2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.7 1995-10-02 15:18:52 adam
8 * New member in recRetrieveCtrl: diagnostic.
10 * Revision 1.6 1995/09/29 15:51:56 adam
11 * First work on multi-way read.
13 * Revision 1.5 1995/09/29 14:01:43 adam
16 * Revision 1.4 1995/09/28 14:22:57 adam
17 * Sort uses smaller temporary files.
19 * Revision 1.3 1995/09/06 16:11:17 adam
20 * Option: only one word key per file.
22 * Revision 1.2 1995/09/04 12:33:42 adam
23 * Various cleanup. YAZ util used instead.
25 * Revision 1.1 1995/09/04 09:10:37 adam
26 * More work on index add/del/update.
27 * Merge sort implemented.
28 * Initial work on z39 server.
42 #define KEY_SIZE (1+sizeof(struct it_key))
43 #define INP_NAME_MAX 8192
44 #define INP_BUF_START 60000
45 #define INP_BUF_ADD 400000
47 static int no_diffs = 0;
48 static int no_updates = 0;
49 static int no_insertions = 0;
50 static int no_iterations = 0;
52 static int read_one (FILE *inf, char *name, char *key)
58 if ((c=getc(inf)) == EOF)
62 for (i = 0; i<KEY_SIZE; i++)
63 ((char *)key)[i] = getc (inf);
68 static int inp (Dict dict, ISAM isam, const char *name)
72 char next_name[INP_NAME_MAX+1];
73 char cur_name[INP_NAME_MAX+1];
74 int key_buf_size = INP_BUF_START;
80 next_key = xmalloc (KEY_SIZE);
81 key_buf = xmalloc (key_buf_size * (KEY_SIZE));
82 if (!(inf = fopen (name, "r")))
84 logf (LOG_FATAL|LOG_ERRNO, "cannot open `%s'", name);
87 more = read_one (inf, cur_name, key_buf);
88 while (more) /* EOF ? */
91 key_buf_ptr = KEY_SIZE;
94 if (!(more = read_one (inf, next_name, next_key)))
96 if (*next_name && strcmp (next_name, cur_name))
98 memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
99 key_buf_ptr += KEY_SIZE;
100 if (key_buf_ptr+KEY_SIZE >= key_buf_size)
103 new_key_buf = xmalloc (key_buf_size + INP_BUF_ADD);
104 memcpy (new_key_buf, key_buf, key_buf_size);
105 key_buf_size += INP_BUF_ADD;
107 key_buf = new_key_buf;
111 nmemb = key_buf_ptr / KEY_SIZE;
112 assert (nmemb*KEY_SIZE == key_buf_ptr);
113 if ((info = dict_lookup (dict, cur_name)))
115 ISAM_P isam_p, isam_p2;
116 logf (LOG_DEBUG, "updating %s", cur_name);
118 memcpy (&isam_p, info+1, sizeof(ISAM_P));
119 isam_p2 = is_merge (isam, isam_p, nmemb, key_buf);
120 if (isam_p2 != isam_p)
121 dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p2);
126 logf (LOG_DEBUG, "inserting %s", cur_name);
128 isam_p = is_merge (isam, 0, nmemb, key_buf);
129 dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p);
131 memcpy (key_buf, next_key, KEY_SIZE);
132 strcpy (cur_name, next_name);
138 void key_input (const char *dict_fname, const char *isam_fname,
139 const char *key_fname, int cache)
144 dict = dict_open (dict_fname, cache, 1);
147 logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
150 isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key));
153 logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);
156 inp (dict, isam, key_fname);
159 logf (LOG_LOG, "Iterations . . .%7d", no_iterations);
160 logf (LOG_LOG, "Distinct words .%7d", no_diffs);
161 logf (LOG_LOG, "Updates. . . . .%7d", no_updates);
162 logf (LOG_LOG, "Insertions . . .%7d", no_insertions);
167 int no; /* file no */
168 off_t offset; /* file offset */
169 unsigned char *buf; /* buffer block */
170 size_t buf_size; /* number of read bytes in block */
