2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.95 1999-05-21 12:00:17 adam
8 * Better diagnostics for extraction process.
10 * Revision 1.94 1999/05/20 12:57:18 adam
11 * Implemented TCL filter. Updated recctrl system.
13 * Revision 1.93 1999/05/15 14:36:38 adam
14 * Updated dictionary. Implemented "compression" of dictionary.
16 * Revision 1.92 1999/03/09 16:27:49 adam
17 * More work on SDRKit integration.
19 * Revision 1.91 1999/02/12 13:29:22 adam
20 * Implemented position-flag for registers.
22 * Revision 1.90 1999/02/02 14:50:52 adam
23 * Updated WIN32 code specific sections. Changed header.
25 * Revision 1.89 1998/10/28 10:54:38 adam
28 * Revision 1.88 1998/10/16 08:14:29 adam
29 * Updated record control system.
31 * Revision 1.87 1998/10/15 13:10:33 adam
32 * Fixed bug in Zebra that caused it to stop indexing when empty
35 * Revision 1.86 1998/10/13 20:33:53 adam
36 * Fixed one log message and change use ordinal to be an unsigned char.
38 * Revision 1.85 1998/09/22 10:03:41 adam
39 * Changed result sets to be persistent in the sense that they can
40 * be re-searched if needed.
41 * Fixed memory leak in rsm_or.
43 * Revision 1.84 1998/06/11 15:42:22 adam
44 * Changed the way use attributes are specified in the recordId
47 * Revision 1.83 1998/06/08 14:43:10 adam
48 * Added suport for EXPLAIN Proxy servers - added settings databasePath
49 * and explainDatabase to facilitate this. Increased maximum number
50 * of databases and attributes in one register.
52 * Revision 1.82 1998/05/20 10:12:15 adam
53 * Implemented automatic EXPLAIN database maintenance.
54 * Modified Zebra to work with ASN.1 compiled version of YAZ.
56 * Revision 1.81 1998/03/11 11:19:04 adam
57 * Changed the way sequence numbers are generated.
59 * Revision 1.80 1998/03/05 08:45:11 adam
60 * New result set model and modular ranking system. Moved towards
61 * descent server API. System information stored as "SGML" records.
63 * Revision 1.79 1998/02/17 10:32:52 adam
64 * Fixed bug: binary files weren't opened with flag b on NT.
66 * Revision 1.78 1998/02/10 12:03:05 adam
69 * Revision 1.77 1998/01/12 15:04:08 adam
70 * The test option (-s) only uses read-lock (and not write lock).
72 * Revision 1.76 1997/10/27 14:33:04 adam
73 * Moved towards generic character mapping depending on "structure"
74 * field in abstract syntax file. Fixed a few memory leaks. Fixed
75 * bug with negative integers when doing searches with relational
78 * Revision 1.75 1997/09/17 12:19:12 adam
79 * Zebra version corresponds to YAZ version 1.4.
80 * Changed Zebra server so that it doesn't depend on global common_resource.
82 * Revision 1.74 1997/09/09 13:38:06 adam
83 * Partial port to WIN95/NT.
85 * Revision 1.73 1997/09/04 13:57:20 adam
86 * New file extract/retrieve method tellf (added).
87 * Added O_BINARY for open calls.
89 * Revision 1.72 1997/07/15 16:32:29 adam
90 * Bug fix: Match handler didn't terminate the resulting string!
92 * Revision 1.71 1997/07/15 16:28:41 adam
93 * Bug fix: storeData didn't work with files with multiple records.
94 * Bug fix: fixed memory management with records; not really well
97 * Revision 1.70 1997/07/01 13:00:42 adam
98 * Bug fix in routine searchRecordKey: uninitialized variables.
100 * Revision 1.69 1997/04/29 09:26:03 adam
101 * Bug fix: generic recordId handling didn't work for compressed internal
104 * Revision 1.68 1997/02/12 20:39:45 adam
105 * Implemented options -f <n> that limits the log to the first <n>
107 * Changed some log messages also.
109 * Revision 1.67 1996/11/15 15:02:14 adam
110 * Minor changes regarding logging.
112 * Revision 1.66 1996/11/14 09:52:21 adam
113 * Strings in record keys bound by IT_MAX_WORD.
115 * Revision 1.65 1996/11/14 08:57:56 adam
116 * Reduction of storeKeys area.
118 * Revision 1.64 1996/11/08 11:10:16 adam
119 * Buffers used during file match got bigger.
120 * Compressed ISAM support everywhere.
121 * Bug fixes regarding masking characters in queries.
122 * Redesigned Regexp-2 queries.
124 * Revision 1.63 1996/10/29 14:09:39 adam
125 * Use of cisam system - enabled if setting isamc is 1.
127 * Revision 1.62 1996/10/11 10:57:01 adam
128 * New module recctrl. Used to manage records (extract/retrieval).
129 * Several files have been moved to the recctrl sub directory.
