2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.85 1998-09-22 10:03:41 adam
8 * Changed result sets to be persistent in the sense that they can
9 * be re-searched if needed.
10 * Fixed memory leak in rsm_or.
12 * Revision 1.84 1998/06/11 15:42:22 adam
13 * Changed the way use attributes are specified in the recordId
16 * Revision 1.83 1998/06/08 14:43:10 adam
17 * Added suport for EXPLAIN Proxy servers - added settings databasePath
18 * and explainDatabase to facilitate this. Increased maximum number
19 * of databases and attributes in one register.
21 * Revision 1.82 1998/05/20 10:12:15 adam
22 * Implemented automatic EXPLAIN database maintenance.
23 * Modified Zebra to work with ASN.1 compiled version of YAZ.
25 * Revision 1.81 1998/03/11 11:19:04 adam
26 * Changed the way sequence numbers are generated.
28 * Revision 1.80 1998/03/05 08:45:11 adam
29 * New result set model and modular ranking system. Moved towards
30 * descent server API. System information stored as "SGML" records.
32 * Revision 1.79 1998/02/17 10:32:52 adam
33 * Fixed bug: binary files weren't opened with flag b on NT.
35 * Revision 1.78 1998/02/10 12:03:05 adam
38 * Revision 1.77 1998/01/12 15:04:08 adam
39 * The test option (-s) only uses read-lock (and not write lock).
41 * Revision 1.76 1997/10/27 14:33:04 adam
42 * Moved towards generic character mapping depending on "structure"
43 * field in abstract syntax file. Fixed a few memory leaks. Fixed
44 * bug with negative integers when doing searches with relational
47 * Revision 1.75 1997/09/17 12:19:12 adam
48 * Zebra version corresponds to YAZ version 1.4.
49 * Changed Zebra server so that it doesn't depend on global common_resource.
51 * Revision 1.74 1997/09/09 13:38:06 adam
52 * Partial port to WIN95/NT.
54 * Revision 1.73 1997/09/04 13:57:20 adam
55 * New file extract/retrieve method tellf (added).
56 * Added O_BINARY for open calls.
58 * Revision 1.72 1997/07/15 16:32:29 adam
59 * Bug fix: Match handler didn't terminate the resulting string!
61 * Revision 1.71 1997/07/15 16:28:41 adam
62 * Bug fix: storeData didn't work with files with multiple records.
63 * Bug fix: fixed memory management with records; not really well
66 * Revision 1.70 1997/07/01 13:00:42 adam
67 * Bug fix in routine searchRecordKey: uninitialized variables.
69 * Revision 1.69 1997/04/29 09:26:03 adam
70 * Bug fix: generic recordId handling didn't work for compressed internal
73 * Revision 1.68 1997/02/12 20:39:45 adam
74 * Implemented options -f <n> that limits the log to the first <n>
76 * Changed some log messages also.
78 * Revision 1.67 1996/11/15 15:02:14 adam
79 * Minor changes regarding logging.
81 * Revision 1.66 1996/11/14 09:52:21 adam
82 * Strings in record keys bound by IT_MAX_WORD.
84 * Revision 1.65 1996/11/14 08:57:56 adam
85 * Reduction of storeKeys area.
87 * Revision 1.64 1996/11/08 11:10:16 adam
88 * Buffers used during file match got bigger.
89 * Compressed ISAM support everywhere.
90 * Bug fixes regarding masking characters in queries.
91 * Redesigned Regexp-2 queries.
93 * Revision 1.63 1996/10/29 14:09:39 adam
94 * Use of cisam system - enabled if setting isamc is 1.
96 * Revision 1.62 1996/10/11 10:57:01 adam
97 * New module recctrl. Used to manage records (extract/retrieval).
98 * Several files have been moved to the recctrl sub directory.
100 * Revision 1.61 1996/06/06 12:08:37 quinn
101 * Added showRecord function
103 * Revision 1.60 1996/06/04 10:18:12 adam
104 * Search/scan uses character mapping module.
106 * Revision 1.59 1996/05/14 15:47:07 adam
107 * Cleanup of various buffer size entities.
109 * Revision 1.58 1996/05/14 06:16:38 adam
110 * Compact use/set bytes used in search service.
112 * Revision 1.57 1996/05/13 14:23:04 adam
113 * Work on compaction of set/use bytes in dictionary.
115 * Revision 1.56 1996/05/09 09:54:42 adam
116 * Server supports maps from one logical attributes to a list of physical
118 * The extraction process doesn't make space consuming 'any' keys.
