2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.93 1999-05-15 14:36:38 adam
8 * Updated dictionary. Implemented "compression" of dictionary.
10 * Revision 1.92 1999/03/09 16:27:49 adam
11 * More work on SDRKit integration.
13 * Revision 1.91 1999/02/12 13:29:22 adam
14 * Implemented position-flag for registers.
16 * Revision 1.90 1999/02/02 14:50:52 adam
17 * Updated WIN32 code specific sections. Changed header.
19 * Revision 1.89 1998/10/28 10:54:38 adam
22 * Revision 1.88 1998/10/16 08:14:29 adam
23 * Updated record control system.
25 * Revision 1.87 1998/10/15 13:10:33 adam
26 * Fixed bug in Zebra that caused it to stop indexing when empty
29 * Revision 1.86 1998/10/13 20:33:53 adam
30 * Fixed one log message and change use ordinal to be an unsigned char.
32 * Revision 1.85 1998/09/22 10:03:41 adam
33 * Changed result sets to be persistent in the sense that they can
34 * be re-searched if needed.
35 * Fixed memory leak in rsm_or.
37 * Revision 1.84 1998/06/11 15:42:22 adam
38 * Changed the way use attributes are specified in the recordId
41 * Revision 1.83 1998/06/08 14:43:10 adam
42 * Added suport for EXPLAIN Proxy servers - added settings databasePath
43 * and explainDatabase to facilitate this. Increased maximum number
44 * of databases and attributes in one register.
46 * Revision 1.82 1998/05/20 10:12:15 adam
47 * Implemented automatic EXPLAIN database maintenance.
48 * Modified Zebra to work with ASN.1 compiled version of YAZ.
50 * Revision 1.81 1998/03/11 11:19:04 adam
51 * Changed the way sequence numbers are generated.
53 * Revision 1.80 1998/03/05 08:45:11 adam
54 * New result set model and modular ranking system. Moved towards
55 * descent server API. System information stored as "SGML" records.
57 * Revision 1.79 1998/02/17 10:32:52 adam
58 * Fixed bug: binary files weren't opened with flag b on NT.
60 * Revision 1.78 1998/02/10 12:03:05 adam
63 * Revision 1.77 1998/01/12 15:04:08 adam
64 * The test option (-s) only uses read-lock (and not write lock).
66 * Revision 1.76 1997/10/27 14:33:04 adam
67 * Moved towards generic character mapping depending on "structure"
68 * field in abstract syntax file. Fixed a few memory leaks. Fixed
69 * bug with negative integers when doing searches with relational
72 * Revision 1.75 1997/09/17 12:19:12 adam
73 * Zebra version corresponds to YAZ version 1.4.
74 * Changed Zebra server so that it doesn't depend on global common_resource.
76 * Revision 1.74 1997/09/09 13:38:06 adam
77 * Partial port to WIN95/NT.
79 * Revision 1.73 1997/09/04 13:57:20 adam
80 * New file extract/retrieve method tellf (added).
81 * Added O_BINARY for open calls.
83 * Revision 1.72 1997/07/15 16:32:29 adam
84 * Bug fix: Match handler didn't terminate the resulting string!
86 * Revision 1.71 1997/07/15 16:28:41 adam
87 * Bug fix: storeData didn't work with files with multiple records.
88 * Bug fix: fixed memory management with records; not really well
91 * Revision 1.70 1997/07/01 13:00:42 adam
92 * Bug fix in routine searchRecordKey: uninitialized variables.
94 * Revision 1.69 1997/04/29 09:26:03 adam
95 * Bug fix: generic recordId handling didn't work for compressed internal
98 * Revision 1.68 1997/02/12 20:39:45 adam
99 * Implemented options -f <n> that limits the log to the first <n>
101 * Changed some log messages also.
103 * Revision 1.67 1996/11/15 15:02:14 adam
104 * Minor changes regarding logging.
106 * Revision 1.66 1996/11/14 09:52:21 adam
107 * Strings in record keys bound by IT_MAX_WORD.
109 * Revision 1.65 1996/11/14 08:57:56 adam
110 * Reduction of storeKeys area.
112 * Revision 1.64 1996/11/08 11:10:16 adam
113 * Buffers used during file match got bigger.
114 * Compressed ISAM support everywhere.
115 * Bug fixes regarding masking characters in queries.
116 * Redesigned Regexp-2 queries.
118 * Revision 1.63 1996/10/29 14:09:39 adam
119 * Use of cisam system - enabled if setting isamc is 1.
121 * Revision 1.62 1996/10/11 10:57:01 adam
122 * New module recctrl. Used to manage records (extract/retrieval).
123 * Several files have been moved to the recctrl sub directory.
125 * Revision 1.61 1996/06/06 12:08:37 quinn
126 * Added showRecord function
128 * Revision 1.60 1996/06/04 10:18:12 adam
129 * Search/scan uses character mapping module.
