1 /* This program compresses a file without losing information.
2 * The "usq" program is required to unsqueeze the file
3 * before it can be used.
4 *
5 * Typical compression rates are between 30 and 50 percent for text files.
6 *
7 * Squeezing a really big file takes a few minutes.
8 *
9 * Useage:
10 * sq [file1] [file2] ... [filen]
11 *
12 * where file1 through filen are the names of the files to be squeezed.
13 * The file type (under CP/M or MS-DOS) is changed to ".SQ"; under UN*X,
14 * ".SQ" is appended to the file name. The original file name is stored
15 * in the squeezed file.
16 *
17 * If no file name is given on the command line you will be
18 * prompted for commands (one at a time). An empty command
19 * terminates the program.
20 *
21 * The transformations compress strings of identical bytes and
22 * then encode each resulting byte value and EOF as bit strings
23 * having lengths in inverse proportion to their frequency of
24 * occurrance in the intermediate input stream. The latter uses
25 * the Huffman algorithm. Decoding information is included in
26 * the squeezed file, so squeezing short files or files with
27 * uniformly distributed byte values will actually increase size.
28 */
29
30 /* CHANGE HISTORY:
31 * 1.3 Close files properly in case of error exit.
32 * 1.4 Break up long introductory lines.
33 * 1.4 Send introduction only to console.
34 * 1.4 Send errors only to console.
35 * 1.5 Fix BUG that caused a rare few squeezed files
36 * to be incorrect and fail the USQ crc check.
37 * The problem was that some 17 bit codes were
38 * generated but are not supported by other code.
39 * THIS IS A MAJOR CHANGE affecting TR2.C and SQ.H and
40 * requires recompilation of all files which are part
41 * of SQ. Two basic changes were made: tree depth is now
42 * used as a tie breaker when weights are equal. This
43 * makes the tree shallower. Although that may always be
44 * sufficient, an error trap was added to cause rescaling
45 * of the counts if any code > 16 bits long is generated.
46 * 1.5 Add debugging displays option '-'.
47 * 1.6 Fixed to work correctly under MP/M II. Also shortened
48 * signon message.
49 * 2.0 New version for use with CI-C86 compiler (CP/M-86 and MS-DOS)
50 * 2.1 Converted for use in MLINK
51 * 2.2 Converted for use with optimizing CI-C86 compiler (MS-DOS)
52 * 3.0 Generalized for UN*X use, changed output file naming convention
53 * 3.3 Modified to work with ULTRIX, as per Tom Reid.
54 */
55
56 /* eject eject */
57
58 /*
59 * The following define MUST be set to the maximum length of a file name
60 * on the system "sq" is being compiled for. If not, "sq" will not be
61 * able to check for whether the output file name it creates is valid
62 * or not.
63 */
64
65 #ifdef TOPS20
66 #define FNM_LEN 79
67 #else
68 #define FNM_LEN 14
69 #endif
70 #define SQMAIN
71
72 #define VERSION "3.3 10/29/86"
73
74 #include <stdio.h>
75 #include "sqcom.h"
76 #include "sq.h"
77 #ifdef TOPS20
78 #include <libt20.h>
79 #include <jsys.h>
80 #endif
81
82 #define FALSE 0
83
84 #ifdef TOPS20
85 struct filename req;
86 #endif
87
main(argc,argv)88 main(argc, argv)
89 int argc;
90 char *argv[];
91 {
92 int i,c;
93 char inparg[128]; /* parameter from input */
94
95 debug = FALSE;
96 printf("File squeezer version %s (original author: R. Greenlaw)\n\n", VERSION);
97
98 /* Process the parameters in order */
99 for(i = 1; i < argc; ++i)
100 obey(argv[i]);
101
102 if(argc < 2) {
103 printf("Enter file names, one line at a time, or type <RETURN> to quit.");
104 do {
105 printf("\n*");
106 #ifdef TOPS20
107 fflush (stdout);
108 #endif
109 for(i = 0; i < 16; ++i) {
110 if((c = getchar()) == EOF)
111 c = '\n'; /* fake empty (exit) command */
112 if((inparg[i] = c) == '\n') {
113 inparg[i] = '\0';
114 break;
115 }
116 }
117 if(inparg[0] != '\0')
118 obey(inparg);
119 } while(inparg[0] != '\0');
