2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.44 1995-12-12 16:00:54 adam
8 * System call sync(2) used after update/commit.
9 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
12 * Revision 1.43 1995/12/11 09:12:46 adam
13 * The rec_get function returns NULL if record doesn't exist - will
14 * happen in the server if the result set records have been deleted since
15 * the creation of the set (i.e. the search).
16 * The server saves a result temporarily if it is 'volatile', i.e. the
17 * set is register dependent.
19 * Revision 1.42 1995/12/07 17:38:46 adam
20 * Work locking mechanisms for concurrent updates/commit.
22 * Revision 1.41 1995/12/06 16:06:42 adam
23 * Better diagnostics. Work on 'real' dictionary deletion.
25 * Revision 1.40 1995/12/05 16:57:40 adam
26 * More work on regular patterns.
28 * Revision 1.39 1995/12/05 13:20:18 adam
29 * Bug fix: file_read sometimes returned early EOF.
31 * Revision 1.38 1995/12/04 17:59:21 adam
32 * More work on regular expression conversion.
34 * Revision 1.37 1995/12/04 14:22:27 adam
35 * Extra arg to recType_byName.
36 * Started work on new regular expression parsed input to
39 * Revision 1.36 1995/11/30 08:34:29 adam
40 * Started work on commit facility.
41 * Changed a few malloc/free to xmalloc/xfree.
43 * Revision 1.35 1995/11/28 14:26:21 adam
44 * Bug fix: recordId with constant wasn't right.
45 * Bug fix: recordId dictionary entry wasn't deleted when needed.
47 * Revision 1.34 1995/11/28 09:09:38 adam
48 * Zebra config renamed.
49 * Use setting 'recordId' to identify record now.
50 * Bug fix in recindex.c: rec_release_blocks was invokeded even
51 * though the blocks were already released.
52 * File traversal properly deletes records when needed.
54 * Revision 1.33 1995/11/27 09:56:20 adam
55 * Record info elements better enumerated. Internal store of records.
57 * Revision 1.32 1995/11/25 10:24:05 adam
58 * More record fields - they are enumerated now.
59 * New options: flagStoreData flagStoreKey.
61 * Revision 1.31 1995/11/24 11:31:35 adam
62 * Commands add & del read filenames from stdin if source directory is
64 * Match criteria supports 'constant' strings.
66 * Revision 1.30 1995/11/22 17:19:16 adam
67 * Record management uses the bfile system.
69 * Revision 1.29 1995/11/21 15:01:14 adam
70 * New general match criteria implemented.
71 * New feature: document groups.
73 * Revision 1.28 1995/11/21 09:20:30 adam
74 * Yet more work on record match.
76 * Revision 1.27 1995/11/20 16:59:45 adam
77 * New update method: the 'old' keys are saved for each records.
79 * Revision 1.26 1995/11/20 11:56:24 adam
80 * Work on new traversal.
82 * Revision 1.25 1995/11/16 15:34:54 adam
83 * Uses new record management system in both indexer and server.
85 * Revision 1.24 1995/11/15 19:13:08 adam
86 * Work on record management.
88 * Revision 1.23 1995/10/27 14:00:10 adam
89 * Implemented detection of database availability.
91 * Revision 1.22 1995/10/17 18:02:07 adam
92 * New feature: databases. Implemented as prefix to words in dictionary.
94 * Revision 1.21 1995/10/10 12:24:38 adam
95 * Temporary sort files are compressed.
97 * Revision 1.20 1995/10/06 13:52:05 adam
98 * Bug fixes. Handler may abort further scanning.
100 * Revision 1.19 1995/10/04 12:55:16 adam
101 * Bug fix in ranked search. Use=Any keys inserted.
103 * Revision 1.18 1995/10/04 09:37:08 quinn
106 * Revision 1.17 1995/10/03 14:28:57 adam
107 * Buffered read in extract works.
109 * Revision 1.16 1995/10/03 14:28:45 adam
110 * Work on more effecient read handler in extract.