120 * Revision 1.55 1996/05/09 07:28:55 quinn
121 * Work towards phrases and multiple registers
123 * Revision 1.54 1996/05/01 13:46:35 adam
124 * First work on multiple records in one file.
125 * New option, -offset, to the "unread" command in the filter module.
127 * Revision 1.53 1996/04/26 12:09:43 adam
128 * Added a few comments.
130 * Revision 1.52 1996/04/25 13:27:57 adam
131 * Function recordExtract modified so that files with no keys (possibly empty)
134 * Revision 1.51 1996/03/19 11:08:42 adam
135 * Bug fix: Log preamble wasn't always turned off after recordExtract.
137 * Revision 1.50 1996/02/12 18:45:36 adam
138 * New fileVerboseFlag in record group control.
140 * Revision 1.49 1996/02/05 12:29:57 adam
141 * Logging reduced a bit.
142 * The remaining running time is estimated during register merge.
144 * Revision 1.48 1996/02/01 20:53:26 adam
145 * The temporary per-record keys are compacted a little, and duplication
146 * of the per-records keys are avoided when they are saved in the record
147 * information buffer.
149 * Revision 1.47 1996/01/17 14:57:48 adam
150 * Prototype changed for reader functions in extract/retrieve. File
151 * is identified by 'void *' instead of 'int.
153 * Revision 1.46 1995/12/15 14:57:16 adam
156 * Revision 1.45 1995/12/15 12:37:41 adam
157 * In addRecordKeyAny: Writes key only when attrSet != -1.
159 * Revision 1.44 1995/12/12 16:00:54 adam
160 * System call sync(2) used after update/commit.
161 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
164 * Revision 1.43 1995/12/11 09:12:46 adam
165 * The rec_get function returns NULL if record doesn't exist - will
166 * happen in the server if the result set records have been deleted since
167 * the creation of the set (i.e. the search).
168 * The server saves a result temporarily if it is 'volatile', i.e. the
169 * set is register dependent.
171 * Revision 1.42 1995/12/07 17:38:46 adam
172 * Work locking mechanisms for concurrent updates/commit.
174 * Revision 1.41 1995/12/06 16:06:42 adam
175 * Better diagnostics. Work on 'real' dictionary deletion.
177 * Revision 1.40 1995/12/05 16:57:40 adam
178 * More work on regular patterns.
180 * Revision 1.39 1995/12/05 13:20:18 adam
181 * Bug fix: file_read sometimes returned early EOF.
183 * Revision 1.38 1995/12/04 17:59:21 adam
184 * More work on regular expression conversion.
186 * Revision 1.37 1995/12/04 14:22:27 adam
187 * Extra arg to recType_byName.
188 * Started work on new regular expression parsed input to
189 * structured records.
191 * Revision 1.36 1995/11/30 08:34:29 adam
192 * Started work on commit facility.
193 * Changed a few malloc/free to xmalloc/xfree.
195 * Revision 1.35 1995/11/28 14:26:21 adam
196 * Bug fix: recordId with constant wasn't right.
197 * Bug fix: recordId dictionary entry wasn't deleted when needed.
199 * Revision 1.34 1995/11/28 09:09:38 adam
200 * Zebra config renamed.
201 * Use setting 'recordId' to identify record now.
202 * Bug fix in recindex.c: rec_release_blocks was invokeded even
203 * though the blocks were already released.
204 * File traversal properly deletes records when needed.
206 * Revision 1.33 1995/11/27 09:56:20 adam
207 * Record info elements better enumerated. Internal store of records.
209 * Revision 1.32 1995/11/25 10:24:05 adam
210 * More record fields - they are enumerated now.
211 * New options: flagStoreData flagStoreKey.
213 * Revision 1.31 1995/11/24 11:31:35 adam
214 * Commands add & del read filenames from stdin if source directory is
216 * Match criteria supports 'constant' strings.
218 * Revision 1.30 1995/11/22 17:19:16 adam
219 * Record management uses the bfile system.
221 * Revision 1.29 1995/11/21 15:01:14 adam
222 * New general match criteria implemented.
223 * New feature: document groups.
225 * Revision 1.28 1995/11/21 09:20:30 adam
226 * Yet more work on record match.
228 * Revision 1.27 1995/11/20 16:59:45 adam
229 * New update method: the 'old' keys are saved for each records.
231 * Revision 1.26 1995/11/20 11:56:24 adam
232 * Work on new traversal.
234 * Revision 1.25 1995/11/16 15:34:54 adam
235 * Uses new record management system in both indexer and server.
237 * Revision 1.24 1995/11/15 19:13:08 adam
238 * Work on record management.
