1 | /**************************************************************************
|
---|
2 | *
|
---|
3 | * mgpp_compression_dict.cpp -- Routines for creating 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 | **************************************************************************/
|
---|
21 |
|
---|
22 | #define _XOPEN_SOURCE 1
|
---|
23 | #define _XOPEN_SOURCE_EXTENDED 1
|
---|
24 |
|
---|
25 | // need this to avoid bizarre compiler problems under VC++ 6.0
|
---|
26 | #if defined (__WIN32__) && !defined (GSDL_USE_IOS_H)
|
---|
27 | # include <iostream>
|
---|
28 | #endif
|
---|
29 |
|
---|
30 | /* getopt is in posix.2, so cygwin should have it in unistd, but doesn't */
|
---|
31 | #if defined (__WIN32__) || defined (__CYGWIN__)
|
---|
32 | # include "getopt_old.h"
|
---|
33 | #else
|
---|
34 | # include <unistd.h>
|
---|
35 | #endif
|
---|
36 |
|
---|
37 | #include "sysfuncs.h"
|
---|
38 | #include "memlib.h"
|
---|
39 | #include "messages.h"
|
---|
40 | #include "local_strings.h"
|
---|
41 | #include "bitio_gen.h"
|
---|
42 | #include "bitio_m_stdio.h"
|
---|
43 | #include "mgheap.h"
|
---|
44 | #include "netorder.h" /* [RPAP - Jan 97: Endian Ordering] */
|
---|
45 |
|
---|
46 | #include "mg_files.h"
|
---|
47 | #include "locallib.h"
|
---|
48 | #include "invf.h"
|
---|
49 | #include "text.h"
|
---|
50 | #include "words.h"
|
---|
51 | #include "mg.h"
|
---|
52 | #include "WordData.h"
|
---|
53 |
|
---|
54 | #define MAXBITS (sizeof(unsigned long) * 8)
|
---|
55 |
|
---|
56 | #define is_power_of_two(a) ((a) != 0 && (((a) & ((a)-1)) == 0))
|
---|
57 |
|
---|
58 | #define MAX_RECALCULATIONS 100
|
---|
59 |
|
---|
60 | typedef struct DictWordData : public WordData
|
---|
61 | {
|
---|
62 | float saving;
|
---|
63 | float char_bit_cost;
|
---|
64 | u_long num_trans;
|
---|
65 | u_char *word;
|
---|
66 | }
|
---|
67 | DictWordData;
|
---|
68 |
|
---|
69 | static DictWordData *Words[2];
|
---|
70 | static u_long Num[2];
|
---|
71 | static u_long chars[2];
|
---|
72 |
|
---|
73 | #define KIND(p) (((p) >= Words[0] && (p) < Words[0] + Num[0]) ? 0 : 1)
|
---|
74 | #define IsWord(p) (((p) >= Words[1] && (p) < Words[1] + Num[1]) ? 1 : 0)
|
---|
75 | #define IsNonWord(p) (((p) >= Words[0] && (p) < Words[0] + Num[0]) ? 1 : 0)
|
---|
76 |
|
---|
77 |
|
---|
78 | typedef struct DictData
|
---|
79 | {
|
---|
80 | DictWordData **wd;
|
---|
81 | u_long num_wds;
|
---|
82 | u_long chars;
|
---|
83 | }
|
---|
84 | DictData;
|
---|
85 |
|
---|
86 | typedef DictData DictInfo[2];
|
---|
87 |
|
---|
88 | static DictInfo keep, discard, all;
|
---|
89 | static compression_stats_header csh;
|
---|
90 |
|
---|
91 | static char *file_name = "";
|
---|
92 | static u_long novel_method = MG_NOVEL_HUFFMAN_CHARS;
|
---|
93 |
|
---|
94 |
|
---|
95 | static void ReadInWords (char *);
|
---|
96 | static void Select_on (int k, heap_comp hc);
|
---|
97 | static void Method3 (int k);
|
---|
98 | static void Select_all (void);
|
---|
99 | static u_long WriteOutWords (char *, u_long, int);
|
---|
100 | static int DecFreqIncWL (void *a, void *b);
|
---|
101 | static int OccuranceOrder (void *a, void *b);
|
---|
102 |
|
---|
103 |
|
---|
104 |
|
---|
105 | int main (int argc, char **argv)
|
---|
106 | {
|
---|
107 | int ch;
|
---|
108 | char type = 'C';
|
---|
109 | char mode = '0';
|
---|
110 | int lookback = 2;
|
---|
111 | double k = 0;
|
---|
112 | u_long mem_reqd;
|
---|
113 | opterr = 0;
|
---|
114 | msg_prefix = argv[0];
|
---|
115 | while ((ch = getopt (argc, argv, "0123CPSf:d:l:hk:HDY")) != -1)
|
---|
116 | switch (ch)
|
---|
117 | {
|
---|
118 | case 'H':
|
---|
119 | novel_method = MG_NOVEL_HUFFMAN_CHARS;
|
---|
120 | break;
|
---|
121 | case 'D':
|
---|
122 | novel_method = MG_NOVEL_DELTA;
|
---|
123 | break;
|
---|
124 | case 'Y':
|
---|
125 | novel_method = MG_NOVEL_HYBRID;
|
---|
126 | break;
|
---|
127 | case 'f': /* input file */
|
---|
128 | file_name = optarg;
|
---|
129 | break;
|
---|
130 | case 'd':
|
---|
131 | set_basepath (optarg);
|
---|
132 | break;
|
---|
133 | case 'C':
|
---|
134 | case 'P':
|
---|
135 | case 'S':
|
---|
136 | type = ch;
|
---|
137 | break;
|
---|
138 | case '0':
|
---|
139 | case '1':
|
---|
140 | case '2':
|
---|
141 | case '3':
|
---|
142 | mode = ch;
|
---|
143 | break;
|
---|
144 | case 'l':
|
---|
145 | lookback = atoi (optarg);
|
---|
146 | if (!is_power_of_two (lookback))
|
---|
147 | FatalError (1, "The lookback value must be a power of 2");
|
---|
148 | lookback = floorlog_2 (lookback);