112 * Revision 1.15 1995/10/02 15:42:53 adam
113 * Extract uses file descriptors instead of FILE pointers.
115 * Revision 1.14 1995/10/02 15:29:13 adam
116 * More logging in file_extract.
118 * Revision 1.13 1995/09/29 14:01:39 adam
121 * Revision 1.12 1995/09/28 14:22:56 adam
122 * Sort uses smaller temporary files.
124 * Revision 1.11 1995/09/28 12:10:31 adam
125 * Bug fixes. Field prefix used in queries.
127 * Revision 1.10 1995/09/28 09:19:41 adam
128 * xfree/xmalloc used everywhere.
129 * Extract/retrieve method seems to work for text records.
131 * Revision 1.9 1995/09/27 12:22:28 adam
132 * More work on extract in record control.
133 * Field name is not in isam keys but in prefix in dictionary words.
135 * Revision 1.8 1995/09/14 07:48:22 adam
136 * Record control management.
138 * Revision 1.7 1995/09/11 13:09:32 adam
139 * More work on relevance feedback.
141 * Revision 1.6 1995/09/08 14:52:27 adam
142 * Minor changes. Dictionary is lower case now.
144 * Revision 1.5 1995/09/06 16:11:16 adam
145 * Option: only one word key per file.
147 * Revision 1.4 1995/09/05 15:28:39 adam
148 * More work on search engine.
150 * Revision 1.3 1995/09/04 12:33:41 adam
151 * Various cleanup. YAZ util used instead.
153 * Revision 1.2 1995/09/04 09:10:34 adam
154 * More work on index add/del/update.
155 * Merge sort implemented.
156 * Initial work on z39 server.
158 * Revision 1.1 1995/09/01 14:06:35 adam
159 * Split of work into more files.
168 #include <alexutil.h>
172 #include "recindex.h"
174 static Dict matchDict;
176 static Records records = NULL;
178 static char **key_buf;
179 static size_t ptr_top;
181 static size_t key_buf_used;
182 static int key_file_no;
184 static int records_inserted = 0;
185 static int records_updated = 0;
186 static int records_deleted = 0;
188 void key_open (int mem)
192 key_buf = xmalloc (mem);
193 ptr_top = mem/sizeof(char*);
199 if (!(matchDict = dict_open (GMATCH_DICT, 50, 1)))
201 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
205 records = rec_open (1);
214 void encode_key_init (struct encode_info *i)
220 char *encode_key_int (int d, char *bp)
229 else if (d <= 4194303)
231 *bp++ = 128 + (d>>16);
232 *bp++ = (d>>8) & 255;
237 *bp++ = 192 + (d>>24);
238 *bp++ = (d>>16) & 255;
239 *bp++ = (d>>8) & 255;
245 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
250 while ((*bp++ = *k++))
252 memcpy (&key, k+1, sizeof(struct it_key));
253 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
254 if (i->sysno != key.sysno)
256 i->sysno = key.sysno;
259 bp = encode_key_int (key.seqno - i->seqno, bp);
260 i->seqno = key.seqno;
261 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
263 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
268 void key_flush (void)
273 struct encode_info encode_info;
279 logf (LOG_LOG, "sorting section %d", key_file_no);
280 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
281 sprintf (out_fname, TEMP_FNAME, key_file_no);
283 if (!(outf = fopen (out_fname, "w")))
285 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
288 logf (LOG_LOG, "writing section %d", key_file_no);
289 prevcp = cp = key_buf[ptr_top-ptr_i];
291 encode_key_init (&encode_info);
292 encode_key_write (cp, &encode_info, outf);
295 cp = key_buf[ptr_top-ptr_i];
296 if (strcmp (cp, prevcp))
298 encode_key_init (&encode_info);
299 encode_key_write (cp, &encode_info, outf);
303 encode_key_write (cp + strlen(cp), &encode_info, outf);
307 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
310 logf (LOG_LOG, "finished section %d", key_file_no);
319 rec_close (&records);
320 dict_close (matchDict);
322 logf (LOG_LOG, "Records inserted %6d", records_inserted);
