2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.18 1995-10-04 09:37:08 quinn
10 * Revision 1.17 1995/10/03 14:28:57 adam
11 * Buffered read in extract works.
13 * Revision 1.16 1995/10/03 14:28:45 adam
14 * Work on more effecient read handler in extract.
16 * Revision 1.15 1995/10/02 15:42:53 adam
17 * Extract uses file descriptors instead of FILE pointers.
19 * Revision 1.14 1995/10/02 15:29:13 adam
20 * More logging in file_extract.
22 * Revision 1.13 1995/09/29 14:01:39 adam
25 * Revision 1.12 1995/09/28 14:22:56 adam
26 * Sort uses smaller temporary files.
28 * Revision 1.11 1995/09/28 12:10:31 adam
29 * Bug fixes. Field prefix used in queries.
31 * Revision 1.10 1995/09/28 09:19:41 adam
32 * xfree/xmalloc used everywhere.
33 * Extract/retrieve method seems to work for text records.
35 * Revision 1.9 1995/09/27 12:22:28 adam
36 * More work on extract in record control.
37 * Field name is not in isam keys but in prefix in dictionary words.
39 * Revision 1.8 1995/09/14 07:48:22 adam
40 * Record control management.
42 * Revision 1.7 1995/09/11 13:09:32 adam
43 * More work on relevance feedback.
45 * Revision 1.6 1995/09/08 14:52:27 adam
46 * Minor changes. Dictionary is lower case now.
48 * Revision 1.5 1995/09/06 16:11:16 adam
49 * Option: only one word key per file.
51 * Revision 1.4 1995/09/05 15:28:39 adam
52 * More work on search engine.
54 * Revision 1.3 1995/09/04 12:33:41 adam
55 * Various cleanup. YAZ util used instead.
57 * Revision 1.2 1995/09/04 09:10:34 adam
58 * More work on index add/del/update.
59 * Merge sort implemented.
60 * Initial work on z39 server.
62 * Revision 1.1 1995/09/01 14:06:35 adam
63 * Split of work into more files.
77 static SYSNO sysno_next;
78 static int sys_idx_fd = -1;
82 static char **key_buf;
83 static size_t ptr_top;
86 static int key_file_no;
88 void key_open (int mem)
94 key_buf = xmalloc (mem);
95 ptr_top = mem/sizeof(char*);
100 if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
102 logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
105 file_key = dict_lookup (file_idx, ".");
107 memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
110 if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
112 logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
117 void key_flush (void)
127 logf (LOG_LOG, "sorting section %d", key_file_no);
128 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
129 sprintf (out_fname, TEMP_FNAME, key_file_no);
131 if (!(outf = fopen (out_fname, "w")))
133 logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
136 logf (LOG_LOG, "writing section %d", key_file_no);
137 prevcp = cp = key_buf[ptr_top-ptr_i];
139 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
141 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
146 cp = key_buf[ptr_top-ptr_i];
147 if (strcmp (cp, prevcp))
149 if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
152 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
159 cp = strlen (cp) + cp;
160 if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
162 logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
169 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
172 logf (LOG_LOG, "finished section %d", key_file_no);
182 dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
183 dict_close (file_idx);
187 static void wordInit (RecWord *p)
191 p->which = Word_String;
194 static void wordAdd (const RecWord *p)
199 if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
202 key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
203 kused += index_word_prefix ((char*)key_buf + kused,
204 p->attrSet, p->attrUse);
208 for (i = 0; p->u.string[i]; i++)
209 ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
210 ((char*)key_buf) [kused++] = '\0';
215 ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
216 key.sysno = key_sysno;
217 key.seqno = p->seqno;
218 memcpy ((char*)key_buf + kused, &key, sizeof(key));
219 kused += sizeof(key);
222 #define FILE_READ_BUF 1
224 static char *file_buf;
225 static int file_offset;
226 static int file_bufsize;
228 static void file_read_start (int fd)
231 file_buf = xmalloc (4096);
232 file_bufsize = read (fd, file_buf, 4096);
235 static void file_read_stop (int fd)
240 static int file_read (int fd, char *buf, size_t count)
242 int l = file_bufsize - file_offset;
248 memcpy (buf, file_buf + file_offset, l);
250 if (count > file_bufsize)
252 if ((r = read (fd, buf + l, count)) == -1)
254 logf (LOG_FATAL|LOG_ERRNO, "read");
261 file_bufsize = r = read (fd, file_buf, 4096);
264 logf (LOG_FATAL|LOG_ERRNO, "read");
270 memcpy (buf + l, file_buf, r);
276 memcpy (buf + l, file_buf, count - l);
280 memcpy (buf, file_buf + file_offset, count);
281 file_offset += count;
285 static int file_read (int fd, char *buf, size_t count)
287 return read (fd, buf, count);
290 void file_extract (int cmd, const char *fname, const char *kname)
296 const char *file_type;
298 struct recExtractCtrl extractCtrl;
301 for (i = strlen(fname); --i >= 0; )
307 else if (fname[i] == '.')
309 strcpy (ext, fname+i+1);
312 sprintf (ext_res, "fileExtension.%s", ext);
313 if (!(file_type = res_get (common_resource, ext_res)))
315 if (!(rt = recType_byName (file_type)))
317 logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
318 file_info = dict_lookup (file_idx, kname);
321 sysno = sysno_next++;
322 dict_insert (file_idx, kname, sizeof(sysno), &sysno);
323 lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
324 write (sys_idx_fd, file_type, strlen (file_type)+1);
325 write (sys_idx_fd, kname, strlen(kname)+1);
328 memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
330 if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
332 logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
335 extractCtrl.subType = "";
336 extractCtrl.init = wordInit;
337 extractCtrl.add = wordAdd;
339 file_read_start (extractCtrl.fd);
341 extractCtrl.readf = file_read;
344 r = (*rt->extract)(&extractCtrl);
346 file_read_stop (extractCtrl.fd);
348 close (extractCtrl.fd);
350 logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);