2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.89 1998-10-28 10:54:38 adam
10 * Revision 1.88 1998/10/16 08:14:29 adam
11 * Updated record control system.
13 * Revision 1.87 1998/10/15 13:10:33 adam
14 * Fixed bug in Zebra that caused it to stop indexing when empty
17 * Revision 1.86 1998/10/13 20:33:53 adam
18 * Fixed one log message and change use ordinal to be an unsigned char.
20 * Revision 1.85 1998/09/22 10:03:41 adam
21 * Changed result sets to be persistent in the sense that they can
22 * be re-searched if needed.
23 * Fixed memory leak in rsm_or.
25 * Revision 1.84 1998/06/11 15:42:22 adam
26 * Changed the way use attributes are specified in the recordId
29 * Revision 1.83 1998/06/08 14:43:10 adam
30 * Added suport for EXPLAIN Proxy servers - added settings databasePath
31 * and explainDatabase to facilitate this. Increased maximum number
32 * of databases and attributes in one register.
34 * Revision 1.82 1998/05/20 10:12:15 adam
35 * Implemented automatic EXPLAIN database maintenance.
36 * Modified Zebra to work with ASN.1 compiled version of YAZ.
38 * Revision 1.81 1998/03/11 11:19:04 adam
39 * Changed the way sequence numbers are generated.
41 * Revision 1.80 1998/03/05 08:45:11 adam
42 * New result set model and modular ranking system. Moved towards
43 * descent server API. System information stored as "SGML" records.
45 * Revision 1.79 1998/02/17 10:32:52 adam
46 * Fixed bug: binary files weren't opened with flag b on NT.
48 * Revision 1.78 1998/02/10 12:03:05 adam
51 * Revision 1.77 1998/01/12 15:04:08 adam
52 * The test option (-s) only uses read-lock (and not write lock).
54 * Revision 1.76 1997/10/27 14:33:04 adam
55 * Moved towards generic character mapping depending on "structure"
56 * field in abstract syntax file. Fixed a few memory leaks. Fixed
57 * bug with negative integers when doing searches with relational
60 * Revision 1.75 1997/09/17 12:19:12 adam
61 * Zebra version corresponds to YAZ version 1.4.
62 * Changed Zebra server so that it doesn't depend on global common_resource.
64 * Revision 1.74 1997/09/09 13:38:06 adam
65 * Partial port to WIN95/NT.
67 * Revision 1.73 1997/09/04 13:57:20 adam
68 * New file extract/retrieve method tellf (added).
69 * Added O_BINARY for open calls.
71 * Revision 1.72 1997/07/15 16:32:29 adam
72 * Bug fix: Match handler didn't terminate the resulting string!
74 * Revision 1.71 1997/07/15 16:28:41 adam
75 * Bug fix: storeData didn't work with files with multiple records.
76 * Bug fix: fixed memory management with records; not really well
79 * Revision 1.70 1997/07/01 13:00:42 adam
80 * Bug fix in routine searchRecordKey: uninitialized variables.
82 * Revision 1.69 1997/04/29 09:26:03 adam
83 * Bug fix: generic recordId handling didn't work for compressed internal
86 * Revision 1.68 1997/02/12 20:39:45 adam
87 * Implemented options -f <n> that limits the log to the first <n>
89 * Changed some log messages also.
91 * Revision 1.67 1996/11/15 15:02:14 adam
92 * Minor changes regarding logging.
94 * Revision 1.66 1996/11/14 09:52:21 adam
95 * Strings in record keys bound by IT_MAX_WORD.
97 * Revision 1.65 1996/11/14 08:57:56 adam
98 * Reduction of storeKeys area.
100 * Revision 1.64 1996/11/08 11:10:16 adam
101 * Buffers used during file match got bigger.
102 * Compressed ISAM support everywhere.
103 * Bug fixes regarding masking characters in queries.
104 * Redesigned Regexp-2 queries.
106 * Revision 1.63 1996/10/29 14:09:39 adam
107 * Use of cisam system - enabled if setting isamc is 1.
109 * Revision 1.62 1996/10/11 10:57:01 adam
110 * New module recctrl. Used to manage records (extract/retrieval).
111 * Several files have been moved to the recctrl sub directory.
113 * Revision 1.61 1996/06/06 12:08:37 quinn
114 * Added showRecord function
116 * Revision 1.60 1996/06/04 10:18:12 adam
117 * Search/scan uses character mapping module.
119 * Revision 1.59 1996/05/14 15:47:07 adam
120 * Cleanup of various buffer size entities.
122 * Revision 1.58 1996/05/14 06:16:38 adam
123 * Compact use/set bytes used in search service.
125 * Revision 1.57 1996/05/13 14:23:04 adam
126 * Work on compaction of set/use bytes in dictionary.