131 * Revision 1.59 1996/05/14 15:47:07 adam
132 * Cleanup of various buffer size entities.
134 * Revision 1.58 1996/05/14 06:16:38 adam
135 * Compact use/set bytes used in search service.
137 * Revision 1.57 1996/05/13 14:23:04 adam
138 * Work on compaction of set/use bytes in dictionary.
140 * Revision 1.56 1996/05/09 09:54:42 adam
141 * Server supports maps from one logical attributes to a list of physical
143 * The extraction process doesn't make space consuming 'any' keys.
145 * Revision 1.55 1996/05/09 07:28:55 quinn
146 * Work towards phrases and multiple registers
148 * Revision 1.54 1996/05/01 13:46:35 adam
149 * First work on multiple records in one file.
150 * New option, -offset, to the "unread" command in the filter module.
152 * Revision 1.53 1996/04/26 12:09:43 adam
153 * Added a few comments.
155 * Revision 1.52 1996/04/25 13:27:57 adam
156 * Function recordExtract modified so that files with no keys (possibly empty)
159 * Revision 1.51 1996/03/19 11:08:42 adam
160 * Bug fix: Log preamble wasn't always turned off after recordExtract.
162 * Revision 1.50 1996/02/12 18:45:36 adam
163 * New fileVerboseFlag in record group control.
165 * Revision 1.49 1996/02/05 12:29:57 adam
166 * Logging reduced a bit.
167 * The remaining running time is estimated during register merge.
169 * Revision 1.48 1996/02/01 20:53:26 adam
170 * The temporary per-record keys are compacted a little, and duplication
171 * of the per-records keys are avoided when they are saved in the record
172 * information buffer.
174 * Revision 1.47 1996/01/17 14:57:48 adam
175 * Prototype changed for reader functions in extract/retrieve. File
176 * is identified by 'void *' instead of 'int.
178 * Revision 1.46 1995/12/15 14:57:16 adam
181 * Revision 1.45 1995/12/15 12:37:41 adam
182 * In addRecordKeyAny: Writes key only when attrSet != -1.
184 * Revision 1.44 1995/12/12 16:00:54 adam
185 * System call sync(2) used after update/commit.
186 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
189 * Revision 1.43 1995/12/11 09:12:46 adam
190 * The rec_get function returns NULL if record doesn't exist - will
191 * happen in the server if the result set records have been deleted since
192 * the creation of the set (i.e. the search).
193 * The server saves a result temporarily if it is 'volatile', i.e. the
194 * set is register dependent.
196 * Revision 1.42 1995/12/07 17:38:46 adam
197 * Work locking mechanisms for concurrent updates/commit.
199 * Revision 1.41 1995/12/06 16:06:42 adam
200 * Better diagnostics. Work on 'real' dictionary deletion.
202 * Revision 1.40 1995/12/05 16:57:40 adam
203 * More work on regular patterns.
205 * Revision 1.39 1995/12/05 13:20:18 adam
206 * Bug fix: file_read sometimes returned early EOF.
208 * Revision 1.38 1995/12/04 17:59:21 adam
209 * More work on regular expression conversion.
211 * Revision 1.37 1995/12/04 14:22:27 adam
212 * Extra arg to recType_byName.
213 * Started work on new regular expression parsed input to
214 * structured records.
216 * Revision 1.36 1995/11/30 08:34:29 adam
217 * Started work on commit facility.
218 * Changed a few malloc/free to xmalloc/xfree.
220 * Revision 1.35 1995/11/28 14:26:21 adam
221 * Bug fix: recordId with constant wasn't right.
222 * Bug fix: recordId dictionary entry wasn't deleted when needed.
224 * Revision 1.34 1995/11/28 09:09:38 adam
225 * Zebra config renamed.
226 * Use setting 'recordId' to identify record now.
227 * Bug fix in recindex.c: rec_release_blocks was invokeded even
228 * though the blocks were already released.
229 * File traversal properly deletes records when needed.
231 * Revision 1.33 1995/11/27 09:56:20 adam
232 * Record info elements better enumerated. Internal store of records.
234 * Revision 1.32 1995/11/25 10:24:05 adam
235 * More record fields - they are enumerated now.
236 * New options: flagStoreData flagStoreKey.
238 * Revision 1.31 1995/11/24 11:31:35 adam
239 * Commands add & del read filenames from stdin if source directory is
241 * Match criteria supports 'constant' strings.
243 * Revision 1.30 1995/11/22 17:19:16 adam
244 * Record management uses the bfile system.
246 * Revision 1.29 1995/11/21 15:01:14 adam
247 * New general match criteria implemented.
248 * New feature: document groups.
250 * Revision 1.28 1995/11/21 09:20:30 adam
251 * Yet more work on record match.
253 * Revision 1.27 1995/11/20 16:59:45 adam
254 * New update method: the 'old' keys are saved for each records.