120 }
121 }
122
123 /* eject eject */
124
obey(p)125 obey(p)
126 char *p;
127 {
128 #ifdef TOPS20
129 char *d, *dir (); /* directory junk */
130 #else
131 char *q;
132 char outfile[128]; /* output file spec. */
133 #endif
134
135 if(*p == '-') {
136 /* toggle debug option */
137 debug = !debug;
138 return;
139 }
140
141 #ifdef TOPS20
142 for (d = dir (p, 0); *d; d = dir (NULL, 0)) {
143 dofil (d);
144 }
145 return (0);
146 }
147
148 int
dofil(p)149 dofil(p)
150 char *p;
151 {
152 char outfile[128]; /* output file spec. */
153 char ftype[40];
154 char fname[40];
155 int r;
156
157 strcpy (req.name, p);
158 if ((r = fullfname(&req, "RAW")) != 0) {
159 if (r == 1) {
160 printf ("JFNS error encountered\n");
161 }
162 printf ("File, %s, not found\n", p);
163 return (-1);
164 }
165 strcpy (fname, req.nm);
166 strcpy (ftype, req.typ);
167 if (strlen (ftype) == 0) {
168 strcpy (ftype, "QQQ");
169 } else {
170 switch(*(ftype+1)) {
171 case '\0':
172 *(ftype+2) = '\0';
173 /* fall thru */
174 default:
175 *(ftype + 1) = 'Q';
176 }
177 }
178 sprintf (outfile, "%.8s.%.3s", fname, ftype);
179
180 squeeze (req.qname, outfile);
181 #else
182 /* Check for ambiguous (wild-card) name */
183 for(q = p; *q != '\0'; ++q)
184 if(*q == '*' || *q == '?') {
185 printf("\nAmbiguous name %s ignored", p);
186 return;
187 }
188 /* First build output file name */
189 strcpy(outfile, p); /* copy input name to output */
190
191 /* Find and change output file suffix */
192
193 if (strlen(outfile) + 3 > FNM_LEN) { /* check for long file name */
194 q = outfile + FNM_LEN - 3;
195 *q = '\0'; /* make room for suffix */
196 }
197 else {
198 q = outfile + strlen(outfile);
199 #ifndef UNIX
200 for(; --q >= outfile;)
201 if (*q == '.') {
202 *q = '\0'; /* delete file type */
203 break;
204 }
205 #else
206 --q;
207 #endif
208 }
209
210 strcat(outfile, ".SQ");
211
212 squeeze(p, outfile);
213 #endif
214 }
215
216 /* eject eject */
217
squeeze(infile,outfile)218 squeeze(infile, outfile)
219 char *infile, *outfile;
220 {
221 #ifdef TOPS20
222 extern int _uioch[];
223 extern int _uioufx[];
224 int jfn;
225 #endif
226 int i, c,c2;
227 FILE *inbuff, *outbuff; /* file buffers */
228
229 printf("%s -> %s: ", infile, outfile);
230 #ifdef TOPS20
231 if(!(inbuff=fopen(infile, "rC-"))) {
232 #else
233 #ifdef ULTRIX
234 if(!(inbuff=fopen(infile, "r"))) {
235 #else
236 if(!(inbuff=fopen(infile, "rb"))) {
237 #endif
238 #endif
239 printf("Can't open %s for input pass 1\n", infile);
240 return;
241 }
242 #ifdef TOPS20
243 jfn = _uioch[_uioufx[inbuff->siofd]];
244 bytsiz = (_gtfdb (jfn, _FBBYV) << FBBSZ_S) & FBBSZ_M;
245 if ((outbuff = fopen (outfile, "w8")) == NULL) {
246 #else
247 #ifdef ULTRIX
248 if(!(outbuff=fopen(outfile, "w"))) {
249 #else
250 if(!(outbuff=fopen(outfile, "wb"))) {
251 #endif
252 #endif
253 printf("Can't create %s\n", outfile);
254 fclose(inbuff);
255 return;
256 }
257
258 /* First pass - get properties of file */
259 crc = 0; /* initialize checksum */
260 printf("analyzing, ");
261 #ifdef TOPS20
262 fflush (stdout);
263 #endif
264 init_ncr();
265 init_huff(inbuff);
266 fclose(inbuff);
267
268 /* Write output file header with decoding info */
269 #ifdef TOPS20
270 wrt_head(outbuff, &req);
271 #else
272 wrt_head(outbuff, infile);
273 #endif
274
275 /* Second pass - encode the file */
276 printf("squeezing,");
277 #ifdef TOPS20
278 fflush (stdout);
279 if(!(inbuff=fopen(infile, "rC-"))) {
280 #else
281 if(!(inbuff=fopen(infile, "rb"))) {
282 #endif
283 printf("Can't open %s for input pass 2\n", infile);
284 goto closeout;
285 }
286 init_ncr(); /* For second pass */
287
288 /* Translate the input file into the output file */
289 while((c = gethuff(inbuff)) != EOF)
290 putce(c, outbuff);
291 oflush(outbuff);
292 printf(" done.\n");
293 closeall:
294 fclose(inbuff);
295 closeout:
296 fclose(outbuff);
297 }
298