2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.24 1995-11-15 19:13:08 adam
8 * Work on record management.
10 * Revision 1.23 1995/10/27 14:00:10 adam
11 * Implemented detection of database availability.
13 * Revision 1.22 1995/10/17 18:02:07 adam
14 * New feature: databases. Implemented as prefix to words in dictionary.
16 * Revision 1.21 1995/10/10 12:24:38 adam
17 * Temporary sort files are compressed.
19 * Revision 1.20 1995/10/06 13:52:05 adam
20 * Bug fixes. Handler may abort further scanning.
22 * Revision 1.19 1995/10/04 12:55:16 adam
23 * Bug fix in ranked search. Use=Any keys inserted.
25 * Revision 1.18 1995/10/04 09:37:08 quinn
28 * Revision 1.17 1995/10/03 14:28:57 adam
29 * Buffered read in extract works.
31 * Revision 1.16 1995/10/03 14:28:45 adam
32 * Work on more effecient read handler in extract.
34 * Revision 1.15 1995/10/02 15:42:53 adam
35 * Extract uses file descriptors instead of FILE pointers.
37 * Revision 1.14 1995/10/02 15:29:13 adam
38 * More logging in file_extract.
40 * Revision 1.13 1995/09/29 14:01:39 adam
43 * Revision 1.12 1995/09/28 14:22:56 adam
44 * Sort uses smaller temporary files.
46 * Revision 1.11 1995/09/28 12:10:31 adam
47 * Bug fixes. Field prefix used in queries.
49 * Revision 1.10 1995/09/28 09:19:41 adam
50 * xfree/xmalloc used everywhere.
51 * Extract/retrieve method seems to work for text records.
53 * Revision 1.9 1995/09/27 12:22:28 adam
54 * More work on extract in record control.
55 * Field name is not in isam keys but in prefix in dictionary words.
57 * Revision 1.8 1995/09/14 07:48:22 adam
58 * Record control management.
60 * Revision 1.7 1995/09/11 13:09:32 adam
61 * More work on relevance feedback.
63 * Revision 1.6 1995/09/08 14:52:27 adam
64 * Minor changes. Dictionary is lower case now.
66 * Revision 1.5 1995/09/06 16:11:16 adam
67 * Option: only one word key per file.
69 * Revision 1.4 1995/09/05 15:28:39 adam
70 * More work on search engine.
72 * Revision 1.3 1995/09/04 12:33:41 adam
73 * Various cleanup. YAZ util used instead.
75 * Revision 1.2 1995/09/04 09:10:34 adam
76 * More work on index add/del/update.
77 * Merge sort implemented.
78 * Initial work on z39 server.
80 * Revision 1.1 1995/09/01 14:06:35 adam
81 * Split of work into more files.
100 static Dict file_idx;
104 static Records records = NULL;
106 static int sys_idx_fd = -1;
107 static SYSNO sysno_next;
111 static int key_sysno;
112 static char *key_databaseName;
113 static char **key_buf;
114 static size_t ptr_top;
117 static int key_file_no;
119 void key_open (int mem)
126 key_buf = xmalloc (mem);
127 ptr_top = mem/sizeof(char*);
132 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
134 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
139 records = rec_open (1);
141 file_key = dict_lookup (file_idx, ".");
143 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
146 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
148 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
160 void encode_key_init (struct encode_info *i)
166 char *encode_key_int (int d, char *bp)
175 else if (d <= 4194303)
177 *bp++ = 128 + (d>>16);
178 *bp++ = (d>>8) & 255;
183 *bp++ = 192 + (d>>24);
184 *bp++ = (d>>16) & 255;
185 *bp++ = (d>>8) & 255;
191 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
196 while ((*bp++ = *k++))
198 memcpy (&key, k+1, sizeof(struct it_key));
199 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
200 if (i->sysno != key.sysno)
202 i->sysno = key.sysno;
205 bp = encode_key_int (key.seqno - i->seqno, bp);
206 i->seqno = key.seqno;
207 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
209 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
214 void key_flush (void)
219 struct encode_info encode_info;
225 logf (LOG_LOG, "sorting section %d", key_file_no);
226 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
