2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.46 1995-12-15 14:57:16 adam
10 * Revision 1.45 1995/12/15 12:37:41 adam
11 * In addRecordKeyAny: Writes key only when attrSet != -1.
13 * Revision 1.44 1995/12/12 16:00:54 adam
14 * System call sync(2) used after update/commit.
15 * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK
18 * Revision 1.43 1995/12/11 09:12:46 adam
19 * The rec_get function returns NULL if record doesn't exist - will
20 * happen in the server if the result set records have been deleted since
21 * the creation of the set (i.e. the search).
22 * The server saves a result temporarily if it is 'volatile', i.e. the
23 * set is register dependent.
25 * Revision 1.42 1995/12/07 17:38:46 adam
26 * Work locking mechanisms for concurrent updates/commit.
28 * Revision 1.41 1995/12/06 16:06:42 adam
29 * Better diagnostics. Work on 'real' dictionary deletion.
31 * Revision 1.40 1995/12/05 16:57:40 adam
32 * More work on regular patterns.
34 * Revision 1.39 1995/12/05 13:20:18 adam
35 * Bug fix: file_read sometimes returned early EOF.
37 * Revision 1.38 1995/12/04 17:59:21 adam
38 * More work on regular expression conversion.
40 * Revision 1.37 1995/12/04 14:22:27 adam
41 * Extra arg to recType_byName.
42 * Started work on new regular expression parsed input to
45 * Revision 1.36 1995/11/30 08:34:29 adam
46 * Started work on commit facility.
47 * Changed a few malloc/free to xmalloc/xfree.
49 * Revision 1.35 1995/11/28 14:26:21 adam
50 * Bug fix: recordId with constant wasn't right.
51 * Bug fix: recordId dictionary entry wasn't deleted when needed.
53 * Revision 1.34 1995/11/28 09:09:38 adam
54 * Zebra config renamed.
55 * Use setting 'recordId' to identify record now.
56 * Bug fix in recindex.c: rec_release_blocks was invokeded even
57 * though the blocks were already released.
58 * File traversal properly deletes records when needed.
60 * Revision 1.33 1995/11/27 09:56:20 adam
61 * Record info elements better enumerated. Internal store of records.
63 * Revision 1.32 1995/11/25 10:24:05 adam
64 * More record fields - they are enumerated now.
65 * New options: flagStoreData flagStoreKey.
67 * Revision 1.31 1995/11/24 11:31:35 adam
68 * Commands add & del read filenames from stdin if source directory is
70 * Match criteria supports 'constant' strings.
72 * Revision 1.30 1995/11/22 17:19:16 adam
73 * Record management uses the bfile system.
75 * Revision 1.29 1995/11/21 15:01:14 adam
76 * New general match criteria implemented.
77 * New feature: document groups.
79 * Revision 1.28 1995/11/21 09:20:30 adam
80 * Yet more work on record match.
82 * Revision 1.27 1995/11/20 16:59:45 adam
83 * New update method: the 'old' keys are saved for each records.
85 * Revision 1.26 1995/11/20 11:56:24 adam
86 * Work on new traversal.
88 * Revision 1.25 1995/11/16 15:34:54 adam
89 * Uses new record management system in both indexer and server.
91 * Revision 1.24 1995/11/15 19:13:08 adam
92 * Work on record management.
94 * Revision 1.23 1995/10/27 14:00:10 adam
95 * Implemented detection of database availability.
97 * Revision 1.22 1995/10/17 18:02:07 adam
98 * New feature: databases. Implemented as prefix to words in dictionary.
100 * Revision 1.21 1995/10/10 12:24:38 adam
101 * Temporary sort files are compressed.
103 * Revision 1.20 1995/10/06 13:52:05 adam
104 * Bug fixes. Handler may abort further scanning.
106 * Revision 1.19 1995/10/04 12:55:16 adam
107 * Bug fix in ranked search. Use=Any keys inserted.
109 * Revision 1.18 1995/10/04 09:37:08 quinn
112 * Revision 1.17 1995/10/03 14:28:57 adam
113 * Buffered read in extract works.
