1 | /**************************************************************************
|
---|
2 | *
|
---|
3 | * mg_compression_dict.c -- Routines for reading the compression dictionary
|
---|
4 | * Copyright (C) 1994 Neil Sharman
|
---|
5 | *
|
---|
6 | * This program is free software; you can redistribute it and/or modify
|
---|
7 | * it under the terms of the GNU General Public License as published by
|
---|
8 | * the Free Software Foundation; either version 2 of the License, or
|
---|
9 | * (at your option) any later version.
|
---|
10 | *
|
---|
11 | * This program is distributed in the hope that it will be useful,
|
---|
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
14 | * GNU General Public License for more details.
|
---|
15 | *
|
---|
16 | * You should have received a copy of the GNU General Public License
|
---|
17 | * along with this program; if not, write to the Free Software
|
---|
18 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
19 | *
|
---|
20 | * $Id: mg_compression_dict.c 3745 2003-02-20 21:20:24Z mdewsnip $
|
---|
21 | *
|
---|
22 | **************************************************************************/
|
---|
23 |
|
---|
24 | #include "sysfuncs.h"
|
---|
25 |
|
---|
26 | #include "memlib.h"
|
---|
27 | #include "messages.h"
|
---|
28 | #include "local_strings.h"
|
---|
29 | #include "bitio_gen.h"
|
---|
30 | #include "bitio_m.h"
|
---|
31 | #include "bitio_m_stdio.h"
|
---|
32 | #include "timing.h"
|
---|
33 | #include "mgheap.h"
|
---|
34 | #include "netorder.h" /* [RPAP - Jan 97: Endian Ordering] */
|
---|
35 |
|
---|
36 | #include "mg_files.h"
|
---|
37 | #include "locallib.h"
|
---|
38 | #include "invf.h"
|
---|
39 | #include "text.h"
|
---|
40 | #include "words.h"
|
---|
41 | #include "mg.h"
|
---|
42 |
|
---|
43 | #define MAXBITS (sizeof(unsigned long) * 8)
|
---|
44 |
|
---|
45 | /*
|
---|
46 | $Log$
|
---|
47 | Revision 1.1 2003/02/20 21:18:24 mdewsnip
|
---|
48 | Addition of MG package for search and retrieval
|
---|
49 |
|
---|
50 | Revision 1.2 1999/12/10 15:04:25 cs025
|
---|
51 | Incorrect use of NTOHUL on a double value (num_bytes).
|
---|
52 |
|
---|
53 | Revision 1.1 1999/08/10 21:18:04 sjboddie
|
---|
54 | renamed mg-1.3d directory mg
|
---|
55 |
|
---|
56 | Revision 1.4 1999/01/15 03:05:51 rjmcnab
|
---|
57 |
|
---|
58 | Renamed lib/heap to be lib/mgheap (it was causing some problems with
|
---|
59 | some versions of the STL libraries which contained a heap.h)
|
---|
60 |
|
---|
61 | Revision 1.3 1998/12/17 09:12:52 rjmcnab
|
---|
62 |
|
---|
63 | Altered mg to process utf-8 encoded Unicode. The main changes
|
---|
64 | are in the parsing of the input, the casefolding, and the stemming.
|
---|
65 |
|
---|
66 | Revision 1.2 1998/11/25 07:55:44 rjmcnab
|
---|
67 |
|
---|
68 | Modified mg to that you can specify the stemmer you want
|
---|
69 | to use via a command line option. You specify it to
|
---|
70 | mg_passes during the build process. The number of the
|
---|
71 | stemmer that you used is stored within the inverted
|
---|
72 | dictionary header and the stemmed dictionary header so
|
---|
73 | the correct stemmer is used in later stages of building
|
---|
74 | and querying.
|
---|
75 |
|
---|
76 | Revision 1.1 1998/11/17 09:34:52 rjmcnab
|
---|
77 | *** empty log message ***
|
---|
78 |
|
---|
79 | * Revision 1.3 1994/10/20 03:56:54 tes
|
---|
80 | * I have rewritten the boolean query optimiser and abstracted out the
|
---|
81 | * components of the boolean query.
|
---|
82 | *
|
---|
83 | * Revision 1.2 1994/09/20 04:41:45 tes
|
---|
84 | * For version 1.1
|
---|
85 | *
|
---|
86 | */
|
---|
87 |
|
---|
88 | static char *RCSID = "$Id: mg_compression_dict.c 3745 2003-02-20 21:20:24Z mdewsnip $";
|
---|
89 |
|
---|
90 | /* #define DEBUG1 */
|
---|
91 |
|
---|
92 | /* #define DEBUG */
|
---|
93 |
|
---|
94 | #define is_power_of_two(a) ((a) != 0 && (((a) & ((a)-1)) == 0))
|
---|
95 |
|
---|
96 | #define MAX_RECALCULATIONS 100
|
---|
97 |
|
---|
98 | typedef struct WordData
|
---|
99 | {
|
---|
100 | u_long freq, occur_num;
|
---|
101 | float saving;
|
---|
102 | float char_bit_cost;
|
---|
103 | u_long num_trans;
|
---|
104 | u_char *word;
|
---|
105 | }
|
---|
106 | WordData;
|
---|
107 |
|
---|
108 | static WordData *Words[2];
|
---|
109 | static u_long Num[2];
|
---|
110 | static u_long chars[2];
|
---|
111 |
|
---|
112 | #define KIND(p) (((p) >= Words[0] && (p) < Words[0] + Num[0]) ? 0 : 1)
|
---|
113 | #define IsWord(p) (((p) >= Words[1] && (p) < Words[1] + Num[1]) ? 1 : 0)
|
---|
114 | #define IsNonWord(p) (((p) >= Words[0] && (p) < Words[0] + Num[0]) ? 1 : 0)
|
---|
115 |
|
---|
116 |
|
---|
117 | typedef struct DictData
|
---|
118 | {
|
---|
119 | WordData **wd;
|
---|
120 | u_long num_wds;
|
---|
121 | u_long chars;
|
---|
122 | }
|
---|
123 | DictData;
|
---|
124 |
|
---|
