2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.27 1995-11-20 16:59:45 adam
8 * New update method: the 'old' keys are saved for each records.
10 * Revision 1.26 1995/11/20 11:56:24 adam
11 * Work on new traversal.
13 * Revision 1.25 1995/11/16 15:34:54 adam
14 * Uses new record management system in both indexer and server.
16 * Revision 1.24 1995/11/15 19:13:08 adam
17 * Work on record management.
19 * Revision 1.23 1995/10/27 14:00:10 adam
20 * Implemented detection of database availability.
22 * Revision 1.22 1995/10/17 18:02:07 adam
23 * New feature: databases. Implemented as prefix to words in dictionary.
25 * Revision 1.21 1995/10/10 12:24:38 adam
26 * Temporary sort files are compressed.
28 * Revision 1.20 1995/10/06 13:52:05 adam
29 * Bug fixes. Handler may abort further scanning.
31 * Revision 1.19 1995/10/04 12:55:16 adam
32 * Bug fix in ranked search. Use=Any keys inserted.
34 * Revision 1.18 1995/10/04 09:37:08 quinn
37 * Revision 1.17 1995/10/03 14:28:57 adam
38 * Buffered read in extract works.
40 * Revision 1.16 1995/10/03 14:28:45 adam
41 * Work on more effecient read handler in extract.
43 * Revision 1.15 1995/10/02 15:42:53 adam
44 * Extract uses file descriptors instead of FILE pointers.
46 * Revision 1.14 1995/10/02 15:29:13 adam
47 * More logging in file_extract.
49 * Revision 1.13 1995/09/29 14:01:39 adam
52 * Revision 1.12 1995/09/28 14:22:56 adam
53 * Sort uses smaller temporary files.
55 * Revision 1.11 1995/09/28 12:10:31 adam
56 * Bug fixes. Field prefix used in queries.
58 * Revision 1.10 1995/09/28 09:19:41 adam
59 * xfree/xmalloc used everywhere.
60 * Extract/retrieve method seems to work for text records.
62 * Revision 1.9 1995/09/27 12:22:28 adam
63 * More work on extract in record control.
64 * Field name is not in isam keys but in prefix in dictionary words.
66 * Revision 1.8 1995/09/14 07:48:22 adam
67 * Record control management.
69 * Revision 1.7 1995/09/11 13:09:32 adam
70 * More work on relevance feedback.
72 * Revision 1.6 1995/09/08 14:52:27 adam
73 * Minor changes. Dictionary is lower case now.
75 * Revision 1.5 1995/09/06 16:11:16 adam
76 * Option: only one word key per file.
78 * Revision 1.4 1995/09/05 15:28:39 adam
79 * More work on search engine.
81 * Revision 1.3 1995/09/04 12:33:41 adam
82 * Various cleanup. YAZ util used instead.
84 * Revision 1.2 1995/09/04 09:10:34 adam
85 * More work on index add/del/update.
86 * Merge sort implemented.
87 * Initial work on z39 server.
89 * Revision 1.1 1995/09/01 14:06:35 adam
90 * Split of work into more files.
