1 c15ce40c 2005-08-10 devnull #include <u.h>
2 c15ce40c 2005-08-10 devnull #include <libc.h>
3 c15ce40c 2005-08-10 devnull #include <bio.h>
4 c15ce40c 2005-08-10 devnull #include <ctype.h>
5 c15ce40c 2005-08-10 devnull #include <mach.h>
8 c15ce40c 2005-08-10 devnull * file - determine type of file
10 c15ce40c 2005-08-10 devnull #define LENDIAN(p) ((p)[0] | ((p)[1]<<8) | ((p)[2]<<16) | ((p)[3]<<24))
12 c15ce40c 2005-08-10 devnull uchar buf[6001];
13 c15ce40c 2005-08-10 devnull short cfreq[140];
14 c15ce40c 2005-08-10 devnull short wfreq[50];
15 c15ce40c 2005-08-10 devnull int nbuf;
16 c15ce40c 2005-08-10 devnull Dir* mbuf;
18 c15ce40c 2005-08-10 devnull char *fname;
19 c15ce40c 2005-08-10 devnull char *slash;
31 c15ce40c 2005-08-10 devnull Clatin = 128,
39 c15ce40c 2005-08-10 devnull char* word;
40 c15ce40c 2005-08-10 devnull int class;
41 c15ce40c 2005-08-10 devnull } dict[] =
43 c15ce40c 2005-08-10 devnull "PATH", Lword,
44 c15ce40c 2005-08-10 devnull "TEXT", Aword,
45 c15ce40c 2005-08-10 devnull "adt", Alword,
46 c15ce40c 2005-08-10 devnull "aggr", Alword,
47 c15ce40c 2005-08-10 devnull "alef", Alword,
48 c15ce40c 2005-08-10 devnull "array", Lword,
49 c15ce40c 2005-08-10 devnull "block", Fword,
50 c15ce40c 2005-08-10 devnull "chan", Alword,
51 c15ce40c 2005-08-10 devnull "char", Cword,
52 c15ce40c 2005-08-10 devnull "common", Fword,
53 c15ce40c 2005-08-10 devnull "con", Lword,
54 c15ce40c 2005-08-10 devnull "data", Fword,
55 c15ce40c 2005-08-10 devnull "dimension", Fword,
56 c15ce40c 2005-08-10 devnull "double", Cword,
57 c15ce40c 2005-08-10 devnull "extern", Cword,
58 c15ce40c 2005-08-10 devnull "bio", I2,
59 c15ce40c 2005-08-10 devnull "float", Cword,
60 c15ce40c 2005-08-10 devnull "fn", Lword,
61 c15ce40c 2005-08-10 devnull "function", Fword,
63 c15ce40c 2005-08-10 devnull "implement", Lword,
64 c15ce40c 2005-08-10 devnull "import", Lword,
65 c15ce40c 2005-08-10 devnull "include", I1,
66 c15ce40c 2005-08-10 devnull "int", Cword,
67 c15ce40c 2005-08-10 devnull "integer", Fword,
68 c15ce40c 2005-08-10 devnull "iota", Lword,
69 c15ce40c 2005-08-10 devnull "libc", I2,
70 c15ce40c 2005-08-10 devnull "long", Cword,
71 c15ce40c 2005-08-10 devnull "module", Lword,
72 c15ce40c 2005-08-10 devnull "real", Fword,
73 c15ce40c 2005-08-10 devnull "ref", Lword,
74 c15ce40c 2005-08-10 devnull "register", Cword,
75 c15ce40c 2005-08-10 devnull "self", Lword,
76 c15ce40c 2005-08-10 devnull "short", Cword,
77 c15ce40c 2005-08-10 devnull "static", Cword,
78 c15ce40c 2005-08-10 devnull "stdio", I2,
79 c15ce40c 2005-08-10 devnull "struct", Cword,
80 c15ce40c 2005-08-10 devnull "subroutine", Fword,
82 c15ce40c 2005-08-10 devnull "void", Cword,
85 c15ce40c 2005-08-10 devnull /* codes for 'mode' field in language structure */