125 | typedef DictData DictInfo[2];
|
---|
126 |
|
---|
127 | static DictInfo keep, discard, all;
|
---|
128 | static compression_stats_header csh;
|
---|
129 |
|
---|
130 | static char *file_name = "";
|
---|
131 | static u_long novel_method = MG_NOVEL_HUFFMAN_CHARS;
|
---|
132 |
|
---|
133 |
|
---|
134 | static void ReadInWords (char *);
|
---|
135 | static void Select_on (int k, heap_comp hc);
|
---|
136 | static void Method3 (int k);
|
---|
137 | static void Select_all (void);
|
---|
138 | static u_long WriteOutWords (char *, u_long, int);
|
---|
139 | static int DecFreqIncWL (void *a, void *b);
|
---|
140 | static int OccuranceOrder (void *a, void *b);
|
---|
141 |
|
---|
142 |
|
---|
143 |
|
---|
144 | int main (int argc, char **argv)
|
---|
145 | {
|
---|
146 | ProgTime StartTime;
|
---|
147 | int ch;
|
---|
148 | char type = 'C';
|
---|
149 | char mode = '0';
|
---|
150 | int lookback = 2;
|
---|
151 | double k = 0;
|
---|
152 | u_long mem_reqd;
|
---|
153 | opterr = 0;
|
---|
154 | msg_prefix = argv[0];
|
---|
155 | while ((ch = getopt (argc, argv, "0123CPSf:d:l:hk:HBDYM")) != -1)
|
---|
156 | switch (ch)
|
---|
157 | {
|
---|
158 | case 'H':
|
---|
159 | novel_method = MG_NOVEL_HUFFMAN_CHARS;
|
---|
160 | break;
|
---|
161 | case 'B':
|
---|
162 | novel_method = MG_NOVEL_BINARY;
|
---|
163 | break;
|
---|
164 | case 'D':
|
---|
165 | novel_method = MG_NOVEL_DELTA;
|
---|
166 | break;
|
---|
167 | case 'Y':
|
---|
168 | novel_method = MG_NOVEL_HYBRID;
|
---|
169 | break;
|
---|
170 | case 'M':
|
---|
171 | novel_method = MG_NOVEL_HYBRID_MTF;
|
---|
172 | break;
|
---|
173 | case 'f': /* input file */
|
---|
174 | file_name = optarg;
|
---|
175 | break;
|
---|
176 | case 'd':
|
---|
177 | set_basepath (optarg);
|
---|
178 | break;
|
---|
179 | case 'C':
|
---|
180 | case 'P':
|
---|
181 | case 'S':
|
---|
182 | type = ch;
|
---|
183 | break;
|
---|
184 | case '0':
|
---|
185 | case '1':
|
---|
186 | case '2':
|
---|
187 | case '3':
|
---|
188 | mode = ch;
|
---|
189 | break;
|
---|
190 | case 'l':
|
---|
191 | lookback = atoi (optarg);
|
---|
192 | if (!is_power_of_two (lookback))
|
---|
193 | FatalError (1, "The lookback value must be a power of 2");
|
---|
194 | lookback = floorlog_2 (lookback);
|
---|
195 | break;
|
---|
196 | case 'k':
|
---|
197 | k = atof (optarg) * 1024;
|
---|
198 | break;
|
---|
199 | case 'h':
|
---|
200 | case '?':
|
---|
201 | fprintf (stderr, "usage: %s [-l lookback] [-f input_file]"
|
---|
202 | "[-d data directory] [-h] [-0|-1|-2|-3] [-C|-P|-S] [-k mem (Kb)]\n",
|
---|
203 | argv[0]);
|
---|
204 | exit (1);
|
---|
205 | }
|
---|
206 | GetTime (&StartTime);
|
---|
207 |
|
---|
208 | ReadInWords (file_name);
|
---|
209 |
|
---|
210 | if (type == 'C')
|
---|
211 | {
|
---|
212 | Select_all ();
|
---|
213 | mem_reqd = WriteOutWords (file_name, MG_COMPLETE_DICTIONARY, lookback);
|
---|
214 | }
|
---|
215 | else
|
---|
216 | {
|
---|
217 | switch (mode)
|
---|
218 | {
|
---|
219 | case '0':
|
---|
220 | Select_all ();
|
---|
221 | break;
|
---|
222 | case '1':
|
---|
223 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
224 | Select_on ((int) k, OccuranceOrder);
|
---|
225 | break;
|
---|
226 | case '2':
|
---|
227 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
228 | Select_on ((int) k, DecFreqIncWL);
|
---|
229 | break;
|
---|
230 | case '3':
|
---|
231 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
232 | Method3 ((int) k);
|
---|
233 | break;
|
---|
234 | }
|
---|
235 | if (type == 'P')
|
---|
236 | {
|
---|
237 | mem_reqd = WriteOutWords (file_name, MG_PARTIAL_DICTIONARY, lookback);
|
---|
238 | }
|
---|
239 | else
|
---|
240 | {
|
---|
241 | mem_reqd = WriteOutWords (file_name, MG_SEED_DICTIONARY, lookback);
|
---|
242 | }
|
---|
243 | }
|
---|
244 |
|
---|
245 | Message ("Num words : %8u -> %8u\n", Num[1], keep[1].num_wds);
|
---|
246 | Message ("Num non-words : %8u -> %8u\n", Num[0], keep[0].num_wds);
|
---|
247 | Message ("Chars of words : %8u -> %8u\n", chars[1], keep[1].chars);
|
---|
248 | Message ("Chars of non-words : %8u -> %8u\n", chars[0], keep[0].chars);
|
---|
249 | Message ("Mem usage : %8u -> %8u\n",
|
---|
250 | (Num[0] + Num[1]) * sizeof (char *) + chars[0] + chars[1],
|
---|
251 | (keep[0].num_wds + keep[1].num_wds) * sizeof (char *) +
|
---|
252 | keep[0].chars + keep[1].chars);
|
---|
253 | Message ("Actual mem required : %8u\n", mem_reqd);
|
---|
254 |
|
---|
255 | Message ("%s", ElapsedTime (&StartTime, NULL));
|
---|
256 | return 0;
|
---|
257 | }
|
---|
258 |
|
---|
259 |
|
---|
260 |
|
---|
261 |
|
---|
262 | static void
|
---|
263 | ReadInWords (char *filename)
|
---|
264 | {
|
---|
265 | FILE *f;
|
---|
266 | unsigned long i;
|
---|
267 | f = open_file (filename, TEXT_STATS_DICT_SUFFIX, "rb",
|
---|
268 | MAGIC_STATS_DICT, MG_ABORT); /* [RPAP - Feb 97: WIN32 Port] */
