2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.29 1995-11-21 15:01:14 adam
8 * New general match criteria implemented.
9 * New feature: document groups.
11 * Revision 1.28 1995/11/21 09:20:30 adam
12 * Yet more work on record match.
14 * Revision 1.27 1995/11/20 16:59:45 adam
15 * New update method: the 'old' keys are saved for each records.
17 * Revision 1.26 1995/11/20 11:56:24 adam
18 * Work on new traversal.
20 * Revision 1.25 1995/11/16 15:34:54 adam
21 * Uses new record management system in both indexer and server.
23 * Revision 1.24 1995/11/15 19:13:08 adam
24 * Work on record management.
26 * Revision 1.23 1995/10/27 14:00:10 adam
27 * Implemented detection of database availability.
29 * Revision 1.22 1995/10/17 18:02:07 adam
30 * New feature: databases. Implemented as prefix to words in dictionary.
32 * Revision 1.21 1995/10/10 12:24:38 adam
33 * Temporary sort files are compressed.
35 * Revision 1.20 1995/10/06 13:52:05 adam
36 * Bug fixes. Handler may abort further scanning.
38 * Revision 1.19 1995/10/04 12:55:16 adam
39 * Bug fix in ranked search. Use=Any keys inserted.
41 * Revision 1.18 1995/10/04 09:37:08 quinn
44 * Revision 1.17 1995/10/03 14:28:57 adam
45 * Buffered read in extract works.
47 * Revision 1.16 1995/10/03 14:28:45 adam
48 * Work on more effecient read handler in extract.
50 * Revision 1.15 1995/10/02 15:42:53 adam
51 * Extract uses file descriptors instead of FILE pointers.
53 * Revision 1.14 1995/10/02 15:29:13 adam
54 * More logging in file_extract.
56 * Revision 1.13 1995/09/29 14:01:39 adam
59 * Revision 1.12 1995/09/28 14:22:56 adam
60 * Sort uses smaller temporary files.
62 * Revision 1.11 1995/09/28 12:10:31 adam
63 * Bug fixes. Field prefix used in queries.
65 * Revision 1.10 1995/09/28 09:19:41 adam
66 * xfree/xmalloc used everywhere.
67 * Extract/retrieve method seems to work for text records.
69 * Revision 1.9 1995/09/27 12:22:28 adam
70 * More work on extract in record control.
71 * Field name is not in isam keys but in prefix in dictionary words.
73 * Revision 1.8 1995/09/14 07:48:22 adam
74 * Record control management.
76 * Revision 1.7 1995/09/11 13:09:32 adam
77 * More work on relevance feedback.
79 * Revision 1.6 1995/09/08 14:52:27 adam
80 * Minor changes. Dictionary is lower case now.
82 * Revision 1.5 1995/09/06 16:11:16 adam
83 * Option: only one word key per file.
85 * Revision 1.4 1995/09/05 15:28:39 adam
86 * More work on search engine.
88 * Revision 1.3 1995/09/04 12:33:41 adam
89 * Various cleanup. YAZ util used instead.
91 * Revision 1.2 1995/09/04 09:10:34 adam
92 * More work on index add/del/update.
93 * Merge sort implemented.
94 * Initial work on z39 server.
96 * Revision 1.1 1995/09/01 14:06:35 adam
97 * Split of work into more files.
106 #include <alexutil.h>
110 #include "recindex.h"
112 static Dict matchDict;
114 static Records records = NULL;
116 static char **key_buf;
117 static size_t ptr_top;
119 static size_t key_buf_used;
120 static int key_file_no;
122 static int records_inserted = 0;
123 static int records_updated = 0;
124 static int records_deleted = 0;
126 #define MATCH_DICT "match"
128 void key_open (int mem)
132 key_buf = xmalloc (mem);
133 ptr_top = mem/sizeof(char*);
139 if (!(matchDict = dict_open (MATCH_DICT, 20, 1)))
141 logf (LOG_FATAL, "dict_open fail of %s", MATCH_DICT);
145 records = rec_open (1);
154 void encode_key_init (struct encode_info *i)
160 char *encode_key_int (int d, char *bp)
169 else if (d <= 4194303)
171 *bp++ = 128 + (d>>16);
172 *bp++ = (d>>8) & 255;
177 *bp++ = 192 + (d>>24);
178 *bp++ = (d>>16) & 255;
179 *bp++ = (d>>8) & 255;
185 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
190 while ((*bp++ = *k++))
192 memcpy (&key, k+1, sizeof(struct it_key));
193 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
194 if (i->sysno != key.sysno)
196 i->sysno = key.sysno;
199 bp = encode_key_int (key.seqno - i->seqno, bp);
200 i->seqno = key.seqno;
201 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
203 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
208 void key_flush (void)
213 struct encode_info encode_info;
219 logf (LOG_LOG, "sorting section %d", key_file_no);
220 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
221 sprintf (out_fname, TEMP_FNAME, key_file_no);