256 * Revision 1.26 1995/11/20 11:56:24 adam
257 * Work on new traversal.
259 * Revision 1.25 1995/11/16 15:34:54 adam
260 * Uses new record management system in both indexer and server.
262 * Revision 1.24 1995/11/15 19:13:08 adam
263 * Work on record management.
265 * Revision 1.23 1995/10/27 14:00:10 adam
266 * Implemented detection of database availability.
268 * Revision 1.22 1995/10/17 18:02:07 adam
269 * New feature: databases. Implemented as prefix to words in dictionary.
271 * Revision 1.21 1995/10/10 12:24:38 adam
272 * Temporary sort files are compressed.
274 * Revision 1.20 1995/10/06 13:52:05 adam
275 * Bug fixes. Handler may abort further scanning.
277 * Revision 1.19 1995/10/04 12:55:16 adam
278 * Bug fix in ranked search. Use=Any keys inserted.
280 * Revision 1.18 1995/10/04 09:37:08 quinn
283 * Revision 1.17 1995/10/03 14:28:57 adam
284 * Buffered read in extract works.
286 * Revision 1.16 1995/10/03 14:28:45 adam
287 * Work on more effecient read handler in extract.
289 * Revision 1.15 1995/10/02 15:42:53 adam
290 * Extract uses file descriptors instead of FILE pointers.
292 * Revision 1.14 1995/10/02 15:29:13 adam
293 * More logging in file_extract.
295 * Revision 1.13 1995/09/29 14:01:39 adam
298 * Revision 1.12 1995/09/28 14:22:56 adam
299 * Sort uses smaller temporary files.
301 * Revision 1.11 1995/09/28 12:10:31 adam
302 * Bug fixes. Field prefix used in queries.
304 * Revision 1.10 1995/09/28 09:19:41 adam
305 * xfree/xmalloc used everywhere.
306 * Extract/retrieve method seems to work for text records.
308 * Revision 1.9 1995/09/27 12:22:28 adam
309 * More work on extract in record control.
310 * Field name is not in isam keys but in prefix in dictionary words.
312 * Revision 1.8 1995/09/14 07:48:22 adam
313 * Record control management.
315 * Revision 1.7 1995/09/11 13:09:32 adam
316 * More work on relevance feedback.
318 * Revision 1.6 1995/09/08 14:52:27 adam
319 * Minor changes. Dictionary is lower case now.
321 * Revision 1.5 1995/09/06 16:11:16 adam
322 * Option: only one word key per file.
324 * Revision 1.4 1995/09/05 15:28:39 adam
325 * More work on search engine.
327 * Revision 1.3 1995/09/04 12:33:41 adam
328 * Various cleanup. YAZ util used instead.
330 * Revision 1.2 1995/09/04 09:10:34 adam
331 * More work on index add/del/update.
332 * Merge sort implemented.
333 * Initial work on z39 server.
335 * Revision 1.1 1995/09/01 14:06:35 adam
336 * Split of work into more files.
360 #include "zebrasdr.h"
363 static Dict matchDict;
365 static Records records = NULL;
366 static SortIdx sortIdx = NULL;
368 static char **key_buf;
369 static size_t ptr_top;
371 static size_t key_buf_used;
372 static int key_file_no;
374 static int records_inserted = 0;
375 static int records_updated = 0;
376 static int records_deleted = 0;
377 static int records_processed = 0;
379 static ZebraExplainInfo zti = NULL;
381 static void logRecord (int showFlag)
385 if (showFlag || !(records_processed % 1000))
387 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
388 records_processed, records_inserted, records_updated,
393 static int explain_extract (void *handle, Record drec, data1_node *n);
395 int key_open (struct recordGroup *rGroup, int mem)
397 BFiles bfs = rGroup->bfs;
398 int rw = rGroup->flagRw;
399 data1_handle dh = rGroup->dh;
401 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
404 key_buf = xmalloc (mem);
405 ptr_top = mem/sizeof(char*);
411 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
413 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
417 records = rec_open (bfs, rw);
420 dict_close (matchDict);
423 zti = zebraExplain_open (records, dh, common_resource,
424 rw, rGroup, explain_extract);
427 rec_close (&records);
428 dict_close (matchDict);
431 sortIdx = sortIdx_open (bfs, 1);
442 void encode_key_init (struct encode_info *i)
449 char *encode_key_int (int d, char *bp)
458 else if (d <= 4194303)
460 *bp++ = 128 + (d>>16);
461 *bp++ = (d>>8) & 255;
466 *bp++ = 192 + (d>>24);
467 *bp++ = (d>>16) & 255;
468 *bp++ = (d>>8) & 255;
474 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
479 while ((*bp++ = *k++))
481 memcpy (&key, k+1, sizeof(struct it_key));
482 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
483 if (i->sysno != key.sysno)
485 i->sysno = key.sysno;
488 else if (!i->seqno && !key.seqno && i->cmd == *k)
490 bp = encode_key_int (key.seqno - i->seqno, bp);
491 i->seqno = key.seqno;
493 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
495 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
503 static int key_y_len;
505 static int key_y_compare (const void *p1, const void *p2)