131 * Revision 1.61 1996/06/06 12:08:37 quinn
132 * Added showRecord function
134 * Revision 1.60 1996/06/04 10:18:12 adam
135 * Search/scan uses character mapping module.
137 * Revision 1.59 1996/05/14 15:47:07 adam
138 * Cleanup of various buffer size entities.
140 * Revision 1.58 1996/05/14 06:16:38 adam
141 * Compact use/set bytes used in search service.
143 * Revision 1.57 1996/05/13 14:23:04 adam
144 * Work on compaction of set/use bytes in dictionary.
146 * Revision 1.56 1996/05/09 09:54:42 adam
147 * Server supports maps from one logical attributes to a list of physical
149 * The extraction process doesn't make space consuming 'any' keys.
151 * Revision 1.55 1996/05/09 07:28:55 quinn
152 * Work towards phrases and multiple registers
154 * Revision 1.54 1996/05/01 13:46:35 adam
155 * First work on multiple records in one file.
156 * New option, -offset, to the "unread" command in the filter module.
158 * Revision 1.53 1996/04/26 12:09:43 adam
159 * Added a few comments.
161 * Revision 1.52 1996/04/25 13:27:57 adam
162 * Function recordExtract modified so that files with no keys (possibly empty)
165 * Revision 1.51 1996/03/19 11:08:42 adam
166 * Bug fix: Log preamble wasn't always turned off after recordExtract.
168 * Revision 1.50 1996/02/12 18:45:36 adam
169 * New fileVerboseFlag in record group control.
171 * Revision 1.49 1996/02/05 12:29:57 adam
172 * Logging reduced a bit.
173 * The remaining running time is estimated during register merge.
175 * Revision 1.48 1996/02/01 20:53:26 adam
176 * The temporary per-record keys are compacted a little, and duplication
177 * of the per-records keys are avoided when they are saved in the record
178 * information buffer.
180 * Revision 1.47 1996/01/17 14:57:48 adam
181 * Prototype changed for reader functions in extract/retrieve. File
182 * is identified by 'void *' instead of 'int.
184 * Revision 1.46 1995/12/15 14:57:16 adam
187 * Revision 1.45 1995/12/15 12:37:41 adam
188 * In addRecordKeyAny: Writes key only when attrSet != -1.
190 * Revision 1.44 1995/12/12 16:00:54 adam
191 * System call sync(2) used after update/commit.
192 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
195 * Revision 1.43 1995/12/11 09:12:46 adam
196 * The rec_get function returns NULL if record doesn't exist - will
197 * happen in the server if the result set records have been deleted since
198 * the creation of the set (i.e. the search).
199 * The server saves a result temporarily if it is 'volatile', i.e. the
200 * set is register dependent.
202 * Revision 1.42 1995/12/07 17:38:46 adam
203 * Work locking mechanisms for concurrent updates/commit.
205 * Revision 1.41 1995/12/06 16:06:42 adam
206 * Better diagnostics. Work on 'real' dictionary deletion.
208 * Revision 1.40 1995/12/05 16:57:40 adam
209 * More work on regular patterns.
211 * Revision 1.39 1995/12/05 13:20:18 adam
212 * Bug fix: file_read sometimes returned early EOF.
214 * Revision 1.38 1995/12/04 17:59:21 adam
215 * More work on regular expression conversion.
217 * Revision 1.37 1995/12/04 14:22:27 adam
218 * Extra arg to recType_byName.
219 * Started work on new regular expression parsed input to
220 * structured records.
222 * Revision 1.36 1995/11/30 08:34:29 adam
223 * Started work on commit facility.
224 * Changed a few malloc/free to xmalloc/xfree.
226 * Revision 1.35 1995/11/28 14:26:21 adam
227 * Bug fix: recordId with constant wasn't right.
228 * Bug fix: recordId dictionary entry wasn't deleted when needed.
230 * Revision 1.34 1995/11/28 09:09:38 adam
231 * Zebra config renamed.
232 * Use setting 'recordId' to identify record now.
233 * Bug fix in recindex.c: rec_release_blocks was invokeded even
234 * though the blocks were already released.
235 * File traversal properly deletes records when needed.
237 * Revision 1.33 1995/11/27 09:56:20 adam
238 * Record info elements better enumerated. Internal store of records.
240 * Revision 1.32 1995/11/25 10:24:05 adam
241 * More record fields - they are enumerated now.
242 * New options: flagStoreData flagStoreKey.
244 * Revision 1.31 1995/11/24 11:31:35 adam
245 * Commands add & del read filenames from stdin if source directory is
247 * Match criteria supports 'constant' strings.
249 * Revision 1.30 1995/11/22 17:19:16 adam
250 * Record management uses the bfile system.
252 * Revision 1.29 1995/11/21 15:01:14 adam
253 * New general match criteria implemented.
254 * New feature: document groups.
256 * Revision 1.28 1995/11/21 09:20:30 adam
257 * Yet more work on record match.
259 * Revision 1.27 1995/11/20 16:59:45 adam
260 * New update method: the 'old' keys are saved for each records.
