the 'old' keys are saved for each records.
269 * Revision 1.26 1995/11/20 11:56:24 adam
270 * Work on new traversal.
272 * Revision 1.25 1995/11/16 15:34:54 adam
273 * Uses new record management system in both indexer and server.
275 * Revision 1.24 1995/11/15 19:13:08 adam
276 * Work on record management.
278 * Revision 1.23 1995/10/27 14:00:10 adam
279 * Implemented detection of database availability.
281 * Revision 1.22 1995/10/17 18:02:07 adam
282 * New feature: databases. Implemented as prefix to words in dictionary.
284 * Revision 1.21 1995/10/10 12:24:38 adam
285 * Temporary sort files are compressed.
287 * Revision 1.20 1995/10/06 13:52:05 adam
288 * Bug fixes. Handler may abort further scanning.
290 * Revision 1.19 1995/10/04 12:55:16 adam
291 * Bug fix in ranked search. Use=Any keys inserted.
293 * Revision 1.18 1995/10/04 09:37:08 quinn
296 * Revision 1.17 1995/10/03 14:28:57 adam
297 * Buffered read in extract works.
299 * Revision 1.16 1995/10/03 14:28:45 adam
300 * Work on more effecient read handler in extract.
302 * Revision 1.15 1995/10/02 15:42:53 adam
303 * Extract uses file descriptors instead of FILE pointers.
305 * Revision 1.14 1995/10/02 15:29:13 adam
306 * More logging in file_extract.
308 * Revision 1.13 1995/09/29 14:01:39 adam
311 * Revision 1.12 1995/09/28 14:22:56 adam
312 * Sort uses smaller temporary files.
314 * Revision 1.11 1995/09/28 12:10:31 adam
315 * Bug fixes. Field prefix used in queries.
317 * Revision 1.10 1995/09/28 09:19:41 adam
318 * xfree/xmalloc used everywhere.
319 * Extract/retrieve method seems to work for text records.
321 * Revision 1.9 1995/09/27 12:22:28 adam
322 * More work on extract in record control.
323 * Field name is not in isam keys but in prefix in dictionary words.
325 * Revision 1.8 1995/09/14 07:48:22 adam
326 * Record control management.
328 * Revision 1.7 1995/09/11 13:09:32 adam
329 * More work on relevance feedback.
331 * Revision 1.6 1995/09/08 14:52:27 adam
332 * Minor changes. Dictionary is lower case now.
334 * Revision 1.5 1995/09/06 16:11:16 adam
335 * Option: only one word key per file.
337 * Revision 1.4 1995/09/05 15:28:39 adam
338 * More work on search engine.
340 * Revision 1.3 1995/09/04 12:33:41 adam
341 * Various cleanup. YAZ util used instead.
343 * Revision 1.2 1995/09/04 09:10:34 adam
344 * More work on index add/del/update.
345 * Merge sort implemented.
346 * Initial work on z39 server.
348 * Revision 1.1 1995/09/01 14:06:35 adam
349 * Split of work into more files.
373 #include "zebrasdr.h"
376 static Dict matchDict;
378 static Records records = NULL;
379 static SortIdx sortIdx = NULL;
381 static char **key_buf;
382 static size_t ptr_top;
384 static size_t key_buf_used;
385 static int key_file_no;
387 static int records_inserted = 0;
388 static int records_updated = 0;
389 static int records_deleted = 0;
390 static int records_processed = 0;
392 static ZebraExplainInfo zti = NULL;
394 static void logRecord (int showFlag)
398 if (showFlag || !(records_processed % 1000))
400 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
401 records_processed, records_inserted, records_updated,
406 static int explain_extract (void *handle, Record drec, data1_node *n);
408 int key_open (struct recordGroup *rGroup, int mem)
410 BFiles bfs = rGroup->bfs;
411 int rw = rGroup->flagRw;
412 data1_handle dh = rGroup->dh;
413 char *recordCompression;
414 int record_compression = REC_COMPRESS_NONE;
416 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
419 key_buf = (char **) xmalloc (mem);
420 ptr_top = mem/sizeof(char*);
426 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
428 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
432 recordCompression = res_get_def (common_resource,
433 "recordCompression", "none");
434 if (!strcmp (recordCompression, "none"))
435 record_compression = REC_COMPRESS_NONE;
436 if (!strcmp (recordCompression, "bzip2"))
437 record_compression = REC_COMPRESS_BZIP2;
438 records = rec_open (bfs, rw, record_compression);