509 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
510 *(char**) p2 + key_y_len + 1)))
512 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
515 static int key_x_compare (const void *p1, const void *p2)
517 return strcmp (*(char**) p1, *(char**) p2);
521 void key_flush (void)
526 struct encode_info encode_info;
535 logf (LOG_LOG, "sorting section %d", key_file_no);
537 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
538 getFnameTmp (out_fname, key_file_no);
540 if (!(outf = fopen (out_fname, "wb")))
542 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
545 logf (LOG_LOG, "writing section %d", key_file_no);
546 prevcp = 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 if (strcmp (cp, prevcp))
555 encode_key_init (&encode_info);
556 encode_key_write (cp, &encode_info, outf);
560 encode_key_write (cp + strlen(cp), &encode_info, outf);
563 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
564 getFnameTmp (out_fname, key_file_no);
566 if (!(outf = fopen (out_fname, "wb")))
568 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
571 logf (LOG_LOG, "writing section %d", key_file_no);
573 prevcp = key_buf[ptr_top-i];
575 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
577 key_y_len = strlen(prevcp)+1;
579 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
580 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
582 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
583 sizeof(char*), key_y_compare);
584 cp = key_buf[ptr_top-ptr_i];
586 encode_key_init (&encode_info);
587 encode_key_write (cp, &encode_info, outf);
590 cp = key_buf[ptr_top-ptr_i];
591 encode_key_write (cp+key_y_len, &encode_info, outf);
595 prevcp = key_buf[ptr_top-ptr_i];
600 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
603 logf (LOG_LOG, "finished section %d", key_file_no);
608 int key_close (struct recordGroup *rGroup)
610 int rw = rGroup->flagRw;
612 zebraExplain_runNumberIncrement (zti, 1);
613 zebraExplain_close (zti, rw, 0);
616 rec_close (&records);
617 dict_close (matchDict);
618 sortIdx_close (sortIdx);
624 static void wordInit (struct recExtractCtrl *p, RecWord *w)
626 w->zebra_maps = p->zebra_maps;
627 w->seqnos = p->seqno;
628 w->attrSet = VAL_BIB1;
633 static struct sortKey {
638 struct sortKey *next;
641 static struct recKeys {
650 static void addIndexString (RecWord *p, const char *string, int length)
653 unsigned char attrSet;
654 unsigned short attrUse;
657 int *pseqno = &p->seqnos[p->reg_type];
659 if (reckeys.buf_used+1024 > reckeys.buf_max)
663 b = xmalloc (reckeys.buf_max += 128000);
664 if (reckeys.buf_used > 0)
665 memcpy (b, reckeys.buf, reckeys.buf_used);
669 dst = reckeys.buf + reckeys.buf_used;
671 attrSet = p->attrSet;
672 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
675 reckeys.prevAttrSet = attrSet;
676 attrUse = p->attrUse;
677 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
680 reckeys.prevAttrUse = attrUse;
682 diff = 1 + *pseqno - reckeys.prevSeqNo;
683 if (diff >= 1 && diff <= 15)
688 reckeys.prevSeqNo = *pseqno;
694 memcpy (dst, &attrSet, sizeof(attrSet));
695 dst += sizeof(attrSet);
699 memcpy (dst, &attrUse, sizeof(attrUse));
700 dst += sizeof(attrUse);
702 *dst++ = p->reg_type;
703 memcpy (dst, string, length);
709 memcpy (dst, pseqno, sizeof(*pseqno));
710 dst += sizeof(*pseqno);
712 reckeys.buf_used = dst - reckeys.buf;
717 static void addSortString (RecWord *p, const char *string, int length)
721 for (sk = sortKeys; sk; sk = sk->next)
722 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
725 sk = xmalloc (sizeof(*sk));
729 sk->string = xmalloc (length);
731 memcpy (sk->string, string, length);
733 sk->attrSet = p->attrSet;
734 sk->attrUse = p->attrUse;
737 static void addString (RecWord *p, const char *string, int length)
740 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
741 addSortString (p, string, length);
743 addIndexString (p, string, length);
746 static void addIncompleteField (RecWord *p)
748 const char *b = p->string;
749 int remain = p->length;
750 const char **map = 0;
753 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
757 char buf[IT_MAX_WORD+1];
761 while (map && *map && **map == *CHR_SPACE)
763 remain = p->length - (b - p->string);
765 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
772 while (map && *map && **map != *CHR_SPACE)
774 const char *cp = *map;
776 while (i < IT_MAX_WORD && *cp)
778 remain = p->length - (b - p->string);
780 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
786 addString (p, buf, i);
790 static void addCompleteField (RecWord *p)
792 const char *b = p->string;