|
---|
149 | break;
|
---|
150 | case 'k':
|
---|
151 | k = atof (optarg) * 1024;
|
---|
152 | break;
|
---|
153 | case 'h':
|
---|
154 | case '?':
|
---|
155 | fprintf (stderr, "usage: %s [-l lookback] [-f input_file]"
|
---|
156 | "[-d data directory] [-h] [-0|-1|-2|-3] [-C|-P|-S] [-k mem (Kb)]\n",
|
---|
157 | argv[0]);
|
---|
158 | exit (1);
|
---|
159 | }
|
---|
160 |
|
---|
161 | ReadInWords (file_name);
|
---|
162 |
|
---|
163 | if (type == 'C')
|
---|
164 | {
|
---|
165 | Select_all ();
|
---|
166 | mem_reqd = WriteOutWords (file_name, MG_COMPLETE_DICTIONARY, lookback);
|
---|
167 | }
|
---|
168 | else
|
---|
169 | {
|
---|
170 | switch (mode)
|
---|
171 | {
|
---|
172 | case '0':
|
---|
173 | Select_all ();
|
---|
174 | break;
|
---|
175 | case '1':
|
---|
176 | #ifndef SILENT
|
---|
177 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
178 | #endif
|
---|
179 | Select_on ((int) k, OccuranceOrder);
|
---|
180 | break;
|
---|
181 | case '2':
|
---|
182 | #ifndef SILENT
|
---|
183 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
184 | #endif
|
---|
185 | Select_on ((int) k, DecFreqIncWL);
|
---|
186 | break;
|
---|
187 | case '3':
|
---|
188 | #ifndef SILENT
|
---|
189 | Message ("Dictionary limit of %.2f Kb", k / 1024);
|
---|
190 | #endif
|
---|
191 | Method3 ((int) k);
|
---|
192 | break;
|
---|
193 | }
|
---|
194 | if (type == 'P')
|
---|
195 | {
|
---|
196 | mem_reqd = WriteOutWords (file_name, MG_PARTIAL_DICTIONARY, lookback);
|
---|
197 | }
|
---|
198 | else
|
---|
199 | {
|
---|
200 | mem_reqd = WriteOutWords (file_name, MG_SEED_DICTIONARY, lookback);
|
---|
201 | }
|
---|
202 | }
|
---|
203 |
|
---|
204 | #ifndef SILENT
|
---|
205 | Message ("Num words : %8u -> %8u\n", Num[1], keep[1].num_wds);
|
---|
206 | Message ("Num non-words : %8u -> %8u\n", Num[0], keep[0].num_wds);
|
---|
207 | Message ("Chars of words : %8u -> %8u\n", chars[1], keep[1].chars);
|
---|
208 | Message ("Chars of non-words : %8u -> %8u\n", chars[0], keep[0].chars);
|
---|
209 | Message ("Mem usage : %8u -> %8u\n",
|
---|
210 | (Num[0] + Num[1]) * sizeof (char *) + chars[0] + chars[1],
|
---|
211 | (keep[0].num_wds + keep[1].num_wds) * sizeof (char *) +
|
---|
212 | keep[0].chars + keep[1].chars);
|
---|
213 | Message ("Actual mem required : %8u\n", mem_reqd);
|
---|
214 | #endif
|
---|
215 | return (0);
|
---|
216 | }
|
---|
217 |
|
---|
218 |
|
---|
219 |
|
---|
220 |
|
---|
221 | static void
|
---|
222 | ReadInWords (char *filename)
|
---|
223 | {
|
---|
224 | FILE *f;
|
---|
225 | unsigned long i;
|
---|
226 | f = open_file (filename, TEXT_STATS_DICT_SUFFIX, "rb",
|
---|
227 | MAGIC_STATS_DICT, MG_ABORT); /* [RPAP - Feb 97: WIN32 Port] */
|
---|
228 |
|
---|
229 | fread (&csh, sizeof (csh), 1, f);
|
---|
230 |
|
---|
231 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
232 | NTOHUL(csh.num_docs);
|
---|
233 | NTOHD(csh.num_bytes);
|
---|
234 |
|
---|
235 | for (i = 0; i < 2; i++)
|
---|
236 | {
|
---|
237 | frags_stats_header fsh;
|
---|
238 | char *buf;
|
---|
239 | DictWordData *wd;
|
---|
240 | chars[i] = 0;
|
---|
241 |
|
---|
242 | fread (&fsh, sizeof (fsh), 1, f);
|
---|
243 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
244 | NTOHUL(fsh.num_frags);
|
---|
245 | NTOHUL(fsh.mem_for_frags);
|
---|
246 |
|
---|
247 | Num[i] = all[i].num_wds = fsh.num_frags;
|
---|
248 |
|
---|
249 | /* The +1 on the following line is to leave room for the esc code. */
|
---|
250 | all[i].wd = (DictWordData **) Xmalloc (sizeof (DictWordData *) * Num[i] + 1);
|
---|
251 |
|
---|
252 | buf = (char *) Xmalloc (fsh.mem_for_frags);
|
---|
253 | wd = Words[i] = (DictWordData *) Xmalloc (sizeof (DictWordData) * Num[i]);
|
---|
254 | unsigned int j;
|
---|
255 | for (j = 0; j < Num[i]; j++, wd++)
|
---|
256 | {
|
---|
257 | int len;
|
---|
258 |
|
---|
259 | // read docCount and wordCount
|
---|
260 | wd->read(f);
|
---|
261 |
|
---|
262 | // read an mgString
|
---|
263 | len = fgetc (f);
|
---|
264 | wd->word = (u_char *) buf;
|
---|
265 | *buf++ = len;
|
---|
266 | fread (buf, len, 1, f);
|
---|
267 | buf += len;
|
---|
268 | all[i].wd[j] = wd;
|
---|
269 | }
|
---|
270 | chars[i] = fsh.mem_for_frags - fsh.num_frags;
|
---|
271 | }
|
---|
272 | fclose (f);
|
---|
273 | }
|
---|
274 |
|
---|
275 |
|
---|
276 | static void
|
---|
277 | Alloc_keep_discard (void)
|
---|
278 | {
|
---|
279 | keep[0].num_wds = 0;
|
---|
280 | keep[0].wd = (DictWordData **) Xmalloc ((Num[0] + 1) * sizeof (DictWordData *));