87 c15ce40c 2005-08-10 devnull Normal = 0,
88 c15ce40c 2005-08-10 devnull First, /* first entry for language spanning several ranges */
89 c15ce40c 2005-08-10 devnull Multi, /* later entries " " " ... */
90 c15ce40c 2005-08-10 devnull Shared, /* codes used in several languages */
95 c15ce40c 2005-08-10 devnull int mode; /* see enum above */
96 c15ce40c 2005-08-10 devnull int count;
98 c15ce40c 2005-08-10 devnull int high;
99 c15ce40c 2005-08-10 devnull char *name;
101 c15ce40c 2005-08-10 devnull } language[] =
103 c15ce40c 2005-08-10 devnull Normal, 0, 0x0080, 0x0080, "Extended Latin",
104 c15ce40c 2005-08-10 devnull Normal, 0, 0x0100, 0x01FF, "Extended Latin",
105 c15ce40c 2005-08-10 devnull Normal, 0, 0x0370, 0x03FF, "Greek",
106 c15ce40c 2005-08-10 devnull Normal, 0, 0x0400, 0x04FF, "Cyrillic",
107 c15ce40c 2005-08-10 devnull Normal, 0, 0x0530, 0x058F, "Armenian",
108 c15ce40c 2005-08-10 devnull Normal, 0, 0x0590, 0x05FF, "Hebrew",
109 c15ce40c 2005-08-10 devnull Normal, 0, 0x0600, 0x06FF, "Arabic",
110 c15ce40c 2005-08-10 devnull Normal, 0, 0x0900, 0x097F, "Devanagari",
111 c15ce40c 2005-08-10 devnull Normal, 0, 0x0980, 0x09FF, "Bengali",
112 c15ce40c 2005-08-10 devnull Normal, 0, 0x0A00, 0x0A7F, "Gurmukhi",
113 c15ce40c 2005-08-10 devnull Normal, 0, 0x0A80, 0x0AFF, "Gujarati",
114 c15ce40c 2005-08-10 devnull Normal, 0, 0x0B00, 0x0B7F, "Oriya",
115 c15ce40c 2005-08-10 devnull Normal, 0, 0x0B80, 0x0BFF, "Tamil",
116 c15ce40c 2005-08-10 devnull Normal, 0, 0x0C00, 0x0C7F, "Telugu",
117 c15ce40c 2005-08-10 devnull Normal, 0, 0x0C80, 0x0CFF, "Kannada",
118 c15ce40c 2005-08-10 devnull Normal, 0, 0x0D00, 0x0D7F, "Malayalam",
119 c15ce40c 2005-08-10 devnull Normal, 0, 0x0E00, 0x0E7F, "Thai",
120 c15ce40c 2005-08-10 devnull Normal, 0, 0x0E80, 0x0EFF, "Lao",
121 c15ce40c 2005-08-10 devnull Normal, 0, 0x1000, 0x105F, "Tibetan",
122 c15ce40c 2005-08-10 devnull Normal, 0, 0x10A0, 0x10FF, "Georgian",
123 c15ce40c 2005-08-10 devnull Normal, 0, 0x3040, 0x30FF, "Japanese",
124 c15ce40c 2005-08-10 devnull Normal, 0, 0x3100, 0x312F, "Chinese",
125 c15ce40c 2005-08-10 devnull First, 0, 0x3130, 0x318F, "Korean",
126 c15ce40c 2005-08-10 devnull Multi, 0, 0x3400, 0x3D2F, "Korean",
127 c15ce40c 2005-08-10 devnull Shared, 0, 0x4e00, 0x9fff, "CJK",
128 c15ce40c 2005-08-10 devnull Normal, 0, 0, 0, 0, /* terminal entry */
134 c15ce40c 2005-08-10 devnull Fascii, /* printable ascii */
135 c15ce40c 2005-08-10 devnull Flatin, /* latin 1*/
136 c15ce40c 2005-08-10 devnull Futf, /* UTf character set */
137 c15ce40c 2005-08-10 devnull Fbinary, /* binary */
138 c15ce40c 2005-08-10 devnull Feascii, /* ASCII with control chars */
139 c15ce40c 2005-08-10 devnull Fnull, /* NULL in file */
140 c15ce40c 2005-08-10 devnull } guess;