240 * Revision 1.23 1995/10/27 14:00:10 adam
241 * Implemented detection of database availability.
243 * Revision 1.22 1995/10/17 18:02:07 adam
244 * New feature: databases. Implemented as prefix to words in dictionary.
246 * Revision 1.21 1995/10/10 12:24:38 adam
247 * Temporary sort files are compressed.
249 * Revision 1.20 1995/10/06 13:52:05 adam
250 * Bug fixes. Handler may abort further scanning.
252 * Revision 1.19 1995/10/04 12:55:16 adam
253 * Bug fix in ranked search. Use=Any keys inserted.
255 * Revision 1.18 1995/10/04 09:37:08 quinn
258 * Revision 1.17 1995/10/03 14:28:57 adam
259 * Buffered read in extract works.
261 * Revision 1.16 1995/10/03 14:28:45 adam
262 * Work on more effecient read handler in extract.
264 * Revision 1.15 1995/10/02 15:42:53 adam
265 * Extract uses file descriptors instead of FILE pointers.
267 * Revision 1.14 1995/10/02 15:29:13 adam
268 * More logging in file_extract.
270 * Revision 1.13 1995/09/29 14:01:39 adam
273 * Revision 1.12 1995/09/28 14:22:56 adam
274 * Sort uses smaller temporary files.
276 * Revision 1.11 1995/09/28 12:10:31 adam
277 * Bug fixes. Field prefix used in queries.
279 * Revision 1.10 1995/09/28 09:19:41 adam
280 * xfree/xmalloc used everywhere.
281 * Extract/retrieve method seems to work for text records.
283 * Revision 1.9 1995/09/27 12:22:28 adam
284 * More work on extract in record control.
285 * Field name is not in isam keys but in prefix in dictionary words.
287 * Revision 1.8 1995/09/14 07:48:22 adam
288 * Record control management.
290 * Revision 1.7 1995/09/11 13:09:32 adam
291 * More work on relevance feedback.
293 * Revision 1.6 1995/09/08 14:52:27 adam
294 * Minor changes. Dictionary is lower case now.
296 * Revision 1.5 1995/09/06 16:11:16 adam
297 * Option: only one word key per file.
299 * Revision 1.4 1995/09/05 15:28:39 adam
300 * More work on search engine.
302 * Revision 1.3 1995/09/04 12:33:41 adam
303 * Various cleanup. YAZ util used instead.
305 * Revision 1.2 1995/09/04 09:10:34 adam
306 * More work on index add/del/update.
307 * Merge sort implemented.
308 * Initial work on z39 server.
310 * Revision 1.1 1995/09/01 14:06:35 adam
311 * Split of work into more files.
330 static Dict matchDict;
332 static Records records = NULL;
333 static SortIdx sortIdx = NULL;
335 static char **key_buf;
336 static size_t ptr_top;
338 static size_t key_buf_used;
339 static int key_file_no;
341 static int records_inserted = 0;
342 static int records_updated = 0;
343 static int records_deleted = 0;
344 static int records_processed = 0;
346 static ZebraExplainInfo zti = NULL;
348 static void logRecord (int showFlag)
352 if (showFlag || !(records_processed % 1000))
354 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
355 records_processed, records_inserted, records_updated,
360 static int explain_extract (void *handle, Record drec, data1_node *n);
362 int key_open (struct recordGroup *rGroup, int mem)
364 BFiles bfs = rGroup->bfs;
365 int rw = rGroup->flagRw;
366 data1_handle dh = rGroup->dh;
368 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
371 key_buf = xmalloc (mem);
372 ptr_top = mem/sizeof(char*);
378 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
380 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
384 records = rec_open (bfs, rw);
387 dict_close (matchDict);
390 zti = zebraExplain_open (records, dh, common_resource,
391 rw, rGroup, explain_extract);
394 rec_close (&records);
395 dict_close (matchDict);
398 sortIdx = sortIdx_open (bfs, 1);
408 void encode_key_init (struct encode_info *i)
414 char *encode_key_int (int d, char *bp)
423 else if (d <= 4194303)
425 *bp++ = 128 + (d>>16);
426 *bp++ = (d>>8) & 255;
431 *bp++ = 192 + (d>>24);
432 *bp++ = (d>>16) & 255;
433 *bp++ = (d>>8) & 255;
439 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
444 while ((*bp++ = *k++))
446 memcpy (&key, k+1, sizeof(struct it_key));
447 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
448 if (i->sysno != key.sysno)
450 i->sysno = key.sysno;
453 bp = encode_key_int (key.seqno - i->seqno, bp);
454 i->seqno = key.seqno;
455 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
457 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
465 static int key_y_len;
467 static int key_y_compare (const void *p1, const void *p2)
471 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
472 *(char**) p2 + key_y_len + 1)))
474 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