|
---|
269 |
|
---|
270 | fread (&csh, sizeof (csh), 1, f);
|
---|
271 |
|
---|
272 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
273 | NTOHUL(csh.num_docs);
|
---|
274 | NTOHD(csh.num_bytes);
|
---|
275 |
|
---|
276 | for (i = 0; i < 2; i++)
|
---|
277 | {
|
---|
278 | frags_stats_header fsh;
|
---|
279 | char *buf;
|
---|
280 | WordData *wd;
|
---|
281 | int j;
|
---|
282 | chars[i] = 0;
|
---|
283 |
|
---|
284 | fread (&fsh, sizeof (fsh), 1, f);
|
---|
285 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
286 | NTOHUL(fsh.num_frags);
|
---|
287 | NTOHUL(fsh.mem_for_frags);
|
---|
288 |
|
---|
289 | Num[i] = all[i].num_wds = fsh.num_frags;
|
---|
290 |
|
---|
291 | /* The +1 on the following line is to leave room for the esc code. */
|
---|
292 | all[i].wd = Xmalloc (sizeof (WordData *) * Num[i] + 1);
|
---|
293 |
|
---|
294 | buf = Xmalloc (fsh.mem_for_frags);
|
---|
295 | wd = Words[i] = Xmalloc (sizeof (WordData) * Num[i]);
|
---|
296 | for (j = 0; j < Num[i]; j++, wd++)
|
---|
297 | {
|
---|
298 | int len;
|
---|
299 | fread (&wd->freq, sizeof (wd->freq), 1, f);
|
---|
300 | NTOHUL(wd->freq); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
301 | fread (&wd->occur_num, sizeof (wd->occur_num), 1, f);
|
---|
302 | NTOHUL(wd->occur_num); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
303 | len = fgetc (f);
|
---|
304 | wd->word = (u_char *) buf;
|
---|
305 | *buf++ = len;
|
---|
306 | fread (buf, len, 1, f);
|
---|
307 | buf += len;
|
---|
308 | all[i].wd[j] = wd;
|
---|
309 | }
|
---|
310 | chars[i] = fsh.mem_for_frags - fsh.num_frags;
|
---|
311 | }
|
---|
312 | fclose (f);
|
---|
313 | }
|
---|
314 |
|
---|
315 |
|
---|
316 | static void
|
---|
317 | Alloc_keep_discard (void)
|
---|
318 | {
|
---|
319 | keep[0].num_wds = 0;
|
---|
320 | keep[0].wd = Xmalloc ((Num[0] + 1) * sizeof (WordData *));
|
---|
321 | keep[1].num_wds = 0;
|
---|
322 | keep[1].wd = Xmalloc ((Num[1] + 1) * sizeof (WordData *));
|
---|
323 | discard[0].num_wds = 0;
|
---|
324 | discard[0].wd = Xmalloc ((Num[0] + 1) * sizeof (WordData *));
|
---|
325 | discard[1].num_wds = 0;
|
---|
326 | discard[1].wd = Xmalloc ((Num[1] + 1) * sizeof (WordData *));
|
---|
327 | }
|
---|
328 |
|
---|
329 |
|
---|
330 | static int
|
---|
331 | sort_comp (const void *a, const void *b)
|
---|
332 | {
|
---|
333 | WordData *aa = *(WordData **) a;
|
---|
334 | WordData *bb = *(WordData **) b;
|
---|
335 | return casecompare (aa->word, bb->word); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
336 | }
|
---|
337 |
|
---|
338 |
|
---|
339 |
|
---|
340 | static void
|
---|
341 | SortAndCount_DictData (DictData * dd)
|
---|
342 | {
|
---|
343 | int i;
|
---|
344 | WordData **wd;
|
---|
345 | qsort (dd->wd, dd->num_wds, sizeof (WordData *), sort_comp);
|
---|
346 | dd->chars = 0;
|
---|
347 | wd = dd->wd;
|
---|
348 | for (i = 0; i < dd->num_wds; i++, wd++)
|
---|
349 | dd->chars += (*wd)->word[0];
|
---|
350 | }
|
---|
351 |
|
---|
352 |
|
---|
353 | static void
|
---|
354 | Select_all (void)
|
---|
355 | {
|
---|
356 | int i;
|
---|
357 | Alloc_keep_discard ();
|
---|
358 | keep[0].num_wds = Num[0];
|
---|
359 | for (i = 0; i < Num[0]; i++)
|
---|
360 | keep[0].wd[i] = Words[0] + i;
|
---|
361 | keep[1].num_wds = Num[1];
|
---|
362 | for (i = 0; i < Num[1]; i++)
|
---|
363 | keep[1].wd[i] = Words[1] + i;
|
---|
364 | SortAndCount_DictData (&keep[0]);
|
---|
365 | SortAndCount_DictData (&keep[1]);
|
---|
366 | }
|
---|
367 |
|
---|
368 |
|
---|
369 |
|
---|
370 |
|
---|
371 | static int
|
---|
372 | DecFreqIncWL (void *a, void *b)
|
---|
373 | {
|
---|
374 | WordData *aa = *(WordData **) a;
|
---|
375 | WordData *bb = *(WordData **) b;
|
---|
376 | if (aa->freq > bb->freq)
|
---|
377 | return -1;
|
---|
378 | else if (aa->freq < bb->freq)
|
---|
379 | return 1;
|
---|
380 | else
|
---|
381 | return bb->word[0] - aa->word[0];
|
---|
382 | }
|
---|
383 |
|
---|
384 |
|
---|
385 | static int
|
---|
386 | OccuranceOrder (void *a, void *b)
|
---|
387 | {
|
---|
388 | WordData *aa = *(WordData **) a;
|
---|
389 | WordData *bb = *(WordData **) b;
|
---|
390 | if (aa->occur_num > bb->occur_num)
|
---|
391 | return 1;
|
---|
392 | else if (aa->occur_num < bb->occur_num)
|
---|
393 | return -1;
|
---|
394 | else
|
---|
395 | return 0;
|
---|
396 | }
|
---|
397 |
|
---|
398 |
|
---|
399 | static void
|
---|
400 | Select_on (int k, heap_comp hc)
|
---|
401 | {
|
---|
402 | int i, num, mem;
|
---|
403 | WordData **wd;
|
---|
404 |
|
---|
405 | Alloc_keep_discard ();
|
---|
406 |
|
---|
407 | num = Num[0] + Num[1];
|
---|
408 | wd = Xmalloc (num * sizeof (WordData *));
|
---|
409 | for (i = 0; i < Num[0]; i++)
|
---|
410 | wd[i] = Words[0] + i;
|
---|
411 | for (i = 0; i < Num[1]; i++)
|
---|
412 | wd[i + Num[0]] = Words[1] + i;
|
---|
413 |
|
---|