441 dict_close (matchDict);
444 zti = zebraExplain_open (records, dh, common_resource,
445 rw, rGroup, explain_extract);
448 rec_close (&records);
449 dict_close (matchDict);
452 sortIdx = sortIdx_open (bfs, 1);
463 void encode_key_init (struct encode_info *i)
470 char *encode_key_int (int d, char *bp)
479 else if (d <= 4194303)
481 *bp++ = 128 + (d>>16);
482 *bp++ = (d>>8) & 255;
487 *bp++ = 192 + (d>>24);
488 *bp++ = (d>>16) & 255;
489 *bp++ = (d>>8) & 255;
495 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
500 while ((*bp++ = *k++))
502 memcpy (&key, k+1, sizeof(struct it_key));
503 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
504 if (i->sysno != key.sysno)
506 i->sysno = key.sysno;
509 else if (!i->seqno && !key.seqno && i->cmd == *k)
511 bp = encode_key_int (key.seqno - i->seqno, bp);
512 i->seqno = key.seqno;
514 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
516 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
524 static int key_y_len;
526 static int key_y_compare (const void *p1, const void *p2)
530 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
531 *(char**) p2 + key_y_len + 1)))
533 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
536 static int key_x_compare (const void *p1, const void *p2)
538 return strcmp (*(char**) p1, *(char**) p2);
542 void key_flush (void)
547 struct encode_info encode_info;
556 logf (LOG_LOG, "sorting section %d", key_file_no);
558 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
559 getFnameTmp (out_fname, key_file_no);
561 if (!(outf = fopen (out_fname, "wb")))
563 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
566 logf (LOG_LOG, "writing section %d", key_file_no);
567 prevcp = cp = key_buf[ptr_top-ptr_i];
569 encode_key_init (&encode_info);
570 encode_key_write (cp, &encode_info, outf);
573 cp = key_buf[ptr_top-ptr_i];
574 if (strcmp (cp, prevcp))
576 encode_key_init (&encode_info);
577 encode_key_write (cp, &encode_info, outf);
581 encode_key_write (cp + strlen(cp), &encode_info, outf);
584 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
585 getFnameTmp (out_fname, key_file_no);
587 if (!(outf = fopen (out_fname, "wb")))
589 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
592 logf (LOG_LOG, "writing section %d", key_file_no);
594 prevcp = key_buf[ptr_top-i];
596 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
598 key_y_len = strlen(prevcp)+1;
600 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
601 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
603 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
604 sizeof(char*), key_y_compare);
605 cp = key_buf[ptr_top-ptr_i];
607 encode_key_init (&encode_info);
608 encode_key_write (cp, &encode_info, outf);
611 cp = key_buf[ptr_top-ptr_i];
612 encode_key_write (cp+key_y_len, &encode_info, outf);
616 prevcp = key_buf[ptr_top-ptr_i];
621 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
624 logf (LOG_LOG, "finished section %d", key_file_no);
629 int key_close (struct recordGroup *rGroup)
631 int rw = rGroup->flagRw;
633 zebraExplain_runNumberIncrement (zti, 1);
634 zebraExplain_close (zti, rw, 0);
637 rec_close (&records);
638 dict_close (matchDict);
639 sortIdx_close (sortIdx);
645 static void wordInit (struct recExtractCtrl *p, RecWord *w)
647 w->zebra_maps = p->zebra_maps;
648 w->seqnos = p->seqno;
649 w->attrSet = VAL_BIB1;
654 static struct sortKey {
659 struct sortKey *next;
662 static struct recKeys {
671 static void addIndexString (RecWord *p, const char *string, int length)
674 unsigned char attrSet;
675 unsigned short attrUse;
678 int *pseqno = &p->seqnos[p->reg_type];
680 if (reckeys.buf_used+1024 > reckeys.buf_max)
684 b = (char *) xmalloc (reckeys.buf_max += 128000);
685 if (reckeys.buf_used > 0)
686 memcpy (b, reckeys.buf, reckeys.buf_used);
690 dst = reckeys.buf + reckeys.buf_used;
692 attrSet = p->attrSet;
693 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
696 reckeys.prevAttrSet = attrSet;