262 * Revision 1.26 1995/11/20 11:56:24 adam
263 * Work on new traversal.
265 * Revision 1.25 1995/11/16 15:34:54 adam
266 * Uses new record management system in both indexer and server.
268 * Revision 1.24 1995/11/15 19:13:08 adam
269 * Work on record management.
271 * Revision 1.23 1995/10/27 14:00:10 adam
272 * Implemented detection of database availability.
274 * Revision 1.22 1995/10/17 18:02:07 adam
275 * New feature: databases. Implemented as prefix to words in dictionary.
277 * Revision 1.21 1995/10/10 12:24:38 adam
278 * Temporary sort files are compressed.
280 * Revision 1.20 1995/10/06 13:52:05 adam
281 * Bug fixes. Handler may abort further scanning.
283 * Revision 1.19 1995/10/04 12:55:16 adam
284 * Bug fix in ranked search. Use=Any keys inserted.
286 * Revision 1.18 1995/10/04 09:37:08 quinn
289 * Revision 1.17 1995/10/03 14:28:57 adam
290 * Buffered read in extract works.
292 * Revision 1.16 1995/10/03 14:28:45 adam
293 * Work on more effecient read handler in extract.
295 * Revision 1.15 1995/10/02 15:42:53 adam
296 * Extract uses file descriptors instead of FILE pointers.
298 * Revision 1.14 1995/10/02 15:29:13 adam
299 * More logging in file_extract.
301 * Revision 1.13 1995/09/29 14:01:39 adam
304 * Revision 1.12 1995/09/28 14:22:56 adam
305 * Sort uses smaller temporary files.
307 * Revision 1.11 1995/09/28 12:10:31 adam
308 * Bug fixes. Field prefix used in queries.
310 * Revision 1.10 1995/09/28 09:19:41 adam
311 * xfree/xmalloc used everywhere.
312 * Extract/retrieve method seems to work for text records.
314 * Revision 1.9 1995/09/27 12:22:28 adam
315 * More work on extract in record control.
316 * Field name is not in isam keys but in prefix in dictionary words.
318 * Revision 1.8 1995/09/14 07:48:22 adam
319 * Record control management.
321 * Revision 1.7 1995/09/11 13:09:32 adam
322 * More work on relevance feedback.
324 * Revision 1.6 1995/09/08 14:52:27 adam
325 * Minor changes. Dictionary is lower case now.
327 * Revision 1.5 1995/09/06 16:11:16 adam
328 * Option: only one word key per file.
330 * Revision 1.4 1995/09/05 15:28:39 adam
331 * More work on search engine.
333 * Revision 1.3 1995/09/04 12:33:41 adam
334 * Various cleanup. YAZ util used instead.
336 * Revision 1.2 1995/09/04 09:10:34 adam
337 * More work on index add/del/update.
338 * Merge sort implemented.
339 * Initial work on z39 server.
341 * Revision 1.1 1995/09/01 14:06:35 adam
342 * Split of work into more files.
366 #include "zebrasdr.h"
369 static Dict matchDict;
371 static Records records = NULL;
372 static SortIdx sortIdx = NULL;
374 static char **key_buf;
375 static size_t ptr_top;
377 static size_t key_buf_used;
378 static int key_file_no;
380 static int records_inserted = 0;
381 static int records_updated = 0;
382 static int records_deleted = 0;
383 static int records_processed = 0;
385 static ZebraExplainInfo zti = NULL;
387 static void logRecord (int showFlag)
391 if (showFlag || !(records_processed % 1000))
393 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
394 records_processed, records_inserted, records_updated,
399 static int explain_extract (void *handle, Record drec, data1_node *n);
401 int key_open (struct recordGroup *rGroup, int mem)
403 BFiles bfs = rGroup->bfs;
404 int rw = rGroup->flagRw;
405 data1_handle dh = rGroup->dh;
407 mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
410 key_buf = xmalloc (mem);
411 ptr_top = mem/sizeof(char*);
417 if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
419 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
423 records = rec_open (bfs, rw);
426 dict_close (matchDict);
429 zti = zebraExplain_open (records, dh, common_resource,
430 rw, rGroup, explain_extract);
433 rec_close (&records);
434 dict_close (matchDict);
437 sortIdx = sortIdx_open (bfs, 1);
448 void encode_key_init (struct encode_info *i)
455 char *encode_key_int (int d, char *bp)
464 else if (d <= 4194303)
466 *bp++ = 128 + (d>>16);
467 *bp++ = (d>>8) & 255;
472 *bp++ = 192 + (d>>24);
473 *bp++ = (d>>16) & 255;
474 *bp++ = (d>>8) & 255;
480 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
485 while ((*bp++ = *k++))
487 memcpy (&key, k+1, sizeof(struct it_key));
488 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
489 if (i->sysno != key.sysno)
491 i->sysno = key.sysno;
494 else if (!i->seqno && !key.seqno && i->cmd == *k)
496 bp = encode_key_int (key.seqno - i->seqno, bp);
497 i->seqno = key.seqno;
499 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
501 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
509 static int key_y_len;
511 static int key_y_compare (const void *p1, const void *p2)