115 * Revision 1.16 1995/10/03 14:28:45 adam
116 * Work on more effecient read handler in extract.
118 * Revision 1.15 1995/10/02 15:42:53 adam
119 * Extract uses file descriptors instead of FILE pointers.
121 * Revision 1.14 1995/10/02 15:29:13 adam
122 * More logging in file_extract.
124 * Revision 1.13 1995/09/29 14:01:39 adam
127 * Revision 1.12 1995/09/28 14:22:56 adam
128 * Sort uses smaller temporary files.
130 * Revision 1.11 1995/09/28 12:10:31 adam
131 * Bug fixes. Field prefix used in queries.
133 * Revision 1.10 1995/09/28 09:19:41 adam
134 * xfree/xmalloc used everywhere.
135 * Extract/retrieve method seems to work for text records.
137 * Revision 1.9 1995/09/27 12:22:28 adam
138 * More work on extract in record control.
139 * Field name is not in isam keys but in prefix in dictionary words.
141 * Revision 1.8 1995/09/14 07:48:22 adam
142 * Record control management.
144 * Revision 1.7 1995/09/11 13:09:32 adam
145 * More work on relevance feedback.
147 * Revision 1.6 1995/09/08 14:52:27 adam
148 * Minor changes. Dictionary is lower case now.
150 * Revision 1.5 1995/09/06 16:11:16 adam
151 * Option: only one word key per file.
153 * Revision 1.4 1995/09/05 15:28:39 adam
154 * More work on search engine.
156 * Revision 1.3 1995/09/04 12:33:41 adam
157 * Various cleanup. YAZ util used instead.
159 * Revision 1.2 1995/09/04 09:10:34 adam
160 * More work on index add/del/update.
161 * Merge sort implemented.
162 * Initial work on z39 server.
164 * Revision 1.1 1995/09/01 14:06:35 adam
165 * Split of work into more files.
174 #include <alexutil.h>
178 #include "recindex.h"
180 static Dict matchDict;
182 static Records records = NULL;
184 static char **key_buf;
185 static size_t ptr_top;
187 static size_t key_buf_used;
188 static int key_file_no;
190 static int records_inserted = 0;
191 static int records_updated = 0;
192 static int records_deleted = 0;
194 void key_open (int mem)
198 key_buf = xmalloc (mem);
199 ptr_top = mem/sizeof(char*);
205 if (!(matchDict = dict_open (GMATCH_DICT, 50, 1)))
207 logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
211 records = rec_open (1);
220 void encode_key_init (struct encode_info *i)
226 char *encode_key_int (int d, char *bp)
235 else if (d <= 4194303)
237 *bp++ = 128 + (d>>16);
238 *bp++ = (d>>8) & 255;
243 *bp++ = 192 + (d>>24);
244 *bp++ = (d>>16) & 255;
245 *bp++ = (d>>8) & 255;
251 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
256 while ((*bp++ = *k++))
258 memcpy (&key, k+1, sizeof(struct it_key));
259 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
260 if (i->sysno != key.sysno)
262 i->sysno = key.sysno;
265 bp = encode_key_int (key.seqno - i->seqno, bp);
266 i->seqno = key.seqno;
267 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
269 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
274 void key_flush (void)
279 struct encode_info encode_info;
285 logf (LOG_LOG, "sorting section %d", key_file_no);
286 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
287 sprintf (out_fname, TEMP_FNAME, key_file_no);
289 if (!(outf = fopen (out_fname, "w")))
291 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
294 logf (LOG_LOG, "writing section %d", key_file_no);
295 prevcp = cp = key_buf[ptr_top-ptr_i];
297 encode_key_init (&encode_info);
298 encode_key_write (cp, &encode_info, outf);
301 cp = key_buf[ptr_top-ptr_i];
302 if (strcmp (cp, prevcp))
304 encode_key_init (&encode_info);
305 encode_key_write (cp, &encode_info, outf);
309 encode_key_write (cp + strlen(cp), &encode_info, outf);
313 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
316 logf (LOG_LOG, "finished section %d", key_file_no);
325 rec_close (&records);
326 dict_close (matchDict);