142 c15ce40c 2005-08-10 devnull void bump_utf_count(Rune);
143 c15ce40c 2005-08-10 devnull int cistrncmp(char*, char*, int);
144 c15ce40c 2005-08-10 devnull void filetype(int);
145 c15ce40c 2005-08-10 devnull int getfontnum(uchar*, uchar**);
146 c15ce40c 2005-08-10 devnull int isas(void);
147 c15ce40c 2005-08-10 devnull int isc(void);
148 c15ce40c 2005-08-10 devnull int isenglish(void);
149 c15ce40c 2005-08-10 devnull int ishp(void);
150 c15ce40c 2005-08-10 devnull int ishtml(void);
151 c15ce40c 2005-08-10 devnull int isrfc822(void);
152 c15ce40c 2005-08-10 devnull int ismbox(void);
153 c15ce40c 2005-08-10 devnull int islimbo(void);
154 c15ce40c 2005-08-10 devnull int ismung(void);
155 c15ce40c 2005-08-10 devnull int isp9bit(void);
156 c15ce40c 2005-08-10 devnull int isp9font(void);
157 c15ce40c 2005-08-10 devnull int isrtf(void);
158 c15ce40c 2005-08-10 devnull int ismsdos(void);
159 c15ce40c 2005-08-10 devnull int iself(void);
160 c15ce40c 2005-08-10 devnull int istring(void);
161 c15ce40c 2005-08-10 devnull int iff(void);
162 c15ce40c 2005-08-10 devnull int long0(void);
163 c15ce40c 2005-08-10 devnull int istar(void);
164 c15ce40c 2005-08-10 devnull int p9bitnum(uchar*);
165 c15ce40c 2005-08-10 devnull int p9subfont(uchar*);
166 c15ce40c 2005-08-10 devnull void print_utf(void);
167 c15ce40c 2005-08-10 devnull void type(char*, int);
168 c15ce40c 2005-08-10 devnull int utf_count(void);
169 c15ce40c 2005-08-10 devnull void wordfreq(void);
171 c15ce40c 2005-08-10 devnull int (*call[])(void) =
173 c15ce40c 2005-08-10 devnull long0, /* recognizable by first 4 bytes */
174 c15ce40c 2005-08-10 devnull istring, /* recognizable by first string */
175 c15ce40c 2005-08-10 devnull iff, /* interchange file format (strings) */
176 c15ce40c 2005-08-10 devnull isrfc822, /* email file */
177 c15ce40c 2005-08-10 devnull ismbox, /* mail box */
178 c15ce40c 2005-08-10 devnull istar, /* recognizable by tar checksum */
179 c15ce40c 2005-08-10 devnull ishtml, /* html keywords */
180 c15ce40c 2005-08-10 devnull /* iscint, /* compiler/assembler intermediate */
181 c15ce40c 2005-08-10 devnull islimbo, /* limbo source */
182 c15ce40c 2005-08-10 devnull isc, /* c & alef compiler key words */
183 c15ce40c 2005-08-10 devnull isas, /* assembler key words */
184 c15ce40c 2005-08-10 devnull ismung, /* entropy compressed/encrypted */
185 c15ce40c 2005-08-10 devnull isp9font, /* plan 9 font */
186 c15ce40c 2005-08-10 devnull isp9bit, /* plan 9 image (as from /dev/window) */
187 c15ce40c 2005-08-10 devnull isenglish, /* char frequency English */
188 c15ce40c 2005-08-10 devnull isrtf, /* rich text format */
189 c15ce40c 2005-08-10 devnull ismsdos, /* msdos exe (virus file attachement) */
190 c15ce40c 2005-08-10 devnull iself, /* ELF (foreign) executable */