515 if ((r = key_compare (*(char**) p1 + key_y_len + 1,
516 *(char**) p2 + key_y_len + 1)))
518 return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len);
521 static int key_x_compare (const void *p1, const void *p2)
523 return strcmp (*(char**) p1, *(char**) p2);
527 void key_flush (void)
532 struct encode_info encode_info;
541 logf (LOG_LOG, "sorting section %d", key_file_no);
543 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
544 getFnameTmp (out_fname, key_file_no);
546 if (!(outf = fopen (out_fname, "wb")))
548 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
551 logf (LOG_LOG, "writing section %d", key_file_no);
552 prevcp = cp = key_buf[ptr_top-ptr_i];
554 encode_key_init (&encode_info);
555 encode_key_write (cp, &encode_info, outf);
558 cp = key_buf[ptr_top-ptr_i];
559 if (strcmp (cp, prevcp))
561 encode_key_init (&encode_info);
562 encode_key_write (cp, &encode_info, outf);
566 encode_key_write (cp + strlen(cp), &encode_info, outf);
569 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
570 getFnameTmp (out_fname, key_file_no);
572 if (!(outf = fopen (out_fname, "wb")))
574 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
577 logf (LOG_LOG, "writing section %d", key_file_no);
579 prevcp = key_buf[ptr_top-i];
581 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
583 key_y_len = strlen(prevcp)+1;
585 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
586 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
588 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
589 sizeof(char*), key_y_compare);
590 cp = key_buf[ptr_top-ptr_i];
592 encode_key_init (&encode_info);
593 encode_key_write (cp, &encode_info, outf);
596 cp = key_buf[ptr_top-ptr_i];
597 encode_key_write (cp+key_y_len, &encode_info, outf);
601 prevcp = key_buf[ptr_top-ptr_i];
606 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
609 logf (LOG_LOG, "finished section %d", key_file_no);
614 int key_close (struct recordGroup *rGroup)
616 int rw = rGroup->flagRw;
618 zebraExplain_runNumberIncrement (zti, 1);
619 zebraExplain_close (zti, rw, 0);
622 rec_close (&records);
623 dict_close (matchDict);
624 sortIdx_close (sortIdx);
630 static void wordInit (struct recExtractCtrl *p, RecWord *w)
632 w->zebra_maps = p->zebra_maps;
633 w->seqnos = p->seqno;
634 w->attrSet = VAL_BIB1;
639 static struct sortKey {
644 struct sortKey *next;
647 static struct recKeys {
656 static void addIndexString (RecWord *p, const char *string, int length)
659 unsigned char attrSet;
660 unsigned short attrUse;
663 int *pseqno = &p->seqnos[p->reg_type];
665 if (reckeys.buf_used+1024 > reckeys.buf_max)
669 b = xmalloc (reckeys.buf_max += 128000);
670 if (reckeys.buf_used > 0)
671 memcpy (b, reckeys.buf, reckeys.buf_used);
675 dst = reckeys.buf + reckeys.buf_used;
677 attrSet = p->attrSet;
678 if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
681 reckeys.prevAttrSet = attrSet;
682 attrUse = p->attrUse;
683 if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
686 reckeys.prevAttrUse = attrUse;
688 diff = 1 + *pseqno - reckeys.prevSeqNo;
689 if (diff >= 1 && diff <= 15)
694 reckeys.prevSeqNo = *pseqno;
700 memcpy (dst, &attrSet, sizeof(attrSet));
701 dst += sizeof(attrSet);
705 memcpy (dst, &attrUse, sizeof(attrUse));
706 dst += sizeof(attrUse);
708 *dst++ = p->reg_type;
709 memcpy (dst, string, length);
715 memcpy (dst, pseqno, sizeof(*pseqno));
716 dst += sizeof(*pseqno);
718 reckeys.buf_used = dst - reckeys.buf;
723 static void addSortString (RecWord *p, const char *string, int length)
727 for (sk = sortKeys; sk; sk = sk->next)
728 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
731 sk = xmalloc (sizeof(*sk));
735 sk->string = xmalloc (length);
737 memcpy (sk->string, string, length);
739 sk->attrSet = p->attrSet;
740 sk->attrUse = p->attrUse;
743 static void addString (RecWord *p, const char *string, int length)
746 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
747 addSortString (p, string, length);
749 addIndexString (p, string, length);
752 static void addIncompleteField (RecWord *p)
754 const char *b = p->string;
755 int remain = p->length;
756 const char **map = 0;
759 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
763 char buf[IT_MAX_WORD+1];
767 while (map && *map && **map == *CHR_SPACE)
769 remain = p->length - (b - p->string);
771 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
778 while (map && *map && **map != *CHR_SPACE)
780 const char *cp = *map;
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, remain);
792 addString (p, buf, i);
796 static void addCompleteField (RecWord *p)
798 const char *b = p->string;
799 char buf[IT_MAX_WORD+1];