|
---|
281 | keep[1].num_wds = 0;
|
---|
282 | keep[1].wd = (DictWordData **) Xmalloc ((Num[1] + 1) * sizeof (DictWordData *));
|
---|
283 | discard[0].num_wds = 0;
|
---|
284 | discard[0].wd = (DictWordData **) Xmalloc ((Num[0] + 1) * sizeof (DictWordData *));
|
---|
285 | discard[1].num_wds = 0;
|
---|
286 | discard[1].wd = (DictWordData **) Xmalloc ((Num[1] + 1) * sizeof (DictWordData *));
|
---|
287 | }
|
---|
288 |
|
---|
289 |
|
---|
290 | static int
|
---|
291 | sort_comp (const void *a, const void *b)
|
---|
292 | {
|
---|
293 | DictWordData *aa = *(DictWordData **) a;
|
---|
294 | DictWordData *bb = *(DictWordData **) b;
|
---|
295 | return casecompare (aa->word, bb->word); /* [RPAP - Jan 97: Stem Index Change] */
|
---|
296 | }
|
---|
297 |
|
---|
298 |
|
---|
299 |
|
---|
300 | static void
|
---|
301 | SortAndCount_DictData (DictData * dd)
|
---|
302 | {
|
---|
303 | unsigned int i;
|
---|
304 | DictWordData **wd;
|
---|
305 | qsort (dd->wd, dd->num_wds, sizeof (DictWordData *), sort_comp);
|
---|
306 | dd->chars = 0;
|
---|
307 | wd = dd->wd;
|
---|
308 | for (i = 0; i < dd->num_wds; i++, wd++)
|
---|
309 | dd->chars += (*wd)->word[0];
|
---|
310 | }
|
---|
311 |
|
---|
312 |
|
---|
313 | static void
|
---|
314 | Select_all (void)
|
---|
315 | {
|
---|
316 | unsigned int i;
|
---|
317 | Alloc_keep_discard ();
|
---|
318 | keep[0].num_wds = Num[0];
|
---|
319 | for (i = 0; i < Num[0]; i++)
|
---|
320 | keep[0].wd[i] = Words[0] + i;
|
---|
321 | keep[1].num_wds = Num[1];
|
---|
322 | for (i = 0; i < Num[1]; i++)
|
---|
323 | keep[1].wd[i] = Words[1] + i;
|
---|
324 | SortAndCount_DictData (&keep[0]);
|
---|
325 | SortAndCount_DictData (&keep[1]);
|
---|
326 | }
|
---|
327 |
|
---|
328 |
|
---|
329 |
|
---|
330 |
|
---|
331 | static int
|
---|
332 | DecFreqIncWL (void *a, void *b)
|
---|
333 | {
|
---|
334 | DictWordData *aa = *(DictWordData **) a;
|
---|
335 | DictWordData *bb = *(DictWordData **) b;
|
---|
336 | if (aa->docCount > bb->docCount)
|
---|
337 | return -1;
|
---|
338 | else if (aa->docCount < bb->docCount)
|
---|
339 | return 1;
|
---|
340 | else
|
---|
341 | return bb->word[0] - aa->word[0];
|
---|
342 | }
|
---|
343 |
|
---|
344 |
|
---|
345 | static int
|
---|
346 | OccuranceOrder (void *a, void *b)
|
---|
347 | {
|
---|
348 | DictWordData *aa = *(DictWordData **) a;
|
---|
349 | DictWordData *bb = *(DictWordData **) b;
|
---|
350 | if (aa->words() > bb->words())
|
---|
351 | return 1;
|
---|
352 | else if (aa->words() < bb->words())
|
---|
353 | return -1;
|
---|
354 | else
|
---|
355 | return 0;
|
---|
356 | }
|
---|
357 |
|
---|
358 |
|
---|
359 | static void
|
---|
360 | Select_on (int k, heap_comp hc)
|
---|
361 | {
|
---|
362 | int i, num, mem;
|
---|
363 | DictWordData **wd;
|
---|
364 |
|
---|
365 | Alloc_keep_discard ();
|
---|
366 |
|
---|
367 | num = Num[0] + Num[1];
|
---|
368 | wd = (DictWordData **) Xmalloc (num * sizeof (DictWordData *));
|
---|
369 | for (i = 0; (unsigned int)i < Num[0]; i++)
|
---|
370 | wd[i] = Words[0] + i;
|
---|
371 | for (i = 0; (unsigned int)i < Num[1]; i++)
|
---|
372 | wd[i + Num[0]] = Words[1] + i;
|
---|
373 |
|
---|
374 | heap_build (wd, sizeof (*wd), num, hc);
|
---|
375 |
|
---|
376 | mem = 0;
|
---|
377 | while (k > mem && num)
|
---|
378 | {
|
---|
379 | int idx;
|
---|
380 | DictWordData *word = wd[0];
|
---|
381 | #ifdef DEBUG
|
---|
382 | fprintf (stderr, "%4d:%4d:%8d :%8d :%8d : \"%s\"\n",
|
---|
383 | keep[0].num_wds, keep[1].num_wds,
|
---|
384 | mem, word->documents(), word->word, word2str (word->word));
|
---|
385 | #endif
|
---|
386 | mem += sizeof (u_char *) + word->word[0];
|
---|
387 | heap_deletehead (wd, sizeof (*wd), &num, hc);
|
---|
388 | idx = KIND (word);
|
---|
389 | keep[idx].wd[keep[idx].num_wds++] = word;
|
---|
390 | }
|
---|
391 |
|
---|
392 | for (i = 0; i < num; i++)
|
---|
393 | {
|
---|
394 | DictWordData *word = wd[i];
|
---|
395 | int idx = KIND (word);
|
---|
396 | discard[idx].wd[discard[idx].num_wds++] = word;
|
---|
397 | }
|
---|
398 | SortAndCount_DictData (&keep[0]);
|
---|
399 | SortAndCount_DictData (&keep[1]);
|
---|
400 | SortAndCount_DictData (&discard[0]);
|
---|
401 | SortAndCount_DictData (&discard[1]);
|
---|
402 |
|
---|
403 | assert (keep[0].num_wds + discard[0].num_wds == Num[0]);
|
---|
404 | assert (keep[1].num_wds + discard[1].num_wds == Num[1]);
|
---|
405 |
|
---|
406 | }
|
---|
407 |
|
---|
408 |
|
---|
409 |
|
---|
410 | static int
|
---|