104 #include "recindex.h"
107 static Dict file_idx;
111 static Records records = NULL;
113 static int sys_idx_fd = -1;
114 static SYSNO sysno_next;
118 static int key_sysno;
119 static const char *key_databaseName;
120 static char **key_buf;
121 static size_t ptr_top;
123 static size_t key_buf_used;
124 static int key_file_no;
126 static int key_del_max;
127 static int key_del_used;
128 static char *key_del_buf;
130 void key_open (int mem)
137 key_buf = xmalloc (mem);
138 ptr_top = mem/sizeof(char*);
147 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
149 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
154 records = rec_open (1);
156 file_key = dict_lookup (file_idx, ".");
158 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
161 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
163 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
175 void encode_key_init (struct encode_info *i)
181 char *encode_key_int (int d, char *bp)
190 else if (d <= 4194303)
192 *bp++ = 128 + (d>>16);
193 *bp++ = (d>>8) & 255;
198 *bp++ = 192 + (d>>24);
199 *bp++ = (d>>16) & 255;
200 *bp++ = (d>>8) & 255;
206 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
211 while ((*bp++ = *k++))
213 memcpy (&key, k+1, sizeof(struct it_key));
214 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
215 if (i->sysno != key.sysno)
217 i->sysno = key.sysno;
220 bp = encode_key_int (key.seqno - i->seqno, bp);
221 i->seqno = key.seqno;
222 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
224 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
229 void key_flush (void)
234 struct encode_info encode_info;
240 logf (LOG_LOG, "sorting section %d", key_file_no);
241 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
242 sprintf (out_fname, TEMP_FNAME, key_file_no);
244 if (!(outf = fopen (out_fname, "w")))
246 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
249 logf (LOG_LOG, "writing section %d", key_file_no);
250 prevcp = cp = key_buf[ptr_top-ptr_i];
252 encode_key_init (&encode_info);
253 encode_key_write (cp, &encode_info, outf);
256 cp = key_buf[ptr_top-ptr_i];
257 if (strcmp (cp, prevcp))
259 encode_key_init (&encode_info);
260 encode_key_write (cp, &encode_info, outf);
264 encode_key_write (cp + strlen(cp), &encode_info, outf);
268 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
271 logf (LOG_LOG, "finished section %d", key_file_no);
281 rec_close (&records);
284 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
286 dict_close (file_idx);
294 static void wordInit (RecWord *p)
298 p->which = Word_String;
301 static void wordAdd (const RecWord *p)
306 if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
309 key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
310 key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
311 p->attrSet, p->attrUse,
316 for (i = 0; p->u.string[i]; i++)
317 ((char*)key_buf) [key_buf_used++] =
318 index_char_cvt (p->u.string[i]);
319 ((char*)key_buf) [key_buf_used++] = '\0';
324 ((char*) key_buf)[key_buf_used++] = ((key_cmd == 'a') ? 1 : 0);
325 key.sysno = key_sysno;
326 key.seqno = p->seqno;
327 memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
328 key_buf_used += sizeof(key);
330 if (key_cmd == 'a' && key_del_used >= 0)
334 if (key_del_used + 1024 > key_del_max)
338 if (!(kbn = malloc (key_del_max += 64000)))
340 logf (LOG_FATAL, "malloc");
344 memcpy (kbn, key_del_buf, key_del_used);
351 for (i = 0; p->u.string[i]; i++)
352 ((char*)key_del_buf) [key_del_used++] = p->u.string[i];
353 ((char*)key_del_buf) [key_del_used++] = '\0';
358 attrSet = p->attrSet;
359 memcpy (key_del_buf + key_del_used, &attrSet, sizeof(attrSet));
360 key_del_used += sizeof(attrSet);
362 attrUse = p->attrUse;
363 memcpy (key_del_buf + key_del_used, &attrUse, sizeof(attrUse));
364 key_del_used += sizeof(attrUse);
366 memcpy (key_del_buf + key_del_used, &p->seqno, sizeof(p->seqno));
367 key_del_used += sizeof(p->seqno);
371 static void wordAddAny (const RecWord *p)
373 if (p->attrSet != 1 || p->attrUse != 1016)
377 memcpy (&w, p, sizeof(w));
385 static char *file_buf;
386 static int file_offset;
387 static int file_bufsize;
389 static void file_read_start (int fd)
392 file_buf = xmalloc (4096);
393 file_bufsize = read (fd, file_buf, 4096);
396 static void file_read_stop (int fd)
401 static int file_read (int fd, char *buf, size_t count)
403 int l = file_bufsize - file_offset;
409 memcpy (buf, file_buf + file_offset, l);
411 if (count > file_bufsize)
413 if ((r = read (fd, buf + l, count)) == -1)
415 logf (LOG_FATAL|LOG_ERRNO, "read");
422 file_bufsize = r = read (fd, file_buf, 4096);
425 logf (LOG_FATAL|LOG_ERRNO, "read");
431 memcpy (buf + l, file_buf, r);
437 memcpy (buf + l, file_buf, count - l);
441 memcpy (buf, file_buf + file_offset, count);
442 file_offset += count;
446 SYSNO file_extract (int cmd, const char *fname, const char *kname,
453 const char *file_type;
455 struct recExtractCtrl extractCtrl;
459 key_databaseName = databaseName;
460 for (i = strlen(fname); --i >= 0; )
466 else if (fname[i] == '.')