128 * Revision 1.56 1996/05/09 09:54:42 adam
129 * Server supports maps from one logical attributes to a list of physical
131 * The extraction process doesn't make space consuming 'any' keys.
133 * Revision 1.55 1996/05/09 07:28:55 quinn
134 * Work towards phrases and multiple registers
136 * Revision 1.54 1996/05/01 13:46:35 adam
137 * First work on multiple records in one file.
138 * New option, -offset, to the "unread" command in the filter module.
140 * Revision 1.53 1996/04/26 12:09:43 adam
141 * Added a few comments.
143 * Revision 1.52 1996/04/25 13:27:57 adam
144 * Function recordExtract modified so that files with no keys (possibly empty)
147 * Revision 1.51 1996/03/19 11:08:42 adam
148 * Bug fix: Log preamble wasn't always turned off after recordExtract.
150 * Revision 1.50 1996/02/12 18:45:36 adam
151 * New fileVerboseFlag in record group control.
153 * Revision 1.49 1996/02/05 12:29:57 adam
154 * Logging reduced a bit.
155 * The remaining running time is estimated during register merge.
157 * Revision 1.48 1996/02/01 20:53:26 adam
158 * The temporary per-record keys are compacted a little, and duplication
159 * of the per-records keys are avoided when they are saved in the record
160 * information buffer.
162 * Revision 1.47 1996/01/17 14:57:48 adam
163 * Prototype changed for reader functions in extract/retrieve. File
164 * is identified by 'void *' instead of 'int.
166 * Revision 1.46 1995/12/15 14:57:16 adam
169 * Revision 1.45 1995/12/15 12:37:41 adam
170 * In addRecordKeyAny: Writes key only when attrSet != -1.
172 * Revision 1.44 1995/12/12 16:00:54 adam
173 * System call sync(2) used after update/commit.
174 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
177 * Revision 1.43 1995/12/11 09:12:46 adam
178 * The rec_get function returns NULL if record doesn't exist - will
179 * happen in the server if the result set records have been deleted since
180 * the creation of the set (i.e. the search).
181 * The server saves a result temporarily if it is 'volatile', i.e. the
182 * set is register dependent.
184 * Revision 1.42 1995/12/07 17:38:46 adam
185 * Work locking mechanisms for concurrent updates/commit.
187 * Revision 1.41 1995/12/06 16:06:42 adam
188 * Better diagnostics. Work on 'real' dictionary deletion.
190 * Revision 1.40 1995/12/05 16:57:40 adam
191 * More work on regular patterns.
193 * Revision 1.39 1995/12/05 13:20:18 adam
194 * Bug fix: file_read sometimes returned early EOF.
196 * Revision 1.38 1995/12/04 17:59:21 adam
197 * More work on regular expression conversion.
199 * Revision 1.37 1995/12/04 14:22:27 adam
200 * Extra arg to recType_byName.
201 * Started work on new regular expression parsed input to
202 * structured records.
204 * Revision 1.36 1995/11/30 08:34:29 adam
205 * Started work on commit facility.
206 * Changed a few malloc/free to xmalloc/xfree.
208 * Revision 1.35 1995/11/28 14:26:21 adam
209 * Bug fix: recordId with constant wasn't right.
210 * Bug fix: recordId dictionary entry wasn't deleted when needed.
212 * Revision 1.34 1995/11/28 09:09:38 adam
213 * Zebra config renamed.
214 * Use setting 'recordId' to identify record now.
215 * Bug fix in recindex.c: rec_release_blocks was invokeded even
216 * though the blocks were already released.
217 * File traversal properly deletes records when needed.
219 * Revision 1.33 1995/11/27 09:56:20 adam
220 * Record info elements better enumerated. Internal store of records.
222 * Revision 1.32 1995/11/25 10:24:05 adam
223 * More record fields - they are enumerated now.
224 * New options: flagStoreData flagStoreKey.
226 * Revision 1.31 1995/11/24 11:31:35 adam
227 * Commands add & del read filenames from stdin if source directory is
229 * Match criteria supports 'constant' strings.
231 * Revision 1.30 1995/11/22 17:19:16 adam
232 * Record management uses the bfile system.
234 * Revision 1.29 1995/11/21 15:01:14 adam
235 * New general match criteria implemented.
236 * New feature: document groups.
238 * Revision 1.28 1995/11/21 09:20:30 adam
239 * Yet more work on record match.
241 * Revision 1.27 1995/11/20 16:59:45 adam
242 * New update method: the 'old' keys are saved for each records.
244 * Revision 1.26 1995/11/20 11:56:24 adam
245 * Work on new traversal.
247 * Revision 1.25 1995/11/16 15:34:54 adam
248 * Uses new record management system in both indexer and server.
250 * Revision 1.24 1995/11/15 19:13:08 adam
251 * Work on record management.