697 attrUse = p->attrUse;
698 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
701 reckeys.prevAttrUse = attrUse;
703 diff = 1 + *pseqno - reckeys.prevSeqNo;
704 if (diff >= 1 && diff <= 15)
709 reckeys.prevSeqNo = *pseqno;
715 memcpy (dst, &attrSet, sizeof(attrSet));
716 dst += sizeof(attrSet);
720 memcpy (dst, &attrUse, sizeof(attrUse));
721 dst += sizeof(attrUse);
723 *dst++ = p->reg_type;
724 memcpy (dst, string, length);
730 memcpy (dst, pseqno, sizeof(*pseqno));
731 dst += sizeof(*pseqno);
733 reckeys.buf_used = dst - reckeys.buf;
738 static void addSortString (RecWord *p, const char *string, int length)
742 for (sk = sortKeys; sk; sk = sk->next)
743 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
746 sk = (struct sortKey *) xmalloc (sizeof(*sk));
750 sk->string = (char *) xmalloc (length);
752 memcpy (sk->string, string, length);
754 sk->attrSet = p->attrSet;
755 sk->attrUse = p->attrUse;
758 static void addString (RecWord *p, const char *string, int length)
761 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
762 addSortString (p, string, length);
764 addIndexString (p, string, length);
767 static void addIncompleteField (RecWord *p)
769 const char *b = p->string;
770 int remain = p->length;
771 const char **map = 0;
774 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
778 char buf[IT_MAX_WORD+1];
782 while (map && *map && **map == *CHR_SPACE)
784 remain = p->length - (b - p->string);
786 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
793 while (map && *map && **map != *CHR_SPACE)
795 const char *cp = *map;
797 while (i < IT_MAX_WORD && *cp)
799 remain = p->length - (b - p->string);
801 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
807 addString (p, buf, i);
811 static void addCompleteField (RecWord *p)
813 const char *b = p->string;
814 char buf[IT_MAX_WORD+1];
815 const char **map = 0;
816 int i = 0, remain = p->length;
819 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
821 while (remain > 0 && i < IT_MAX_WORD)
823 while (map && *map && **map == *CHR_SPACE)
825 remain = p->length - (b - p->string);
827 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
834 if (i && i < IT_MAX_WORD)
835 buf[i++] = *CHR_SPACE;
836 while (map && *map && **map != *CHR_SPACE)
838 const char *cp = *map;
840 if (i >= IT_MAX_WORD)
842 while (i < IT_MAX_WORD && *cp)
844 remain = p->length - (b - p->string);
846 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
854 addString (p, buf, i);
857 static void addRecordKey (RecWord *p)
859 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
860 addCompleteField (p);
862 addIncompleteField(p);
865 static void flushSortKeys (SYSNO sysno, int cmd)
867 struct sortKey *sk = sortKeys;
869 sortIdx_sysno (sortIdx, sysno);
872 struct sortKey *sk_next = sk->next;
873 sortIdx_type (sortIdx, sk->attrUse);
874 sortIdx_add (sortIdx, sk->string, sk->length);
882 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
884 unsigned char attrSet = (unsigned char) -1;
885 unsigned short attrUse = (unsigned short) -1;
889 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
890 while (off < reckeys->buf_used)
892 const char *src = reckeys->buf + off;
900 memcpy (&attrSet, src, sizeof(attrSet));
901 src += sizeof(attrSet);
905 memcpy (&attrUse, src, sizeof(attrUse));
906 src += sizeof(attrUse);
908 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
911 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
913 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
915 ch = zebraExplain_addSU (zti, attrSet, attrUse);
917 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
920 ((char*)key_buf) [key_buf_used++] = *src++;
922 ((char*)key_buf) [key_buf_used++] = '\0';
923 ((char*) key_buf)[key_buf_used++] = cmd;
926 seqno += ((lead>>2) & 15)-1;
929 memcpy (&seqno, src, sizeof(seqno));
930 src += sizeof(seqno);
934 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
935 key_buf_used += sizeof(key);
936 off = src - reckeys->buf;
938 assert (off == reckeys->buf_used);