328 logf (LOG_LOG, "Records inserted %6d", records_inserted);
329 logf (LOG_LOG, "Records updated %6d", records_updated);
330 logf (LOG_LOG, "Records deleted %6d", records_deleted);
334 static void wordInit (RecWord *p)
338 p->which = Word_String;
347 static void addRecordKey (const RecWord *p)
354 if (reckeys.buf_used+1024 > reckeys.buf_max)
358 b = xmalloc (reckeys.buf_max += 65000);
359 if (reckeys.buf_used > 0)
360 memcpy (b, reckeys.buf, reckeys.buf_used);
364 dst = reckeys.buf + reckeys.buf_used;
368 attrSet = p->attrSet;
369 memcpy (dst, &attrSet, sizeof(attrSet));
370 dst += sizeof(attrSet);
372 attrUse = p->attrUse;
373 memcpy (dst, &attrUse, sizeof(attrUse));
374 dst += sizeof(attrUse);
376 for (i = 0; p->u.string[i]; i++)
377 *dst++ = p->u.string[i];
380 memcpy (dst, &p->seqno, sizeof(p->seqno));
381 dst += sizeof(p->seqno);
387 reckeys.buf_used = dst - reckeys.buf;
390 static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
391 const char *databaseName)
394 while (off < reckeys->buf_used)
396 const char *src = reckeys->buf + off;
401 memcpy (&attrSet, src, sizeof(attrSet));
402 src += sizeof(attrSet);
404 memcpy (&attrUse, src, sizeof(attrUse));
405 src += sizeof(attrUse);
407 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
410 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
411 key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
412 attrSet, attrUse, databaseName);
414 ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
416 ((char*)key_buf) [key_buf_used++] = '\0';
418 ((char*) key_buf)[key_buf_used++] = cmd;
420 memcpy (&key.seqno, src, sizeof(key.seqno));
421 src += sizeof(key.seqno);
423 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
424 key_buf_used += sizeof(key);
425 off = src - reckeys->buf;
427 assert (off == reckeys->buf_used);
430 static const char **searchRecordKey (struct recKeys *reckeys,
431 int attrSetS, int attrUseS)
433 static const char *ws[32];
438 for (i = 0; i<32; i++)
441 while (off < reckeys->buf_used)
443 const char *src = reckeys->buf + off;
449 memcpy (&attrSet, src, sizeof(attrSet));
450 src += sizeof(attrSet);
452 memcpy (&attrUse, src, sizeof(attrUse));
453 src += sizeof(attrUse);
459 memcpy (&seqno, src, sizeof(seqno));
460 src += sizeof(seqno);
463 logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart);
465 if (attrUseS == attrUse && attrSetS == attrSet)
472 woff = seqno - startSeq;
473 if (woff >= 0 && woff < 31)
477 off = src - reckeys->buf;
479 assert (off == reckeys->buf_used);
483 static void addRecordKeyAny (const RecWord *p)
485 if (p->attrSet != 1 || p->attrUse != 1016)
489 memcpy (&w, p, sizeof(w));
494 if (p->attrSet != -1)
499 #define FILE_READ_BUFSIZE 4096
501 static int file_noread;
502 #if FILE_READ_BUFSIZE
503 static char *file_buf;
504 static int file_offset;
505 static int file_bufsize;
508 static void file_read_start (int fd)
511 #if FILE_READ_BUFSIZE
513 file_buf = xmalloc (FILE_READ_BUFSIZE);
514 file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
518 static void file_read_stop (int fd)
520 #if FILE_READ_BUFSIZE
526 static int file_read (int fd, char *buf, size_t count)
528 #if FILE_READ_BUFSIZE
529 int l = file_bufsize - file_offset;
535 memcpy (buf, file_buf + file_offset, l);
537 if (count > FILE_READ_BUFSIZE)
539 if ((r = read (fd, buf + l, count)) == -1)
541 logf (LOG_FATAL|LOG_ERRNO, "read");
549 file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
552 logf (LOG_FATAL|LOG_ERRNO, "read");
558 memcpy (buf + l, file_buf, r);
565 memcpy (buf + l, file_buf, count - l);
566 file_noread += count;
570 memcpy (buf, file_buf + file_offset, count);
571 file_offset += count;
572 file_noread += count;
576 r = read (fd, buf, count);
583 static int atois (const char **s)