253 * Revision 1.23 1995/10/27 14:00:10 adam
254 * Implemented detection of database availability.
256 * Revision 1.22 1995/10/17 18:02:07 adam
257 * New feature: databases. Implemented as prefix to words in dictionary.
259 * Revision 1.21 1995/10/10 12:24:38 adam
260 * Temporary sort files are compressed.
262 * Revision 1.20 1995/10/06 13:52:05 adam
263 * Bug fixes. Handler may abort further scanning.
265 * Revision 1.19 1995/10/04 12:55:16 adam
266 * Bug fix in ranked search. Use=Any keys inserted.
268 * Revision 1.18 1995/10/04 09:37:08 quinn
271 * Revision 1.17 1995/10/03 14:28:57 adam
272 * Buffered read in extract works.
274 * Revision 1.16 1995/10/03 14:28:45 adam
275 * Work on more effecient read handler in extract.
277 * Revision 1.15 1995/10/02 15:42:53 adam
278 * Extract uses file descriptors instead of FILE pointers.
280 * Revision 1.14 1995/10/02 15:29:13 adam
281 * More logging in file_extract.
283 * Revision 1.13 1995/09/29 14:01:39 adam
286 * Revision 1.12 1995/09/28 14:22:56 adam
287 * Sort uses smaller temporary files.
289 * Revision 1.11 1995/09/28 12:10:31 adam
290 * Bug fixes. Field prefix used in queries.
292 * Revision 1.10 1995/09/28 09:19:41 adam
293 * xfree/xmalloc used everywhere.
294 * Extract/retrieve method seems to work for text records.
296 * Revision 1.9 1995/09/27 12:22:28 adam
297 * More work on extract in record control.
298 * Field name is not in isam keys but in prefix in dictionary words.
300 * Revision 1.8 1995/09/14 07:48:22 adam
301 * Record control management.
303 * Revision 1.7 1995/09/11 13:09:32 adam
304 * More work on relevance feedback.
306 * Revision 1.6 1995/09/08 14:52:27 adam
307 * Minor changes. Dictionary is lower case now.
309 * Revision 1.5 1995/09/06 16:11:16 adam
310 * Option: only one word key per file.
312 * Revision 1.4 1995/09/05 15:28:39 adam
313 * More work on search engine.
315 * Revision 1.3 1995/09/04 12:33:41 adam
316 * Various cleanup. YAZ util used instead.
318 * Revision 1.2 1995/09/04 09:10:34 adam
319 * More work on index add/del/update.
320 * Merge sort implemented.
321 * Initial work on z39 server.
323 * Revision 1.1 1995/09/01 14:06:35 adam
324 * Split of work into more files.
348 #include "zebrasdr.h"
351 static Dict matchDict;
353 static Records records = NULL;
354 static SortIdx sortIdx = NULL;
356 static char **key_buf;
357 static size_t ptr_top;
359 static size_t key_buf_used;
360 static int key_file_no;
362 static int records_inserted = 0;
363 static int records_updated = 0;
364 static int records_deleted = 0;
365 static int records_processed = 0;
367 static ZebraExplainInfo zti = NULL;
369 static void logRecord (int showFlag)
373 if (showFlag || !(records_processed % 1000))
375 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
376 records_processed, records_inserted, records_updated,
381 static int explain_extract (void *handle, Record drec, data1_node *n);
383 int key_open (struct recordGroup *rGroup, int mem)
385 BFiles bfs = rGroup->bfs;
386 int rw = rGroup->flagRw;
387 data1_handle dh = rGroup->dh;
389 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
392 key_buf = xmalloc (mem);
393 ptr_top = mem/sizeof(char*);
399 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
401 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
405 records = rec_open (bfs, rw);
408 dict_close (matchDict);
411 zti = zebraExplain_open (records, dh, common_resource,
412 rw, rGroup, explain_extract);
415 rec_close (&records);
416 dict_close (matchDict);
419 sortIdx = sortIdx_open (bfs, 1);
429 void encode_key_init (struct encode_info *i)
435 char *encode_key_int (int d, char *bp)
444 else if (d <= 4194303)
446 *bp++ = 128 + (d>>16);
447 *bp++ = (d>>8) & 255;
452 *bp++ = 192 + (d>>24);
453 *bp++ = (d>>16) & 255;
454 *bp++ = (d>>8) & 255;
460 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
465 while ((*bp++ = *k++))
467 memcpy (&key, k+1, sizeof(struct it_key));
468 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
469 if (i->sysno != key.sysno)
471 i->sysno = key.sysno;
474 bp = encode_key_int (key.seqno - i->seqno, bp);
475 i->seqno = key.seqno;
476 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
478 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
486 static int key_y_len;
488 static int key_y_compare (const void *p1, const void *p2)
492 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
493 *(char**) p2 + key_y_len + 1)))
495 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
498 static int key_x_compare (const void *p1, const void *p2)