941 static const char **searchRecordKey (struct recKeys *reckeys,
942 int attrSetS, int attrUseS)
944 static const char *ws[32];
952 for (i = 0; i<32; i++)
955 while (off < reckeys->buf_used)
958 const char *src = reckeys->buf + off;
966 memcpy (&attrSet, src, sizeof(attrSet));
967 src += sizeof(attrSet);
971 memcpy (&attrUse, src, sizeof(attrUse));
972 src += sizeof(attrUse);
978 seqno += ((lead>>2) & 15)-1;
981 memcpy (&seqno, src, sizeof(seqno));
982 src += sizeof(seqno);
984 if (attrUseS == attrUse && attrSetS == attrSet)
991 woff = seqno - startSeq;
992 if (woff >= 0 && woff < 31)
996 off = src - reckeys->buf;
998 assert (off == reckeys->buf_used);
1002 struct file_read_info {
1003 off_t file_max; /* maximum offset so far */
1004 off_t file_offset; /* current offset */
1005 off_t file_moffset; /* offset of rec/rec boundary */
1012 static struct file_read_info *file_read_start (int fd)
1014 struct file_read_info *fi = (struct file_read_info *)
1015 xmalloc (sizeof(*fi));
1019 fi->file_moffset = 0;
1025 static void file_read_stop (struct file_read_info *fi)
1030 static off_t file_seek (void *handle, off_t offset)
1032 struct file_read_info *p = (struct file_read_info *) handle;
1033 p->file_offset = offset;
1036 return lseek (p->fd, offset, SEEK_SET);
1039 static off_t file_tell (void *handle)
1041 struct file_read_info *p = (struct file_read_info *) handle;
1042 return p->file_offset;
1045 static int file_read (void *handle, char *buf, size_t count)
1047 struct file_read_info *p = (struct file_read_info *) handle;
1053 if (r > p->sdrmax - p->file_offset)
1054 r = p->sdrmax - p->file_offset;
1056 memcpy (buf, p->sdrbuf + p->file_offset, r);
1059 r = read (fd, buf, count);
1062 p->file_offset += r;
1063 if (p->file_offset > p->file_max)
1064 p->file_max = p->file_offset;
1069 static void file_begin (void *handle)
1071 struct file_read_info *p = (struct file_read_info *) handle;
1073 p->file_offset = p->file_moffset;
1074 if (!p->sdrbuf && p->file_moffset)
1075 lseek (p->fd, p->file_moffset, SEEK_SET);
1079 static void file_end (void *handle, off_t offset)
1081 struct file_read_info *p = (struct file_read_info *) handle;
1083 assert (p->file_more == 0);
1085 p->file_moffset = offset;
1088 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1089 const char *fname, const char *spec)
1091 static char dstBuf[2048];
1093 const char *s = spec;
1094 static const char **w;
1098 while (*s == ' ' || *s == '\t')
1104 char attset_str[64], attname_str[64];
1105 data1_attset *attset;
1108 int attSet = 1, attUse = 1;
1112 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1114 attset_str[i++] = *s;
1115 attset_str[i] = '\0';
1120 for (i = 0; *s && *s != ')'; s++)
1122 attname_str[i++] = *s;
1123 attname_str[i] = '\0';
1126 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1129 attSet = attset->reference;
1130 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1132 attUse = att->value;
1134 attUse = atoi (attname_str);
1136 w = searchRecordKey (reckeys, attSet, attUse);
1141 for (i = 0; i<32; i++)
1146 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1147 spec, rGroup->groupName ? rGroup->groupName : "none");
1152 for (i = 0; i<32; i++)
1153 if (matchFlag[i] && w[i])
1161 dst += strlen(w[i]);
1165 logf (LOG_WARN, "Record didn't contain match"
1166 " fields in (%s,%s)", attset_str, attname_str);
1174 const char *spec_src = NULL;
1175 const char *s1 = ++s;
1176 while (*s1 && *s1 != ' ' && *s1 != '\t')
1182 memcpy (special, s, spec_len);
1183 special[spec_len] = '\0';
1186 if (!strcmp (special, "group"))
1187 spec_src = rGroup->groupName;
1188 else if (!strcmp (special, "database"))
1189 spec_src = rGroup->databaseName;
1190 else if (!strcmp (special, "filename"))
1192 else if (!strcmp (special, "type"))
1193 spec_src = rGroup->recordType;
1198 strcpy (dst, spec_src);
1199 dst += strlen (spec_src);
1202 else if (*s == '\"' || *s == '\'')
1204 int stopMarker = *s++;
1208 while (*s && *s != stopMarker)