194 c15ce40c 2005-08-10 devnull int mime;
196 c15ce40c 2005-08-10 devnull #define OCTET "application/octet-stream\n"
197 c15ce40c 2005-08-10 devnull #define PLAIN "text/plain\n"
200 c15ce40c 2005-08-10 devnull main(int argc, char *argv[])
202 c15ce40c 2005-08-10 devnull int i, j, maxlen;
203 c15ce40c 2005-08-10 devnull char *cp;
206 c15ce40c 2005-08-10 devnull ARGBEGIN{
207 c15ce40c 2005-08-10 devnull case 'm':
208 c15ce40c 2005-08-10 devnull mime = 1;
210 c15ce40c 2005-08-10 devnull default:
211 c15ce40c 2005-08-10 devnull fprint(2, "usage: file [-m] [file...]\n");
212 c15ce40c 2005-08-10 devnull exits("usage");
213 c15ce40c 2005-08-10 devnull }ARGEND;
215 c15ce40c 2005-08-10 devnull maxlen = 0;
216 c15ce40c 2005-08-10 devnull if(mime == 0 || argc > 1){
217 c15ce40c 2005-08-10 devnull for(i = 0; i < argc; i++) {
218 c15ce40c 2005-08-10 devnull for (j = 0, cp = argv[i]; *cp; j++, cp += chartorune(&r, cp))
220 c15ce40c 2005-08-10 devnull if(j > maxlen)
221 c15ce40c 2005-08-10 devnull maxlen = j;
224 c15ce40c 2005-08-10 devnull if (argc <= 0) {
225 c15ce40c 2005-08-10 devnull if(!mime)
226 c15ce40c 2005-08-10 devnull print ("stdin: ");
227 c15ce40c 2005-08-10 devnull filetype(0);
230 c15ce40c 2005-08-10 devnull for(i = 0; i < argc; i++)
231 c15ce40c 2005-08-10 devnull type(argv[i], maxlen);
233 c15ce40c 2005-08-10 devnull exits(0);
237 c15ce40c 2005-08-10 devnull type(char *file, int nlen)
241 c15ce40c 2005-08-10 devnull char *p;
243 c15ce40c 2005-08-10 devnull if(nlen > 0){
244 c15ce40c 2005-08-10 devnull slash = 0;
245 c15ce40c 2005-08-10 devnull for (i = 0, p = file; *p; i++) {
246 c15ce40c 2005-08-10 devnull if (*p == '/') /* find rightmost slash */
247 c15ce40c 2005-08-10 devnull slash = p;
248 c15ce40c 2005-08-10 devnull p += chartorune(&r, p); /* count runes */
250 c15ce40c 2005-08-10 devnull print("%s:%*s",file, nlen-i+1, "");
252 c15ce40c 2005-08-10 devnull fname = file;
253 c15ce40c 2005-08-10 devnull if ((fd = open(file, OREAD)) < 0) {
254 c15ce40c 2005-08-10 devnull print("cannot open\n");
257 c15ce40c 2005-08-10 devnull filetype(fd);
258 c15ce40c 2005-08-10 devnull close(fd);
262 c15ce40c 2005-08-10 devnull filetype(int fd)
265 c15ce40c 2005-08-10 devnull int i, f, n;
266 c15ce40c 2005-08-10 devnull char *p, *eob;
268 c15ce40c 2005-08-10 devnull free(mbuf);
269 c15ce40c 2005-08-10 devnull mbuf = dirfstat(fd);
270 c15ce40c 2005-08-10 devnull if(mbuf == nil){
271 c15ce40c 2005-08-10 devnull print("cannot stat: %r\n");
274 c15ce40c 2005-08-10 devnull if(mbuf->mode & DMDIR) {
275 c15ce40c 2005-08-10 devnull print(mime ? "text/directory\n" : "directory\n");
278 c15ce40c 2005-08-10 devnull if(mbuf->type != 'M' && mbuf->type != '|') {