477 static int key_x_compare (const void *p1, const void *p2)
479 return strcmp (*(char**) p1, *(char**) p2);
483 void key_flush (void)
488 struct encode_info encode_info;
497 logf (LOG_LOG, "sorting section %d", key_file_no);
499 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
500 getFnameTmp (out_fname, key_file_no);
502 if (!(outf = fopen (out_fname, "wb")))
504 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
507 logf (LOG_LOG, "writing section %d", key_file_no);
508 prevcp = cp = key_buf[ptr_top-ptr_i];
510 encode_key_init (&encode_info);
511 encode_key_write (cp, &encode_info, outf);
514 cp = key_buf[ptr_top-ptr_i];
515 if (strcmp (cp, prevcp))
517 encode_key_init (&encode_info);
518 encode_key_write (cp, &encode_info, outf);
522 encode_key_write (cp + strlen(cp), &encode_info, outf);
525 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
526 getFnameTmp (out_fname, key_file_no);
528 if (!(outf = fopen (out_fname, "wb")))
530 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
533 logf (LOG_LOG, "writing section %d", key_file_no);
535 prevcp = key_buf[ptr_top-i];
537 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
539 key_y_len = strlen(prevcp)+1;
541 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
542 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
544 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
545 sizeof(char*), key_y_compare);
546 cp = key_buf[ptr_top-ptr_i];
548 encode_key_init (&encode_info);
549 encode_key_write (cp, &encode_info, outf);
552 cp = key_buf[ptr_top-ptr_i];
553 encode_key_write (cp+key_y_len, &encode_info, outf);
557 prevcp = key_buf[ptr_top-ptr_i];
562 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
565 logf (LOG_LOG, "finished section %d", key_file_no);
570 int key_close (struct recordGroup *rGroup)
572 int rw = rGroup->flagRw;
574 zebraExplain_runNumberIncrement (zti, 1);
575 zebraExplain_close (zti, rw, 0);
578 rec_close (&records);
579 dict_close (matchDict);
580 sortIdx_close (sortIdx);
586 static void wordInit (struct recExtractCtrl *p, RecWord *w)
588 w->zebra_maps = p->zebra_maps;
589 w->seqnos = p->seqno;
590 w->attrSet = VAL_BIB1;
595 static struct sortKey {
600 struct sortKey *next;
603 static struct recKeys {
612 static void addIndexString (RecWord *p, const char *string, int length)
619 int *pseqno = &p->seqnos[p->reg_type];
621 if (reckeys.buf_used+1024 > reckeys.buf_max)
625 b = xmalloc (reckeys.buf_max += 128000);
626 if (reckeys.buf_used > 0)
627 memcpy (b, reckeys.buf, reckeys.buf_used);
631 dst = reckeys.buf + reckeys.buf_used;
633 attrSet = p->attrSet;
634 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
637 reckeys.prevAttrSet = attrSet;
638 attrUse = p->attrUse;
639 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
642 reckeys.prevAttrUse = attrUse;
644 diff = 1 + *pseqno - reckeys.prevSeqNo;
645 if (diff >= 1 && diff <= 15)
650 reckeys.prevSeqNo = *pseqno;
656 memcpy (dst, &attrSet, sizeof(attrSet));
657 dst += sizeof(attrSet);
661 memcpy (dst, &attrUse, sizeof(attrUse));
662 dst += sizeof(attrUse);
664 *dst++ = p->reg_type;
665 memcpy (dst, string, length);
671 memcpy (dst, pseqno, sizeof(*pseqno));
672 dst += sizeof(*pseqno);
674 reckeys.buf_used = dst - reckeys.buf;
678 static void addSortString (RecWord *p, const char *string, int length)
682 for (sk = sortKeys; sk; sk = sk->next)
683 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
686 sk = xmalloc (sizeof(*sk));
690 sk->string = xmalloc (length);
692 memcpy (sk->string, string, length);
694 sk->attrSet = p->attrSet;
695 sk->attrUse = p->attrUse;
698 static void addString (RecWord *p, const char *string, int length)
701 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
702 addSortString (p, string, length);
704 addIndexString (p, string, length);
707 static void addIncompleteField (RecWord *p)
709 const char *b = p->string;
710 int remain = p->length;
711 const char **map = 0;
714 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
718 char buf[IT_MAX_WORD+1];
722 while (map && *map && **map == *CHR_SPACE)
724 remain = p->length - (b - p->string);
726 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
733 while (map && *map && **map != *CHR_SPACE)
735 const char *cp = *map;
737 while (i < IT_MAX_WORD && *cp)
739 remain = p->length - (b - p->string);
741 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
747 addString (p, buf, i);
751 static void addCompleteField (RecWord *p)