793 char buf[IT_MAX_WORD+1];
794 const char **map = 0;
795 int i = 0, remain = p->length;
798 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
800 while (remain > 0 && i < IT_MAX_WORD)
802 while (map && *map && **map == *CHR_SPACE)
804 remain = p->length - (b - p->string);
806 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
813 if (i && i < IT_MAX_WORD)
814 buf[i++] = *CHR_SPACE;
815 while (map && *map && **map != *CHR_SPACE)
817 const char *cp = *map;
819 if (i >= IT_MAX_WORD)
821 while (i < IT_MAX_WORD && *cp)
823 remain = p->length - (b - p->string);
825 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
833 addString (p, buf, i);
836 static void addRecordKey (RecWord *p)
838 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
839 addCompleteField (p);
841 addIncompleteField(p);
844 static void flushSortKeys (SYSNO sysno, int cmd)
846 struct sortKey *sk = sortKeys;
848 sortIdx_sysno (sortIdx, sysno);
851 struct sortKey *sk_next = sk->next;
852 sortIdx_type (sortIdx, sk->attrUse);
853 sortIdx_add (sortIdx, sk->string, sk->length);
861 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
863 unsigned char attrSet = (unsigned char) -1;
864 unsigned short attrUse = (unsigned short) -1;
868 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
869 while (off < reckeys->buf_used)
871 const char *src = reckeys->buf + off;
879 memcpy (&attrSet, src, sizeof(attrSet));
880 src += sizeof(attrSet);
884 memcpy (&attrUse, src, sizeof(attrUse));
885 src += sizeof(attrUse);
887 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
890 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
892 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
894 ch = zebraExplain_addSU (zti, attrSet, attrUse);
896 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
899 ((char*)key_buf) [key_buf_used++] = *src++;
901 ((char*)key_buf) [key_buf_used++] = '\0';
902 ((char*) key_buf)[key_buf_used++] = cmd;
905 seqno += ((lead>>2) & 15)-1;
908 memcpy (&seqno, src, sizeof(seqno));
909 src += sizeof(seqno);
913 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
914 key_buf_used += sizeof(key);
915 off = src - reckeys->buf;
917 assert (off == reckeys->buf_used);
920 static const char **searchRecordKey (struct recKeys *reckeys,
921 int attrSetS, int attrUseS)
923 static const char *ws[32];
931 for (i = 0; i<32; i++)
934 while (off < reckeys->buf_used)
937 const char *src = reckeys->buf + off;
945 memcpy (&attrSet, src, sizeof(attrSet));
946 src += sizeof(attrSet);
950 memcpy (&attrUse, src, sizeof(attrUse));
951 src += sizeof(attrUse);
957 seqno += ((lead>>2) & 15)-1;
960 memcpy (&seqno, src, sizeof(seqno));
961 src += sizeof(seqno);
963 if (attrUseS == attrUse && attrSetS == attrSet)
970 woff = seqno - startSeq;
971 if (woff >= 0 && woff < 31)
975 off = src - reckeys->buf;
977 assert (off == reckeys->buf_used);
981 struct file_read_info {
982 off_t file_max; /* maximum offset so far */
983 off_t file_offset; /* current offset */
984 off_t file_moffset; /* offset of rec/rec boundary */
991 static struct file_read_info *file_read_start (int fd)
993 struct file_read_info *fi = xmalloc (sizeof(*fi));
997 fi->file_moffset = 0;
1003 static void file_read_stop (struct file_read_info *fi)
1008 static off_t file_seek (void *handle, off_t offset)
1010 struct file_read_info *p = handle;
1011 p->file_offset = offset;
1014 return lseek (p->fd, offset, SEEK_SET);
1017 static off_t file_tell (void *handle)
1019 struct file_read_info *p = handle;
1020 return p->file_offset;
1023 static int file_read (void *handle, char *buf, size_t count)
1025 struct file_read_info *p = handle;
1031 if (r > p->sdrmax - p->file_offset)
1032 r = p->sdrmax - p->file_offset;
1034 memcpy (buf, p->sdrbuf + p->file_offset, r);
1037 r = read (fd, buf, count);
1040 p->file_offset += r;
1041 if (p->file_offset > p->file_max)
1042 p->file_max = p->file_offset;
1047 static void file_begin (void *handle)
1049 struct file_read_info *p = handle;
1051 p->file_offset = p->file_moffset;
1052 if (!p->sdrbuf && p->file_moffset)
1053 lseek (p->fd, p->file_moffset, SEEK_SET);
1057 static void file_end (void *handle, off_t offset)
1059 struct file_read_info *p = handle;
1061 assert (p->file_more == 0);
1063 p->file_moffset = offset;
1066 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1067 const char *fname, const char *spec)
1069 static char dstBuf[2048];
1071 const char *s = spec;
1072 static const char **w;
1076 while (*s == ' ' || *s == '\t')
1082 char attset_str[64], attname_str[64];
1083 data1_attset *attset;
1086 int attSet = 1, attUse = 1;