468 strcpy (ext, fname+i+1);
471 sprintf (ext_res, "fileExtension.%s", ext);
472 if (!(file_type = res_get (common_resource, ext_res)))
474 if (!(rt = recType_byName (file_type)))
476 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
477 file_info = dict_lookup (file_idx, kname);
481 Record rec = rec_new (records);
484 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
485 rec->info[0] = rec_strdup (file_type, &rec->size[0]);
486 rec->info[1] = rec_strdup (kname, &rec->size[1]);
487 rec_put (records, &rec);
489 sysno = sysno_next++;
490 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
491 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
492 write (sys_idx_fd, file_type, strlen (file_type)+1);
493 write (sys_idx_fd, kname, strlen(kname)+1);
497 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
499 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
501 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
504 extractCtrl.subType = "";
505 extractCtrl.init = wordInit;
506 extractCtrl.add = wordAddAny;
508 file_read_start (extractCtrl.fd);
510 extractCtrl.readf = file_read;
513 r = (*rt->extract)(&extractCtrl);
515 file_read_stop (extractCtrl.fd);
517 close (extractCtrl.fd);
519 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
523 int fileExtract (SYSNO *sysno, const char *fname, const char *databaseName,
529 const char *file_type;
530 struct recExtractCtrl extractCtrl;
534 logf (LOG_DEBUG, "fileExtractAdd %s", fname);
537 for (i = strlen(fname); --i >= 0; )
543 else if (fname[i] == '.')
545 strcpy (ext, fname+i+1);
548 sprintf (ext_res, "fileExtension.%s", ext);
549 if (!(file_type = res_get (common_resource, ext_res)))
551 if (!(rt = recType_byName (file_type)))
554 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
556 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
560 extractCtrl.subType = "";
561 extractCtrl.init = wordInit;
562 extractCtrl.add = wordAddAny;
566 logf (LOG_LOG, "add record %s", fname);
567 rec = rec_new (records);
569 rec->info[0] = rec_strdup (file_type, &rec->size[0]);
570 rec->info[1] = rec_strdup (fname, &rec->size[1]);
571 rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
579 logf (LOG_LOG, "delete record %s", fname);
581 logf (LOG_LOG, "update record %s", fname);
582 rec = rec_get (records, *sysno);
586 key_databaseName = rec->info[3];
588 for (off = 0; off < rec->size[2]; )
594 rw.which = Word_String;
595 rw.u.string = kb + off;
599 memcpy (&attrSet, kb + off, sizeof(attrSet));
600 off += sizeof(attrSet);
601 memcpy (&attrUse, kb + off, sizeof(attrUse));
602 off += sizeof(attrUse);
603 memcpy (&rw.seqno, kb + off, sizeof(rw.seqno));
604 off += sizeof(rw.seqno);
605 rw.attrSet = attrSet;
606 rw.attrUse = attrUse;
608 (*extractCtrl.add) (&rw);
610 assert (off == rec->size[2]);
616 rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
622 rec_del (records, *sysno);
628 key_databaseName = databaseName;
632 file_read_start (extractCtrl.fd);
633 extractCtrl.readf = file_read;
634 r = (*rt->extract)(&extractCtrl);
635 file_read_stop (extractCtrl.fd);
637 close (extractCtrl.fd);
641 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
644 if (key_del_used > 0)
646 rec->size[2] = key_del_used;
647 rec->info[2] = malloc (rec->size[2]);
648 memcpy (rec->info[2], key_del_buf, rec->size[2]);
650 rec_put (records, &rec);