753 const char *b = p->string;
754 char buf[IT_MAX_WORD+1];
755 const char **map = 0;
756 int i = 0, remain = p->length;
759 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
761 while (remain > 0 && i < IT_MAX_WORD)
763 while (map && *map && **map == *CHR_SPACE)
765 remain = p->length - (b - p->string);
767 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
774 if (i && i < IT_MAX_WORD)
775 buf[i++] = *CHR_SPACE;
776 while (map && *map && **map != *CHR_SPACE)
778 const char *cp = *map;
780 if (i >= IT_MAX_WORD)
782 while (i < IT_MAX_WORD && *cp)
784 remain = p->length - (b - p->string);
786 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
794 addString (p, buf, i);
797 static void addRecordKey (RecWord *p)
799 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
800 addCompleteField (p);
802 addIncompleteField(p);
805 static void flushSortKeys (SYSNO sysno, int cmd)
807 struct sortKey *sk = sortKeys;
809 sortIdx_sysno (sortIdx, sysno);
812 struct sortKey *sk_next = sk->next;
813 sortIdx_type (sortIdx, sk->attrUse);
814 sortIdx_add (sortIdx, sk->string, sk->length);
822 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
829 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
830 while (off < reckeys->buf_used)
832 const char *src = reckeys->buf + off;
840 memcpy (&attrSet, src, sizeof(attrSet));
841 src += sizeof(attrSet);
845 memcpy (&attrUse, src, sizeof(attrUse));
846 src += sizeof(attrUse);
848 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
851 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
853 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
855 ch = zebraExplain_addSU (zti, attrSet, attrUse);
857 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
860 ((char*)key_buf) [key_buf_used++] = *src++;
862 ((char*)key_buf) [key_buf_used++] = '\0';
863 ((char*) key_buf)[key_buf_used++] = cmd;
866 seqno += ((lead>>2) & 15)-1;
869 memcpy (&seqno, src, sizeof(seqno));
870 src += sizeof(seqno);
874 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
875 key_buf_used += sizeof(key);
876 off = src - reckeys->buf;
878 assert (off == reckeys->buf_used);
881 static const char **searchRecordKey (struct recKeys *reckeys,
882 int attrSetS, int attrUseS)
884 static const char *ws[32];
892 for (i = 0; i<32; i++)
895 while (off < reckeys->buf_used)
898 const char *src = reckeys->buf + off;
906 memcpy (&attrSet, src, sizeof(attrSet));
907 src += sizeof(attrSet);
911 memcpy (&attrUse, src, sizeof(attrUse));
912 src += sizeof(attrUse);
918 seqno += ((lead>>2) & 15)-1;
921 memcpy (&seqno, src, sizeof(seqno));
922 src += sizeof(seqno);
924 if (attrUseS == attrUse && attrSetS == attrSet)
931 woff = seqno - startSeq;
932 if (woff >= 0 && woff < 31)
936 off = src - reckeys->buf;
938 assert (off == reckeys->buf_used);
942 struct file_read_info {
950 static struct file_read_info *file_read_start (int fd)
952 struct file_read_info *fi = xmalloc (sizeof(*fi));
956 fi->file_moffset = 0;
960 static void file_read_stop (struct file_read_info *fi)
965 static off_t file_seek (void *handle, off_t offset)
967 struct file_read_info *p = handle;
968 p->file_offset = offset;
969 return lseek (p->fd, offset, SEEK_SET);
972 static off_t file_tell (void *handle)
974 struct file_read_info *p = handle;
975 return p->file_offset;
978 static int file_read (void *handle, char *buf, size_t count)
980 struct file_read_info *p = handle;
983 r = read (fd, buf, count);
987 if (p->file_offset > p->file_max)
988 p->file_max = p->file_offset;
993 static void file_begin (void *handle)
995 struct file_read_info *p = handle;
997 p->file_offset = p->file_moffset;
999 lseek (p->fd, p->file_moffset, SEEK_SET);
1003 static void file_end (void *handle, off_t offset)
1005 struct file_read_info *p = handle;
1007 assert (p->file_more == 0);
1009 p->file_moffset = offset;
1012 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1013 const char *fname, const char *spec)
1015 static char dstBuf[2048];
1017 const char *s = spec;
1018 static const char **w;
1022 while (*s == ' ' || *s == '\t')
1028 char attset_str[64], attname_str[64];
1029 data1_attset *attset;
1032 int attSet = 1, attUse = 1;
1036 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1038 attset_str[i++] = *s;
1039 attset_str[i] = '\0';
1044 for (i = 0; *s && *s != ')'; s++)
1046 attname_str[i++] = *s;
1047 attname_str[i] = '\0';
1050 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1053 attSet = attset->reference;