323 logf (LOG_LOG, "Records updated %6d", records_updated);
324 logf (LOG_LOG, "Records deleted %6d", records_deleted);
328 static void wordInit (RecWord *p)
332 p->which = Word_String;
341 static void addRecordKey (const RecWord *p)
348 if (reckeys.buf_used+1024 > reckeys.buf_max)
352 b = xmalloc (reckeys.buf_max += 65000);
353 if (reckeys.buf_used > 0)
354 memcpy (b, reckeys.buf, reckeys.buf_used);
358 dst = reckeys.buf + reckeys.buf_used;
362 attrSet = p->attrSet;
363 memcpy (dst, &attrSet, sizeof(attrSet));
364 dst += sizeof(attrSet);
366 attrUse = p->attrUse;
367 memcpy (dst, &attrUse, sizeof(attrUse));
368 dst += sizeof(attrUse);
370 for (i = 0; p->u.string[i]; i++)
371 *dst++ = p->u.string[i];
374 memcpy (dst, &p->seqno, sizeof(p->seqno));
375 dst += sizeof(p->seqno);
381 reckeys.buf_used = dst - reckeys.buf;
384 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
385 const char *databaseName)
388 while (off < reckeys->buf_used)
390 const char *src = reckeys->buf + off;
395 memcpy (&attrSet, src, sizeof(attrSet));
396 src += sizeof(attrSet);
398 memcpy (&attrUse, src, sizeof(attrUse));
399 src += sizeof(attrUse);
401 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
404 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
405 key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
406 attrSet, attrUse, databaseName);
408 ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
410 ((char*)key_buf) [key_buf_used++] = '\0';
412 ((char*) key_buf)[key_buf_used++] = cmd;
414 memcpy (&key.seqno, src, sizeof(key.seqno));
415 src += sizeof(key.seqno);
417 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
418 key_buf_used += sizeof(key);
419 off = src - reckeys->buf;
421 assert (off == reckeys->buf_used);
424 static const char **searchRecordKey (struct recKeys *reckeys,
425 int attrSetS, int attrUseS)
427 static const char *ws[32];
432 for (i = 0; i<32; i++)
435 while (off < reckeys->buf_used)
437 const char *src = reckeys->buf + off;
443 memcpy (&attrSet, src, sizeof(attrSet));
444 src += sizeof(attrSet);
446 memcpy (&attrUse, src, sizeof(attrUse));
447 src += sizeof(attrUse);
453 memcpy (&seqno, src, sizeof(seqno));
454 src += sizeof(seqno);
457 logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart);
459 if (attrUseS == attrUse && attrSetS == attrSet)
466 woff = seqno - startSeq;
467 if (woff >= 0 && woff < 31)
471 off = src - reckeys->buf;
473 assert (off == reckeys->buf_used);
477 static void addRecordKeyAny (const RecWord *p)
479 if (p->attrSet != 1 || p->attrUse != 1016)
483 memcpy (&w, p, sizeof(w));
492 #define FILE_READ_BUFSIZE 4096
494 static int file_noread;
495 #if FILE_READ_BUFSIZE
496 static char *file_buf;
497 static int file_offset;
498 static int file_bufsize;
501 static void file_read_start (int fd)
504 #if FILE_READ_BUFSIZE
506 file_buf = xmalloc (FILE_READ_BUFSIZE);
507 file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
511 static void file_read_stop (int fd)
513 #if FILE_READ_BUFSIZE
519 static int file_read (int fd, char *buf, size_t count)
521 #if FILE_READ_BUFSIZE
522 int l = file_bufsize - file_offset;
528 memcpy (buf, file_buf + file_offset, l);
530 if (count > FILE_READ_BUFSIZE)
532 if ((r = read (fd, buf + l, count)) == -1)
534 logf (LOG_FATAL|LOG_ERRNO, "read");
542 file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
545 logf (LOG_FATAL|LOG_ERRNO, "read");
551 memcpy (buf + l, file_buf, r);
558 memcpy (buf + l, file_buf, count - l);
559 file_noread += count;
563 memcpy (buf, file_buf + file_offset, count);
564 file_offset += count;
565 file_noread += count;
569 r = read (fd, buf, count);
576 static int atois (const char **s)
579 while ( (c=**s) >= '0' && c <= '9')