1211 tmpString[i++] = *s++;
1215 tmpString[i] = '\0';
1216 strcpy (dst, tmpString);
1217 dst += strlen (tmpString);
1221 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1222 spec, rGroup->groupName ? rGroup->groupName : "none");
1229 logf (LOG_WARN, "No match criteria for record %s in group %s",
1230 fname, rGroup->groupName ? rGroup->groupName : "none");
1237 struct recordLogInfo {
1240 struct recordGroup *rGroup;
1243 static void recordLogPreamble (int level, const char *msg, void *info)
1245 struct recordLogInfo *p = (struct recordLogInfo *) info;
1246 FILE *outf = log_file ();
1248 if (level & LOG_LOG)
1250 fprintf (outf, "File %s, offset %d, type %s\n",
1251 p->rGroup->recordType, p->recordOffset, p->fname);
1252 log_event_start (NULL, NULL);
1255 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1257 zebraExplain_addSchema (zti, oid);
1260 static int recordExtract (SYSNO *sysno, const char *fname,
1261 struct recordGroup *rGroup, int deleteFlag,
1262 struct file_read_info *fi,
1263 RecType recType, char *subType, void *clientData)
1265 RecordAttr *recordAttr;
1270 struct recordLogInfo logInfo;
1271 off_t recordOffset = 0;
1275 struct recExtractCtrl extractCtrl;
1277 /* we are going to read from a file, so prepare the extraction */
1280 reckeys.buf_used = 0;
1281 reckeys.prevAttrUse = -1;
1282 reckeys.prevAttrSet = -1;
1283 reckeys.prevSeqNo = 0;
1285 recordOffset = fi->file_moffset;
1286 extractCtrl.offset = fi->file_moffset;
1287 extractCtrl.readf = file_read;
1288 extractCtrl.seekf = file_seek;
1289 extractCtrl.tellf = file_tell;
1290 extractCtrl.endf = file_end;
1291 extractCtrl.fh = fi;
1292 extractCtrl.subType = subType;
1293 extractCtrl.init = wordInit;
1294 extractCtrl.addWord = addRecordKey;
1295 extractCtrl.addSchema = addSchema;
1296 extractCtrl.dh = rGroup->dh;
1297 for (i = 0; i<256; i++)
1299 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1300 extractCtrl.seqno[i] = 1;
1302 extractCtrl.seqno[i] = 0;
1304 extractCtrl.zebra_maps = rGroup->zebra_maps;
1305 extractCtrl.flagShowRecords = !rGroup->flagRw;
1307 if (!rGroup->flagRw)
1308 printf ("File: %s %ld\n", fname, (long) recordOffset);
1310 logInfo.fname = fname;
1311 logInfo.recordOffset = recordOffset;
1312 logInfo.rGroup = rGroup;
1313 log_event_start (recordLogPreamble, &logInfo);
1315 r = (*recType->extract)(clientData, &extractCtrl);
1317 log_event_start (NULL, NULL);
1319 if (r == RECCTRL_EXTRACT_EOF)
1321 else if (r == RECCTRL_EXTRACT_ERROR)
1323 /* error occured during extraction ... */
1324 if (rGroup->flagRw &&
1325 records_processed < rGroup->fileVerboseLimit)
1327 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1328 fname, (long) recordOffset);
1332 if (reckeys.buf_used == 0)
1334 /* the extraction process returned no information - the record
1335 is probably empty - unless flagShowRecords is in use */
1336 if (!rGroup->flagRw)
1338 logf (LOG_WARN, "No keys generated for file %s", fname);
1339 logf (LOG_WARN, " The file is probably empty");
1344 /* perform match if sysno not known and if match criteria is specified */
1351 if (rGroup->recordId && *rGroup->recordId)
1355 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1359 rinfo = dict_lookup (matchDict, matchStr);
1361 memcpy (sysno, rinfo+1, sizeof(*sysno));
1365 logf (LOG_WARN, "Bad match criteria");
1376 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1377 fname, (long) recordOffset);
1378 logf (LOG_WARN, "cannot delete record above (seems new)");
1381 if (records_processed < rGroup->fileVerboseLimit)
1382 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1383 fname, (long) recordOffset);
1384 rec = rec_new (records);
1386 *sysno = rec->sysno;
1388 recordAttr = rec_init_attr (zti, rec);
1392 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1394 flushRecordKeys (*sysno, 1, &reckeys);
1395 flushSortKeys (*sysno, 1);
1401 /* record already exists */
1402 struct recKeys delkeys;
1404 rec = rec_get (records, *sysno);
1407 recordAttr = rec_init_attr (zti, rec);