411 | BigSaving (void *a, void *b)
|
---|
412 | {
|
---|
413 | DictWordData *aa = *(DictWordData **) a;
|
---|
414 | DictWordData *bb = *(DictWordData **) b;
|
---|
415 | return (aa->saving > bb->saving) ? -1 : (aa->saving < bb->saving);
|
---|
416 | }
|
---|
417 |
|
---|
418 | static int
|
---|
419 | SmallSaving (void *a, void *b)
|
---|
420 | {
|
---|
421 | DictWordData *aa = *(DictWordData **) a;
|
---|
422 | DictWordData *bb = *(DictWordData **) b;
|
---|
423 | return (aa->saving < bb->saving) ? -1 : (aa->saving > bb->saving);
|
---|
424 | }
|
---|
425 |
|
---|
426 |
|
---|
427 | static void
|
---|
428 | CalcCharCounts (DictWordData ** wd, int num,
|
---|
429 | char *char_lens[2], char *len_lens[2],
|
---|
430 | u_long escape[2])
|
---|
431 | {
|
---|
432 | long char_freqs[2][256];
|
---|
433 | long len_freqs[2][16];
|
---|
434 | huff_data hd;
|
---|
435 | int i;
|
---|
436 | escape[0] = 0;
|
---|
437 | escape[1] = 0;
|
---|
438 | memset (char_freqs, '\0', sizeof (char_freqs));
|
---|
439 | memset (len_freqs, '\0', sizeof (len_freqs));
|
---|
440 | for (i = 0; i < num; i++, wd++)
|
---|
441 | {
|
---|
442 | u_long freq = (*wd)->documents();
|
---|
443 | u_char *buf = (*wd)->word;
|
---|
444 | int len = *buf++;
|
---|
445 | int idx = KIND (*wd);
|
---|
446 | len_freqs[idx][len] += freq;
|
---|
447 | escape[idx] += freq;
|
---|
448 | for (; len; len--, buf++)
|
---|
449 | char_freqs[idx][(u_long) (*buf)] += freq;
|
---|
450 | }
|
---|
451 | Generate_Huffman_Data (256, char_freqs[0], &hd, NULL);
|
---|
452 | char_lens[0] = hd.clens;
|
---|
453 | Generate_Huffman_Data (256, char_freqs[1], &hd, NULL);
|
---|
454 | char_lens[1] = hd.clens;
|
---|
455 |
|
---|
456 | Generate_Huffman_Data (16, len_freqs[0], &hd, NULL);
|
---|
457 | len_lens[0] = hd.clens;
|
---|
458 | Generate_Huffman_Data (16, len_freqs[1], &hd, NULL);
|
---|
459 | len_lens[1] = hd.clens;
|
---|
460 | }
|
---|
461 |
|
---|
462 |
|
---|
463 |
|
---|
464 |
|
---|
465 |
|
---|
466 | inline void CalcBitCostForWordData(DictWordData **word, int num,
|
---|
467 | double freqs_trans[2], u_long escape[2],
|
---|
468 | char * char_lens[2], char *len_lens[2],
|
---|
469 | double esc[2], int num_trans)
|
---|
470 | {
|
---|
471 | int j;
|
---|
472 |
|
---|
473 | for (j = 0; j < num; j++, word++)
|
---|
474 | {
|
---|
475 | float cbc, wbc;
|
---|
476 | u_char *buf = (*word)->word;
|
---|
477 | int len = *buf++;
|
---|
478 | u_long freq = (*word)->documents();
|
---|
479 | int idx = KIND (*word);
|
---|
480 |
|
---|
481 | cbc = len_lens[idx][len];
|
---|
482 | for (; len; len--, buf++)
|
---|
483 | cbc += char_lens[idx][(u_long) (*buf)];
|
---|
484 |
|
---|
485 | (*word)->char_bit_cost = (cbc + esc[idx]) * freq;
|
---|
486 |
|
---|
487 | wbc = -log2 (freq / (freqs_trans[idx] + escape[idx])) * freq;
|
---|
488 |
|
---|
489 | (*word)->saving = ((*word)->char_bit_cost - wbc) /
|
---|
490 | (sizeof (char *) + (*word)->word[0]);
|
---|
491 |
|
---|
492 | (*word)->num_trans = num_trans;
|
---|
493 | }
|
---|
494 | }
|
---|
495 |
|
---|
496 |
|
---|
497 |
|
---|
498 | void
|
---|
499 | CalcBitCost (DictWordData ** discard_word, int discard_num,
|
---|
500 | DictWordData ** keep_word, int keep_num, double freqs_trans[2],
|
---|
501 | u_long escape[2], int num_trans)
|
---|
502 | {
|
---|
503 | char *char_lens[2];
|
---|
504 | char *len_lens[2];
|
---|
505 | double esc[2];
|
---|
506 | CalcCharCounts (discard_word, discard_num, char_lens, len_lens, escape);
|
---|
507 | esc[0] = -log2 (escape[0] / (freqs_trans[0] + escape[0]));
|
---|
508 | esc[1] = -log2 (escape[1] / (freqs_trans[1] + escape[1]));
|
---|
509 | CalcBitCostForWordData(discard_word, discard_num, freqs_trans, escape,
|
---|
510 | char_lens, len_lens, esc, num_trans);
|
---|
511 | CalcBitCostForWordData(keep_word, keep_num, freqs_trans, escape,
|
---|
512 | char_lens, len_lens, esc, num_trans);
|
---|
513 | Xfree (char_lens[0]);
|
---|
514 | Xfree (char_lens[1]);
|
---|
515 | Xfree (len_lens[0]);
|
---|
516 | Xfree (len_lens[1]);
|
---|
517 | }
|
---|
518 |
|
---|
519 | inline void m3_transferWord(DictWordData **toHeap, int *toNum, heap_comp toSaving,
|
---|
520 | DictWordData **fromHeap, int *fromNum, heap_comp fromSaving,
|
---|
521 | int *num_trans, double freqs_trans[], int *mem, double *total,
|
---|
522 | int *count)
|
---|
523 | {
|
---|
524 | /* Transfer the word at the top of the keep heap to the
|
---|
525 | discard heap. */
|
---|
526 | DictWordData *word = fromHeap[0];
|