581 val = val*10 + c - '0';
587 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
591 static char dstBuf[2048];
593 const char *s = spec;
594 static const char **w;
599 while (*s == ' ' || *s == '\t')
606 int attrSet, attrUse;
610 attrSet = atois (&s);
613 logf (LOG_WARN, "Missing , in match criteria %s in group %s",
614 spec, rGroup->groupName ? rGroup->groupName : "none");
618 attrUse = atois (&s);
619 w = searchRecordKey (reckeys, attrSet, attrUse);
624 for (i = 0; i<32; i++)
629 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
630 spec, rGroup->groupName ? rGroup->groupName : "none");
635 for (i = 0; i<32; i++)
636 if (matchFlag[i] && w[i])
648 logf (LOG_WARN, "Record in file %s didn't contain match"
649 " fields in (%d,%d)", fname, attrSet, attrUse);
657 const char *spec_src = NULL;
658 const char *s1 = ++s;
659 while (*s1 && *s1 != ' ' && *s1 != '\t')
665 memcpy (special, s, spec_len);
666 special[spec_len] = '\0';
669 if (!strcmp (special, "group"))
670 spec_src = rGroup->groupName;
671 else if (!strcmp (special, "database"))
672 spec_src = rGroup->databaseName;
673 else if (!strcmp (special, "filename"))
675 else if (!strcmp (special, "type"))
676 spec_src = rGroup->recordType;
681 strcpy (dst, spec_src);
682 dst += strlen (spec_src);
685 else if (*s == '\"' || *s == '\'')
687 int stopMarker = *s++;
691 while (*s && *s != stopMarker)
694 tmpString[i++] = *s++;
699 strcpy (dst, tmpString);
700 dst += strlen (tmpString);
704 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
705 spec, rGroup->groupName ? rGroup->groupName : "none");
712 logf (LOG_WARN, "No match criteria for record %s in group %s",
713 fname, rGroup->groupName ? rGroup->groupName : "none");
719 static int recordExtract (SYSNO *sysno, const char *fname,
720 struct recordGroup *rGroup, int deleteFlag,
721 int fd, RecType recType, char *subType)
723 struct recExtractCtrl extractCtrl;
733 extractCtrl.subType = subType;
734 extractCtrl.init = wordInit;
735 extractCtrl.add = addRecordKeyAny;
737 reckeys.buf_used = 0;
738 extractCtrl.readf = file_read;
739 r = (*recType->extract)(&extractCtrl);
743 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
748 /* perform match if sysno not known and if match criteria is specified */
755 if (rGroup->recordId && *rGroup->recordId)
759 matchStr = fileMatchStr (&reckeys, rGroup, fname,
763 rinfo = dict_lookup (matchDict, matchStr);
765 memcpy (sysno, rinfo+1, sizeof(*sysno));
769 logf (LOG_WARN, "Record not inserted");
780 logf (LOG_LOG, "? %s", fname);
783 logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
784 rec = rec_new (records);
789 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
791 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
797 struct recKeys delkeys;
799 rec = rec_get (records, *sysno);
801 delkeys.buf_used = rec->size[recInfo_delKeys];
802 delkeys.buf = rec->info[recInfo_delKeys];
803 flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]);
806 if (!delkeys.buf_used)
808 logf (LOG_WARN, "cannot delete %s: storeKeys false",
813 logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname);
816 dict_delete (matchDict, matchStr);
817 rec_del (records, &rec);
823 if (!delkeys.buf_used)
825 logf (LOG_WARN, "cannot update %s: storeKeys false",
830 logf (LOG_LOG, "update %s %s", rGroup->recordType,
832 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
837 xfree (rec->info[recInfo_fileType]);
838 rec->info[recInfo_fileType] =
839 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
841 xfree (rec->info[recInfo_filename]);
842 rec->info[recInfo_filename] =
843 rec_strdup (fname, &rec->size[recInfo_filename]);
845 xfree (rec->info[recInfo_delKeys]);