1409 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1411 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1412 fname, (long) recordOffset);
1417 delkeys.buf_used = rec->size[recInfo_delKeys];
1418 delkeys.buf = rec->info[recInfo_delKeys];
1419 flushSortKeys (*sysno, 0);
1420 flushRecordKeys (*sysno, 0, &delkeys);
1423 /* record going to be deleted */
1424 if (!delkeys.buf_used)
1426 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1427 fname, (long) recordOffset);
1428 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1432 if (records_processed < rGroup->fileVerboseLimit)
1433 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1434 fname, (long) recordOffset);
1437 dict_delete (matchDict, matchStr);
1438 rec_del (records, &rec);
1446 /* record going to be updated */
1447 if (!delkeys.buf_used)
1449 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1450 fname, (long) recordOffset);
1451 logf (LOG_WARN, "cannot update file above, storeKeys false");
1455 if (records_processed < rGroup->fileVerboseLimit)
1456 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1457 fname, (long) recordOffset);
1458 flushRecordKeys (*sysno, 1, &reckeys);
1463 /* update file type */
1464 xfree (rec->info[recInfo_fileType]);
1465 rec->info[recInfo_fileType] =
1466 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1468 /* update filename */
1469 xfree (rec->info[recInfo_filename]);
1470 rec->info[recInfo_filename] =
1471 rec_strdup (fname, &rec->size[recInfo_filename]);
1473 /* update delete keys */
1474 xfree (rec->info[recInfo_delKeys]);
1475 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1478 rec->size[recInfo_delKeys] = reckeys.buf_used;
1479 rec->info[recInfo_delKeys] = reckeys.buf;
1481 reckeys.buf_max = 0;
1483 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1484 rec->size[recInfo_delKeys] = reckeys.buf_used;
1485 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1486 rec->size[recInfo_delKeys]);
1491 rec->info[recInfo_delKeys] = NULL;
1492 rec->size[recInfo_delKeys] = 0;
1495 /* save file size of original record */
1496 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1497 recordAttr->recordSize = fi->file_moffset - recordOffset;
1498 if (!recordAttr->recordSize)
1499 recordAttr->recordSize = fi->file_max - recordOffset;
1500 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1502 /* set run-number for this record */
1503 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1505 /* update store data */
1506 xfree (rec->info[recInfo_storeData]);
1507 if (rGroup->flagStoreData == 1)
1509 rec->size[recInfo_storeData] = recordAttr->recordSize;
1510 rec->info[recInfo_storeData] = (char *)
1511 xmalloc (recordAttr->recordSize);
1512 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1514 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1515 (long) recordOffset, fname);
1518 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1519 < recordAttr->recordSize)
1521 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1522 recordAttr->recordSize, fname);
1528 rec->info[recInfo_storeData] = NULL;
1529 rec->size[recInfo_storeData] = 0;
1531 /* update database name */
1532 xfree (rec->info[recInfo_databaseName]);
1533 rec->info[recInfo_databaseName] =
1534 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1537 recordAttr->recordOffset = recordOffset;
1539 /* commit this record */
1540 rec_put (records, &rec);
1545 int fileExtract (SYSNO *sysno, const char *fname,
1546 const struct recordGroup *rGroupP, int deleteFlag)
1554 struct recordGroup rGroupM;
1555 struct recordGroup *rGroup = &rGroupM;
1556 struct file_read_info *fi;
1559 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1561 if (!rGroup->groupName || !*rGroup->groupName)
1564 sprintf (gprefix, "%s.", rGroup->groupName);
1566 logf (LOG_DEBUG, "fileExtract %s", fname);
1568 /* determine file extension */
1570 for (i = strlen(fname); --i >= 0; )
1571 if (fname[i] == '/')
1573 else if (fname[i] == '.')