1090 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1092 attset_str[i++] = *s;
1093 attset_str[i] = '\0';
1098 for (i = 0; *s && *s != ')'; s++)
1100 attname_str[i++] = *s;
1101 attname_str[i] = '\0';
1104 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1107 attSet = attset->reference;
1108 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1110 attUse = att->value;
1112 attUse = atoi (attname_str);
1114 w = searchRecordKey (reckeys, attSet, attUse);
1119 for (i = 0; i<32; i++)
1124 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1125 spec, rGroup->groupName ? rGroup->groupName : "none");
1130 for (i = 0; i<32; i++)
1131 if (matchFlag[i] && w[i])
1139 dst += strlen(w[i]);
1143 logf (LOG_WARN, "Record didn't contain match"
1144 " fields in (%s,%s)", attset_str, attname_str);
1152 const char *spec_src = NULL;
1153 const char *s1 = ++s;
1154 while (*s1 && *s1 != ' ' && *s1 != '\t')
1160 memcpy (special, s, spec_len);
1161 special[spec_len] = '\0';
1164 if (!strcmp (special, "group"))
1165 spec_src = rGroup->groupName;
1166 else if (!strcmp (special, "database"))
1167 spec_src = rGroup->databaseName;
1168 else if (!strcmp (special, "filename"))
1170 else if (!strcmp (special, "type"))
1171 spec_src = rGroup->recordType;
1176 strcpy (dst, spec_src);
1177 dst += strlen (spec_src);
1180 else if (*s == '\"' || *s == '\'')
1182 int stopMarker = *s++;
1186 while (*s && *s != stopMarker)
1189 tmpString[i++] = *s++;
1193 tmpString[i] = '\0';
1194 strcpy (dst, tmpString);
1195 dst += strlen (tmpString);
1199 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1200 spec, rGroup->groupName ? rGroup->groupName : "none");
1207 logf (LOG_WARN, "No match criteria for record %s in group %s",
1208 fname, rGroup->groupName ? rGroup->groupName : "none");
1215 struct recordLogInfo {
1218 struct recordGroup *rGroup;
1221 static void recordLogPreamble (int level, const char *msg, void *info)
1223 struct recordLogInfo *p = info;
1224 FILE *outf = log_file ();
1226 if (level & LOG_LOG)
1228 fprintf (outf, "File %s, offset %d, type %s\n",
1229 p->rGroup->recordType, p->recordOffset, p->fname);
1230 log_event_start (NULL, NULL);
1233 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1235 zebraExplain_addSchema (zti, oid);
1238 static int recordExtract (SYSNO *sysno, const char *fname,
1239 struct recordGroup *rGroup, int deleteFlag,
1240 struct file_read_info *fi,
1241 RecType recType, char *subType)
1243 RecordAttr *recordAttr;
1248 struct recordLogInfo logInfo;
1249 off_t recordOffset = 0;
1253 struct recExtractCtrl extractCtrl;
1255 /* we are going to read from a file, so prepare the extraction */
1258 reckeys.buf_used = 0;
1259 reckeys.prevAttrUse = -1;
1260 reckeys.prevAttrSet = -1;
1261 reckeys.prevSeqNo = 0;
1263 recordOffset = fi->file_moffset;
1264 extractCtrl.offset = fi->file_moffset;
1265 extractCtrl.readf = file_read;
1266 extractCtrl.seekf = file_seek;
1267 extractCtrl.tellf = file_tell;
1268 extractCtrl.endf = file_end;
1269 extractCtrl.fh = fi;
1270 extractCtrl.subType = subType;
1271 extractCtrl.init = wordInit;
1272 extractCtrl.addWord = addRecordKey;
1273 extractCtrl.addSchema = addSchema;
1274 extractCtrl.dh = rGroup->dh;
1275 for (i = 0; i<256; i++)
1277 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1278 extractCtrl.seqno[i] = 1;
1280 extractCtrl.seqno[i] = 0;
1282 extractCtrl.zebra_maps = rGroup->zebra_maps;
1283 extractCtrl.flagShowRecords = !rGroup->flagRw;
1285 if (!rGroup->flagRw)
1286 printf ("File: %s %ld\n", fname, (long) recordOffset);
1288 logInfo.fname = fname;
1289 logInfo.recordOffset = recordOffset;
1290 logInfo.rGroup = rGroup;
1291 log_event_start (recordLogPreamble, &logInfo);
1293 r = (*recType->extract)(&extractCtrl);
1295 log_event_start (NULL, NULL);
1299 /* error occured during extraction ... */
1300 if (rGroup->flagRw &&
1301 records_processed < rGroup->fileVerboseLimit)
1303 logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
1304 fname, (long) recordOffset, r);
1308 if (reckeys.buf_used == 0)
1310 /* the extraction process returned no information - the record
1311 is probably empty - unless flagShowRecords is in use */
1312 if (!rGroup->flagRw)
1314 logf (LOG_WARN, "No keys generated for file %s", fname);
1315 logf (LOG_WARN, " The file is probably empty");
1320 /* perform match if sysno not known and if match criteria is specified */
1327 if (rGroup->recordId && *rGroup->recordId)
1331 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1335 rinfo = dict_lookup (matchDict, matchStr);
1337 memcpy (sysno, rinfo+1, sizeof(*sysno));
1341 logf (LOG_WARN, "Bad match criteria");