414 | heap_build (wd, sizeof (*wd), num, hc);
|
---|
415 |
|
---|
416 | mem = 0;
|
---|
417 | while (k > mem && num)
|
---|
418 | {
|
---|
419 | int idx;
|
---|
420 | WordData *word = wd[0];
|
---|
421 | #ifdef DEBUG
|
---|
422 | fprintf (stderr, "%4d:%4d:%8d :%8d :%8d : \"%s\"\n",
|
---|
423 | keep[0].num_wds, keep[1].num_wds,
|
---|
424 | mem, word->freq, word->occur_num, word2str (word->word));
|
---|
425 | #endif
|
---|
426 | mem += sizeof (u_char *) + word->word[0];
|
---|
427 | heap_deletehead (wd, sizeof (*wd), &num, hc);
|
---|
428 | idx = KIND (word);
|
---|
429 | keep[idx].wd[keep[idx].num_wds++] = word;
|
---|
430 | }
|
---|
431 |
|
---|
432 | for (i = 0; i < num; i++)
|
---|
433 | {
|
---|
434 | WordData *word = wd[i];
|
---|
435 | int idx = KIND (word);
|
---|
436 | discard[idx].wd[discard[idx].num_wds++] = word;
|
---|
437 | }
|
---|
438 | SortAndCount_DictData (&keep[0]);
|
---|
439 | SortAndCount_DictData (&keep[1]);
|
---|
440 | SortAndCount_DictData (&discard[0]);
|
---|
441 | SortAndCount_DictData (&discard[1]);
|
---|
442 |
|
---|
443 | assert (keep[0].num_wds + discard[0].num_wds == Num[0]);
|
---|
444 | assert (keep[1].num_wds + discard[1].num_wds == Num[1]);
|
---|
445 |
|
---|
446 | }
|
---|
447 |
|
---|
448 |
|
---|
449 |
|
---|
450 | static int
|
---|
451 | BigSaving (void *a, void *b)
|
---|
452 | {
|
---|
453 | WordData *aa = *(WordData **) a;
|
---|
454 | WordData *bb = *(WordData **) b;
|
---|
455 | return (aa->saving > bb->saving) ? -1 : (aa->saving < bb->saving);
|
---|
456 | }
|
---|
457 |
|
---|
458 | static int
|
---|
459 | SmallSaving (void *a, void *b)
|
---|
460 | {
|
---|
461 | WordData *aa = *(WordData **) a;
|
---|
462 | WordData *bb = *(WordData **) b;
|
---|
463 | return (aa->saving < bb->saving) ? -1 : (aa->saving > bb->saving);
|
---|
464 | }
|
---|
465 |
|
---|
466 |
|
---|
467 | static void
|
---|
468 | CalcCharCounts (WordData ** wd, int num,
|
---|
469 | char *char_lens[2], char *len_lens[2],
|
---|
470 | u_long escape[2])
|
---|
471 | {
|
---|
472 | long char_freqs[2][256];
|
---|
473 | long len_freqs[2][16];
|
---|
474 | huff_data hd;
|
---|
475 | int i;
|
---|
476 | escape[0] = 0;
|
---|
477 | escape[1] = 0;
|
---|
478 | bzero ((char *) char_freqs, sizeof (char_freqs));
|
---|
479 | bzero ((char *) len_freqs, sizeof (len_freqs));
|
---|
480 | for (i = 0; i < num; i++, wd++)
|
---|
481 | {
|
---|
482 | u_long freq = (*wd)->freq;
|
---|
483 | u_char *buf = (*wd)->word;
|
---|
484 | int len = *buf++;
|
---|
485 | int idx = KIND (*wd);
|
---|
486 | len_freqs[idx][len] += freq;
|
---|
487 | escape[idx] += freq;
|
---|
488 | for (; len; len--, buf++)
|
---|
489 | char_freqs[idx][(u_long) (*buf)] += freq;
|
---|
490 | }
|
---|
491 | Generate_Huffman_Data (256, char_freqs[0], &hd, NULL);
|
---|
492 | char_lens[0] = hd.clens;
|
---|
493 | Generate_Huffman_Data (256, char_freqs[1], &hd, NULL);
|
---|
494 | char_lens[1] = hd.clens;
|
---|
495 |
|
---|
496 | Generate_Huffman_Data (16, len_freqs[0], &hd, NULL);
|
---|
497 | len_lens[0] = hd.clens;
|
---|
498 | Generate_Huffman_Data (16, len_freqs[1], &hd, NULL);
|
---|
499 | len_lens[1] = hd.clens;
|
---|
500 | }
|
---|
501 |
|
---|
502 |
|
---|
503 |
|
---|
504 |
|
---|
505 |
|
---|
506 |
|
---|
507 |
|
---|
508 |
|
---|
509 |
|
---|
510 | void
|
---|
511 | CalcBitCost (WordData ** discard_word, int discard_num,
|
---|
512 | WordData ** keep_word, int keep_num, double freqs_trans[2],
|
---|
513 | u_long escape[2], int num_trans)
|
---|
514 | {
|
---|
515 | char *char_lens[2];
|
---|
516 | char *len_lens[2];
|
---|
517 | double esc[2];
|
---|
518 | int j;
|
---|
519 | CalcCharCounts (discard_word, discard_num, char_lens, len_lens, escape);
|
---|
520 | esc[0] = -log2 (escape[0] / (freqs_trans[0] + escape[0]));
|
---|
521 | esc[1] = -log2 (escape[1] / (freqs_trans[1] + escape[1]));
|
---|
522 | for (j = 0; j < discard_num; j++, discard_word++)
|
---|
523 | {
|
---|
524 | float cbc, wbc;
|
---|
525 | u_char *buf = (*discard_word)->word;
|
---|
526 | int len = *buf++;
|
---|
527 | u_long freq = (*discard_word)->freq;
|
---|
528 | int idx = KIND (*discard_word);
|
---|
529 |
|
---|
530 | cbc = len_lens[idx][len];
|
---|
531 | for (; len; len--, buf++)
|
---|
532 | cbc += char_lens[idx][(u_long) (*buf)];
|
---|
533 |
|
---|
534 | (*discard_word)->char_bit_cost = (cbc + esc[idx]) * freq;
|
---|
535 |
|
---|
536 | wbc = -log2 (freq / (freqs_trans[idx] + escape[idx])) * freq;
|
---|
537 |
|
---|
538 | (*discard_word)->saving = ((*discard_word)->char_bit_cost - wbc) /
|
---|
539 | (sizeof (char *) + (*discard_word)->word[0]);
|
---|
540 |
|
---|
541 | (*discard_word)->num_trans = num_trans;
|
---|
542 | }
|
---|
543 | for (j = 0; j < keep_num; j++, keep_word++)
|
---|
544 | {
|
---|
545 | float cbc, wbc;
|
---|
546 | u_char *buf = (*keep_word)->word;