500 return strcmp (*(char**) p1, *(char**) p2);
504 void key_flush (void)
509 struct encode_info encode_info;
518 logf (LOG_LOG, "sorting section %d", key_file_no);
520 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
521 getFnameTmp (out_fname, key_file_no);
523 if (!(outf = fopen (out_fname, "wb")))
525 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
528 logf (LOG_LOG, "writing section %d", key_file_no);
529 prevcp = cp = key_buf[ptr_top-ptr_i];
531 encode_key_init (&encode_info);
532 encode_key_write (cp, &encode_info, outf);
535 cp = key_buf[ptr_top-ptr_i];
536 if (strcmp (cp, prevcp))
538 encode_key_init (&encode_info);
539 encode_key_write (cp, &encode_info, outf);
543 encode_key_write (cp + strlen(cp), &encode_info, outf);
546 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
547 getFnameTmp (out_fname, key_file_no);
549 if (!(outf = fopen (out_fname, "wb")))
551 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
554 logf (LOG_LOG, "writing section %d", key_file_no);
556 prevcp = key_buf[ptr_top-i];
558 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
560 key_y_len = strlen(prevcp)+1;
562 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
563 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
565 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
566 sizeof(char*), key_y_compare);
567 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 encode_key_write (cp+key_y_len, &encode_info, outf);
578 prevcp = key_buf[ptr_top-ptr_i];
583 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
586 logf (LOG_LOG, "finished section %d", key_file_no);
591 int key_close (struct recordGroup *rGroup)
593 int rw = rGroup->flagRw;
595 zebraExplain_runNumberIncrement (zti, 1);
596 zebraExplain_close (zti, rw, 0);
599 rec_close (&records);
600 dict_close (matchDict);
601 sortIdx_close (sortIdx);
607 static void wordInit (struct recExtractCtrl *p, RecWord *w)
609 w->zebra_maps = p->zebra_maps;
610 w->seqnos = p->seqno;
611 w->attrSet = VAL_BIB1;
616 static struct sortKey {
621 struct sortKey *next;
624 static struct recKeys {
633 static void addIndexString (RecWord *p, const char *string, int length)
636 unsigned char attrSet;
637 unsigned short attrUse;
640 int *pseqno = &p->seqnos[p->reg_type];
642 if (reckeys.buf_used+1024 > reckeys.buf_max)
646 b = xmalloc (reckeys.buf_max += 128000);
647 if (reckeys.buf_used > 0)
648 memcpy (b, reckeys.buf, reckeys.buf_used);
652 dst = reckeys.buf + reckeys.buf_used;
654 attrSet = p->attrSet;
655 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
658 reckeys.prevAttrSet = attrSet;
659 attrUse = p->attrUse;
660 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
663 reckeys.prevAttrUse = attrUse;
665 diff = 1 + *pseqno - reckeys.prevSeqNo;
666 if (diff >= 1 && diff <= 15)
671 reckeys.prevSeqNo = *pseqno;
677 memcpy (dst, &attrSet, sizeof(attrSet));
678 dst += sizeof(attrSet);
682 memcpy (dst, &attrUse, sizeof(attrUse));
683 dst += sizeof(attrUse);
685 *dst++ = p->reg_type;
686 memcpy (dst, string, length);
692 memcpy (dst, pseqno, sizeof(*pseqno));
693 dst += sizeof(*pseqno);
695 reckeys.buf_used = dst - reckeys.buf;
699 static void addSortString (RecWord *p, const char *string, int length)
703 for (sk = sortKeys; sk; sk = sk->next)
704 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
707 sk = xmalloc (sizeof(*sk));
711 sk->string = xmalloc (length);
713 memcpy (sk->string, string, length);
715 sk->attrSet = p->attrSet;
716 sk->attrUse = p->attrUse;
719 static void addString (RecWord *p, const char *string, int length)
722 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
723 addSortString (p, string, length);
725 addIndexString (p, string, length);
728 static void addIncompleteField (RecWord *p)
730 const char *b = p->string;
731 int remain = p->length;
732 const char **map = 0;
735 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
739 char buf[IT_MAX_WORD+1];
743 while (map && *map && **map == *CHR_SPACE)
745 remain = p->length - (b - p->string);
747 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
754 while (map && *map && **map != *CHR_SPACE)
756 const char *cp = *map;
758 while (i < IT_MAX_WORD && *cp)
760 remain = p->length - (b - p->string);
762 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
768 addString (p, buf, i);
772 static void addCompleteField (RecWord *p)
774 const char *b = p->string;
775 char buf[IT_MAX_WORD+1];
776 const char **map = 0;
777 int i = 0, remain = p->length;
780 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