227 sprintf (out_fname, TEMP_FNAME, key_file_no);
229 if (!(outf = fopen (out_fname, "w")))
231 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
234 logf (LOG_LOG, "writing section %d", key_file_no);
235 prevcp = cp = key_buf[ptr_top-ptr_i];
237 encode_key_init (&encode_info);
238 encode_key_write (cp, &encode_info, outf);
241 cp = key_buf[ptr_top-ptr_i];
242 if (strcmp (cp, prevcp))
244 encode_key_init (&encode_info);
245 encode_key_write (cp, &encode_info, outf);
249 encode_key_write (cp + strlen(cp), &encode_info, outf);
253 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
256 logf (LOG_LOG, "finished section %d", key_file_no);
266 rec_close (&records);
269 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
271 dict_close (file_idx);
275 static void wordInit (RecWord *p)
279 p->which = Word_String;
282 static void wordAdd (const RecWord *p)
287 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
290 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
291 kused += index_word_prefix ((char*)key_buf + kused,
292 p->attrSet, p->attrUse,
297 for (i = 0; p->u.string[i]; i++)
298 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
299 ((char*)key_buf) [kused++] = '\0';
304 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
305 key.sysno = key_sysno;
306 key.seqno = p->seqno;
307 memcpy ((char*)key_buf + kused, &key, sizeof(key));
308 kused += sizeof(key);
311 static void wordAddAny (const RecWord *p)
313 if (p->attrSet != 1 || p->attrUse != 1016)
317 memcpy (&w, p, sizeof(w));
326 #define FILE_READ_BUF 1
328 static char *file_buf;
329 static int file_offset;
330 static int file_bufsize;
332 static void file_read_start (int fd)
335 file_buf = xmalloc (4096);
336 file_bufsize = read (fd, file_buf, 4096);
339 static void file_read_stop (int fd)
344 static int file_read (int fd, char *buf, size_t count)
346 int l = file_bufsize - file_offset;
352 memcpy (buf, file_buf + file_offset, l);
354 if (count > file_bufsize)
356 if ((r = read (fd, buf + l, count)) == -1)
358 logf (LOG_FATAL|LOG_ERRNO, "read");
365 file_bufsize = r = read (fd, file_buf, 4096);
368 logf (LOG_FATAL|LOG_ERRNO, "read");
374 memcpy (buf + l, file_buf, r);
380 memcpy (buf + l, file_buf, count - l);
384 memcpy (buf, file_buf + file_offset, count);
385 file_offset += count;
389 static int file_read (int fd, char *buf, size_t count)
391 return read (fd, buf, count);
394 void file_extract (int cmd, const char *fname, const char *kname,
401 const char *file_type;
403 struct recExtractCtrl extractCtrl;
406 key_databaseName = databaseName;
407 for (i = strlen(fname); --i >= 0; )
413 else if (fname[i] == '.')
415 strcpy (ext, fname+i+1);
418 sprintf (ext_res, "fileExtension.%s", ext);
419 if (!(file_type = res_get (common_resource, ext_res)))
421 if (!(rt = recType_byName (file_type)))
423 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
424 file_info = dict_lookup (file_idx, kname);
428 Record rec = rec_new (records);
431 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
432 rec->info[0] = rec_strdup (file_type);
433 rec->info[1] = rec_strdup (kname);
434 rec_put (records, rec);
436 sysno = sysno_next++;
437 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
438 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
439 write (sys_idx_fd, file_type, strlen (file_type)+1);
440 write (sys_idx_fd, kname, strlen(kname)+1);
444 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
446 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
448 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
451 extractCtrl.subType = "";
452 extractCtrl.init = wordInit;
453 extractCtrl.add = wordAddAny;
455 file_read_start (extractCtrl.fd);
457 extractCtrl.readf = file_read;
460 r = (*rt->extract)(&extractCtrl);
462 file_read_stop (extractCtrl.fd);
464 close (extractCtrl.fd);
466 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);