2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.21 1995-10-10 12:24:38 adam
8 * Temporary sort files are compressed.
10 * Revision 1.20 1995/10/06 13:52:05 adam
11 * Bug fixes. Handler may abort further scanning.
13 * Revision 1.19 1995/10/04 12:55:16 adam
14 * Bug fix in ranked search. Use=Any keys inserted.
16 * Revision 1.18 1995/10/04 09:37:08 quinn
19 * Revision 1.17 1995/10/03 14:28:57 adam
20 * Buffered read in extract works.
22 * Revision 1.16 1995/10/03 14:28:45 adam
23 * Work on more effecient read handler in extract.
25 * Revision 1.15 1995/10/02 15:42:53 adam
26 * Extract uses file descriptors instead of FILE pointers.
28 * Revision 1.14 1995/10/02 15:29:13 adam
29 * More logging in file_extract.
31 * Revision 1.13 1995/09/29 14:01:39 adam
34 * Revision 1.12 1995/09/28 14:22:56 adam
35 * Sort uses smaller temporary files.
37 * Revision 1.11 1995/09/28 12:10:31 adam
38 * Bug fixes. Field prefix used in queries.
40 * Revision 1.10 1995/09/28 09:19:41 adam
41 * xfree/xmalloc used everywhere.
42 * Extract/retrieve method seems to work for text records.
44 * Revision 1.9 1995/09/27 12:22:28 adam
45 * More work on extract in record control.
46 * Field name is not in isam keys but in prefix in dictionary words.
48 * Revision 1.8 1995/09/14 07:48:22 adam
49 * Record control management.
51 * Revision 1.7 1995/09/11 13:09:32 adam
52 * More work on relevance feedback.
54 * Revision 1.6 1995/09/08 14:52:27 adam
55 * Minor changes. Dictionary is lower case now.
57 * Revision 1.5 1995/09/06 16:11:16 adam
58 * Option: only one word key per file.
60 * Revision 1.4 1995/09/05 15:28:39 adam
61 * More work on search engine.
63 * Revision 1.3 1995/09/04 12:33:41 adam
64 * Various cleanup. YAZ util used instead.
66 * Revision 1.2 1995/09/04 09:10:34 adam
67 * More work on index add/del/update.
68 * Merge sort implemented.
69 * Initial work on z39 server.
71 * Revision 1.1 1995/09/01 14:06:35 adam
72 * Split of work into more files.
86 static SYSNO sysno_next;
87 static int sys_idx_fd = -1;
91 static char **key_buf;
92 static size_t ptr_top;
95 static int key_file_no;
97 void key_open (int mem)
103 key_buf = xmalloc (mem);
104 ptr_top = mem/sizeof(char*);
109 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
111 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
114 file_key = dict_lookup (file_idx, ".");
116 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
119 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
121 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
132 void encode_key_init (struct encode_info *i)
138 char *encode_key_int (int d, char *bp)
147 else if (d <= 4194303)
149 *bp++ = 128 + (d>>16);
150 *bp++ = (d>>8) & 255;
155 *bp++ = 192 + (d>>24);
156 *bp++ = (d>>16) & 255;
157 *bp++ = (d>>8) & 255;
163 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
168 while ((*bp++ = *k++))
170 memcpy (&key, k+1, sizeof(struct it_key));
171 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
172 if (i->sysno != key.sysno)
174 i->sysno = key.sysno;
177 bp = encode_key_int (key.seqno - i->seqno, bp);
178 i->seqno = key.seqno;
179 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
181 logf (LOG_FATAL|LOG_ERRNO, "fwrite");
186 void key_flush (void)
191 struct encode_info encode_info;
197 logf (LOG_LOG, "sorting section %d", key_file_no);
198 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
199 sprintf (out_fname, TEMP_FNAME, key_file_no);
201 if (!(outf = fopen (out_fname, "w")))
203 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
206 logf (LOG_LOG, "writing section %d", key_file_no);
207 prevcp = cp = key_buf[ptr_top-ptr_i];
209 encode_key_init (&encode_info);
210 encode_key_write (cp, &encode_info, outf);
213 cp = key_buf[ptr_top-ptr_i];
214 if (strcmp (cp, prevcp))
216 encode_key_init (&encode_info);
217 encode_key_write (cp, &encode_info, outf);
221 encode_key_write (cp + strlen(cp), &encode_info, outf);
225 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
228 logf (LOG_LOG, "finished section %d", key_file_no);
238 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
239 dict_close (file_idx);
243 static void wordInit (RecWord *p)
247 p->which = Word_String;
250 static void wordAdd (const RecWord *p)
255 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
258 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
259 kused += index_word_prefix ((char*)key_buf + kused,
260 p->attrSet, p->attrUse);
264 for (i = 0; p->u.string[i]; i++)
265 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
266 ((char*)key_buf) [kused++] = '\0';
271 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
272 key.sysno = key_sysno;
273 key.seqno = p->seqno;
274 memcpy ((char*)key_buf + kused, &key, sizeof(key));
275 kused += sizeof(key);
278 static void wordAddAny (const RecWord *p)
280 if (p->attrSet != 1 || p->attrUse != 1016)
284 memcpy (&w, p, sizeof(w));
293 #define FILE_READ_BUF 1
295 static char *file_buf;
296 static int file_offset;
297 static int file_bufsize;
299 static void file_read_start (int fd)
302 file_buf = xmalloc (4096);
303 file_bufsize = read (fd, file_buf, 4096);
306 static void file_read_stop (int fd)
311 static int file_read (int fd, char *buf, size_t count)
313 int l = file_bufsize - file_offset;
319 memcpy (buf, file_buf + file_offset, l);
321 if (count > file_bufsize)
323 if ((r = read (fd, buf + l, count)) == -1)
325 logf (LOG_FATAL|LOG_ERRNO, "read");
332 file_bufsize = r = read (fd, file_buf, 4096);
335 logf (LOG_FATAL|LOG_ERRNO, "read");
341 memcpy (buf + l, file_buf, r);
347 memcpy (buf + l, file_buf, count - l);
351 memcpy (buf, file_buf + file_offset, count);
352 file_offset += count;
356 static int file_read (int fd, char *buf, size_t count)
358 return read (fd, buf, count);
361 void file_extract (int cmd, const char *fname, const char *kname)
367 const char *file_type;
369 struct recExtractCtrl extractCtrl;
372 for (i = strlen(fname); --i >= 0; )
378 else if (fname[i] == '.')
380 strcpy (ext, fname+i+1);
383 sprintf (ext_res, "fileExtension.%s", ext);
384 if (!(file_type = res_get (common_resource, ext_res)))
386 if (!(rt = recType_byName (file_type)))
388 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
389 file_info = dict_lookup (file_idx, kname);
392 sysno = sysno_next++;
393 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
394 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
395 write (sys_idx_fd, file_type, strlen (file_type)+1);
396 write (sys_idx_fd, kname, strlen(kname)+1);
399 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
401 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
403 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
406 extractCtrl.subType = "";
407 extractCtrl.init = wordInit;
408 extractCtrl.add = wordAddAny;
410 file_read_start (extractCtrl.fd);
412 extractCtrl.readf = file_read;
415 r = (*rt->extract)(&extractCtrl);
417 file_read_stop (extractCtrl.fd);
419 close (extractCtrl.fd);
421 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);