1054 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1056 attUse = att->value;
1058 attUse = atoi (attname_str);
1060 w = searchRecordKey (reckeys, attSet, attUse);
1065 for (i = 0; i<32; i++)
1070 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1071 spec, rGroup->groupName ? rGroup->groupName : "none");
1076 for (i = 0; i<32; i++)
1077 if (matchFlag[i] && w[i])
1085 dst += strlen(w[i]);
1089 logf (LOG_WARN, "Record didn't contain match"
1090 " fields in (%s,%s)", attset_str, attname_str);
1098 const char *spec_src = NULL;
1099 const char *s1 = ++s;
1100 while (*s1 && *s1 != ' ' && *s1 != '\t')
1106 memcpy (special, s, spec_len);
1107 special[spec_len] = '\0';
1110 if (!strcmp (special, "group"))
1111 spec_src = rGroup->groupName;
1112 else if (!strcmp (special, "database"))
1113 spec_src = rGroup->databaseName;
1114 else if (!strcmp (special, "filename"))
1116 else if (!strcmp (special, "type"))
1117 spec_src = rGroup->recordType;
1122 strcpy (dst, spec_src);
1123 dst += strlen (spec_src);
1126 else if (*s == '\"' || *s == '\'')
1128 int stopMarker = *s++;
1132 while (*s && *s != stopMarker)
1135 tmpString[i++] = *s++;
1139 tmpString[i] = '\0';
1140 strcpy (dst, tmpString);
1141 dst += strlen (tmpString);
1145 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1146 spec, rGroup->groupName ? rGroup->groupName : "none");
1153 logf (LOG_WARN, "No match criteria for record %s in group %s",
1154 fname, rGroup->groupName ? rGroup->groupName : "none");
1161 struct recordLogInfo {
1164 struct recordGroup *rGroup;
1167 static void recordLogPreamble (int level, const char *msg, void *info)
1169 struct recordLogInfo *p = info;
1170 FILE *outf = log_file ();
1172 if (level & LOG_LOG)
1174 fprintf (outf, "File %s, offset %d, type %s\n",
1175 p->rGroup->recordType, p->recordOffset, p->fname);
1176 log_event_start (NULL, NULL);
1179 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1181 zebraExplain_addSchema (zti, oid);
1184 static int recordExtract (SYSNO *sysno, const char *fname,
1185 struct recordGroup *rGroup, int deleteFlag,
1186 struct file_read_info *fi,
1187 RecType recType, char *subType)
1189 RecordAttr *recordAttr;
1194 struct recordLogInfo logInfo;
1195 off_t recordOffset = 0;
1199 struct recExtractCtrl extractCtrl;
1201 /* we are going to read from a file, so prepare the extraction */
1204 reckeys.buf_used = 0;
1205 reckeys.prevAttrUse = -1;
1206 reckeys.prevAttrSet = -1;
1207 reckeys.prevSeqNo = 0;
1209 recordOffset = fi->file_moffset;
1210 extractCtrl.offset = fi->file_moffset;
1211 extractCtrl.readf = file_read;
1212 extractCtrl.seekf = file_seek;
1213 extractCtrl.tellf = file_tell;
1214 extractCtrl.endf = file_end;
1215 extractCtrl.fh = fi;
1216 extractCtrl.subType = subType;
1217 extractCtrl.init = wordInit;
1218 extractCtrl.addWord = addRecordKey;
1219 extractCtrl.addSchema = addSchema;
1220 extractCtrl.dh = rGroup->dh;
1221 for (i = 0; i<256; i++)
1222 extractCtrl.seqno[i] = 0;
1223 extractCtrl.zebra_maps = rGroup->zebra_maps;
1224 extractCtrl.flagShowRecords = !rGroup->flagRw;
1226 if (!rGroup->flagRw)
1227 printf ("File: %s %ld\n", fname, (long) recordOffset);
1229 logInfo.fname = fname;
1230 logInfo.recordOffset = recordOffset;
1231 logInfo.rGroup = rGroup;
1232 log_event_start (recordLogPreamble, &logInfo);
1234 r = (*recType->extract)(&extractCtrl);
1236 log_event_start (NULL, NULL);
1240 /* error occured during extraction ... */
1241 if (rGroup->flagRw &&
1242 records_processed < rGroup->fileVerboseLimit)
1244 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1245 fname, (long) recordOffset, r);
1249 if (reckeys.buf_used == 0)
1251 /* the extraction process returned no information - the record
1252 is probably empty - unless flagShowRecords is in use */
1253 if (!rGroup->flagRw)
1255 logf (LOG_WARN, "No keys generated for file %s", fname);
1256 logf (LOG_WARN, " The file is probably empty");
1261 /* perform match if sysno not known and if match criteria is specified */
1268 if (rGroup->recordId && *rGroup->recordId)
1272 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1276 rinfo = dict_lookup (matchDict, matchStr);
1278 memcpy (sysno, rinfo+1, sizeof(*sysno));
1282 logf (LOG_WARN, "Bad match criteria");
1293 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1294 fname, (long) recordOffset);
1295 logf (LOG_WARN, "cannot delete record above (seems new)");
1298 if (records_processed < rGroup->fileVerboseLimit)