|
---|
547 | int len = *buf++;
|
---|
548 | u_long freq = (*keep_word)->freq;
|
---|
549 | int idx = KIND (*keep_word);
|
---|
550 |
|
---|
551 | cbc = len_lens[idx][len];
|
---|
552 | for (; len; len--, buf++)
|
---|
553 | cbc += char_lens[idx][(u_long) (*buf)];
|
---|
554 |
|
---|
555 | (*keep_word)->char_bit_cost = (cbc + esc[idx]) * freq;
|
---|
556 |
|
---|
557 | wbc = -log2 (freq / (freqs_trans[idx] + escape[idx])) * freq;
|
---|
558 |
|
---|
559 | (*keep_word)->saving = ((*keep_word)->char_bit_cost - wbc) /
|
---|
560 | (sizeof (char *) + (*keep_word)->word[0]);
|
---|
561 |
|
---|
562 | (*keep_word)->num_trans = num_trans;
|
---|
563 | }
|
---|
564 | Xfree (char_lens[0]);
|
---|
565 | Xfree (char_lens[1]);
|
---|
566 | Xfree (len_lens[0]);
|
---|
567 | Xfree (len_lens[1]);
|
---|
568 | }
|
---|
569 |
|
---|
570 |
|
---|
571 |
|
---|
572 | static void
|
---|
573 | Method3 (int k)
|
---|
574 | {
|
---|
575 | int i, keep_num, discard_num, mem, num_trans, recalc_reqd;
|
---|
576 | int keep_to_discard = 0;
|
---|
577 | int discard_to_keep = 0;
|
---|
578 | int recalcs = 0;
|
---|
579 | double freqs_trans[2], total;
|
---|
580 | u_long escape[2];
|
---|
581 | WordData **keep_heap, **discard_heap;
|
---|
582 |
|
---|
583 | freqs_trans[0] = freqs_trans[1] = 0;
|
---|
584 | num_trans = 0;
|
---|
585 |
|
---|
586 | discard_num = Num[0] + Num[1];
|
---|
587 | discard_heap = Xmalloc (discard_num * sizeof (WordData *));
|
---|
588 |
|
---|
589 | keep_num = 0;
|
---|
590 | keep_heap = Xmalloc (discard_num * sizeof (WordData *));
|
---|
591 |
|
---|
592 |
|
---|
593 | for (i = 0; i < Num[0]; i++)
|
---|
594 | discard_heap[i] = Words[0] + i;
|
---|
595 | for (i = 0; i < Num[1]; i++)
|
---|
596 | discard_heap[i + Num[0]] = Words[1] + i;
|
---|
597 |
|
---|
598 | heap_build (discard_heap, sizeof (*discard_heap), discard_num, DecFreqIncWL);
|
---|
599 |
|
---|
600 | mem = 0;
|
---|
601 | while (k > mem && discard_num)
|
---|
602 | {
|
---|
603 | WordData *word = discard_heap[0];
|
---|
604 | mem += sizeof (u_char *) + word->word[0];
|
---|
605 | heap_deletehead (discard_heap, sizeof (word), &discard_num, DecFreqIncWL);
|
---|
606 | keep_heap[keep_num++] = word;
|
---|
607 | freqs_trans[KIND (word)] += word->freq;
|
---|
608 | num_trans++;
|
---|
609 | }
|
---|
610 |
|
---|
611 | CalcBitCost (discard_heap, discard_num, keep_heap, keep_num,
|
---|
612 | freqs_trans, escape, num_trans);
|
---|
613 | heap_build (discard_heap, sizeof (*discard_heap), discard_num, BigSaving);
|
---|
614 | heap_build (keep_heap, sizeof (*keep_heap), keep_num, SmallSaving);
|
---|
615 |
|
---|
616 | total = 0;
|
---|
617 | recalc_reqd = 0;
|
---|
618 | while (keep_num && discard_num)
|
---|
619 | {
|
---|
620 | if ((keep_num && keep_heap[0]->num_trans < num_trans) ||
|
---|
621 | (discard_num && discard_heap[0]->num_trans < num_trans))
|
---|
622 | {
|
---|
623 | if (keep_num && keep_heap[0]->num_trans < num_trans)
|
---|
624 | {
|
---|
625 | WordData *word = keep_heap[0];
|
---|
626 | int idx = KIND (word);
|
---|
627 | float wbc;
|
---|
628 | #ifdef DEBUG1
|
---|
629 | fprintf (stderr, "KEEP \"%s\" %.2f ->", word2str (word->word),
|
---|
630 | word->saving);
|
---|
631 | #endif
|
---|
632 | wbc = -log2 (word->freq / (freqs_trans[idx] + escape[idx])) *
|
---|
633 | word->freq;
|
---|
634 | word->saving = (word->char_bit_cost - wbc) /
|
---|
635 | (sizeof (char *) + word->word[0]);
|
---|
636 | #ifdef DEBUG1
|
---|
637 | fprintf (stderr, " %.2f\n", word->saving);
|
---|
638 | #endif
|
---|
639 | word->num_trans = num_trans;
|
---|
640 | heap_changedhead (keep_heap, sizeof (word), keep_num, SmallSaving);
|
---|
641 | }
|
---|
642 |
|
---|
643 | if (discard_num && discard_heap[0]->num_trans < num_trans)
|
---|
644 | {
|
---|
645 | WordData *word = discard_heap[0];
|
---|
646 | int idx = KIND (word);
|
---|
647 | float wbc;
|
---|
648 | #ifdef DEBUG1
|
---|
649 | fprintf (stderr, "DISCARD \"%s\" %.2f ->", word2str (word->word),
|
---|
650 | word->saving);
|
---|
651 | #endif
|
---|
652 | wbc = -log2 (word->freq / (freqs_trans[idx] + escape[idx])) *
|
---|
653 | word->freq;
|
---|
654 | word->saving = (word->char_bit_cost - wbc) /
|
---|
655 | (sizeof (char *) + word->word[0]);
|
---|
656 | #ifdef DEBUG1
|
---|
657 | fprintf (stderr, " %.2f\n", word->saving);
|
---|
658 | #endif
|
---|
659 | word->num_trans = num_trans;
|
---|
660 | heap_changedhead (discard_heap, sizeof (word),
|
---|
661 | discard_num, BigSaving);
|
---|
662 | }
|
---|
663 | }
|
---|
664 | else if (keep_heap[0]->saving < discard_heap[0]->saving)
|
---|
665 | {
|
---|
666 | assert (keep_num && discard_num);
|
---|
667 | if (keep_num && mem + sizeof (char *) + discard_heap[0]->word[0] > k)
|
---|
668 | {
|
---|
669 | /* Transfer the word at the top of the keep heap to the
|
---|
670 | discard heap. */