223 if (!(outf = fopen (out_fname, "w")))
225 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
228 logf (LOG_LOG, "writing section %d", key_file_no);
229 prevcp = cp = key_buf[ptr_top-ptr_i];
231 encode_key_init (&encode_info);
232 encode_key_write (cp, &encode_info, outf);
235 cp = key_buf[ptr_top-ptr_i];
236 if (strcmp (cp, prevcp))
238 encode_key_init (&encode_info);
239 encode_key_write (cp, &encode_info, outf);
243 encode_key_write (cp + strlen(cp), &encode_info, outf);
247 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
250 logf (LOG_LOG, "finished section %d", key_file_no);
259 rec_close (&records);
260 dict_close (matchDict);
262 logf (LOG_LOG, "Records inserted %6d", records_inserted);
263 logf (LOG_LOG, "Records updated %6d", records_updated);
264 logf (LOG_LOG, "Records deleted %6d", records_deleted);
268 static void wordInit (RecWord *p)
272 p->which = Word_String;
281 static void addRecordKey (const RecWord *p)
288 if (reckeys.buf_used+1024 > reckeys.buf_max)
292 b = malloc (reckeys.buf_max += 65000);
293 if (reckeys.buf_used > 0)
294 memcpy (b, reckeys.buf, reckeys.buf_used);
298 dst = reckeys.buf + reckeys.buf_used;
302 attrSet = p->attrSet;
303 memcpy (dst, &attrSet, sizeof(attrSet));
304 dst += sizeof(attrSet);
306 attrUse = p->attrUse;
307 memcpy (dst, &attrUse, sizeof(attrUse));
308 dst += sizeof(attrUse);
310 for (i = 0; p->u.string[i]; i++)
311 *dst++ = p->u.string[i];
314 memcpy (dst, &p->seqno, sizeof(p->seqno));
315 dst += sizeof(p->seqno);
321 reckeys.buf_used = dst - reckeys.buf;
324 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
325 const char *databaseName)
328 while (off < reckeys->buf_used)
330 const char *src = reckeys->buf + off;
335 memcpy (&attrSet, src, sizeof(attrSet));
336 src += sizeof(attrSet);
338 memcpy (&attrUse, src, sizeof(attrUse));
339 src += sizeof(attrUse);
341 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
344 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
345 key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
346 attrSet, attrUse, databaseName);
348 ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
350 ((char*)key_buf) [key_buf_used++] = '\0';
352 ((char*) key_buf)[key_buf_used++] = cmd;
354 memcpy (&key.seqno, src, sizeof(key.seqno));
355 src += sizeof(key.seqno);
357 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
358 key_buf_used += sizeof(key);
359 off = src - reckeys->buf;
361 assert (off == reckeys->buf_used);
364 static const char **searchRecordKey (struct recKeys *reckeys,
365 int attrSetS, int attrUseS)
367 static const char *ws[32];
372 for (i = 0; i<32; i++)
375 while (off < reckeys->buf_used)
377 const char *src = reckeys->buf + off;
383 memcpy (&attrSet, src, sizeof(attrSet));
384 src += sizeof(attrSet);
386 memcpy (&attrUse, src, sizeof(attrUse));
387 src += sizeof(attrUse);
393 memcpy (&seqno, src, sizeof(seqno));
394 src += sizeof(seqno);
397 logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart);
399 if (attrUseS == attrUse && attrSetS == attrSet)
406 woff = seqno - startSeq;
407 if (woff >= 0 && woff < 31)
411 off = src - reckeys->buf;
413 assert (off == reckeys->buf_used);
417 static void addRecordKeyAny (const RecWord *p)
419 if (p->attrSet != 1 || p->attrUse != 1016)
423 memcpy (&w, p, sizeof(w));
431 static char *file_buf;
432 static int file_offset;
433 static int file_bufsize;
435 static void file_read_start (int fd)
438 file_buf = xmalloc (4096);
439 file_bufsize = read (fd, file_buf, 4096);
442 static void file_read_stop (int fd)
447 static int file_read (int fd, char *buf, size_t count)
449 int l = file_bufsize - file_offset;
455 memcpy (buf, file_buf + file_offset, l);
457 if (count > file_bufsize)
459 if ((r = read (fd, buf + l, count)) == -1)
461 logf (LOG_FATAL|LOG_ERRNO, "read");
468 file_bufsize = r = read (fd, file_buf, 4096);
471 logf (LOG_FATAL|LOG_ERRNO, "read");
477 memcpy (buf + l, file_buf, r);
483 memcpy (buf + l, file_buf, count - l);
487 memcpy (buf, file_buf + file_offset, count);
488 file_offset += count;
492 static int atois (const char **s)
495 while ( (c=**s) >= '0' && c <= '9')
497 val = val*10 + c - '0';
503 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
505 const char *recordType,
508 static char dstBuf[2048];
510 const char *s = spec;