1299 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1300 fname, (long) recordOffset);
1301 rec = rec_new (records);
1303 *sysno = rec->sysno;
1305 recordAttr = rec_init_attr (zti, rec);
1309 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1311 flushRecordKeys (*sysno, 1, &reckeys);
1312 flushSortKeys (*sysno, 1);
1318 /* record already exists */
1319 struct recKeys delkeys;
1321 rec = rec_get (records, *sysno);
1324 recordAttr = rec_init_attr (zti, rec);
1326 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1328 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1329 fname, (long) recordOffset);
1334 delkeys.buf_used = rec->size[recInfo_delKeys];
1335 delkeys.buf = rec->info[recInfo_delKeys];
1336 flushSortKeys (*sysno, 0);
1337 flushRecordKeys (*sysno, 0, &delkeys);
1340 /* record going to be deleted */
1341 if (!delkeys.buf_used)
1343 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1344 fname, (long) recordOffset);
1345 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1349 if (records_processed < rGroup->fileVerboseLimit)
1350 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1351 fname, (long) recordOffset);
1354 dict_delete (matchDict, matchStr);
1355 rec_del (records, &rec);
1363 /* record going to be updated */
1364 if (!delkeys.buf_used)
1366 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1367 fname, (long) recordOffset);
1368 logf (LOG_WARN, "cannot update file above, storeKeys false");
1372 if (records_processed < rGroup->fileVerboseLimit)
1373 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1374 fname, (long) recordOffset);
1375 flushRecordKeys (*sysno, 1, &reckeys);
1380 /* update file type */
1381 xfree (rec->info[recInfo_fileType]);
1382 rec->info[recInfo_fileType] =
1383 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1385 /* update filename */
1386 xfree (rec->info[recInfo_filename]);
1387 rec->info[recInfo_filename] =
1388 rec_strdup (fname, &rec->size[recInfo_filename]);
1390 /* update delete keys */
1391 xfree (rec->info[recInfo_delKeys]);
1392 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1395 rec->size[recInfo_delKeys] = reckeys.buf_used;
1396 rec->info[recInfo_delKeys] = reckeys.buf;
1398 reckeys.buf_max = 0;
1400 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1401 rec->size[recInfo_delKeys] = reckeys.buf_used;
1402 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1403 rec->size[recInfo_delKeys]);
1408 rec->info[recInfo_delKeys] = NULL;
1409 rec->size[recInfo_delKeys] = 0;
1412 /* save file size of original record */
1413 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1414 recordAttr->recordSize = fi->file_moffset - recordOffset;
1415 if (!recordAttr->recordSize)
1416 recordAttr->recordSize = fi->file_max - recordOffset;
1417 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1419 /* set run-number for this record */
1420 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1422 /* update store data */
1423 xfree (rec->info[recInfo_storeData]);
1424 if (rGroup->flagStoreData == 1)
1426 rec->size[recInfo_storeData] = recordAttr->recordSize;
1427 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1428 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1430 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", fname,
1431 (long) recordOffset);
1434 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1435 < recordAttr->recordSize)
1437 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1438 recordAttr->recordSize, fname);
1444 rec->info[recInfo_storeData] = NULL;
1445 rec->size[recInfo_storeData] = 0;
1447 /* update database name */
1448 xfree (rec->info[recInfo_databaseName]);
1449 rec->info[recInfo_databaseName] =
1450 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1453 recordAttr->recordOffset = recordOffset;
1455 /* commit this record */
1456 rec_put (records, &rec);
1461 int fileExtract (SYSNO *sysno, const char *fname,
1462 const struct recordGroup *rGroupP, int deleteFlag)
1470 struct recordGroup rGroupM;
1471 struct recordGroup *rGroup = &rGroupM;
1472 struct file_read_info *fi;
1474 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1476 if (!rGroup->groupName || !*rGroup->groupName)
1479 sprintf (gprefix, "%s.", rGroup->groupName);
1481 logf (LOG_DEBUG, "fileExtract %s", fname);
1483 /* determine file extension */
1485 for (i = strlen(fname); --i >= 0; )
1486 if (fname[i] == '/')
1488 else if (fname[i] == '.')