|
---|
671 | WordData *word = keep_heap[0];
|
---|
672 | int idx = KIND (word);
|
---|
673 | heap_deletehead (keep_heap, sizeof (word), &keep_num, SmallSaving);
|
---|
674 | discard_heap[discard_num] = word;
|
---|
675 | heap_additem (discard_heap, sizeof (word), &discard_num,
|
---|
676 | BigSaving);
|
---|
677 | freqs_trans[idx] -= word->freq;
|
---|
678 | mem -= sizeof (u_char *) + word->word[0];
|
---|
679 | num_trans++;
|
---|
680 | total += word->saving;
|
---|
681 | keep_to_discard++;
|
---|
682 | #ifdef DEBUG
|
---|
683 | fprintf (stderr,
|
---|
684 | "KEEP -> DISCARD %8d :%8d :%8.0f :%8.0f : \"%s\"\n",
|
---|
685 | mem, word->freq, word->char_bit_cost,
|
---|
686 | word->saving, word2str (word->word));
|
---|
687 | #endif
|
---|
688 | }
|
---|
689 | else
|
---|
690 | {
|
---|
691 | /* Transfer the word at the top of the discard heap to the
|
---|
692 | keep heap. */
|
---|
693 | WordData *word = discard_heap[0];
|
---|
694 | int idx = KIND (word);
|
---|
695 | heap_deletehead (discard_heap, sizeof (word),
|
---|
696 | &discard_num, BigSaving);
|
---|
697 | keep_heap[keep_num] = word;
|
---|
698 | heap_additem (keep_heap, sizeof (word), &keep_num, SmallSaving);
|
---|
699 | freqs_trans[idx] += word->freq;
|
---|
700 | mem += sizeof (u_char *) + word->word[0];
|
---|
701 | num_trans++;
|
---|
702 | total += word->saving;
|
---|
703 | discard_to_keep++;
|
---|
704 | #ifdef DEBUG
|
---|
705 | fprintf (stderr,
|
---|
706 | "DISCARD -> KEEP %8d :%8d :%8.0f :%8.0f : \"%s\"\n",
|
---|
707 | mem, word->freq, word->char_bit_cost,
|
---|
708 | word->saving, word2str (word->word));
|
---|
709 | #endif
|
---|
710 | }
|
---|
711 |
|
---|
712 | recalc_reqd = 1;
|
---|
713 |
|
---|
714 | }
|
---|
715 | else
|
---|
716 | {
|
---|
717 | if (recalc_reqd == 0)
|
---|
718 | break;
|
---|
719 | #ifdef DEBUG1
|
---|
720 | fprintf (stderr, "--------------\n");
|
---|
721 | #endif
|
---|
722 | if (recalcs == MAX_RECALCULATIONS)
|
---|
723 | break;
|
---|
724 | CalcBitCost (discard_heap, discard_num, keep_heap, keep_num,
|
---|
725 | freqs_trans, escape, num_trans);
|
---|
726 | heap_build (discard_heap, sizeof (*discard_heap),
|
---|
727 | discard_num, BigSaving);
|
---|
728 | heap_build (keep_heap, sizeof (keep_heap), keep_num, SmallSaving);
|
---|
729 | recalc_reqd = 0;
|
---|
730 | recalcs++;
|
---|
731 | }
|
---|
732 | }
|
---|
733 |
|
---|
734 | Alloc_keep_discard ();
|
---|
735 |
|
---|
736 | for (i = 0; i < discard_num; i++)
|
---|
737 | {
|
---|
738 | WordData *word = discard_heap[i];
|
---|
739 | int idx = KIND (word);
|
---|
740 | assert (IsWord (word) || IsNonWord (word));
|
---|
741 | discard[idx].wd[discard[idx].num_wds++] = word;
|
---|
742 | }
|
---|
743 | for (i = 0; i < keep_num; i++)
|
---|
744 | {
|
---|
745 | WordData *word = keep_heap[i];
|
---|
746 | int idx = KIND (word);
|
---|
747 | assert (IsWord (word) || IsNonWord (word));
|
---|
748 | keep[idx].wd[keep[idx].num_wds++] = word;
|
---|
749 | }
|
---|
750 | SortAndCount_DictData (&keep[0]);
|
---|
751 | SortAndCount_DictData (&keep[1]);
|
---|
752 | SortAndCount_DictData (&discard[0]);
|
---|
753 | SortAndCount_DictData (&discard[1]);
|
---|
754 |
|
---|
755 | assert (keep[0].num_wds + discard[0].num_wds == Num[0]);
|
---|
756 | assert (keep[1].num_wds + discard[1].num_wds == Num[1]);
|
---|
757 | Message ("Keep -> Discard : %8d", keep_to_discard);
|
---|
758 | Message ("Discard -> Keep : %8d", discard_to_keep);
|
---|
759 | Message ("Huffman Recalculations : %8d", recalcs);
|
---|
760 | if (recalcs == MAX_RECALCULATIONS)
|
---|
761 | Message ("WARNING: The number of recalculations == %d", MAX_RECALCULATIONS);
|
---|
762 | }
|
---|
763 |
|
---|
764 |
|
---|
765 |
|
---|
766 |
|
---|
767 |
|
---|
768 |
|
---|
769 |
|
---|
770 |
|
---|
771 |
|
---|
772 |
|
---|
773 | /****************************************************************************
|
---|
774 | ***** *****
|
---|
775 | ***** Dictionary saving code *****
|
---|
776 | ***** *****
|
---|
777 | ****************************************************************************/
|
---|
778 |
|
---|
779 |
|
---|
780 |
|
---|
781 | static void
|
---|
782 | Write_cdh (FILE * f, compression_dict_header * cdh)
|
---|
783 | {
|
---|
784 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
785 | int i;
|
---|
786 | compression_dict_header tmp = *cdh;
|
---|
787 | HTONUL(tmp.dict_type);
|
---|
788 | HTONUL(tmp.novel_method);
|
---|
789 | for (i = 0; i < TEXT_PARAMS; i++)
|
---|
790 | HTONUL(tmp.params[i]);
|
---|
791 | HTONUL(tmp.num_words[0]);
|
---|
792 | HTONUL(tmp.num_words[1]);
|
---|
793 | HTONUL(tmp.num_word_chars[0]);
|
---|
794 | HTONUL(tmp.num_word_chars[1]);
|
---|
795 | HTONUL(tmp.lookback);
|
---|
796 |
|
---|
797 | fwrite (&tmp, sizeof (tmp), 1, f);
|
---|
798 | }
|
---|
799 |
|
---|
800 |
|
---|
801 | static void
|