1352 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1353 fname, (long) recordOffset);
1354 logf (LOG_WARN, "cannot delete record above (seems new)");
1357 if (records_processed < rGroup->fileVerboseLimit)
1358 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1359 fname, (long) recordOffset);
1360 rec = rec_new (records);
1362 *sysno = rec->sysno;
1364 recordAttr = rec_init_attr (zti, rec);
1368 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1370 flushRecordKeys (*sysno, 1, &reckeys);
1371 flushSortKeys (*sysno, 1);
1377 /* record already exists */
1378 struct recKeys delkeys;
1380 rec = rec_get (records, *sysno);
1383 recordAttr = rec_init_attr (zti, rec);
1385 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1387 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1388 fname, (long) recordOffset);
1393 delkeys.buf_used = rec->size[recInfo_delKeys];
1394 delkeys.buf = rec->info[recInfo_delKeys];
1395 flushSortKeys (*sysno, 0);
1396 flushRecordKeys (*sysno, 0, &delkeys);
1399 /* record going to be deleted */
1400 if (!delkeys.buf_used)
1402 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1403 fname, (long) recordOffset);
1404 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1408 if (records_processed < rGroup->fileVerboseLimit)
1409 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1410 fname, (long) recordOffset);
1413 dict_delete (matchDict, matchStr);
1414 rec_del (records, &rec);
1422 /* record going to be updated */
1423 if (!delkeys.buf_used)
1425 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1426 fname, (long) recordOffset);
1427 logf (LOG_WARN, "cannot update file above, storeKeys false");
1431 if (records_processed < rGroup->fileVerboseLimit)
1432 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1433 fname, (long) recordOffset);
1434 flushRecordKeys (*sysno, 1, &reckeys);
1439 /* update file type */
1440 xfree (rec->info[recInfo_fileType]);
1441 rec->info[recInfo_fileType] =
1442 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1444 /* update filename */
1445 xfree (rec->info[recInfo_filename]);
1446 rec->info[recInfo_filename] =
1447 rec_strdup (fname, &rec->size[recInfo_filename]);
1449 /* update delete keys */
1450 xfree (rec->info[recInfo_delKeys]);
1451 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1454 rec->size[recInfo_delKeys] = reckeys.buf_used;
1455 rec->info[recInfo_delKeys] = reckeys.buf;
1457 reckeys.buf_max = 0;
1459 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1460 rec->size[recInfo_delKeys] = reckeys.buf_used;
1461 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1462 rec->size[recInfo_delKeys]);
1467 rec->info[recInfo_delKeys] = NULL;
1468 rec->size[recInfo_delKeys] = 0;
1471 /* save file size of original record */
1472 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1473 recordAttr->recordSize = fi->file_moffset - recordOffset;
1474 if (!recordAttr->recordSize)
1475 recordAttr->recordSize = fi->file_max - recordOffset;
1476 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1478 /* set run-number for this record */
1479 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1481 /* update store data */
1482 xfree (rec->info[recInfo_storeData]);
1483 if (rGroup->flagStoreData == 1)
1485 rec->size[recInfo_storeData] = recordAttr->recordSize;
1486 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1487 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1489 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1490 (long) recordOffset, fname);
1493 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1494 < recordAttr->recordSize)
1496 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1497 recordAttr->recordSize, fname);
1503 rec->info[recInfo_storeData] = NULL;
1504 rec->size[recInfo_storeData] = 0;
1506 /* update database name */
1507 xfree (rec->info[recInfo_databaseName]);
1508 rec->info[recInfo_databaseName] =
1509 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1512 recordAttr->recordOffset = recordOffset;
1514 /* commit this record */
1515 rec_put (records, &rec);
1520 int fileExtract (SYSNO *sysno, const char *fname,
1521 const struct recordGroup *rGroupP, int deleteFlag)
1529 struct recordGroup rGroupM;
1530 struct recordGroup *rGroup = &rGroupM;
1531 struct file_read_info *fi;
1533 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1535 if (!rGroup->groupName || !*rGroup->groupName)
1538 sprintf (gprefix, "%s.", rGroup->groupName);
1540 logf (LOG_DEBUG, "fileExtract %s", fname);
1542 /* determine file extension */
1544 for (i = strlen(fname); --i >= 0; )
1545 if (fname[i] == '/')
1547 else if (fname[i] == '.')