---|
527 | int idx = KIND (word);
|
---|
528 | heap_deletehead (fromHeap, sizeof (word), fromNum, fromSaving);
|
---|
529 | toHeap[*toNum] = word;
|
---|
530 | heap_additem (toHeap, sizeof (word), toNum, toSaving);
|
---|
531 | freqs_trans[idx] -= word->documents();
|
---|
532 | *mem = (*mem) - sizeof (u_char *) + word->word[0];
|
---|
533 | *num_trans += 1;
|
---|
534 | *total += word->saving;
|
---|
535 | *count += 1;
|
---|
536 | #ifdef DEBUG
|
---|
537 | fprintf (stderr,
|
---|
538 | "KEEP -> DISCARD %8d :%8d :%8.0f :%8.0f : \"%s\"\n",
|
---|
539 | *mem, word->documents(), word->char_bit_cost,
|
---|
540 | word->saving, word2str (word->word));
|
---|
541 | #endif
|
---|
542 | }
|
---|
543 |
|
---|
544 | inline void m3_storeWord(DictWordData **wordHeap, int wordNum, int num_trans,
|
---|
545 | double freqs_trans[], u_long escape[], heap_comp hc)
|
---|
546 | {
|
---|
547 | DictWordData *word = wordHeap[0];
|
---|
548 | int idx = KIND (word);
|
---|
549 | float wbc;
|
---|
550 | #ifdef DEBUG1
|
---|
551 | fprintf (stderr, "KEEP \"%s\" %.2f ->", word2str (word->word),
|
---|
552 | word->saving);
|
---|
553 | #endif
|
---|
554 | wbc = -log2 (word->documents() / (freqs_trans[idx] + escape[idx])) *
|
---|
555 | word->documents();
|
---|
556 | word->saving = (word->char_bit_cost - wbc) /
|
---|
557 | (sizeof (char *) + word->word[0]);
|
---|
558 | #ifdef DEBUG1
|
---|
559 | fprintf (stderr, " %.2f\n", word->saving);
|
---|
560 | #endif
|
---|
561 | word->num_trans = num_trans;
|
---|
562 | heap_changedhead (wordHeap, sizeof (word), wordNum, hc);
|
---|
563 | }
|
---|
564 |
|
---|
565 |
|
---|
566 | static void
|
---|
567 | Method3 (int k)
|
---|
568 | {
|
---|
569 | int i, keep_num, discard_num, mem, num_trans, recalc_reqd;
|
---|
570 | int keep_to_discard = 0;
|
---|
571 | int discard_to_keep = 0;
|
---|
572 | int recalcs = 0;
|
---|
573 | double freqs_trans[2], total;
|
---|
574 | u_long escape[2];
|
---|
575 | DictWordData **keep_heap, **discard_heap;
|
---|
576 |
|
---|
577 | freqs_trans[0] = freqs_trans[1] = 0;
|
---|
578 | num_trans = 0;
|
---|
579 |
|
---|
580 | discard_num = Num[0] + Num[1];
|
---|
581 | discard_heap = (DictWordData **) Xmalloc (discard_num * sizeof (DictWordData *));
|
---|
582 |
|
---|
583 | keep_num = 0;
|
---|
584 | keep_heap = (DictWordData **) Xmalloc (discard_num * sizeof (DictWordData *));
|
---|
585 |
|
---|
586 |
|
---|
587 | for (i = 0; (unsigned int)i < Num[0]; i++)
|
---|
588 | discard_heap[i] = Words[0] + i;
|
---|
589 | for (i = 0; (unsigned int)i < Num[1]; i++)
|
---|
590 | discard_heap[i + Num[0]] = Words[1] + i;
|
---|
591 |
|
---|
592 | heap_build (discard_heap, sizeof (*discard_heap), discard_num, DecFreqIncWL);
|
---|
593 |
|
---|
594 | mem = 0;
|
---|
595 | while (k > mem && discard_num)
|
---|
596 | {
|
---|
597 | DictWordData *word = discard_heap[0];
|
---|
598 | mem += sizeof (u_char *) + word->word[0];
|
---|
599 | heap_deletehead (discard_heap, sizeof (word), &discard_num, DecFreqIncWL);
|
---|
600 | keep_heap[keep_num++] = word;
|
---|
601 | freqs_trans[KIND (word)] += word->documents();
|
---|
602 | num_trans++;
|
---|
603 | }
|
---|
604 |
|
---|
605 | CalcBitCost (discard_heap, discard_num, keep_heap, keep_num,
|
---|
606 | freqs_trans, escape, num_trans);
|
---|
607 | heap_build (discard_heap, sizeof (*discard_heap), discard_num, BigSaving);
|
---|
608 | heap_build (keep_heap, sizeof (*keep_heap), keep_num, SmallSaving);
|
---|
609 |
|
---|
610 | total = 0;
|
---|
611 | recalc_reqd = 0;
|
---|
612 | while (keep_num && discard_num)
|
---|
613 | {
|
---|
614 | if ((keep_num && keep_heap[0]->num_trans < (unsigned)num_trans) ||
|
---|
615 | (discard_num && discard_heap[0]->num_trans < (unsigned)num_trans))
|
---|
616 | {
|
---|
617 | if (keep_num && keep_heap[0]->num_trans < (unsigned)num_trans)
|
---|
618 | {
|
---|
619 | m3_storeWord(keep_heap, keep_num, num_trans, freqs_trans, escape, SmallSaving);
|
---|
620 | }
|
---|
621 |
|
---|
622 | if (discard_num && discard_heap[0]->num_trans < (unsigned)num_trans)
|
---|
623 | {
|
---|
624 | m3_storeWord(discard_heap, discard_num, num_trans, freqs_trans, escape, BigSaving);
|
---|
625 | }
|
---|
626 | }
|
---|
627 | else if (keep_heap[0]->saving < discard_heap[0]->saving)
|
---|
628 | {
|
---|
629 | assert (keep_num && discard_num);
|
---|
630 | if (keep_num && mem + sizeof (char *) + discard_heap[0]->word[0] > (unsigned)k)
|
---|
631 | {
|
---|
632 | /* Transfer the word at the top of the keep heap to the