171 size_t chunk; /* number of bytes allocated */
172 size_t buf_ptr; /* current position in buffer */
173 char *prev_name; /* last word read */
176 void key_file_chunk_read (struct key_file *f)
180 sprintf (fname, TEMP_FNAME, f->no);
181 fd = open (fname, O_RDONLY);
184 logf (LOG_FATAL|LOG_ERRNO, "cannot open %s", fname);
187 if (lseek (fd, f->offset, SEEK_SET) == -1)
189 logf (LOG_FATAL|LOG_ERRNO, "cannot seek %s", fname);
192 while (f->chunk - nr > 0)
194 r = read (fd, f->buf + nr, f->chunk - nr);
201 logf (LOG_FATAL|LOG_ERRNO, "read of %s", fname);
207 void key_file_init (struct key_file *f)
209 f->buf = xmalloc (f->chunk);
210 f->prev_name = xmalloc (256);
211 *f->prev_name = '\0';
214 int key_file_getc (struct key_file *f)
216 if (f->buf_ptr < f->buf_size)
217 return f->buf[(f->buf_ptr)++];
218 if (f->buf_size < f->chunk)
220 f->offset += f->buf_size;
221 key_file_chunk_read (f);
222 if (f->buf_ptr < f->buf_size)
223 return f->buf[(f->buf_ptr)++];
228 int key_file_read (struct key_file *f, char *name, char *key)
232 c = key_file_getc (f);
234 strcpy (name, f->prev_name);
241 while ((name[i++] = key_file_getc (f)))
243 strcpy (f->prev_name, name);
245 for (i = 0; i<KEY_SIZE; i++)
246 key[i++] = key_file_getc (f);
250 int inp2 (Dict dict, ISAM isam, const char *name)
254 char next_name[INP_NAME_MAX+1];
255 char cur_name[INP_NAME_MAX+1];
256 int key_buf_size = INP_BUF_START;
262 next_key = xmalloc (KEY_SIZE);
263 key_buf = xmalloc (key_buf_size * (KEY_SIZE));
264 if (!(inf = fopen (name, "r")))
266 logf (LOG_FATAL|LOG_ERRNO, "cannot open `%s'", name);
269 more = read_one (inf, cur_name, key_buf);
270 while (more) /* EOF ? */
273 key_buf_ptr = KEY_SIZE;
276 if (!(more = read_one (inf, next_name, next_key)))
278 if (*next_name && strcmp (next_name, cur_name))
280 memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
281 key_buf_ptr += KEY_SIZE;
282 if (key_buf_ptr+KEY_SIZE >= key_buf_size)
285 new_key_buf = xmalloc (key_buf_size + INP_BUF_ADD);
286 memcpy (new_key_buf, key_buf, key_buf_size);
287 key_buf_size += INP_BUF_ADD;
289 key_buf = new_key_buf;
293 nmemb = key_buf_ptr / KEY_SIZE;
294 assert (nmemb*KEY_SIZE == key_buf_ptr);
295 if ((info = dict_lookup (dict, cur_name)))
297 ISAM_P isam_p, isam_p2;
298 logf (LOG_DEBUG, "updating %s", cur_name);
300 memcpy (&isam_p, info+1, sizeof(ISAM_P));
301 isam_p2 = is_merge (isam, isam_p, nmemb, key_buf);
302 if (isam_p2 != isam_p)
303 dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p2);
308 logf (LOG_DEBUG, "inserting %s", cur_name);
310 isam_p = is_merge (isam, 0, nmemb, key_buf);
311 dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p);
313 memcpy (key_buf, next_key, KEY_SIZE);
314 strcpy (cur_name, next_name);
320 void key_input2 (const char *dict_fname, const char *isam_fname,
321 int nkeys, int cache)
326 dict = dict_open (dict_fname, cache, 1);
329 logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
332 isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key));
335 logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);
338 inp2 (dict, isam, NULL);
341 logf (LOG_LOG, "Iterations . . .%7d", no_iterations);
342 logf (LOG_LOG, "Distinct words .%7d", no_diffs);
343 logf (LOG_LOG, "Updates. . . . .%7d", no_updates);
344 logf (LOG_LOG, "Insertions . . .%7d", no_insertions);