782 while (remain > 0 && i < IT_MAX_WORD)
784 while (map && *map && **map == *CHR_SPACE)
786 remain = p->length - (b - p->string);
788 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
795 if (i && i < IT_MAX_WORD)
796 buf[i++] = *CHR_SPACE;
797 while (map && *map && **map != *CHR_SPACE)
799 const char *cp = *map;
801 if (i >= IT_MAX_WORD)
803 while (i < IT_MAX_WORD && *cp)
805 remain = p->length - (b - p->string);
807 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
815 addString (p, buf, i);
818 static void addRecordKey (RecWord *p)
820 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
821 addCompleteField (p);
823 addIncompleteField(p);
826 static void flushSortKeys (SYSNO sysno, int cmd)
828 struct sortKey *sk = sortKeys;
830 sortIdx_sysno (sortIdx, sysno);
833 struct sortKey *sk_next = sk->next;
834 sortIdx_type (sortIdx, sk->attrUse);
835 sortIdx_add (sortIdx, sk->string, sk->length);
843 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
845 unsigned char attrSet = (unsigned char) -1;
846 unsigned short attrUse = (unsigned short) -1;
850 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
851 while (off < reckeys->buf_used)
853 const char *src = reckeys->buf + off;
861 memcpy (&attrSet, src, sizeof(attrSet));
862 src += sizeof(attrSet);
866 memcpy (&attrUse, src, sizeof(attrUse));
867 src += sizeof(attrUse);
869 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
872 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
874 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
876 ch = zebraExplain_addSU (zti, attrSet, attrUse);
878 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
881 ((char*)key_buf) [key_buf_used++] = *src++;
883 ((char*)key_buf) [key_buf_used++] = '\0';
884 ((char*) key_buf)[key_buf_used++] = cmd;
887 seqno += ((lead>>2) & 15)-1;
890 memcpy (&seqno, src, sizeof(seqno));
891 src += sizeof(seqno);
895 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
896 key_buf_used += sizeof(key);
897 off = src - reckeys->buf;
899 assert (off == reckeys->buf_used);
902 static const char **searchRecordKey (struct recKeys *reckeys,
903 int attrSetS, int attrUseS)
905 static const char *ws[32];
913 for (i = 0; i<32; i++)
916 while (off < reckeys->buf_used)
919 const char *src = reckeys->buf + off;
927 memcpy (&attrSet, src, sizeof(attrSet));
928 src += sizeof(attrSet);
932 memcpy (&attrUse, src, sizeof(attrUse));
933 src += sizeof(attrUse);
939 seqno += ((lead>>2) & 15)-1;
942 memcpy (&seqno, src, sizeof(seqno));
943 src += sizeof(seqno);
945 if (attrUseS == attrUse && attrSetS == attrSet)
952 woff = seqno - startSeq;
953 if (woff >= 0 && woff < 31)
957 off = src - reckeys->buf;
959 assert (off == reckeys->buf_used);
963 struct file_read_info {
964 off_t file_max; /* maximum offset so far */
965 off_t file_offset; /* current offset */
966 off_t file_moffset; /* offset of rec/rec boundary */
973 static struct file_read_info *file_read_start (int fd)
975 struct file_read_info *fi = xmalloc (sizeof(*fi));
979 fi->file_moffset = 0;
985 static void file_read_stop (struct file_read_info *fi)
990 static off_t file_seek (void *handle, off_t offset)
992 struct file_read_info *p = handle;
993 p->file_offset = offset;
996 return lseek (p->fd, offset, SEEK_SET);
999 static off_t file_tell (void *handle)
1001 struct file_read_info *p = handle;
1002 return p->file_offset;
1005 static int file_read (void *handle, char *buf, size_t count)
1007 struct file_read_info *p = handle;
1013 if (r > p->sdrmax - p->file_offset)
1014 r = p->sdrmax - p->file_offset;
1016 memcpy (buf, p->sdrbuf + p->file_offset, r);
1019 r = read (fd, buf, count);
1022 p->file_offset += r;
1023 if (p->file_offset > p->file_max)
1024 p->file_max = p->file_offset;
1029 static void file_begin (void *handle)
1031 struct file_read_info *p = handle;
1033 p->file_offset = p->file_moffset;
1034 if (!p->sdrbuf && p->file_moffset)
1035 lseek (p->fd, p->file_moffset, SEEK_SET);
1039 static void file_end (void *handle, off_t offset)
1041 struct file_read_info *p = handle;
1043 assert (p->file_more == 0);
1045 p->file_moffset = offset;
1048 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1049 const char *fname, const char *spec)
1051 static char dstBuf[2048];
1053 const char *s = spec;
1054 static const char **w;
1058 while (*s == ' ' || *s == '\t')
1064 char attset_str[64], attname_str[64];
1065 data1_attset *attset;
1068 int attSet = 1, attUse = 1;
1072 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1074 attset_str[i++] = *s;