|
---|
633 | discard heap. */
|
---|
634 | m3_transferWord(discard_heap, &discard_num, BigSaving,
|
---|
635 | keep_heap, &keep_num, SmallSaving,
|
---|
636 | &num_trans, freqs_trans, &mem, &total, &keep_to_discard);
|
---|
637 | }
|
---|
638 | else
|
---|
639 | {
|
---|
640 | /* Transfer the word at the top of the discard heap to the
|
---|
641 | keep heap. */
|
---|
642 | m3_transferWord(keep_heap, &keep_num, SmallSaving,
|
---|
643 | discard_heap, &discard_num, BigSaving,
|
---|
644 | &num_trans, freqs_trans, &mem, &total, &discard_to_keep);
|
---|
645 | }
|
---|
646 |
|
---|
647 | recalc_reqd = 1;
|
---|
648 |
|
---|
649 | }
|
---|
650 | else
|
---|
651 | {
|
---|
652 | if (recalc_reqd == 0)
|
---|
653 | break;
|
---|
654 | #ifdef DEBUG1
|
---|
655 | fprintf (stderr, "--------------\n");
|
---|
656 | #endif
|
---|
657 | if (recalcs == MAX_RECALCULATIONS)
|
---|
658 | break;
|
---|
659 | CalcBitCost (discard_heap, discard_num, keep_heap, keep_num,
|
---|
660 | freqs_trans, escape, num_trans);
|
---|
661 | heap_build (discard_heap, sizeof (*discard_heap),
|
---|
662 | discard_num, BigSaving);
|
---|
663 | heap_build (keep_heap, sizeof (keep_heap), keep_num, SmallSaving);
|
---|
664 | recalc_reqd = 0;
|
---|
665 | recalcs++;
|
---|
666 | }
|
---|
667 | }
|
---|
668 |
|
---|
669 | Alloc_keep_discard ();
|
---|
670 |
|
---|
671 | for (i = 0; i < discard_num; i++)
|
---|
672 | {
|
---|
673 | DictWordData *word = discard_heap[i];
|
---|
674 | int idx = KIND (word);
|
---|
675 | assert (IsWord (word) || IsNonWord (word));
|
---|
676 | discard[idx].wd[discard[idx].num_wds++] = word;
|
---|
677 | }
|
---|
678 | for (i = 0; i < keep_num; i++)
|
---|
679 | {
|
---|
680 | DictWordData *word = keep_heap[i];
|
---|
681 | int idx = KIND (word);
|
---|
682 | assert (IsWord (word) || IsNonWord (word));
|
---|
683 | keep[idx].wd[keep[idx].num_wds++] = word;
|
---|
684 | }
|
---|
685 | SortAndCount_DictData (&keep[0]);
|
---|
686 | SortAndCount_DictData (&keep[1]);
|
---|
687 | SortAndCount_DictData (&discard[0]);
|
---|
688 | SortAndCount_DictData (&discard[1]);
|
---|
689 |
|
---|
690 | assert (keep[0].num_wds + discard[0].num_wds == Num[0]);
|
---|
691 | assert (keep[1].num_wds + discard[1].num_wds == Num[1]);
|
---|
692 | #ifndef SILENT
|
---|
693 | Message ("Keep -> Discard : %8d", keep_to_discard);
|
---|
694 | Message ("Discard -> Keep : %8d", discard_to_keep);
|
---|
695 | Message ("Huffman Recalculations : %8d", recalcs);
|
---|
696 | if (recalcs == MAX_RECALCULATIONS)
|
---|
697 | Message ("WARNING: The number of recalculations == %d", MAX_RECALCULATIONS);
|
---|
698 | #endif
|
---|
699 | }
|
---|
700 |
|
---|
701 |
|
---|
702 |
|
---|
703 |
|
---|
704 |
|
---|
705 |
|
---|
706 |
|
---|
707 |
|
---|
708 |
|
---|
709 |
|
---|
710 | /****************************************************************************
|
---|
711 | ***** *****
|
---|
712 | ***** Dictionary saving code *****
|
---|
713 | ***** *****
|
---|
714 | ****************************************************************************/
|
---|
715 |
|
---|
716 |
|
---|
717 |
|
---|
718 | static void
|
---|
719 | Write_cdh (FILE * f, compression_dict_header * cdh)
|
---|
720 | {
|
---|
721 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
722 | int i;
|
---|
723 | compression_dict_header tmp = *cdh;
|
---|
724 | HTONUL(tmp.dict_type);
|
---|
725 | HTONUL(tmp.novel_method);
|
---|
726 | for (i = 0; i < TEXT_PARAMS; i++)
|
---|
727 | HTONUL(tmp.params[i]);
|
---|
728 | HTONUL(tmp.num_words[0]);
|
---|
729 | HTONUL(tmp.num_words[1]);
|
---|
730 | HTONUL(tmp.num_word_chars[0]);
|
---|
731 | HTONUL(tmp.num_word_chars[1]);
|
---|
732 | HTONUL(tmp.lookback);
|
---|
733 |
|
---|
734 | fwrite (&tmp, sizeof (tmp), 1, f);
|
---|
735 | }
|
---|
736 |
|
---|
737 |
|
---|
738 | static void
|
---|
739 | Write_words (FILE * f, DictData * dd)
|
---|
740 | {
|
---|
741 | unsigned int i;
|
---|
742 | u_char *curr, *prev = NULL;
|
---|
743 | for (i = 0; i < dd->num_wds; i++)
|
---|
744 | {
|
---|
745 | int len;
|
---|
746 | curr = dd->wd[i]->word;
|
---|
747 | if (prev)
|
---|
748 | /* look for prefix match with prev string */
|
---|
749 | len = prefixlen (prev, curr);
|
---|
750 | else
|
---|
751 | len = 0;
|
---|
752 | fputc ((len << 4) + (curr[0] - len), f);
|
---|
753 | fwrite (curr + len + 1, sizeof (u_char), curr[0] - len, f);
|
---|
754 | prev = curr;
|
---|
755 | }
|
---|
756 |
|
---|
757 | }
|
---|
758 |
|
---|
759 |
|
---|
760 | static int