1575 strcpy (ext, fname+i+1);
1578 /* determine file type - depending on extension */
1579 if (!rGroup->recordType)
1581 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1582 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1584 sprintf (ext_res, "%srecordType", gprefix);
1585 rGroup->recordType = res_get (common_resource, ext_res);
1588 if (!rGroup->recordType)
1590 if (records_processed < rGroup->fileVerboseLimit)
1591 logf (LOG_LOG, "? %s", fname);
1594 if (!*rGroup->recordType)
1597 recType_byName (rGroup->recTypes, rGroup->recordType, subType,
1600 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1604 /* determine match criteria */
1605 if (!rGroup->recordId)
1607 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1608 rGroup->recordId = res_get (common_resource, ext_res);
1611 /* determine database name */
1612 if (!rGroup->databaseName)
1614 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1615 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1617 sprintf (ext_res, "%sdatabase", gprefix);
1618 rGroup->databaseName = res_get (common_resource, ext_res);
1621 if (!rGroup->databaseName)
1622 rGroup->databaseName = "Default";
1624 /* determine if explain database */
1626 sprintf (ext_res, "%sexplainDatabase", gprefix);
1627 rGroup->explainDatabase =
1628 atoi (res_get_def (common_resource, ext_res, "0"));
1630 /* announce database */
1631 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1633 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1634 rGroup->explainDatabase))
1638 if (rGroup->flagStoreData == -1)
1641 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1642 if (!(sval = res_get (common_resource, ext_res)))
1644 sprintf (ext_res, "%sstoreData", gprefix);
1645 sval = res_get (common_resource, ext_res);
1648 rGroup->flagStoreData = atoi (sval);
1650 if (rGroup->flagStoreData == -1)
1651 rGroup->flagStoreData = 0;
1653 if (rGroup->flagStoreKeys == -1)
1657 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1658 if (!(sval = res_get (common_resource, ext_res)))
1660 sprintf (ext_res, "%sstoreKeys", gprefix);
1661 sval = res_get (common_resource, ext_res);
1664 rGroup->flagStoreKeys = atoi (sval);
1666 if (rGroup->flagStoreKeys == -1)
1667 rGroup->flagStoreKeys = 0;
1673 char xname[128], *xp;
1675 strncpy (xname, fname, 127);
1676 if (!(xp = strchr (xname, '.')))
1679 if (strcmp (xp+1, "sdr.bits"))
1682 h = zebraSdr_open (xname);
1685 logf (LOG_WARN, "sdr open %s", xname);
1694 segmentno = zebraSdr_segment (h, 0);
1695 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1701 r = zebraSdr_read (h, &buf);
1706 fi = file_read_start (0);
1712 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1714 } while (r && !sysno && fi->file_more);
1715 file_read_stop (fi);
1722 if (sysno && deleteFlag)
1726 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1728 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1732 fi = file_read_start (fd);
1736 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1737 recType, subType, clientData);
1738 } while (r && !sysno && fi->file_more);
1739 file_read_stop (fi);
1745 static int explain_extract (void *handle, Record rec, data1_node *n)
1747 struct recordGroup *rGroup = (struct recordGroup*) handle;
1748 struct recExtractCtrl extractCtrl;
1751 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1753 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1757 reckeys.buf_used = 0;
1758 reckeys.prevAttrUse = -1;
1759 reckeys.prevAttrSet = -1;
1760 reckeys.prevSeqNo = 0;
1762 extractCtrl.init = wordInit;
1763 extractCtrl.addWord = addRecordKey;
1764 extractCtrl.addSchema = addSchema;
1765 extractCtrl.dh = rGroup->dh;
1766 for (i = 0; i<256; i++)
1767 extractCtrl.seqno[i] = 0;
1768 extractCtrl.zebra_maps = rGroup->zebra_maps;
1769 extractCtrl.flagShowRecords = !rGroup->flagRw;
1771 grs_extract_tree(&extractCtrl, n);
1773 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1775 if (rec->size[recInfo_delKeys])
1777 struct recKeys delkeys;
1779 delkeys.buf_used = rec->size[recInfo_delKeys];
1780 delkeys.buf = rec->info[recInfo_delKeys];
1781 flushSortKeys (rec->sysno, 0);
1782 flushRecordKeys (rec->sysno, 0, &delkeys);
1784 flushRecordKeys (rec->sysno, 1, &reckeys);
1785 flushSortKeys (rec->sysno, 1);
1787 xfree (rec->info[recInfo_delKeys]);
1788 rec->size[recInfo_delKeys] = reckeys.buf_used;
1789 rec->info[recInfo_delKeys] = reckeys.buf;
1791 reckeys.buf_max = 0;