1490 strcpy (ext, fname+i+1);
1493 /* determine file type - depending on extension */
1494 if (!rGroup->recordType)
1496 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1497 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1499 sprintf (ext_res, "%srecordType", gprefix);
1500 rGroup->recordType = res_get (common_resource, ext_res);
1503 if (!rGroup->recordType)
1505 if (records_processed < rGroup->fileVerboseLimit)
1506 logf (LOG_LOG, "? %s", fname);
1509 if (!*rGroup->recordType)
1511 if (!(recType = recType_byName (rGroup->recordType, subType)))
1513 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1517 /* determine match criteria */
1518 if (!rGroup->recordId)
1520 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1521 rGroup->recordId = res_get (common_resource, ext_res);
1524 /* determine database name */
1525 if (!rGroup->databaseName)
1527 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1528 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1530 sprintf (ext_res, "%sdatabase", gprefix);
1531 rGroup->databaseName = res_get (common_resource, ext_res);
1534 if (!rGroup->databaseName)
1535 rGroup->databaseName = "Default";
1537 /* determine if explain database */
1539 sprintf (ext_res, "%sexplainDatabase", gprefix);
1540 rGroup->explainDatabase =
1541 atoi (res_get_def (common_resource, ext_res, "0"));
1543 /* announce database */
1544 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1546 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1547 rGroup->explainDatabase))
1551 if (rGroup->flagStoreData == -1)
1554 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1555 if (!(sval = res_get (common_resource, ext_res)))
1557 sprintf (ext_res, "%sstoreData", gprefix);
1558 sval = res_get (common_resource, ext_res);
1561 rGroup->flagStoreData = atoi (sval);
1563 if (rGroup->flagStoreData == -1)
1564 rGroup->flagStoreData = 0;
1566 if (rGroup->flagStoreKeys == -1)
1570 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1571 if (!(sval = res_get (common_resource, ext_res)))
1573 sprintf (ext_res, "%sstoreKeys", gprefix);
1574 sval = res_get (common_resource, ext_res);
1577 rGroup->flagStoreKeys = atoi (sval);
1579 if (rGroup->flagStoreKeys == -1)
1580 rGroup->flagStoreKeys = 0;
1582 if (sysno && deleteFlag)
1586 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1588 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1592 fi = file_read_start (fd);
1596 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1598 } while (r && !sysno && fi->file_more);
1599 file_read_stop (fi);
1605 static int explain_extract (void *handle, Record rec, data1_node *n)
1607 struct recordGroup *rGroup = (struct recordGroup*) handle;
1608 struct recExtractCtrl extractCtrl;
1611 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1613 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1617 reckeys.buf_used = 0;
1618 reckeys.prevAttrUse = -1;
1619 reckeys.prevAttrSet = -1;
1620 reckeys.prevSeqNo = 0;
1622 extractCtrl.init = wordInit;
1623 extractCtrl.addWord = addRecordKey;
1624 extractCtrl.addSchema = addSchema;
1625 extractCtrl.dh = rGroup->dh;
1626 for (i = 0; i<256; i++)
1627 extractCtrl.seqno[i] = 0;
1628 extractCtrl.zebra_maps = rGroup->zebra_maps;
1629 extractCtrl.flagShowRecords = !rGroup->flagRw;
1631 grs_extract_tree(&extractCtrl, n);
1633 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1635 if (rec->size[recInfo_delKeys])
1637 struct recKeys delkeys;
1639 delkeys.buf_used = rec->size[recInfo_delKeys];
1640 delkeys.buf = rec->info[recInfo_delKeys];
1641 flushSortKeys (rec->sysno, 0);
1642 flushRecordKeys (rec->sysno, 0, &delkeys);
1644 flushRecordKeys (rec->sysno, 1, &reckeys);
1645 flushSortKeys (rec->sysno, 1);
1647 xfree (rec->info[recInfo_delKeys]);
1648 rec->size[recInfo_delKeys] = reckeys.buf_used;
1649 rec->info[recInfo_delKeys] = reckeys.buf;
1651 reckeys.buf_max = 0;