800 const char **map = 0;
801 int i = 0, remain = p->length;
804 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
806 while (remain > 0 && i < IT_MAX_WORD)
808 while (map && *map && **map == *CHR_SPACE)
810 remain = p->length - (b - p->string);
812 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
819 if (i && i < IT_MAX_WORD)
820 buf[i++] = *CHR_SPACE;
821 while (map && *map && **map != *CHR_SPACE)
823 const char *cp = *map;
825 if (i >= IT_MAX_WORD)
827 while (i < IT_MAX_WORD && *cp)
829 remain = p->length - (b - p->string);
831 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
839 addString (p, buf, i);
842 static void addRecordKey (RecWord *p)
844 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
845 addCompleteField (p);
847 addIncompleteField(p);
850 static void flushSortKeys (SYSNO sysno, int cmd)
852 struct sortKey *sk = sortKeys;
854 sortIdx_sysno (sortIdx, sysno);
857 struct sortKey *sk_next = sk->next;
858 sortIdx_type (sortIdx, sk->attrUse);
859 sortIdx_add (sortIdx, sk->string, sk->length);
867 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
869 unsigned char attrSet = (unsigned char) -1;
870 unsigned short attrUse = (unsigned short) -1;
874 zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
875 while (off < reckeys->buf_used)
877 const char *src = reckeys->buf + off;
885 memcpy (&attrSet, src, sizeof(attrSet));
886 src += sizeof(attrSet);
890 memcpy (&attrUse, src, sizeof(attrUse));
891 src += sizeof(attrUse);
893 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
896 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
898 ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
900 ch = zebraExplain_addSU (zti, attrSet, attrUse);
902 key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
905 ((char*)key_buf) [key_buf_used++] = *src++;
907 ((char*)key_buf) [key_buf_used++] = '\0';
908 ((char*) key_buf)[key_buf_used++] = cmd;
911 seqno += ((lead>>2) & 15)-1;
914 memcpy (&seqno, src, sizeof(seqno));
915 src += sizeof(seqno);
919 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
920 key_buf_used += sizeof(key);
921 off = src - reckeys->buf;
923 assert (off == reckeys->buf_used);
926 static const char **searchRecordKey (struct recKeys *reckeys,
927 int attrSetS, int attrUseS)
929 static const char *ws[32];
937 for (i = 0; i<32; i++)
940 while (off < reckeys->buf_used)
943 const char *src = reckeys->buf + off;
951 memcpy (&attrSet, src, sizeof(attrSet));
952 src += sizeof(attrSet);
956 memcpy (&attrUse, src, sizeof(attrUse));
957 src += sizeof(attrUse);
963 seqno += ((lead>>2) & 15)-1;
966 memcpy (&seqno, src, sizeof(seqno));
967 src += sizeof(seqno);
969 if (attrUseS == attrUse && attrSetS == attrSet)
976 woff = seqno - startSeq;
977 if (woff >= 0 && woff < 31)
981 off = src - reckeys->buf;
983 assert (off == reckeys->buf_used);
987 struct file_read_info {
988 off_t file_max; /* maximum offset so far */
989 off_t file_offset; /* current offset */
990 off_t file_moffset; /* offset of rec/rec boundary */
997 static struct file_read_info *file_read_start (int fd)
999 struct file_read_info *fi = xmalloc (sizeof(*fi));
1003 fi->file_moffset = 0;
1009 static void file_read_stop (struct file_read_info *fi)
1014 static off_t file_seek (void *handle, off_t offset)
1016 struct file_read_info *p = handle;
1017 p->file_offset = offset;
1020 return lseek (p->fd, offset, SEEK_SET);
1023 static off_t file_tell (void *handle)
1025 struct file_read_info *p = handle;
1026 return p->file_offset;
1029 static int file_read (void *handle, char *buf, size_t count)
1031 struct file_read_info *p = handle;
1037 if (r > p->sdrmax - p->file_offset)
1038 r = p->sdrmax - p->file_offset;
1040 memcpy (buf, p->sdrbuf + p->file_offset, r);
1043 r = read (fd, buf, count);
1046 p->file_offset += r;
1047 if (p->file_offset > p->file_max)
1048 p->file_max = p->file_offset;
1053 static void file_begin (void *handle)
1055 struct file_read_info *p = handle;
1057 p->file_offset = p->file_moffset;
1058 if (!p->sdrbuf && p->file_moffset)
1059 lseek (p->fd, p->file_moffset, SEEK_SET);
1063 static void file_end (void *handle, off_t offset)
1065 struct file_read_info *p = handle;
1067 assert (p->file_more == 0);
1069 p->file_moffset = offset;
1072 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
1073 const char *fname, const char *spec)
1075 static char dstBuf[2048];
1077 const char *s = spec;
1078 static const char **w;
1082 while (*s == ' ' || *s == '\t')
1088 char attset_str[64], attname_str[64];
1089 data1_attset *attset;
1092 int attSet = 1, attUse = 1;
1096 for (i = 0; *s && *s != ',' && *s != ')'; s++)
1098 attset_str[i++] = *s;
1099 attset_str[i] = '\0';