1075 attset_str[i] = '\0';
1080 for (i = 0; *s && *s != ')'; s++)
1082 attname_str[i++] = *s;
1083 attname_str[i] = '\0';
1086 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1089 attSet = attset->reference;
1090 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1092 attUse = att->value;
1094 attUse = atoi (attname_str);
1096 w = searchRecordKey (reckeys, attSet, attUse);
1101 for (i = 0; i<32; i++)
1106 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1107 spec, rGroup->groupName ? rGroup->groupName : "none");
1112 for (i = 0; i<32; i++)
1113 if (matchFlag[i] && w[i])
1121 dst += strlen(w[i]);
1125 logf (LOG_WARN, "Record didn't contain match"
1126 " fields in (%s,%s)", attset_str, attname_str);
1134 const char *spec_src = NULL;
1135 const char *s1 = ++s;
1136 while (*s1 && *s1 != ' ' && *s1 != '\t')
1142 memcpy (special, s, spec_len);
1143 special[spec_len] = '\0';
1146 if (!strcmp (special, "group"))
1147 spec_src = rGroup->groupName;
1148 else if (!strcmp (special, "database"))
1149 spec_src = rGroup->databaseName;
1150 else if (!strcmp (special, "filename"))
1152 else if (!strcmp (special, "type"))
1153 spec_src = rGroup->recordType;
1158 strcpy (dst, spec_src);
1159 dst += strlen (spec_src);
1162 else if (*s == '\"' || *s == '\'')
1164 int stopMarker = *s++;
1168 while (*s && *s != stopMarker)
1171 tmpString[i++] = *s++;
1175 tmpString[i] = '\0';
1176 strcpy (dst, tmpString);
1177 dst += strlen (tmpString);
1181 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1182 spec, rGroup->groupName ? rGroup->groupName : "none");
1189 logf (LOG_WARN, "No match criteria for record %s in group %s",
1190 fname, rGroup->groupName ? rGroup->groupName : "none");
1197 struct recordLogInfo {
1200 struct recordGroup *rGroup;
1203 static void recordLogPreamble (int level, const char *msg, void *info)
1205 struct recordLogInfo *p = info;
1206 FILE *outf = log_file ();
1208 if (level & LOG_LOG)
1210 fprintf (outf, "File %s, offset %d, type %s\n",
1211 p->rGroup->recordType, p->recordOffset, p->fname);
1212 log_event_start (NULL, NULL);
1215 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1217 zebraExplain_addSchema (zti, oid);
1220 static int recordExtract (SYSNO *sysno, const char *fname,
1221 struct recordGroup *rGroup, int deleteFlag,
1222 struct file_read_info *fi,
1223 RecType recType, char *subType)
1225 RecordAttr *recordAttr;
1230 struct recordLogInfo logInfo;
1231 off_t recordOffset = 0;
1235 struct recExtractCtrl extractCtrl;
1237 /* we are going to read from a file, so prepare the extraction */
1240 reckeys.buf_used = 0;
1241 reckeys.prevAttrUse = -1;
1242 reckeys.prevAttrSet = -1;
1243 reckeys.prevSeqNo = 0;
1245 recordOffset = fi->file_moffset;
1246 extractCtrl.offset = fi->file_moffset;
1247 extractCtrl.readf = file_read;
1248 extractCtrl.seekf = file_seek;
1249 extractCtrl.tellf = file_tell;
1250 extractCtrl.endf = file_end;
1251 extractCtrl.fh = fi;
1252 extractCtrl.subType = subType;
1253 extractCtrl.init = wordInit;
1254 extractCtrl.addWord = addRecordKey;
1255 extractCtrl.addSchema = addSchema;
1256 extractCtrl.dh = rGroup->dh;
1257 for (i = 0; i<256; i++)
1258 extractCtrl.seqno[i] = 0;
1259 extractCtrl.zebra_maps = rGroup->zebra_maps;
1260 extractCtrl.flagShowRecords = !rGroup->flagRw;
1262 if (!rGroup->flagRw)
1263 printf ("File: %s %ld\n", fname, (long) recordOffset);
1265 logInfo.fname = fname;
1266 logInfo.recordOffset = recordOffset;
1267 logInfo.rGroup = rGroup;
1268 log_event_start (recordLogPreamble, &logInfo);
1270 r = (*recType->extract)(&extractCtrl);
1272 log_event_start (NULL, NULL);
1276 /* error occured during extraction ... */
1277 if (rGroup->flagRw &&
1278 records_processed < rGroup->fileVerboseLimit)
1280 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1281 fname, (long) recordOffset, r);
1285 if (reckeys.buf_used == 0)
1287 /* the extraction process returned no information - the record
1288 is probably empty - unless flagShowRecords is in use */
1289 if (!rGroup->flagRw)
1291 logf (LOG_WARN, "No keys generated for file %s", fname);
1292 logf (LOG_WARN, " The file is probably empty");
1297 /* perform match if sysno not known and if match criteria is specified */
1304 if (rGroup->recordId && *rGroup->recordId)
1308 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1312 rinfo = dict_lookup (matchDict, matchStr);
1314 memcpy (sysno, rinfo+1, sizeof(*sysno));
1318 logf (LOG_WARN, "Bad match criteria");
1329 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1330 fname, (long) recordOffset);