846 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
848 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
849 rec->size[recInfo_delKeys] = reckeys.buf_used;
850 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
851 rec->size[recInfo_delKeys]);
855 rec->info[recInfo_delKeys] = NULL;
856 rec->size[recInfo_delKeys] = 0;
859 xfree (rec->info[recInfo_storeData]);
860 if (rGroup->flagStoreData == 1)
862 rec->size[recInfo_storeData] = file_noread;
863 rec->info[recInfo_storeData] = xmalloc (file_noread);
864 #if FILE_READ_BUFSIZE
865 if (file_noread < FILE_READ_BUFSIZE)
866 memcpy (rec->info[recInfo_storeData], file_buf, file_noread);
870 if (lseek (fd, 0L, SEEK_SET) < 0)
872 logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname);
875 if (read (fd, rec->info[recInfo_storeData], file_noread)
878 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
886 rec->info[recInfo_storeData] = NULL;
887 rec->size[recInfo_storeData] = 0;
889 xfree (rec->info[recInfo_databaseName]);
890 rec->info[recInfo_databaseName] =
891 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
893 rec_put (records, &rec);
897 int fileExtract (SYSNO *sysno, const char *fname,
898 const struct recordGroup *rGroupP, int deleteFlag)
906 struct recordGroup rGroupM;
907 struct recordGroup *rGroup = &rGroupM;
909 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
911 if (!rGroup->groupName || !*rGroup->groupName)
914 sprintf (gprefix, "%s.", rGroup->groupName);
916 logf (LOG_DEBUG, "fileExtract %s", fname);
918 /* determine file extension */
919 for (i = strlen(fname); --i >= 0; )
925 else if (fname[i] == '.')
927 strcpy (ext, fname+i+1);
930 /* determine file type - depending on extension */
931 if (!rGroup->recordType)
933 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
934 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
936 sprintf (ext_res, "%srecordType", gprefix);
937 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
939 logf (LOG_LOG, "? %s", fname);
944 if (!rGroup->recordType)
946 logf (LOG_LOG, "? record %s", fname);
949 if (!(recType = recType_byName (rGroup->recordType, subType)))
951 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
955 /* determine match criteria */
956 if (!rGroup->recordId)
958 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
959 rGroup->recordId = res_get (common_resource, ext_res);
962 /* determine database name */
963 if (!rGroup->databaseName)
965 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
966 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
968 sprintf (ext_res, "%sdatabase", gprefix);
969 rGroup->databaseName = res_get (common_resource, ext_res);
972 if (!rGroup->databaseName)
973 rGroup->databaseName = "Default";
975 if (rGroup->flagStoreData == -1)
978 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
979 if (!(sval = res_get (common_resource, ext_res)))
981 sprintf (ext_res, "%sstoreData", gprefix);
982 sval = res_get (common_resource, ext_res);
985 rGroup->flagStoreData = atoi (sval);
987 if (rGroup->flagStoreData == -1)
988 rGroup->flagStoreData = 0;
990 if (rGroup->flagStoreKeys == -1)
994 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
995 if (!(sval = res_get (common_resource, ext_res)))
997 sprintf (ext_res, "%sstoreKeys", gprefix);
998 sval = res_get (common_resource, ext_res);
1001 rGroup->flagStoreKeys = atoi (sval);
1003 if (rGroup->flagStoreKeys == -1)
1004 rGroup->flagStoreKeys = 0;
1006 if (sysno && deleteFlag)
1010 if ((fd = open (fname, O_RDONLY)) == -1)
1012 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1016 file_read_start (fd);
1017 recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType);
1018 file_read_stop (fd);