2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.14 1995-10-02 15:29:13 adam
8 * More logging in file_extract.
10 * Revision 1.13 1995/09/29 14:01:39 adam
13 * Revision 1.12 1995/09/28 14:22:56 adam
14 * Sort uses smaller temporary files.
16 * Revision 1.11 1995/09/28 12:10:31 adam
17 * Bug fixes. Field prefix used in queries.
19 * Revision 1.10 1995/09/28 09:19:41 adam
20 * xfree/xmalloc used everywhere.
21 * Extract/retrieve method seems to work for text records.
23 * Revision 1.9 1995/09/27 12:22:28 adam
24 * More work on extract in record control.
25 * Field name is not in isam keys but in prefix in dictionary words.
27 * Revision 1.8 1995/09/14 07:48:22 adam
28 * Record control management.
30 * Revision 1.7 1995/09/11 13:09:32 adam
31 * More work on relevance feedback.
33 * Revision 1.6 1995/09/08 14:52:27 adam
34 * Minor changes. Dictionary is lower case now.
36 * Revision 1.5 1995/09/06 16:11:16 adam
37 * Option: only one word key per file.
39 * Revision 1.4 1995/09/05 15:28:39 adam
40 * More work on search engine.
42 * Revision 1.3 1995/09/04 12:33:41 adam
43 * Various cleanup. YAZ util used instead.
45 * Revision 1.2 1995/09/04 09:10:34 adam
46 * More work on index add/del/update.
47 * Merge sort implemented.
48 * Initial work on z39 server.
50 * Revision 1.1 1995/09/01 14:06:35 adam
51 * Split of work into more files.
65 static SYSNO sysno_next;
66 static int sys_idx_fd = -1;
70 static char **key_buf;
71 static size_t ptr_top;
74 static int key_file_no;
76 void key_open (int mem)
82 key_buf = xmalloc (mem);
83 ptr_top = mem/sizeof(char*);
88 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
90 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
93 file_key = dict_lookup (file_idx, ".");
95 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
98 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
100 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
105 void key_flush (void)
115 logf (LOG_LOG, "sorting section %d", key_file_no);
116 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
117 sprintf (out_fname, TEMP_FNAME, key_file_no);
119 if (!(outf = fopen (out_fname, "w")))
121 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
124 logf (LOG_LOG, "writing section %d", key_file_no);
125 prevcp = cp = key_buf[ptr_top-ptr_i];
127 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
129 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
134 cp = key_buf[ptr_top-ptr_i];
135 if (strcmp (cp, prevcp))
137 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
140 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
147 cp = strlen (cp) + cp;
148 if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
150 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
157 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
160 logf (LOG_LOG, "finished section %d", key_file_no);
170 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
171 dict_close (file_idx);
175 static void wordInit (RecWord *p)
179 p->which = Word_String;
182 static void wordAdd (const RecWord *p)
187 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
190 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
191 kused += index_word_prefix ((char*)key_buf + kused,
192 p->attrSet, p->attrUse);
196 for (i = 0; p->u.string[i]; i++)
197 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
198 ((char*)key_buf) [kused++] = '\0';
203 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
204 key.sysno = key_sysno;
205 key.seqno = p->seqno;
206 memcpy ((char*)key_buf + kused, &key, sizeof(key));
207 kused += sizeof(key);
210 void file_extract (int cmd, const char *fname, const char *kname)
216 const char *file_type;
219 struct recExtractCtrl extractCtrl;
222 for (i = strlen(fname); --i >= 0; )
228 else if (fname[i] == '.')
230 strcpy (ext, fname+i+1);
233 sprintf (ext_res, "fileExtension.%s", ext);
234 if (!(file_type = res_get (common_resource, ext_res)))
236 if (!(rt = recType_byName (file_type)))
238 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
239 file_info = dict_lookup (file_idx, kname);
242 sysno = sysno_next++;
243 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
244 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
245 write (sys_idx_fd, file_type, strlen (file_type)+1);
246 write (sys_idx_fd, kname, strlen(kname)+1);
249 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
251 if (!(inf = fopen (fname, "r")))
253 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
256 extractCtrl.inf = inf;
257 extractCtrl.subType = "";
258 extractCtrl.init = wordInit;
259 extractCtrl.add = wordAdd;
262 r = (*rt->extract)(&extractCtrl);
265 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);