1104 for (i = 0; *s && *s != ')'; s++)
1106 attname_str[i++] = *s;
1107 attname_str[i] = '\0';
1110 if ((attset = data1_get_attset (rGroup->dh, attset_str)))
1113 attSet = attset->reference;
1114 att = data1_getattbyname(rGroup->dh, attset, attname_str);
1116 attUse = att->value;
1118 attUse = atoi (attname_str);
1120 w = searchRecordKey (reckeys, attSet, attUse);
1125 for (i = 0; i<32; i++)
1130 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
1131 spec, rGroup->groupName ? rGroup->groupName : "none");
1136 for (i = 0; i<32; i++)
1137 if (matchFlag[i] && w[i])
1145 dst += strlen(w[i]);
1149 logf (LOG_WARN, "Record didn't contain match"
1150 " fields in (%s,%s)", attset_str, attname_str);
1158 const char *spec_src = NULL;
1159 const char *s1 = ++s;
1160 while (*s1 && *s1 != ' ' && *s1 != '\t')
1166 memcpy (special, s, spec_len);
1167 special[spec_len] = '\0';
1170 if (!strcmp (special, "group"))
1171 spec_src = rGroup->groupName;
1172 else if (!strcmp (special, "database"))
1173 spec_src = rGroup->databaseName;
1174 else if (!strcmp (special, "filename"))
1176 else if (!strcmp (special, "type"))
1177 spec_src = rGroup->recordType;
1182 strcpy (dst, spec_src);
1183 dst += strlen (spec_src);
1186 else if (*s == '\"' || *s == '\'')
1188 int stopMarker = *s++;
1192 while (*s && *s != stopMarker)
1195 tmpString[i++] = *s++;
1199 tmpString[i] = '\0';
1200 strcpy (dst, tmpString);
1201 dst += strlen (tmpString);
1205 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
1206 spec, rGroup->groupName ? rGroup->groupName : "none");
1213 logf (LOG_WARN, "No match criteria for record %s in group %s",
1214 fname, rGroup->groupName ? rGroup->groupName : "none");
1221 struct recordLogInfo {
1224 struct recordGroup *rGroup;
1227 static void recordLogPreamble (int level, const char *msg, void *info)
1229 struct recordLogInfo *p = info;
1230 FILE *outf = log_file ();
1232 if (level & LOG_LOG)
1234 fprintf (outf, "File %s, offset %d, type %s\n",
1235 p->rGroup->recordType, p->recordOffset, p->fname);
1236 log_event_start (NULL, NULL);
1239 void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
1241 zebraExplain_addSchema (zti, oid);
1244 static int recordExtract (SYSNO *sysno, const char *fname,
1245 struct recordGroup *rGroup, int deleteFlag,
1246 struct file_read_info *fi,
1247 RecType recType, char *subType, void *clientData)
1249 RecordAttr *recordAttr;
1254 struct recordLogInfo logInfo;
1255 off_t recordOffset = 0;
1259 struct recExtractCtrl extractCtrl;
1261 /* we are going to read from a file, so prepare the extraction */
1264 reckeys.buf_used = 0;
1265 reckeys.prevAttrUse = -1;
1266 reckeys.prevAttrSet = -1;
1267 reckeys.prevSeqNo = 0;
1269 recordOffset = fi->file_moffset;
1270 extractCtrl.offset = fi->file_moffset;
1271 extractCtrl.readf = file_read;
1272 extractCtrl.seekf = file_seek;
1273 extractCtrl.tellf = file_tell;
1274 extractCtrl.endf = file_end;
1275 extractCtrl.fh = fi;
1276 extractCtrl.subType = subType;
1277 extractCtrl.init = wordInit;
1278 extractCtrl.addWord = addRecordKey;
1279 extractCtrl.addSchema = addSchema;
1280 extractCtrl.dh = rGroup->dh;
1281 for (i = 0; i<256; i++)
1283 if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
1284 extractCtrl.seqno[i] = 1;
1286 extractCtrl.seqno[i] = 0;
1288 extractCtrl.zebra_maps = rGroup->zebra_maps;
1289 extractCtrl.flagShowRecords = !rGroup->flagRw;
1291 if (!rGroup->flagRw)
1292 printf ("File: %s %ld\n", fname, (long) recordOffset);
1294 logInfo.fname = fname;
1295 logInfo.recordOffset = recordOffset;
1296 logInfo.rGroup = rGroup;
1297 log_event_start (recordLogPreamble, &logInfo);
1299 r = (*recType->extract)(clientData, &extractCtrl);
1301 log_event_start (NULL, NULL);
1303 if (r == RECCTRL_EXTRACT_EOF)
1305 else if (r == RECCTRL_EXTRACT_ERROR)
1307 /* error occured during extraction ... */
1308 if (rGroup->flagRw &&
1309 records_processed < rGroup->fileVerboseLimit)
1311 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
1312 fname, (long) recordOffset);
1316 if (reckeys.buf_used == 0)
1318 /* the extraction process returned no information - the record
1319 is probably empty - unless flagShowRecords is in use */
1320 if (!rGroup->flagRw)
1322 logf (LOG_WARN, "No keys generated for file %s", fname);
1323 logf (LOG_WARN, " The file is probably empty");
1328 /* perform match if sysno not known and if match criteria is specified */
1335 if (rGroup->recordId && *rGroup->recordId)
1339 matchStr = fileMatchStr (&reckeys, rGroup, fname,
1343 rinfo = dict_lookup (matchDict, matchStr);
1345 memcpy (sysno, rinfo+1, sizeof(*sysno));
1349 logf (LOG_WARN, "Bad match criteria");