1331 logf (LOG_WARN, "cannot delete record above (seems new)");
1334 if (records_processed < rGroup->fileVerboseLimit)
1335 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1336 fname, (long) recordOffset);
1337 rec = rec_new (records);
1339 *sysno = rec->sysno;
1341 recordAttr = rec_init_attr (zti, rec);
1345 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1347 flushRecordKeys (*sysno, 1, &reckeys);
1348 flushSortKeys (*sysno, 1);
1354 /* record already exists */
1355 struct recKeys delkeys;
1357 rec = rec_get (records, *sysno);
1360 recordAttr = rec_init_attr (zti, rec);
1362 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1364 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1365 fname, (long) recordOffset);
1370 delkeys.buf_used = rec->size[recInfo_delKeys];
1371 delkeys.buf = rec->info[recInfo_delKeys];
1372 flushSortKeys (*sysno, 0);
1373 flushRecordKeys (*sysno, 0, &delkeys);
1376 /* record going to be deleted */
1377 if (!delkeys.buf_used)
1379 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1380 fname, (long) recordOffset);
1381 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1385 if (records_processed < rGroup->fileVerboseLimit)
1386 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1387 fname, (long) recordOffset);
1390 dict_delete (matchDict, matchStr);
1391 rec_del (records, &rec);
1399 /* record going to be updated */
1400 if (!delkeys.buf_used)
1402 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1403 fname, (long) recordOffset);
1404 logf (LOG_WARN, "cannot update file above, storeKeys false");
1408 if (records_processed < rGroup->fileVerboseLimit)
1409 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1410 fname, (long) recordOffset);
1411 flushRecordKeys (*sysno, 1, &reckeys);
1416 /* update file type */
1417 xfree (rec->info[recInfo_fileType]);
1418 rec->info[recInfo_fileType] =
1419 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1421 /* update filename */
1422 xfree (rec->info[recInfo_filename]);
1423 rec->info[recInfo_filename] =
1424 rec_strdup (fname, &rec->size[recInfo_filename]);
1426 /* update delete keys */
1427 xfree (rec->info[recInfo_delKeys]);
1428 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1431 rec->size[recInfo_delKeys] = reckeys.buf_used;
1432 rec->info[recInfo_delKeys] = reckeys.buf;
1434 reckeys.buf_max = 0;
1436 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1437 rec->size[recInfo_delKeys] = reckeys.buf_used;
1438 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1439 rec->size[recInfo_delKeys]);
1444 rec->info[recInfo_delKeys] = NULL;
1445 rec->size[recInfo_delKeys] = 0;
1448 /* save file size of original record */
1449 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1450 recordAttr->recordSize = fi->file_moffset - recordOffset;
1451 if (!recordAttr->recordSize)
1452 recordAttr->recordSize = fi->file_max - recordOffset;
1453 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1455 /* set run-number for this record */
1456 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1458 /* update store data */
1459 xfree (rec->info[recInfo_storeData]);
1460 if (rGroup->flagStoreData == 1)
1462 rec->size[recInfo_storeData] = recordAttr->recordSize;
1463 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1464 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1466 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1467 (long) recordOffset, fname);
1470 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1471 < recordAttr->recordSize)
1473 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1474 recordAttr->recordSize, fname);
1480 rec->info[recInfo_storeData] = NULL;
1481 rec->size[recInfo_storeData] = 0;
1483 /* update database name */
1484 xfree (rec->info[recInfo_databaseName]);
1485 rec->info[recInfo_databaseName] =
1486 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1489 recordAttr->recordOffset = recordOffset;
1491 /* commit this record */
1492 rec_put (records, &rec);
1497 int fileExtract (SYSNO *sysno, const char *fname,
1498 const struct recordGroup *rGroupP, int deleteFlag)
1506 struct recordGroup rGroupM;
1507 struct recordGroup *rGroup = &rGroupM;
1508 struct file_read_info *fi;
1510 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1512 if (!rGroup->groupName || !*rGroup->groupName)
1515 sprintf (gprefix, "%s.", rGroup->groupName);
1517 logf (LOG_DEBUG, "fileExtract %s", fname);
1519 /* determine file extension */
1521 for (i = strlen(fname); --i >= 0; )
1522 if (fname[i] == '/')
1524 else if (fname[i] == '.')
