2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.16 1995-10-03 14:28:45 adam
8 * Work on more effecient read handler in extract.
10 * Revision 1.15 1995/10/02 15:42:53 adam
11 * Extract uses file descriptors instead of FILE pointers.
13 * Revision 1.14 1995/10/02 15:29:13 adam
14 * More logging in file_extract.
16 * Revision 1.13 1995/09/29 14:01:39 adam
19 * Revision 1.12 1995/09/28 14:22:56 adam
20 * Sort uses smaller temporary files.
22 * Revision 1.11 1995/09/28 12:10:31 adam
23 * Bug fixes. Field prefix used in queries.
25 * Revision 1.10 1995/09/28 09:19:41 adam
26 * xfree/xmalloc used everywhere.
27 * Extract/retrieve method seems to work for text records.
29 * Revision 1.9 1995/09/27 12:22:28 adam
30 * More work on extract in record control.
31 * Field name is not in isam keys but in prefix in dictionary words.
33 * Revision 1.8 1995/09/14 07:48:22 adam
34 * Record control management.
36 * Revision 1.7 1995/09/11 13:09:32 adam
37 * More work on relevance feedback.
39 * Revision 1.6 1995/09/08 14:52:27 adam
40 * Minor changes. Dictionary is lower case now.
42 * Revision 1.5 1995/09/06 16:11:16 adam
43 * Option: only one word key per file.
45 * Revision 1.4 1995/09/05 15:28:39 adam
46 * More work on search engine.
48 * Revision 1.3 1995/09/04 12:33:41 adam
49 * Various cleanup. YAZ util used instead.
51 * Revision 1.2 1995/09/04 09:10:34 adam
52 * More work on index add/del/update.
53 * Merge sort implemented.
54 * Initial work on z39 server.
56 * Revision 1.1 1995/09/01 14:06:35 adam
57 * Split of work into more files.
71 static SYSNO sysno_next;
72 static int sys_idx_fd = -1;
76 static char **key_buf;
77 static size_t ptr_top;
80 static int key_file_no;
82 void key_open (int mem)
88 key_buf = xmalloc (mem);
89 ptr_top = mem/sizeof(char*);
94 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
96 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
99 file_key = dict_lookup (file_idx, ".");
101 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
104 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
106 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
111 void key_flush (void)
121 logf (LOG_LOG, "sorting section %d", key_file_no);
122 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
123 sprintf (out_fname, TEMP_FNAME, key_file_no);
125 if (!(outf = fopen (out_fname, "w")))
127 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
130 logf (LOG_LOG, "writing section %d", key_file_no);
131 prevcp = cp = key_buf[ptr_top-ptr_i];
133 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
135 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
140 cp = key_buf[ptr_top-ptr_i];
141 if (strcmp (cp, prevcp))
143 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
146 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
153 cp = strlen (cp) + cp;
154 if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
156 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
163 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
166 logf (LOG_LOG, "finished section %d", key_file_no);
176 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
177 dict_close (file_idx);
181 static void wordInit (RecWord *p)
185 p->which = Word_String;
188 static void wordAdd (const RecWord *p)
193 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
196 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
197 kused += index_word_prefix ((char*)key_buf + kused,
198 p->attrSet, p->attrUse);
202 for (i = 0; p->u.string[i]; i++)
203 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
204 ((char*)key_buf) [kused++] = '\0';
209 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
210 key.sysno = key_sysno;
211 key.seqno = p->seqno;
212 memcpy ((char*)key_buf + kused, &key, sizeof(key));
213 kused += sizeof(key);
216 static char *file_buf;
217 static int file_offset;
218 static int file_bufsize;
220 static int file_read (int fd, char *buf, size_t count)
222 if (file_offset + count <= file_bufsize)
224 memcpy (buf, file_buf + file_offset, count);
225 file_offset += count;
228 else if (file_bufsize > 0)
230 int l = file_bufsize - file_offset;
232 memcpy (buf, file_buf + file_offset, l);
233 if (count - l > file_bufsize)
235 r = read (fd, buf, count);
237 file_bufsize = read (fd, file_buf, file_bufsize);
243 r = read (fd, file_buf, file_bufsize);
246 logf (LOG_FATAL|LOG_ERRNO, "read");
249 memcpy (buf + l, file_buf, count - l);
250 file_offset = count - l;
253 return read (fd, buf, count);
256 void file_extract (int cmd, const char *fname, const char *kname)
262 const char *file_type;
264 struct recExtractCtrl extractCtrl;
267 for (i = strlen(fname); --i >= 0; )
273 else if (fname[i] == '.')
275 strcpy (ext, fname+i+1);
278 sprintf (ext_res, "fileExtension.%s", ext);
279 if (!(file_type = res_get (common_resource, ext_res)))
281 if (!(rt = recType_byName (file_type)))
283 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
284 file_info = dict_lookup (file_idx, kname);
287 sysno = sysno_next++;
288 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
289 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
290 write (sys_idx_fd, file_type, strlen (file_type)+1);
291 write (sys_idx_fd, kname, strlen(kname)+1);
294 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
296 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
298 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
301 extractCtrl.subType = "";
302 extractCtrl.init = wordInit;
303 extractCtrl.add = wordAdd;
304 extractCtrl.readf = file_read;
307 r = (*rt->extract)(&extractCtrl);
308 close (extractCtrl.fd);
310 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);