511 static const char **w;
516 while (*s == ' ' || *s == '\t')
523 int attrSet, attrUse;
527 attrSet = atois (&s);
530 logf (LOG_WARN, "Missing , in match criteria %s in group %s",
531 spec, rGroup->groupName ? rGroup->groupName : "none");
535 attrUse = atois (&s);
536 w = searchRecordKey (reckeys, attrSet, attrUse);
541 for (i = 0; i<32; i++)
546 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
547 spec, rGroup->groupName ? rGroup->groupName : "none");
552 for (i = 0; i<32; i++)
553 if (matchFlag[i] && w[i])
565 logf (LOG_WARN, "Record in file %s didn't contain match"
566 " fields in (%d,%d)", fname, attrSet, attrUse);
574 const char *spec_src = NULL;
575 const char *s1 = ++s;
576 while (*s1 && *s1 != ' ' && *s1 != '\t')
582 memcpy (special, s, spec_len);
583 special[spec_len] = '\0';
586 if (strcmp (special, "group"))
587 spec_src = rGroup->groupName;
588 else if (strcmp (special, "database"))
589 spec_src = rGroup->databaseName;
590 else if (strcmp (special, "filename"))
592 else if (strcmp (special, "type"))
593 spec_src = recordType;
598 strcpy (dst, spec_src);
599 dst += strlen(spec_src);
604 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
605 spec, rGroup->groupName ? rGroup->groupName : "none");
612 logf (LOG_WARN, "No match criteria for record %s in group %s",
613 fname, rGroup->groupName ? rGroup->groupName : "none");
619 int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
627 const char *file_type;
628 const char *file_match;
629 struct recExtractCtrl extractCtrl;
634 if (!rGroup->groupName || !*rGroup->groupName)
637 sprintf (gprefix, "%s.", rGroup->groupName);
639 logf (LOG_DEBUG, "fileExtractAdd %s", fname);
641 /* determine file extension */
642 for (i = strlen(fname); --i >= 0; )
648 else if (fname[i] == '.')
650 strcpy (ext, fname+i+1);
653 /* determine file type - depending on extension */
654 sprintf (ext_res, "%sfileExtension.%s", gprefix, ext);
655 if (!(file_type = res_get (common_resource, ext_res)))
657 if (!(recType = recType_byName (file_type)))
660 /* determine match criteria */
661 sprintf (ext_res, "%sfileMatch.%s", gprefix, ext);
662 file_match = res_get (common_resource, ext_res);
665 sprintf (ext_res, "%sfileMatch", gprefix);
666 file_match = res_get (common_resource, ext_res);
669 /* determine database name */
670 if (!rGroup->databaseName)
672 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
673 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
675 sprintf (ext_res, "%sdatabase", gprefix);
676 rGroup->databaseName = res_get (common_resource, ext_res);
679 if (!rGroup->databaseName)
680 rGroup->databaseName = "Default";
682 /* open input file */
683 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
685 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
690 extractCtrl.subType = "";
691 extractCtrl.init = wordInit;
692 extractCtrl.add = addRecordKeyAny;
694 reckeys.buf_used = 0;
695 file_read_start (extractCtrl.fd);
696 extractCtrl.readf = file_read;
697 r = (*recType->extract)(&extractCtrl);
698 file_read_stop (extractCtrl.fd);
699 close (extractCtrl.fd);
703 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
707 /* perform match if sysno not known and if match criteria is specified */
710 if (!sysno && file_match)
715 matchStr = fileMatchStr(&reckeys, rGroup, fname, file_type,
719 rinfo = dict_lookup (matchDict, matchStr);
721 memcpy (sysno, rinfo+1, sizeof(*sysno));
727 logf (LOG_WARN, "Record not inserted");
735 logf (LOG_LOG, "add record %s", fname);
736 rec = rec_new (records);
740 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
741 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
747 struct recKeys delkeys;
749 rec = rec_get (records, *sysno);
751 delkeys.buf_used = rec->size[2];
752 delkeys.buf = rec->info[2];
753 flushRecordKeys (*sysno, 0, &delkeys, rec->info[3]);
754 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
759 rec->info[0] = rec_strdup (file_type, &rec->size[0]);
762 rec->info[1] = rec_strdup (fname, &rec->size[1]);
765 if (reckeys.buf_used > 0)
767 rec->info[2] = malloc (reckeys.buf_used);
768 rec->size[2] = reckeys.buf_used;
769 memcpy (rec->info[2], reckeys.buf, rec->size[2]);
777 rec->info[3] = rec_strdup (rGroup->databaseName, &rec->size[3]);
779 rec_put (records, &rec);