1526 strcpy (ext, fname+i+1);
1529 /* determine file type - depending on extension */
1530 if (!rGroup->recordType)
1532 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1533 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1535 sprintf (ext_res, "%srecordType", gprefix);
1536 rGroup->recordType = res_get (common_resource, ext_res);
1539 if (!rGroup->recordType)
1541 if (records_processed < rGroup->fileVerboseLimit)
1542 logf (LOG_LOG, "? %s", fname);
1545 if (!*rGroup->recordType)
1548 recType_byName (rGroup->recTypes, rGroup->recordType, subType)))
1550 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1554 /* determine match criteria */
1555 if (!rGroup->recordId)
1557 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1558 rGroup->recordId = res_get (common_resource, ext_res);
1561 /* determine database name */
1562 if (!rGroup->databaseName)
1564 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1565 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1567 sprintf (ext_res, "%sdatabase", gprefix);
1568 rGroup->databaseName = res_get (common_resource, ext_res);
1571 if (!rGroup->databaseName)
1572 rGroup->databaseName = "Default";
1574 /* determine if explain database */
1576 sprintf (ext_res, "%sexplainDatabase", gprefix);
1577 rGroup->explainDatabase =
1578 atoi (res_get_def (common_resource, ext_res, "0"));
1580 /* announce database */
1581 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1583 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1584 rGroup->explainDatabase))
1588 if (rGroup->flagStoreData == -1)
1591 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1592 if (!(sval = res_get (common_resource, ext_res)))
1594 sprintf (ext_res, "%sstoreData", gprefix);
1595 sval = res_get (common_resource, ext_res);
1598 rGroup->flagStoreData = atoi (sval);
1600 if (rGroup->flagStoreData == -1)
1601 rGroup->flagStoreData = 0;
1603 if (rGroup->flagStoreKeys == -1)
1607 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1608 if (!(sval = res_get (common_resource, ext_res)))
1610 sprintf (ext_res, "%sstoreKeys", gprefix);
1611 sval = res_get (common_resource, ext_res);
1614 rGroup->flagStoreKeys = atoi (sval);
1616 if (rGroup->flagStoreKeys == -1)
1617 rGroup->flagStoreKeys = 0;
1623 char xname[128], *xp;
1625 strncpy (xname, fname, 127);
1626 if ((xp = strchr (xname, '.')))
1629 h = zebraSdr_open (xname);
1632 logf (LOG_WARN, "sdr open %s", xname);
1641 segmentno = zebraSdr_segment (h, 0);
1642 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1643 logf (LOG_LOG, "SDR: %s", sdr_name);
1649 r = zebraSdr_read (h, &buf);
1654 fi = file_read_start (0);
1660 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1662 } while (r && !sysno && fi->file_more);
1663 file_read_stop (fi);
1670 if (sysno && deleteFlag)
1674 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1676 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1680 fi = file_read_start (fd);
1684 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1686 } while (r && !sysno && fi->file_more);
1687 file_read_stop (fi);
1693 static int explain_extract (void *handle, Record rec, data1_node *n)
1695 struct recordGroup *rGroup = (struct recordGroup*) handle;
1696 struct recExtractCtrl extractCtrl;
1699 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1701 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1705 reckeys.buf_used = 0;
1706 reckeys.prevAttrUse = -1;
1707 reckeys.prevAttrSet = -1;
1708 reckeys.prevSeqNo = 0;
1710 extractCtrl.init = wordInit;
1711 extractCtrl.addWord = addRecordKey;
1712 extractCtrl.addSchema = addSchema;
1713 extractCtrl.dh = rGroup->dh;
1714 for (i = 0; i<256; i++)
1715 extractCtrl.seqno[i] = 0;
1716 extractCtrl.zebra_maps = rGroup->zebra_maps;
1717 extractCtrl.flagShowRecords = !rGroup->flagRw;
1719 grs_extract_tree(&extractCtrl, n);
1721 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1723 if (rec->size[recInfo_delKeys])
1725 struct recKeys delkeys;
1727 delkeys.buf_used = rec->size[recInfo_delKeys];
1728 delkeys.buf = rec->info[recInfo_delKeys];
1729 flushSortKeys (rec->sysno, 0);
1730 flushRecordKeys (rec->sysno, 0, &delkeys);
1732 flushRecordKeys (rec->sysno, 1, &reckeys);
1733 flushSortKeys (rec->sysno, 1);
1735 xfree (rec->info[recInfo_delKeys]);
1736 rec->size[recInfo_delKeys] = reckeys.buf_used;
1737 rec->info[recInfo_delKeys] = reckeys.buf;
1739 reckeys.buf_max = 0;