2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.12 1995-09-28 14:22:56 adam
8 * Sort uses smaller temporary files.
10 * Revision 1.11 1995/09/28 12:10:31 adam
11 * Bug fixes. Field prefix used in queries.
13 * Revision 1.10 1995/09/28 09:19:41 adam
14 * xfree/xmalloc used everywhere.
15 * Extract/retrieve method seems to work for text records.
17 * Revision 1.9 1995/09/27 12:22:28 adam
18 * More work on extract in record control.
19 * Field name is not in isam keys but in prefix in dictionary words.
21 * Revision 1.8 1995/09/14 07:48:22 adam
22 * Record control management.
24 * Revision 1.7 1995/09/11 13:09:32 adam
25 * More work on relevance feedback.
27 * Revision 1.6 1995/09/08 14:52:27 adam
28 * Minor changes. Dictionary is lower case now.
30 * Revision 1.5 1995/09/06 16:11:16 adam
31 * Option: only one word key per file.
33 * Revision 1.4 1995/09/05 15:28:39 adam
34 * More work on search engine.
36 * Revision 1.3 1995/09/04 12:33:41 adam
37 * Various cleanup. YAZ util used instead.
39 * Revision 1.2 1995/09/04 09:10:34 adam
40 * More work on index add/del/update.
41 * Merge sort implemented.
42 * Initial work on z39 server.
44 * Revision 1.1 1995/09/01 14:06:35 adam
45 * Split of work into more files.
59 static SYSNO sysno_next;
60 static int sys_idx_fd = -1;
64 static char **key_buf;
65 static size_t ptr_top;
68 static int key_file_no;
70 static int sort_compare (const void *p1, const void *p2)
74 char *cp1 = *(char**) p1;
75 char *cp2 = *(char**) p2;
77 if ((r = strcmp (cp1, cp2)))
80 if ((r = key_compare (cp1+l, cp2+l)))
82 return cp1[l+sizeof(struct it_key)] -
83 cp2[l+sizeof(struct it_key)];
86 void key_open (int mem)
92 key_buf = xmalloc (mem);
93 ptr_top = mem/sizeof(char*);
98 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
100 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
103 file_key = dict_lookup (file_idx, ".");
105 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
108 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
110 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
115 void key_flush (void)
125 logf (LOG_LOG, "sorting section %d", key_file_no);
126 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), sort_compare);
127 sprintf (out_fname, TEMP_FNAME, key_file_no);
130 if (!(outf = fopen (out_fname, "w")))
132 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
135 logf (LOG_LOG, "writing section %d", key_file_no);
136 prevcp = cp = key_buf[ptr_top-ptr_i];
138 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
140 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
145 cp = key_buf[ptr_top-ptr_i];
146 if (strcmp (cp, prevcp))
148 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
151 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
158 cp = strlen (cp) + cp;
159 if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
161 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
168 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
171 logf (LOG_LOG, "finished section %d", key_file_no);
181 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
182 dict_close (file_idx);
186 static void wordInit (RecWord *p)
190 p->which = Word_String;
193 static void wordAdd (const RecWord *p)
198 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
201 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
202 kused += index_word_prefix ((char*)key_buf + kused,
203 p->attrSet, p->attrUse);
207 for (i = 0; p->u.string[i]; i++)
208 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
209 ((char*)key_buf) [kused++] = '\0';
214 key.sysno = key_sysno;
215 key.seqno = p->seqno;
216 memcpy ((char*)key_buf + kused, &key, sizeof(key));
217 kused += sizeof(key);
219 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
222 void file_extract (int cmd, const char *fname, const char *kname)
228 const char *file_type;
231 struct recExtractCtrl extractCtrl;
234 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
235 for (i = strlen(fname); --i >= 0; )
241 else if (fname[i] == '.')
243 strcpy (ext, fname+i+1);
246 sprintf (ext_res, "fileExtension.%s", ext);
247 if (!(file_type = res_get (common_resource, ext_res)))
249 if (!(rt = recType_byName (file_type)))
251 file_info = dict_lookup (file_idx, kname);
254 sysno = sysno_next++;
255 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
256 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
257 write (sys_idx_fd, file_type, strlen (file_type)+1);
258 write (sys_idx_fd, kname, strlen(kname)+1);
261 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
263 if (!(inf = fopen (fname, "r")))
265 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
268 extractCtrl.inf = inf;
269 extractCtrl.subType = "";
270 extractCtrl.init = wordInit;
271 extractCtrl.add = wordAdd;
274 (*rt->extract)(&extractCtrl);