586 while ( (c=**s) >= '0' && c <= '9')
588 val = val*10 + c - '0';
594 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
598 static char dstBuf[2048];
600 const char *s = spec;
601 static const char **w;
606 while (*s == ' ' || *s == '\t')
613 int attrSet, attrUse;
617 attrSet = atois (&s);
620 logf (LOG_WARN, "Missing , in match criteria %s in group %s",
621 spec, rGroup->groupName ? rGroup->groupName : "none");
625 attrUse = atois (&s);
626 w = searchRecordKey (reckeys, attrSet, attrUse);
631 for (i = 0; i<32; i++)
636 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
637 spec, rGroup->groupName ? rGroup->groupName : "none");
642 for (i = 0; i<32; i++)
643 if (matchFlag[i] && w[i])
655 logf (LOG_WARN, "Record in file %s didn't contain match"
656 " fields in (%d,%d)", fname, attrSet, attrUse);
664 const char *spec_src = NULL;
665 const char *s1 = ++s;
666 while (*s1 && *s1 != ' ' && *s1 != '\t')
672 memcpy (special, s, spec_len);
673 special[spec_len] = '\0';
676 if (!strcmp (special, "group"))
677 spec_src = rGroup->groupName;
678 else if (!strcmp (special, "database"))
679 spec_src = rGroup->databaseName;
680 else if (!strcmp (special, "filename"))
682 else if (!strcmp (special, "type"))
683 spec_src = rGroup->recordType;
688 strcpy (dst, spec_src);
689 dst += strlen (spec_src);
692 else if (*s == '\"' || *s == '\'')
694 int stopMarker = *s++;
698 while (*s && *s != stopMarker)
701 tmpString[i++] = *s++;
706 strcpy (dst, tmpString);
707 dst += strlen (tmpString);
711 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
712 spec, rGroup->groupName ? rGroup->groupName : "none");
719 logf (LOG_WARN, "No match criteria for record %s in group %s",
720 fname, rGroup->groupName ? rGroup->groupName : "none");
726 static int recordExtract (SYSNO *sysno, const char *fname,
727 struct recordGroup *rGroup, int deleteFlag,
728 int fd, RecType recType, char *subType)
730 struct recExtractCtrl extractCtrl;
740 extractCtrl.subType = subType;
741 extractCtrl.init = wordInit;
742 extractCtrl.add = addRecordKeyAny;
744 reckeys.buf_used = 0;
745 extractCtrl.readf = file_read;
746 r = (*recType->extract)(&extractCtrl);
750 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
755 /* perform match if sysno not known and if match criteria is specified */
762 if (rGroup->recordId && *rGroup->recordId)
766 matchStr = fileMatchStr (&reckeys, rGroup, fname,
770 rinfo = dict_lookup (matchDict, matchStr);
772 memcpy (sysno, rinfo+1, sizeof(*sysno));
776 logf (LOG_WARN, "Record not inserted");
787 logf (LOG_LOG, "? %s", fname);
790 logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
791 rec = rec_new (records);
796 dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
798 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
804 struct recKeys delkeys;
806 rec = rec_get (records, *sysno);
808 delkeys.buf_used = rec->size[recInfo_delKeys];
809 delkeys.buf = rec->info[recInfo_delKeys];
810 flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]);
813 if (!delkeys.buf_used)
815 logf (LOG_WARN, "cannot delete %s: storeKeys false",
820 logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname);
823 dict_delete (matchDict, matchStr);
824 rec_del (records, &rec);
830 if (!delkeys.buf_used)
832 logf (LOG_WARN, "cannot update %s: storeKeys false",
837 logf (LOG_LOG, "update %s %s", rGroup->recordType,
839 flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
844 xfree (rec->info[recInfo_fileType]);
845 rec->info[recInfo_fileType] =
846 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
848 xfree (rec->info[recInfo_filename]);
849 rec->info[recInfo_filename] =
850 rec_strdup (fname, &rec->size[recInfo_filename]);
852 xfree (rec->info[recInfo_delKeys]);