---|
802 | Write_words (FILE * f, DictData * dd)
|
---|
803 | {
|
---|
804 | int i;
|
---|
805 | u_char *curr, *prev = NULL;
|
---|
806 | for (i = 0; i < dd->num_wds; i++)
|
---|
807 | {
|
---|
808 | int len;
|
---|
809 | curr = dd->wd[i]->word;
|
---|
810 | if (prev)
|
---|
811 | /* look for prefix match with prev string */
|
---|
812 | len = prefixlen (prev, curr);
|
---|
813 | else
|
---|
814 | len = 0;
|
---|
815 | fputc ((len << 4) + (curr[0] - len), f);
|
---|
816 | fwrite (curr + len + 1, sizeof (u_char), curr[0] - len, f);
|
---|
817 | prev = curr;
|
---|
818 | }
|
---|
819 |
|
---|
820 | }
|
---|
821 |
|
---|
822 |
|
---|
823 | static int
|
---|
824 | Uncompressed_size (DictData * dd)
|
---|
825 | {
|
---|
826 | int i, us;
|
---|
827 | for (us = i = 0; i < dd->num_wds; i++)
|
---|
828 | us += dd->wd[i]->word[0];
|
---|
829 | return us;
|
---|
830 | }
|
---|
831 |
|
---|
832 |
|
---|
833 | static u_long
|
---|
834 | Write_data (FILE * f, DictData * dd, int lookback)
|
---|
835 | {
|
---|
836 | u_long mem_reqd;
|
---|
837 | huff_data *hd;
|
---|
838 | int i;
|
---|
839 | u_long us = dd->num_wds;
|
---|
840 | long *freqs;
|
---|
841 | u_long huff_words_size[MAX_HUFFCODE_LEN + 1];
|
---|
842 | u_long lencounts[MAX_HUFFCODE_LEN + 1];
|
---|
843 | u_char *lastword[MAX_HUFFCODE_LEN + 1];
|
---|
844 |
|
---|
845 | if (!(freqs = Xmalloc ((dd->num_wds) * sizeof (*freqs))))
|
---|
846 | FatalError (1, "Unable to allocate memory for freqs");
|
---|
847 |
|
---|
848 | for (i = 0; i < dd->num_wds; i++)
|
---|
849 | {
|
---|
850 | freqs[i] = dd->wd[i]->freq;
|
---|
851 | us += dd->wd[i]->word[0];
|
---|
852 | }
|
---|
853 |
|
---|
854 | if (!(hd = Generate_Huffman_Data (dd->num_wds, freqs, NULL, NULL)))
|
---|
855 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
856 |
|
---|
857 | Xfree (freqs), freqs = NULL;
|
---|
858 |
|
---|
859 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
860 | FatalError (1, "Unable to write huffman data");
|
---|
861 |
|
---|
862 | HTONUL(us); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
863 | fwrite (&us, sizeof (us), 1, f);
|
---|
864 | NTOHUL(us); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
865 |
|
---|
866 |
|
---|
867 | /* Calculate the amount of memory that will be required to store the text for
|
---|
868 | each different huffman code len. Every 1<<lookback words for each different
|
---|
869 | codelen length will not be prefixed by previous strings. */
|
---|
870 |
|
---|
871 |
|
---|
872 | bzero ((char *) &huff_words_size, sizeof (huff_words_size));
|
---|
873 | bzero ((char *) &lencounts, sizeof (lencounts));
|
---|
874 |
|
---|
875 | mem_reqd = 0;
|
---|
876 |
|
---|
877 | for (i = 0; i < dd->num_wds; i++)
|
---|
878 | {
|
---|
879 | int codelen = hd->clens[i];
|
---|
880 | u_char *word = dd->wd[i]->word;
|
---|
881 |
|
---|
882 | if (!codelen)
|
---|
883 | FatalError (1, "The length of a code for a word was zero");
|
---|
884 |
|
---|
885 | huff_words_size[codelen] += word[0] + 1;
|
---|
886 | mem_reqd += word[0] + (lookback != 0);
|
---|
887 | #if 0
|
---|
888 | if ((lencounts[codelen] & ((1 << lookback) - 1)) == 0)
|
---|
889 | lastword[codelen] = word;
|
---|
890 | else
|
---|
891 | huff_words_size[codelen] -= prefixlen (lastword[codelen], word);
|
---|
892 | #else
|
---|
893 | if ((lencounts[codelen] & ((1 << lookback) - 1)) != 0)
|
---|
894 | {
|
---|
895 | int save = prefixlen (lastword[codelen], word);
|
---|
896 | mem_reqd -= save;
|
---|
897 | huff_words_size[codelen] -= save;
|
---|
898 | }
|
---|
899 | else
|
---|
900 | {
|
---|
901 | mem_reqd += sizeof (u_char *);
|
---|
902 | }
|
---|
903 | lastword[codelen] = word;
|
---|
904 | #endif
|
---|
905 | lencounts[codelen]++;
|
---|
906 | }
|
---|
907 |
|
---|
908 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
909 | for (i = hd->mincodelen; i < hd->maxcodelen + 1; i++)
|
---|
910 | HTONUL(huff_words_size[i]);
|
---|
911 |
|
---|
912 | fwrite (huff_words_size + hd->mincodelen, sizeof (*huff_words_size),
|
---|
913 | hd->maxcodelen - hd->mincodelen + 1, f);
|
---|
914 |
|
---|
915 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
916 | for (i = hd->mincodelen; i < hd->maxcodelen + 1; i++)
|
---|
917 | NTOHUL(huff_words_size[i]);
|
---|
918 |
|
---|
919 | Write_words (f, dd);
|
---|
920 |
|
---|
921 | Xfree (hd->clens);
|
---|
922 | Xfree (hd);
|
---|
923 |
|
---|
924 | return mem_reqd;
|
---|
925 | }
|
---|
926 |
|
---|
927 |
|
---|
928 |
|
---|
929 | static void
|
---|
930 | Write_charfreqs (FILE * f, DictData * dd, int words,
|
---|
931 | int zero_freq_permitted)
|
---|
932 | {
|
---|
933 | int j;
|
---|
934 | long freqs[256];
|
---|
935 | WordData **wd = dd->wd;
|
---|
936 | huff_data *hd;
|
---|
937 |
|
---|
938 | bzero ((char *) freqs, sizeof (freqs));
|
---|
939 |
|
---|
940 | for (j = 0; j < dd->num_wds; j++, wd++)
