2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.26 1995-11-20 11:56:24 adam
8 * Work on new traversal.
10 * Revision 1.25 1995/11/16 15:34:54 adam
11 * Uses new record management system in both indexer and server.
13 * Revision 1.24 1995/11/15 19:13:08 adam
14 * Work on record management.
16 * Revision 1.23 1995/10/27 14:00:10 adam
17 * Implemented detection of database availability.
19 * Revision 1.22 1995/10/17 18:02:07 adam
20 * New feature: databases. Implemented as prefix to words in dictionary.
22 * Revision 1.21 1995/10/10 12:24:38 adam
23 * Temporary sort files are compressed.
25 * Revision 1.20 1995/10/06 13:52:05 adam
26 * Bug fixes. Handler may abort further scanning.
28 * Revision 1.19 1995/10/04 12:55:16 adam
29 * Bug fix in ranked search. Use=Any keys inserted.
31 * Revision 1.18 1995/10/04 09:37:08 quinn
34 * Revision 1.17 1995/10/03 14:28:57 adam
35 * Buffered read in extract works.
37 * Revision 1.16 1995/10/03 14:28:45 adam
38 * Work on more effecient read handler in extract.
40 * Revision 1.15 1995/10/02 15:42:53 adam
41 * Extract uses file descriptors instead of FILE pointers.
43 * Revision 1.14 1995/10/02 15:29:13 adam
44 * More logging in file_extract.
46 * Revision 1.13 1995/09/29 14:01:39 adam
49 * Revision 1.12 1995/09/28 14:22:56 adam
50 * Sort uses smaller temporary files.
52 * Revision 1.11 1995/09/28 12:10:31 adam
53 * Bug fixes. Field prefix used in queries.
55 * Revision 1.10 1995/09/28 09:19:41 adam
56 * xfree/xmalloc used everywhere.
57 * Extract/retrieve method seems to work for text records.
59 * Revision 1.9 1995/09/27 12:22:28 adam
60 * More work on extract in record control.
61 * Field name is not in isam keys but in prefix in dictionary words.
63 * Revision 1.8 1995/09/14 07:48:22 adam
64 * Record control management.
66 * Revision 1.7 1995/09/11 13:09:32 adam
67 * More work on relevance feedback.
69 * Revision 1.6 1995/09/08 14:52:27 adam
70 * Minor changes. Dictionary is lower case now.
72 * Revision 1.5 1995/09/06 16:11:16 adam
73 * Option: only one word key per file.
75 * Revision 1.4 1995/09/05 15:28:39 adam
76 * More work on search engine.
78 * Revision 1.3 1995/09/04 12:33:41 adam
79 * Various cleanup. YAZ util used instead.
81 * Revision 1.2 1995/09/04 09:10:34 adam
82 * More work on index add/del/update.
83 * Merge sort implemented.
84 * Initial work on z39 server.
86 * Revision 1.1 1995/09/01 14:06:35 adam
87 * Split of work into more files.
