2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.13 1995-09-29 14:01:39 adam
10 * Revision 1.12 1995/09/28 14:22:56 adam
11 * Sort uses smaller temporary files.
13 * Revision 1.11 1995/09/28 12:10:31 adam
14 * Bug fixes. Field prefix used in queries.
16 * Revision 1.10 1995/09/28 09:19:41 adam
17 * xfree/xmalloc used everywhere.
18 * Extract/retrieve method seems to work for text records.
20 * Revision 1.9 1995/09/27 12:22:28 adam
21 * More work on extract in record control.
22 * Field name is not in isam keys but in prefix in dictionary words.
24 * Revision 1.8 1995/09/14 07:48:22 adam
25 * Record control management.
27 * Revision 1.7 1995/09/11 13:09:32 adam
28 * More work on relevance feedback.
30 * Revision 1.6 1995/09/08 14:52:27 adam
31 * Minor changes. Dictionary is lower case now.
33 * Revision 1.5 1995/09/06 16:11:16 adam
34 * Option: only one word key per file.
36 * Revision 1.4 1995/09/05 15:28:39 adam
37 * More work on search engine.
39 * Revision 1.3 1995/09/04 12:33:41 adam
40 * Various cleanup. YAZ util used instead.
42 * Revision 1.2 1995/09/04 09:10:34 adam
43 * More work on index add/del/update.
44 * Merge sort implemented.
45 * Initial work on z39 server.
47 * Revision 1.1 1995/09/01 14:06:35 adam
48 * Split of work into more files.
62 static SYSNO sysno_next;
63 static int sys_idx_fd = -1;
67 static char **key_buf;
68 static size_t ptr_top;
71 static int key_file_no;
73 void key_open (int mem)
79 key_buf = xmalloc (mem);
80 ptr_top = mem/sizeof(char*);
85 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
87 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
90 file_key = dict_lookup (file_idx, ".");
92 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
95 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
97 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
102 void key_flush (void)
112 logf (LOG_LOG, "sorting section %d", key_file_no);
113 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
114 sprintf (out_fname, TEMP_FNAME, key_file_no);
116 if (!(outf = fopen (out_fname, "w")))
118 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
121 logf (LOG_LOG, "writing section %d", key_file_no);
122 prevcp = cp = key_buf[ptr_top-ptr_i];
124 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
126 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
131 cp = key_buf[ptr_top-ptr_i];
132 if (strcmp (cp, prevcp))
134 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
137 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
144 cp = strlen (cp) + cp;
145 if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
147 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
154 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
157 logf (LOG_LOG, "finished section %d", key_file_no);
167 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
168 dict_close (file_idx);
172 static void wordInit (RecWord *p)
176 p->which = Word_String;
179 static void wordAdd (const RecWord *p)
184 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
187 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
188 kused += index_word_prefix ((char*)key_buf + kused,
189 p->attrSet, p->attrUse);
193 for (i = 0; p->u.string[i]; i++)
194 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
195 ((char*)key_buf) [kused++] = '\0';
200 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
201 key.sysno = key_sysno;
202 key.seqno = p->seqno;
203 memcpy ((char*)key_buf + kused, &key, sizeof(key));
204 kused += sizeof(key);
207 void file_extract (int cmd, const char *fname, const char *kname)
213 const char *file_type;
216 struct recExtractCtrl extractCtrl;
219 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
220 for (i = strlen(fname); --i >= 0; )
226 else if (fname[i] == '.')
228 strcpy (ext, fname+i+1);
231 sprintf (ext_res, "fileExtension.%s", ext);
232 if (!(file_type = res_get (common_resource, ext_res)))
234 if (!(rt = recType_byName (file_type)))
236 file_info = dict_lookup (file_idx, kname);
239 sysno = sysno_next++;
240 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
241 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
242 write (sys_idx_fd, file_type, strlen (file_type)+1);
243 write (sys_idx_fd, kname, strlen(kname)+1);
246 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
248 if (!(inf = fopen (fname, "r")))
250 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
253 extractCtrl.inf = inf;
254 extractCtrl.subType = "";
255 extractCtrl.init = wordInit;
256 extractCtrl.add = wordAdd;
259 (*rt->extract)(&extractCtrl);