853 if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
855 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
856 rec->size[recInfo_delKeys] = reckeys.buf_used;
857 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
858 rec->size[recInfo_delKeys]);
862 rec->info[recInfo_delKeys] = NULL;
863 rec->size[recInfo_delKeys] = 0;
866 xfree (rec->info[recInfo_storeData]);
867 if (rGroup->flagStoreData == 1)
869 rec->size[recInfo_storeData] = file_noread;
870 rec->info[recInfo_storeData] = xmalloc (file_noread);
871 #if FILE_READ_BUFSIZE
872 if (file_noread < FILE_READ_BUFSIZE)
873 memcpy (rec->info[recInfo_storeData], file_buf, file_noread);
877 if (lseek (fd, 0L, SEEK_SET) < 0)
879 logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname);
882 if (read (fd, rec->info[recInfo_storeData], file_noread)
885 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
893 rec->info[recInfo_storeData] = NULL;
894 rec->size[recInfo_storeData] = 0;
896 xfree (rec->info[recInfo_databaseName]);
897 rec->info[recInfo_databaseName] =
898 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
900 rec_put (records, &rec);
904 int fileExtract (SYSNO *sysno, const char *fname,
905 const struct recordGroup *rGroupP, int deleteFlag)
913 struct recordGroup rGroupM;
914 struct recordGroup *rGroup = &rGroupM;
916 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
918 if (!rGroup->groupName || !*rGroup->groupName)
921 sprintf (gprefix, "%s.", rGroup->groupName);
923 logf (LOG_DEBUG, "fileExtract %s", fname);
925 /* determine file extension */
926 for (i = strlen(fname); --i >= 0; )
932 else if (fname[i] == '.')
934 strcpy (ext, fname+i+1);
937 /* determine file type - depending on extension */
938 if (!rGroup->recordType)
940 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
941 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
943 sprintf (ext_res, "%srecordType", gprefix);
944 if (!(rGroup->recordType = res_get (common_resource, ext_res)))
946 logf (LOG_LOG, "? %s", fname);
951 if (!rGroup->recordType)
953 logf (LOG_LOG, "? record %s", fname);
956 if (!(recType = recType_byName (rGroup->recordType, subType)))
958 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
962 /* determine match criteria */
963 if (!rGroup->recordId)
965 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
966 rGroup->recordId = res_get (common_resource, ext_res);
969 /* determine database name */
970 if (!rGroup->databaseName)
972 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
973 if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
975 sprintf (ext_res, "%sdatabase", gprefix);
976 rGroup->databaseName = res_get (common_resource, ext_res);
979 if (!rGroup->databaseName)
980 rGroup->databaseName = "Default";
982 if (rGroup->flagStoreData == -1)
985 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
986 if (!(sval = res_get (common_resource, ext_res)))
988 sprintf (ext_res, "%sstoreData", gprefix);
989 sval = res_get (common_resource, ext_res);
992 rGroup->flagStoreData = atoi (sval);
994 if (rGroup->flagStoreData == -1)
995 rGroup->flagStoreData = 0;
997 if (rGroup->flagStoreKeys == -1)
1001 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
1002 if (!(sval = res_get (common_resource, ext_res)))
1004 sprintf (ext_res, "%sstoreKeys", gprefix);
1005 sval = res_get (common_resource, ext_res);
1008 rGroup->flagStoreKeys = atoi (sval);
1010 if (rGroup->flagStoreKeys == -1)
1011 rGroup->flagStoreKeys = 0;
1013 if (sysno && deleteFlag)
1017 if ((fd = open (fname, O_RDONLY)) == -1)
1019 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
1023 file_read_start (fd);
1024 recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType);
1025 file_read_stop (fd);