1360 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1361 fname, (long) recordOffset);
1362 logf (LOG_WARN, "cannot delete record above (seems new)");
1365 if (records_processed < rGroup->fileVerboseLimit)
1366 logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
1367 fname, (long) recordOffset);
1368 rec = rec_new (records);
1370 *sysno = rec->sysno;
1372 recordAttr = rec_init_attr (zti, rec);
1376 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
1378 flushRecordKeys (*sysno, 1, &reckeys);
1379 flushSortKeys (*sysno, 1);
1385 /* record already exists */
1386 struct recKeys delkeys;
1388 rec = rec_get (records, *sysno);
1391 recordAttr = rec_init_attr (zti, rec);
1393 if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
1395 logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
1396 fname, (long) recordOffset);
1401 delkeys.buf_used = rec->size[recInfo_delKeys];
1402 delkeys.buf = rec->info[recInfo_delKeys];
1403 flushSortKeys (*sysno, 0);
1404 flushRecordKeys (*sysno, 0, &delkeys);
1407 /* record going to be deleted */
1408 if (!delkeys.buf_used)
1410 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1411 fname, (long) recordOffset);
1412 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1416 if (records_processed < rGroup->fileVerboseLimit)
1417 logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
1418 fname, (long) recordOffset);
1421 dict_delete (matchDict, matchStr);
1422 rec_del (records, &rec);
1430 /* record going to be updated */
1431 if (!delkeys.buf_used)
1433 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1434 fname, (long) recordOffset);
1435 logf (LOG_WARN, "cannot update file above, storeKeys false");
1439 if (records_processed < rGroup->fileVerboseLimit)
1440 logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
1441 fname, (long) recordOffset);
1442 flushRecordKeys (*sysno, 1, &reckeys);
1447 /* update file type */
1448 xfree (rec->info[recInfo_fileType]);
1449 rec->info[recInfo_fileType] =
1450 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
1452 /* update filename */
1453 xfree (rec->info[recInfo_filename]);
1454 rec->info[recInfo_filename] =
1455 rec_strdup (fname, &rec->size[recInfo_filename]);
1457 /* update delete keys */
1458 xfree (rec->info[recInfo_delKeys]);
1459 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
1462 rec->size[recInfo_delKeys] = reckeys.buf_used;
1463 rec->info[recInfo_delKeys] = reckeys.buf;
1465 reckeys.buf_max = 0;
1467 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
1468 rec->size[recInfo_delKeys] = reckeys.buf_used;
1469 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
1470 rec->size[recInfo_delKeys]);
1475 rec->info[recInfo_delKeys] = NULL;
1476 rec->size[recInfo_delKeys] = 0;
1479 /* save file size of original record */
1480 zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
1481 recordAttr->recordSize = fi->file_moffset - recordOffset;
1482 if (!recordAttr->recordSize)
1483 recordAttr->recordSize = fi->file_max - recordOffset;
1484 zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
1486 /* set run-number for this record */
1487 recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
1489 /* update store data */
1490 xfree (rec->info[recInfo_storeData]);
1491 if (rGroup->flagStoreData == 1)
1493 rec->size[recInfo_storeData] = recordAttr->recordSize;
1494 rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
1495 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1497 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1498 (long) recordOffset, fname);
1501 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1502 < recordAttr->recordSize)
1504 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1505 recordAttr->recordSize, fname);
1511 rec->info[recInfo_storeData] = NULL;
1512 rec->size[recInfo_storeData] = 0;
1514 /* update database name */
1515 xfree (rec->info[recInfo_databaseName]);
1516 rec->info[recInfo_databaseName] =
1517 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
1520 recordAttr->recordOffset = recordOffset;
1522 /* commit this record */
1523 rec_put (records, &rec);
1528 int fileExtract (SYSNO *sysno, const char *fname,
1529 const struct recordGroup *rGroupP, int deleteFlag)
1537 struct recordGroup rGroupM;
1538 struct recordGroup *rGroup = &rGroupM;
1539 struct file_read_info *fi;
1542 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
1544 if (!rGroup->groupName || !*rGroup->groupName)
1547 sprintf (gprefix, "%s.", rGroup->groupName);
1549 logf (LOG_DEBUG, "fileExtract %s", fname);
1551 /* determine file extension */
1553 for (i = strlen(fname); --i >= 0; )
1554 if (fname[i] == '/')
1556 else if (fname[i] == '.')