101 #include "recindex.h"
104 static Dict file_idx;
108 static Records records = NULL;
110 static int sys_idx_fd = -1;
111 static SYSNO sysno_next;
115 static int key_sysno;
116 static char *key_databaseName;
117 static char **key_buf;
118 static size_t ptr_top;
121 static int key_file_no;
123 void key_open (int mem)
130 key_buf = xmalloc (mem);
131 ptr_top = mem/sizeof(char*);
136 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
138 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
143 records = rec_open (1);
145 file_key = dict_lookup (file_idx, ".");
147 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
150 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
152 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
164 void encode_key_init (struct encode_info *i)
170 char *encode_key_int (int d, char *bp)
179 else if (d <= 4194303)
181 *bp++ = 128 + (d>>16);
182 *bp++ = (d>>8) & 255;
187 *bp++ = 192 + (d>>24);
188 *bp++ = (d>>16) & 255;
189 *bp++ = (d>>8) & 255;
195 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
200 while ((*bp++ = *k++))
202 memcpy (&key, k+1, sizeof(struct it_key));
203 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
204 if (i->sysno != key.sysno)
206 i->sysno = key.sysno;
209 bp = encode_key_int (key.seqno - i->seqno, bp);
210 i->seqno = key.seqno;
211 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
213 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
218 void key_flush (void)
223 struct encode_info encode_info;
229 logf (LOG_LOG, "sorting section %d", key_file_no);
230 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
231 sprintf (out_fname, TEMP_FNAME, key_file_no);
233 if (!(outf = fopen (out_fname, "w")))
235 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
238 logf (LOG_LOG, "writing section %d", key_file_no);
239 prevcp = cp = key_buf[ptr_top-ptr_i];
241 encode_key_init (&encode_info);
242 encode_key_write (cp, &encode_info, outf);
245 cp = key_buf[ptr_top-ptr_i];
246 if (strcmp (cp, prevcp))
248 encode_key_init (&encode_info);
249 encode_key_write (cp, &encode_info, outf);
253 encode_key_write (cp + strlen(cp), &encode_info, outf);
257 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
260 logf (LOG_LOG, "finished section %d", key_file_no);
270 rec_close (&records);
273 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
275 dict_close (file_idx);
279 static void wordInit (RecWord *p)
283 p->which = Word_String;
286 static void wordAdd (const RecWord *p)
291 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
294 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
295 kused += index_word_prefix ((char*)key_buf + kused,
296 p->attrSet, p->attrUse,
301 for (i = 0; p->u.string[i]; i++)
302 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
303 ((char*)key_buf) [kused++] = '\0';
308 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
309 key.sysno = key_sysno;
310 key.seqno = p->seqno;
311 memcpy ((char*)key_buf + kused, &key, sizeof(key));
312 kused += sizeof(key);
315 static void wordAddAny (const RecWord *p)
317 if (p->attrSet != 1 || p->attrUse != 1016)
321 memcpy (&w, p, sizeof(w));
330 #define FILE_READ_BUF 1
332 static char *file_buf;
333 static int file_offset;
334 static int file_bufsize;
336 static void file_read_start (int fd)
339 file_buf = xmalloc (4096);
340 file_bufsize = read (fd, file_buf, 4096);
343 static void file_read_stop (int fd)
348 static int file_read (int fd, char *buf, size_t count)
350 int l = file_bufsize - file_offset;
356 memcpy (buf, file_buf + file_offset, l);
358 if (count > file_bufsize)
360 if ((r = read (fd, buf + l, count)) == -1)
362 logf (LOG_FATAL|LOG_ERRNO, "read");
369 file_bufsize = r = read (fd, file_buf, 4096);
372 logf (LOG_FATAL|LOG_ERRNO, "read");
378 memcpy (buf + l, file_buf, r);
384 memcpy (buf + l, file_buf, count - l);
388 memcpy (buf, file_buf + file_offset, count);
389 file_offset += count;
393 static int file_read (int fd, char *buf, size_t count)
395 return read (fd, buf, count);
398 SYSNO file_extract (int cmd, const char *fname, const char *kname,
405 const char *file_type;
407 struct recExtractCtrl extractCtrl;
410 key_databaseName = databaseName;
411 for (i = strlen(fname); --i >= 0; )
417 else if (fname[i] == '.')
419 strcpy (ext, fname+i+1);
422 sprintf (ext_res, "fileExtension.%s", ext);
423 if (!(file_type = res_get (common_resource, ext_res)))
425 if (!(rt = recType_byName (file_type)))
427 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
428 file_info = dict_lookup (file_idx, kname);
432 Record rec = rec_new (records);
435 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
436 rec->info[0] = rec_strdup (file_type);
437 rec->info[1] = rec_strdup (kname);
438 rec_put (records, rec);
441 sysno = sysno_next++;
442 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
443 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
444 write (sys_idx_fd, file_type, strlen (file_type)+1);
445 write (sys_idx_fd, kname, strlen(kname)+1);
449 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
451 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
453 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
456 extractCtrl.subType = "";
457 extractCtrl.init = wordInit;
458 extractCtrl.add = wordAddAny;
460 file_read_start (extractCtrl.fd);
462 extractCtrl.readf = file_read;
465 r = (*rt->extract)(&extractCtrl);
467 file_read_stop (extractCtrl.fd);
469 close (extractCtrl.fd);
471 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);