|
---|
761 | Uncompressed_size (DictData * dd)
|
---|
762 | {
|
---|
763 | unsigned int i, us;
|
---|
764 | for (us = i = 0; i < dd->num_wds; i++)
|
---|
765 | us += dd->wd[i]->word[0];
|
---|
766 | return us;
|
---|
767 | }
|
---|
768 |
|
---|
769 |
|
---|
770 | static u_long
|
---|
771 | Write_data (FILE * f, DictData * dd, int lookback)
|
---|
772 | {
|
---|
773 | u_long mem_reqd;
|
---|
774 | huff_data *hd;
|
---|
775 | int i;
|
---|
776 | u_long us = dd->num_wds;
|
---|
777 | long *freqs;
|
---|
778 | u_long huff_words_size[MAX_HUFFCODE_LEN + 1];
|
---|
779 | u_long lencounts[MAX_HUFFCODE_LEN + 1];
|
---|
780 | u_char *lastword[MAX_HUFFCODE_LEN + 1];
|
---|
781 |
|
---|
782 | if (!(freqs = new long [dd->num_wds]))
|
---|
783 | FatalError (1, "Unable to allocate memory for freqs");
|
---|
784 |
|
---|
785 | for (i = 0; (unsigned)i < dd->num_wds; i++)
|
---|
786 | {
|
---|
787 | freqs[i] = dd->wd[i]->documents();
|
---|
788 | us += dd->wd[i]->word[0];
|
---|
789 | }
|
---|
790 |
|
---|
791 | if (!(hd = Generate_Huffman_Data (dd->num_wds, freqs, NULL, NULL)))
|
---|
792 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
793 |
|
---|
794 | delete (freqs);
|
---|
795 | freqs = NULL;
|
---|
796 |
|
---|
797 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
798 | FatalError (1, "Unable to write huffman data");
|
---|
799 |
|
---|
800 | HTONUL(us); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
801 | fwrite (&us, sizeof (us), 1, f);
|
---|
802 | NTOHUL(us); /* [RPAP - Jan 97: Endian Ordering] */
|
---|
803 |
|
---|
804 |
|
---|
805 | /* Calculate the amount of memory that will be required to store the text for
|
---|
806 | each different huffman code len. Every 1<<lookback words for each different
|
---|
807 | codelen length will not be prefixed by previous strings. */
|
---|
808 |
|
---|
809 |
|
---|
810 | memset (&huff_words_size, '\0', sizeof (huff_words_size));
|
---|
811 | memset (&lencounts, '\0', sizeof (lencounts));
|
---|
812 |
|
---|
813 | mem_reqd = 0;
|
---|
814 |
|
---|
815 | for (i = 0; (unsigned)i < dd->num_wds; i++)
|
---|
816 | {
|
---|
817 | int codelen = hd->clens[i];
|
---|
818 | u_char *word = dd->wd[i]->word;
|
---|
819 |
|
---|
820 | if (!codelen)
|
---|
821 | FatalError (1, "The length of a code for a word was zero");
|
---|
822 |
|
---|
823 | huff_words_size[codelen] += word[0] + 1;
|
---|
824 | mem_reqd += word[0] + (lookback != 0);
|
---|
825 | #if 0
|
---|
826 | if ((lencounts[codelen] & ((1 << lookback) - 1)) == 0)
|
---|
827 | lastword[codelen] = word;
|
---|
828 | else
|
---|
829 | huff_words_size[codelen] -= prefixlen (lastword[codelen], word);
|
---|
830 | #else
|
---|
831 | if ((lencounts[codelen] & ((1 << lookback) - 1)) != 0)
|
---|
832 | {
|
---|
833 | int save = prefixlen (lastword[codelen], word);
|
---|
834 | mem_reqd -= save;
|
---|
835 | huff_words_size[codelen] -= save;
|
---|
836 | }
|
---|
837 | else
|
---|
838 | {
|
---|
839 | mem_reqd += sizeof (u_char *);
|
---|
840 | }
|
---|
841 | lastword[codelen] = word;
|
---|
842 | #endif
|
---|
843 | lencounts[codelen]++;
|
---|
844 | }
|
---|
845 |
|
---|
846 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
847 | for (i = hd->mincodelen; i < hd->maxcodelen + 1; i++)
|
---|
848 | HTONUL(huff_words_size[i]);
|
---|
849 |
|
---|
850 | fwrite (huff_words_size + hd->mincodelen, sizeof (*huff_words_size),
|
---|
851 | hd->maxcodelen - hd->mincodelen + 1, f);
|
---|
852 |
|
---|
853 | /* [RPAP - Jan 97: Endian Ordering] */
|
---|
854 | for (i = hd->mincodelen; i < hd->maxcodelen + 1; i++)
|
---|
855 | NTOHUL(huff_words_size[i]);
|
---|
856 |
|
---|
857 | Write_words (f, dd);
|
---|
858 |
|
---|
859 | delete hd->clens;
|
---|
860 | delete hd;
|
---|
861 |
|
---|
862 | return mem_reqd;
|
---|
863 | }
|
---|
864 |
|
---|
865 |
|
---|
866 |
|
---|
867 | static void
|
---|
868 | Write_charfreqs (FILE * f, DictData * dd, int words,
|
---|
869 | int zero_freq_permitted)
|
---|
870 | {
|
---|
871 | unsigned int j;
|
---|
872 | long freqs[256];
|
---|
873 | DictWordData **wd = dd->wd;
|
---|
874 | huff_data *hd;
|
---|
875 |
|
---|
876 | memset (freqs, '\0', sizeof (freqs));
|
---|
877 |
|
---|
878 | for (j = 0; j < dd->num_wds; j++, wd++)
|
---|
879 | {
|
---|
880 | u_char *buf = (*wd)->word;
|
---|
881 | int len = *buf++;
|
---|
882 | for (; len; len--, buf++)
|
---|
883 | freqs[(u_long) (*buf)] += (*wd)->documents();
|
---|
884 | }
|
---|
885 |
|
---|
886 | if (!zero_freq_permitted)
|
---|