1549 strcpy (ext, fname+i+1);
1552 /* determine file type - depending on extension */
1553 if (!rGroup->recordType)
1555 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1556 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1558 sprintf (ext_res, "%srecordType", gprefix);
1559 rGroup->recordType = res_get (common_resource, ext_res);
1562 if (!rGroup->recordType)
1564 if (records_processed < rGroup->fileVerboseLimit)
1565 logf (LOG_LOG, "? %s", fname);
1568 if (!*rGroup->recordType)
1571 recType_byName (rGroup->recTypes, rGroup->recordType, subType)))
1573 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1577 /* determine match criteria */
1578 if (!rGroup->recordId)
1580 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1581 rGroup->recordId = res_get (common_resource, ext_res);
1584 /* determine database name */
1585 if (!rGroup->databaseName)
1587 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1588 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1590 sprintf (ext_res, "%sdatabase", gprefix);
1591 rGroup->databaseName = res_get (common_resource, ext_res);
1594 if (!rGroup->databaseName)
1595 rGroup->databaseName = "Default";
1597 /* determine if explain database */
1599 sprintf (ext_res, "%sexplainDatabase", gprefix);
1600 rGroup->explainDatabase =
1601 atoi (res_get_def (common_resource, ext_res, "0"));
1603 /* announce database */
1604 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1606 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1607 rGroup->explainDatabase))
1611 if (rGroup->flagStoreData == -1)
1614 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1615 if (!(sval = res_get (common_resource, ext_res)))
1617 sprintf (ext_res, "%sstoreData", gprefix);
1618 sval = res_get (common_resource, ext_res);
1621 rGroup->flagStoreData = atoi (sval);
1623 if (rGroup->flagStoreData == -1)
1624 rGroup->flagStoreData = 0;
1626 if (rGroup->flagStoreKeys == -1)
1630 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1631 if (!(sval = res_get (common_resource, ext_res)))
1633 sprintf (ext_res, "%sstoreKeys", gprefix);
1634 sval = res_get (common_resource, ext_res);
1637 rGroup->flagStoreKeys = atoi (sval);
1639 if (rGroup->flagStoreKeys == -1)
1640 rGroup->flagStoreKeys = 0;
1646 char xname[128], *xp;
1648 strncpy (xname, fname, 127);
1649 if (!(xp = strchr (xname, '.')))
1652 if (strcmp (xp+1, "sdr.bits"))
1655 h = zebraSdr_open (xname);
1658 logf (LOG_WARN, "sdr open %s", xname);
1667 segmentno = zebraSdr_segment (h, 0);
1668 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1674 r = zebraSdr_read (h, &buf);
1679 fi = file_read_start (0);
1685 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1687 } while (r && !sysno && fi->file_more);
1688 file_read_stop (fi);
1695 if (sysno && deleteFlag)
1699 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1701 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1705 fi = file_read_start (fd);
1709 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1711 } while (r && !sysno && fi->file_more);
1712 file_read_stop (fi);
1718 static int explain_extract (void *handle, Record rec, data1_node *n)
1720 struct recordGroup *rGroup = (struct recordGroup*) handle;
1721 struct recExtractCtrl extractCtrl;
1724 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1726 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1730 reckeys.buf_used = 0;
1731 reckeys.prevAttrUse = -1;
1732 reckeys.prevAttrSet = -1;
1733 reckeys.prevSeqNo = 0;
1735 extractCtrl.init = wordInit;
1736 extractCtrl.addWord = addRecordKey;
1737 extractCtrl.addSchema = addSchema;
1738 extractCtrl.dh = rGroup->dh;
1739 for (i = 0; i<256; i++)
1740 extractCtrl.seqno[i] = 0;
1741 extractCtrl.zebra_maps = rGroup->zebra_maps;
1742 extractCtrl.flagShowRecords = !rGroup->flagRw;
1744 grs_extract_tree(&extractCtrl, n);
1746 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1748 if (rec->size[recInfo_delKeys])
1750 struct recKeys delkeys;
1752 delkeys.buf_used = rec->size[recInfo_delKeys];
1753 delkeys.buf = rec->info[recInfo_delKeys];
1754 flushSortKeys (rec->sysno, 0);
1755 flushRecordKeys (rec->sysno, 0, &delkeys);
1757 flushRecordKeys (rec->sysno, 1, &reckeys);
1758 flushSortKeys (rec->sysno, 1);
1760 xfree (rec->info[recInfo_delKeys]);
1761 rec->size[recInfo_delKeys] = reckeys.buf_used;
1762 rec->info[recInfo_delKeys] = reckeys.buf;
1764 reckeys.buf_max = 0;