1558 strcpy (ext, fname+i+1);
1561 /* determine file type - depending on extension */
1562 if (!rGroup->recordType)
1564 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
1565 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
1567 sprintf (ext_res, "%srecordType", gprefix);
1568 rGroup->recordType = res_get (common_resource, ext_res);
1571 if (!rGroup->recordType)
1573 if (records_processed < rGroup->fileVerboseLimit)
1574 logf (LOG_LOG, "? %s", fname);
1577 if (!*rGroup->recordType)
1580 recType_byName (rGroup->recTypes, rGroup->recordType, subType,
1583 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
1587 /* determine match criteria */
1588 if (!rGroup->recordId)
1590 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
1591 rGroup->recordId = res_get (common_resource, ext_res);
1594 /* determine database name */
1595 if (!rGroup->databaseName)
1597 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
1598 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
1600 sprintf (ext_res, "%sdatabase", gprefix);
1601 rGroup->databaseName = res_get (common_resource, ext_res);
1604 if (!rGroup->databaseName)
1605 rGroup->databaseName = "Default";
1607 /* determine if explain database */
1609 sprintf (ext_res, "%sexplainDatabase", gprefix);
1610 rGroup->explainDatabase =
1611 atoi (res_get_def (common_resource, ext_res, "0"));
1613 /* announce database */
1614 if (zebraExplain_curDatabase (zti, rGroup->databaseName))
1616 if (zebraExplain_newDatabase (zti, rGroup->databaseName,
1617 rGroup->explainDatabase))
1621 if (rGroup->flagStoreData == -1)
1624 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
1625 if (!(sval = res_get (common_resource, ext_res)))
1627 sprintf (ext_res, "%sstoreData", gprefix);
1628 sval = res_get (common_resource, ext_res);
1631 rGroup->flagStoreData = atoi (sval);
1633 if (rGroup->flagStoreData == -1)
1634 rGroup->flagStoreData = 0;
1636 if (rGroup->flagStoreKeys == -1)
1640 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1641 if (!(sval = res_get (common_resource, ext_res)))
1643 sprintf (ext_res, "%sstoreKeys", gprefix);
1644 sval = res_get (common_resource, ext_res);
1647 rGroup->flagStoreKeys = atoi (sval);
1649 if (rGroup->flagStoreKeys == -1)
1650 rGroup->flagStoreKeys = 0;
1656 char xname[128], *xp;
1658 strncpy (xname, fname, 127);
1659 if (!(xp = strchr (xname, '.')))
1662 if (strcmp (xp+1, "sdr.bits"))
1665 h = zebraSdr_open (xname);
1668 logf (LOG_WARN, "sdr open %s", xname);
1677 segmentno = zebraSdr_segment (h, 0);
1678 sprintf (sdr_name, "%%%s.%d", xname, segmentno);
1684 r = zebraSdr_read (h, &buf);
1689 fi = file_read_start (0);
1695 r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
1697 } while (r && !sysno && fi->file_more);
1698 file_read_stop (fi);
1705 if (sysno && deleteFlag)
1709 if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
1711 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1715 fi = file_read_start (fd);
1719 r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
1720 recType, subType, clientData);
1721 } while (r && !sysno && fi->file_more);
1722 file_read_stop (fi);
1728 static int explain_extract (void *handle, Record rec, data1_node *n)
1730 struct recordGroup *rGroup = (struct recordGroup*) handle;
1731 struct recExtractCtrl extractCtrl;
1734 if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
1736 if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
1740 reckeys.buf_used = 0;
1741 reckeys.prevAttrUse = -1;
1742 reckeys.prevAttrSet = -1;
1743 reckeys.prevSeqNo = 0;
1745 extractCtrl.init = wordInit;
1746 extractCtrl.addWord = addRecordKey;
1747 extractCtrl.addSchema = addSchema;
1748 extractCtrl.dh = rGroup->dh;
1749 for (i = 0; i<256; i++)
1750 extractCtrl.seqno[i] = 0;
1751 extractCtrl.zebra_maps = rGroup->zebra_maps;
1752 extractCtrl.flagShowRecords = !rGroup->flagRw;
1754 grs_extract_tree(&extractCtrl, n);
1756 logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
1758 if (rec->size[recInfo_delKeys])
1760 struct recKeys delkeys;
1762 delkeys.buf_used = rec->size[recInfo_delKeys];
1763 delkeys.buf = rec->info[recInfo_delKeys];
1764 flushSortKeys (rec->sysno, 0);
1765 flushRecordKeys (rec->sysno, 0, &delkeys);
1767 flushRecordKeys (rec->sysno, 1, &reckeys);
1768 flushSortKeys (rec->sysno, 1);
1770 xfree (rec->info[recInfo_delKeys]);
1771 rec->size[recInfo_delKeys] = reckeys.buf_used;
1772 rec->info[recInfo_delKeys] = reckeys.buf;
1774 reckeys.buf_max = 0;