2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.23 1995-10-27 14:00:10 adam
8 * Implemented detection of database availability.
10 * Revision 1.22 1995/10/17 18:02:07 adam
11 * New feature: databases. Implemented as prefix to words in dictionary.
13 * Revision 1.21 1995/10/10 12:24:38 adam
14 * Temporary sort files are compressed.
16 * Revision 1.20 1995/10/06 13:52:05 adam
17 * Bug fixes. Handler may abort further scanning.
19 * Revision 1.19 1995/10/04 12:55:16 adam
20 * Bug fix in ranked search. Use=Any keys inserted.
22 * Revision 1.18 1995/10/04 09:37:08 quinn
25 * Revision 1.17 1995/10/03 14:28:57 adam
26 * Buffered read in extract works.
28 * Revision 1.16 1995/10/03 14:28:45 adam
29 * Work on more effecient read handler in extract.
31 * Revision 1.15 1995/10/02 15:42:53 adam
32 * Extract uses file descriptors instead of FILE pointers.
34 * Revision 1.14 1995/10/02 15:29:13 adam
35 * More logging in file_extract.
37 * Revision 1.13 1995/09/29 14:01:39 adam
40 * Revision 1.12 1995/09/28 14:22:56 adam
41 * Sort uses smaller temporary files.
43 * Revision 1.11 1995/09/28 12:10:31 adam
44 * Bug fixes. Field prefix used in queries.
46 * Revision 1.10 1995/09/28 09:19:41 adam
47 * xfree/xmalloc used everywhere.
48 * Extract/retrieve method seems to work for text records.
50 * Revision 1.9 1995/09/27 12:22:28 adam
51 * More work on extract in record control.
52 * Field name is not in isam keys but in prefix in dictionary words.
54 * Revision 1.8 1995/09/14 07:48:22 adam
55 * Record control management.
57 * Revision 1.7 1995/09/11 13:09:32 adam
58 * More work on relevance feedback.
60 * Revision 1.6 1995/09/08 14:52:27 adam
61 * Minor changes. Dictionary is lower case now.
63 * Revision 1.5 1995/09/06 16:11:16 adam
64 * Option: only one word key per file.
66 * Revision 1.4 1995/09/05 15:28:39 adam
67 * More work on search engine.
69 * Revision 1.3 1995/09/04 12:33:41 adam
70 * Various cleanup. YAZ util used instead.
72 * Revision 1.2 1995/09/04 09:10:34 adam
73 * More work on index add/del/update.
74 * Merge sort implemented.
75 * Initial work on z39 server.
77 * Revision 1.1 1995/09/01 14:06:35 adam
78 * Split of work into more files.
92 static SYSNO sysno_next;
93 static int sys_idx_fd = -1;
97 static char *key_databaseName;
98 static char **key_buf;
99 static size_t ptr_top;
102 static int key_file_no;
104 void key_open (int mem)
110 key_buf = xmalloc (mem);
111 ptr_top = mem/sizeof(char*);
116 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
118 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
121 file_key = dict_lookup (file_idx, ".");
123 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
126 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
128 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
139 void encode_key_init (struct encode_info *i)
145 char *encode_key_int (int d, char *bp)
154 else if (d <= 4194303)
156 *bp++ = 128 + (d>>16);
157 *bp++ = (d>>8) & 255;
162 *bp++ = 192 + (d>>24);
163 *bp++ = (d>>16) & 255;
164 *bp++ = (d>>8) & 255;
170 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
175 while ((*bp++ = *k++))
177 memcpy (&key, k+1, sizeof(struct it_key));
178 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
179 if (i->sysno != key.sysno)
181 i->sysno = key.sysno;
184 bp = encode_key_int (key.seqno - i->seqno, bp);
185 i->seqno = key.seqno;
186 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
188 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
193 void key_flush (void)
198 struct encode_info encode_info;
204 logf (LOG_LOG, "sorting section %d", key_file_no);
205 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
206 sprintf (out_fname, TEMP_FNAME, key_file_no);
208 if (!(outf = fopen (out_fname, "w")))
210 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
213 logf (LOG_LOG, "writing section %d", key_file_no);
214 prevcp = cp = key_buf[ptr_top-ptr_i];
216 encode_key_init (&encode_info);
217 encode_key_write (cp, &encode_info, outf);
220 cp = key_buf[ptr_top-ptr_i];
221 if (strcmp (cp, prevcp))
223 encode_key_init (&encode_info);
224 encode_key_write (cp, &encode_info, outf);
228 encode_key_write (cp + strlen(cp), &encode_info, outf);
232 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
235 logf (LOG_LOG, "finished section %d", key_file_no);
245 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
246 dict_close (file_idx);
250 static void wordInit (RecWord *p)
254 p->which = Word_String;
257 static void wordAdd (const RecWord *p)
262 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
265 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
266 kused += index_word_prefix ((char*)key_buf + kused,
267 p->attrSet, p->attrUse,
272 for (i = 0; p->u.string[i]; i++)
273 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
274 ((char*)key_buf) [kused++] = '\0';
279 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
280 key.sysno = key_sysno;
281 key.seqno = p->seqno;
282 memcpy ((char*)key_buf + kused, &key, sizeof(key));
283 kused += sizeof(key);
286 static void wordAddAny (const RecWord *p)
288 if (p->attrSet != 1 || p->attrUse != 1016)
292 memcpy (&w, p, sizeof(w));
301 #define FILE_READ_BUF 1
303 static char *file_buf;
304 static int file_offset;
305 static int file_bufsize;
307 static void file_read_start (int fd)
310 file_buf = xmalloc (4096);
311 file_bufsize = read (fd, file_buf, 4096);
314 static void file_read_stop (int fd)
319 static int file_read (int fd, char *buf, size_t count)
321 int l = file_bufsize - file_offset;
327 memcpy (buf, file_buf + file_offset, l);
329 if (count > file_bufsize)
331 if ((r = read (fd, buf + l, count)) == -1)
333 logf (LOG_FATAL|LOG_ERRNO, "read");
340 file_bufsize = r = read (fd, file_buf, 4096);
343 logf (LOG_FATAL|LOG_ERRNO, "read");
349 memcpy (buf + l, file_buf, r);
355 memcpy (buf + l, file_buf, count - l);
359 memcpy (buf, file_buf + file_offset, count);
360 file_offset += count;
364 static int file_read (int fd, char *buf, size_t count)
366 return read (fd, buf, count);
369 void file_extract (int cmd, const char *fname, const char *kname,
376 const char *file_type;
378 struct recExtractCtrl extractCtrl;
381 key_databaseName = databaseName;
382 for (i = strlen(fname); --i >= 0; )
388 else if (fname[i] == '.')
390 strcpy (ext, fname+i+1);
393 sprintf (ext_res, "fileExtension.%s", ext);
394 if (!(file_type = res_get (common_resource, ext_res)))
396 if (!(rt = recType_byName (file_type)))
398 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
399 file_info = dict_lookup (file_idx, kname);
402 sysno = sysno_next++;
403 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
404 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
405 write (sys_idx_fd, file_type, strlen (file_type)+1);
406 write (sys_idx_fd, kname, strlen(kname)+1);
409 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
411 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
413 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
416 extractCtrl.subType = "";
417 extractCtrl.init = wordInit;
418 extractCtrl.add = wordAddAny;
420 file_read_start (extractCtrl.fd);
422 extractCtrl.readf = file_read;
425 r = (*rt->extract)(&extractCtrl);
427 file_read_stop (extractCtrl.fd);
429 close (extractCtrl.fd);
431 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);