887 | for (j = 0; j < 256; j++)
|
---|
888 | if (!freqs[j] && PESINAWORD (j) == words)
|
---|
889 | freqs[j] = 1;
|
---|
890 |
|
---|
891 | if (!(hd = Generate_Huffman_Data (256, freqs, NULL, NULL)))
|
---|
892 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
893 |
|
---|
894 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
895 | FatalError (1, "Unable to write huffman data");
|
---|
896 |
|
---|
897 | delete hd->clens;
|
---|
898 | delete hd;
|
---|
899 | }
|
---|
900 |
|
---|
901 |
|
---|
902 |
|
---|
903 |
|
---|
904 | static void
|
---|
905 | Write_wordlens (FILE * f, DictData * dd, int zero_freq_permitted)
|
---|
906 | {
|
---|
907 | unsigned int j;
|
---|
908 | long freqs[16];
|
---|
909 | DictWordData **wd = dd->wd;
|
---|
910 | huff_data *hd;
|
---|
911 |
|
---|
912 | memset (freqs, '\0', sizeof (freqs));
|
---|
913 |
|
---|
914 | for (j = 0; j < dd->num_wds; j++, wd++)
|
---|
915 | freqs[(*wd)->word[0]] += (*wd)->documents();
|
---|
916 |
|
---|
917 | if (!zero_freq_permitted)
|
---|
918 | for (j = 0; j < 16; j++)
|
---|
919 | if (!freqs[j])
|
---|
920 | freqs[j] = 1;
|
---|
921 |
|
---|
922 | if (!(hd = Generate_Huffman_Data (16, freqs, NULL, NULL)))
|
---|
923 | FatalError (1, "Unable to allocate memory for huffman data");
|
---|
924 |
|
---|
925 | if (Write_Huffman_Data (f, hd) == -1)
|
---|
926 | FatalError (1, "Unable to write huffman data");
|
---|
927 |
|
---|
928 |
|
---|
929 | delete hd->clens;
|
---|
930 | delete hd;
|
---|
931 | }
|
---|
932 |
|
---|
933 |
|
---|
934 |
|
---|
935 | static u_long
|
---|
936 | WriteOutWords (char *file_name, u_long type, int lookback)
|
---|
937 | {
|
---|
938 | FILE *f;
|
---|
939 | int i;
|
---|
940 | u_long mem_reqd = 0;
|
---|
941 |
|
---|
942 | compression_dict_header cdh;
|
---|
943 | f = create_file (file_name, TEXT_DICT_SUFFIX, "w+b",
|
---|
944 | MAGIC_DICT, MG_ABORT); /* [RPAP - Feb 97: WIN32 Port] */
|
---|
945 |
|
---|
946 | memset (&cdh, '\0', sizeof(cdh));
|
---|
947 |
|
---|
948 | cdh.dict_type = type;
|
---|
949 | cdh.novel_method = (type == MG_SEED_DICTIONARY) ? novel_method :
|
---|
950 | MG_NOVEL_HUFFMAN_CHARS;
|
---|
951 |
|
---|
952 | cdh.num_words[1] = keep[1].num_wds;
|
---|
953 | cdh.num_words[0] = keep[0].num_wds;
|
---|
954 | cdh.num_word_chars[1] = Uncompressed_size (&keep[1]);
|
---|
955 | cdh.num_word_chars[0] = Uncompressed_size (&keep[0]);
|
---|
956 | cdh.lookback = lookback;
|
---|
957 |
|
---|
958 | Write_cdh (f, &cdh);
|
---|
959 |
|
---|
960 | for (i = 0; i < 2; i++)
|
---|
961 | switch (type)
|
---|
962 | {
|
---|
963 | case MG_COMPLETE_DICTIONARY:
|
---|
964 | {
|
---|
965 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
966 | }
|
---|
967 | break;
|
---|
968 | case MG_PARTIAL_DICTIONARY:
|
---|
969 | {
|
---|
970 | if (keep[i].num_wds)
|
---|
971 | {
|
---|
972 | int j;
|
---|
973 | DictWordData esc;
|
---|
974 | esc.docCount = 0;
|
---|
975 | esc.word = (u_char *) "";
|
---|
976 | keep[i].wd[keep[i].num_wds++] = &esc;
|
---|
977 | for (j = 0; (unsigned)j < discard[i].num_wds; j++)
|
---|
978 | esc.docCount += discard[i].wd[j]->documents();
|
---|
979 | if (!esc.docCount)
|
---|
980 | esc.docCount++;
|
---|
981 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
982 | }
|
---|
983 | Write_charfreqs (f, &discard[i], i, 1);
|
---|
984 | Write_wordlens (f, &discard[i], 1);
|
---|
985 | }
|
---|
986 | break;
|
---|
987 | case MG_SEED_DICTIONARY:
|
---|
988 | {
|
---|
989 | if (keep[i].num_wds)
|
---|
990 | {
|
---|
991 | int j;
|
---|
992 | DictWordData esc;
|
---|
993 | esc.docCount = 0;
|
---|
994 | esc.word = (u_char *) "";
|
---|
995 | keep[i].wd[keep[i].num_wds++] = &esc;
|
---|
996 | for (j = 0; (unsigned)j < all[i].num_wds; j++)
|
---|
997 | if (all[i].wd[j]->documents() == 1)
|
---|
998 | esc.docCount++;
|
---|
999 | if (!esc.docCount)
|
---|
1000 | esc.docCount++;
|
---|
1001 | mem_reqd += Write_data (f, &keep[i], lookback);
|
---|
1002 | }
|
---|
1003 | switch (novel_method)
|
---|
1004 | {
|
---|
1005 | case MG_NOVEL_HUFFMAN_CHARS:
|
---|
1006 | Write_charfreqs (f, &all[i], i, 0);
|
---|
1007 | Write_wordlens (f, &all[i], 0);
|
---|
1008 | break;
|
---|
1009 | case MG_NOVEL_DELTA:
|
---|
1010 | break;
|
---|
1011 | case MG_NOVEL_HYBRID:
|
---|
1012 | break;
|
---|
1013 | default:
|
---|
1014 | FatalError (1, "Bad novel method");
|
---|
1015 | }
|
---|
1016 | }
|
---|
1017 | break;
|
---|
1018 | }
|
---|
1019 | fclose (f);
|
---|
1020 | return mem_reqd;
|
---|
1021 | }
|
---|