|
---|
941 | {
|
---|
942 | u_char *buf = (*wd)->word;
|
---|
943 | int len = *buf++;
|
---|
944 | for (; len; len--, buf++)
|
---|
945 | freqs[(u_long) (*buf)] += (*wd)->freq;
|
---|
946 | }
|
---|
947 |
|
---|
948 | if (!zero_freq_permitted)
|
---|
949 | for (j = 0; j < 256; j++)
|
---|
950 | if (!freqs[j] && PESINAWORD (j) == words)
|
---|
951 | freqs[j] = 1;
|
---|
952 |
|
---|
953 | if (!(hd = Generate_Huffman_Data (256, freqs, NULL, NULL)))
|
---|
954 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
955 |
|
---|
956 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
957 | FatalError (1, "Unable to write huffman data");
|
---|
958 |
|
---|
959 | Xfree (hd->clens);
|
---|
960 | Xfree (hd);
|
---|
961 | }
|
---|
962 |
|
---|
963 |
|
---|
964 |
|
---|
965 |
|
---|
966 | static void
|
---|
967 | Write_wordlens (FILE * f, DictData * dd, int zero_freq_permitted)
|
---|
968 | {
|
---|
969 | int j;
|
---|
970 | long freqs[16];
|
---|
971 | WordData **wd = dd->wd;
|
---|
972 | huff_data *hd;
|
---|
973 |
|
---|
974 | bzero ((char *) freqs, sizeof (freqs));
|
---|
975 |
|
---|
976 | for (j = 0; j < dd->num_wds; j++, wd++)
|
---|
977 | freqs[(*wd)->word[0]] += (*wd)->freq;
|
---|
978 |
|
---|
979 | if (!zero_freq_permitted)
|
---|
980 | for (j = 0; j < 16; j++)
|
---|
981 | if (!freqs[j])
|
---|
982 | freqs[j] = 1;
|
---|
983 |
|
---|
984 | if (!(hd = Generate_Huffman_Data (16, freqs, NULL, NULL)))
|
---|
985 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
986 |
|
---|
987 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
988 | FatalError (1, "Unable to write huffman data");
|
---|
989 |
|
---|
990 |
|
---|
991 | Xfree (hd->clens);
|
---|
992 | Xfree (hd);
|
---|
993 | }
|
---|
994 |
|
---|
995 |
|
---|
996 |
|
---|
997 | static u_long
|
---|
998 | WriteOutWords (char *file_name, u_long type, int lookback)
|
---|
999 | {
|
---|
1000 | FILE *f;
|
---|
1001 | int i;
|
---|
1002 | u_long mem_reqd = 0;
|
---|
1003 |
|
---|
1004 | compression_dict_header cdh;
|
---|
1005 | f = create_file (file_name, TEXT_DICT_SUFFIX, "w+b",
|
---|
1006 | MAGIC_DICT, MG_ABORT); /* [RPAP - Feb 97: WIN32 Port] */
|
---|
1007 |
|
---|
1008 | bzero((char *) &cdh, sizeof(cdh));
|
---|
1009 |
|
---|
1010 | cdh.dict_type = type;
|
---|
1011 | cdh.novel_method = (type == MG_SEED_DICTIONARY) ? novel_method :
|
---|
1012 | MG_NOVEL_HUFFMAN_CHARS;
|
---|
1013 |
|
---|
1014 | cdh.num_words[1] = keep[1].num_wds;
|
---|
1015 | cdh.num_words[0] = keep[0].num_wds;
|
---|
1016 | cdh.num_word_chars[1] = Uncompressed_size (&keep[1]);
|
---|
1017 | cdh.num_word_chars[0] = Uncompressed_size (&keep[0]);
|
---|
1018 | cdh.lookback = lookback;
|
---|
1019 |
|
---|
1020 | Write_cdh (f, &cdh);
|
---|
1021 |
|
---|
1022 | for (i = 0; i < 2; i++)
|
---|
1023 | switch (type)
|
---|
1024 | {
|
---|
1025 | case MG_COMPLETE_DICTIONARY:
|
---|
1026 | {
|
---|
1027 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
1028 | }
|
---|
1029 | break;
|
---|
1030 | case MG_PARTIAL_DICTIONARY:
|
---|
1031 | {
|
---|
1032 | if (keep[i].num_wds)
|
---|
1033 | {
|
---|
1034 | int j;
|
---|
1035 | WordData esc;
|
---|
1036 | esc.freq = 0;
|
---|
1037 | esc.word = (u_char *) "";
|
---|
1038 | keep[i].wd[keep[i].num_wds++] = &esc;
|
---|
1039 | for (j = 0; j < discard[i].num_wds; j++)
|
---|
1040 | esc.freq += discard[i].wd[j]->freq;
|
---|
1041 | if (!esc.freq)
|
---|
1042 | esc.freq++;
|
---|
1043 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
1044 | }
|
---|
1045 | Write_charfreqs (f, &discard[i], i, 1);
|
---|
1046 | Write_wordlens (f, &discard[i], 1);
|
---|
1047 | }
|
---|
1048 | break;
|
---|
1049 | case MG_SEED_DICTIONARY:
|
---|
1050 | {
|
---|
1051 | if (keep[i].num_wds)
|
---|
1052 | {
|
---|
1053 | int j;
|
---|
1054 | WordData esc;
|
---|
1055 | esc.freq = 0;
|
---|
1056 | esc.word = (u_char *) "";
|
---|
1057 | keep[i].wd[keep[i].num_wds++] = &esc;
|
---|
1058 | for (j = 0; j < all[i].num_wds; j++)
|
---|
1059 | if (all[i].wd[j]->freq == 1)
|
---|
1060 | esc.freq++;
|
---|
1061 | if (!esc.freq)
|
---|
1062 | esc.freq++;
|
---|
1063 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
1064 | }
|
---|
1065 | switch (novel_method)
|
---|
1066 | {
|
---|
1067 | case MG_NOVEL_HUFFMAN_CHARS:
|
---|
1068 | Write_charfreqs (f, &all[i], i, 0);
|
---|
1069 | Write_wordlens (f, &all[i], 0);
|
---|
1070 | break;
|
---|
1071 | case MG_NOVEL_BINARY:
|
---|
1072 | break;
|
---|
1073 | case MG_NOVEL_DELTA:
|
---|
1074 | break;
|
---|
1075 | case MG_NOVEL_HYBRID:
|
---|
1076 | break;
|
---|
1077 | case MG_NOVEL_HYBRID_MTF:
|
---|
1078 | break;
|
---|
1079 | default:
|
---|
1080 | FatalError (1, "Bad novel method");
|
---|
1081 | }
|
---|
1082 | }
|
---|
1083 | break;
|
---|
1084 | }
|
---|
1085 | fclose (f);
|
---|
1086 | return mem_reqd;
|
---|
1087 | }
|
---|