279 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "special file #%c/%s\n",
280 c15ce40c 2005-08-10 devnull mbuf->type, mbuf->name);
283 c15ce40c 2005-08-10 devnull nbuf = read(fd, buf, sizeof(buf)-1);
285 c15ce40c 2005-08-10 devnull if(nbuf < 0) {
286 c15ce40c 2005-08-10 devnull print("cannot read\n");
289 c15ce40c 2005-08-10 devnull if(nbuf == 0) {
290 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "empty file\n");
293 c15ce40c 2005-08-10 devnull buf[nbuf] = 0;
296 c15ce40c 2005-08-10 devnull * build histogram table
298 c15ce40c 2005-08-10 devnull memset(cfreq, 0, sizeof(cfreq));
299 c15ce40c 2005-08-10 devnull for (i = 0; language[i].name; i++)
300 c15ce40c 2005-08-10 devnull language[i].count = 0;
301 c15ce40c 2005-08-10 devnull eob = (char *)buf+nbuf;
302 c15ce40c 2005-08-10 devnull for(n = 0, p = (char *)buf; p < eob; n++) {
303 c15ce40c 2005-08-10 devnull if (!fullrune(p, eob-p) && eob-p < UTFmax)
305 c15ce40c 2005-08-10 devnull p += chartorune(&r, p);
306 c15ce40c 2005-08-10 devnull if (r == 0)
307 c15ce40c 2005-08-10 devnull f = Cnull;
308 c15ce40c 2005-08-10 devnull else if (r <= 0x7f) {
309 c15ce40c 2005-08-10 devnull if (!isprint(r) && !isspace(r))
310 c15ce40c 2005-08-10 devnull f = Ceascii; /* ASCII control char */
311 c15ce40c 2005-08-10 devnull else f = r;
312 c15ce40c 2005-08-10 devnull } else if (r == 0x080) {
313 c15ce40c 2005-08-10 devnull bump_utf_count(r);
314 c15ce40c 2005-08-10 devnull f = Cutf;
315 c15ce40c 2005-08-10 devnull } else if (r < 0xA0)
316 c15ce40c 2005-08-10 devnull f = Cbinary; /* Invalid Runes */
317 c15ce40c 2005-08-10 devnull else if (r <= 0xff)
318 c15ce40c 2005-08-10 devnull f = Clatin; /* Latin 1 */
320 c15ce40c 2005-08-10 devnull bump_utf_count(r);
321 c15ce40c 2005-08-10 devnull f = Cutf; /* UTF extension */
323 c15ce40c 2005-08-10 devnull cfreq[f]++; /* ASCII chars peg directly */
326 c15ce40c 2005-08-10 devnull * gross classify
328 c15ce40c 2005-08-10 devnull if (cfreq[Cbinary])
329 c15ce40c 2005-08-10 devnull guess = Fbinary;
330 c15ce40c 2005-08-10 devnull else if (cfreq[Cutf])
331 c15ce40c 2005-08-10 devnull guess = Futf;
332 c15ce40c 2005-08-10 devnull else if (cfreq[Clatin])
333 c15ce40c 2005-08-10 devnull guess = Flatin;
334 c15ce40c 2005-08-10 devnull else if (cfreq[Ceascii])
335 c15ce40c 2005-08-10 devnull guess = Feascii;
336 c15ce40c 2005-08-10 devnull else if (cfreq[Cnull] == n) {
337 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "first block all null bytes\n");
340 c15ce40c 2005-08-10 devnull else guess = Fascii;
342 c15ce40c 2005-08-10 devnull * lookup dictionary words
344 c15ce40c 2005-08-10 devnull memset(wfreq, 0, sizeof(wfreq));
345 c15ce40c 2005-08-10 devnull if(guess == Fascii || guess == Flatin || guess == Futf)
346 c15ce40c 2005-08-10 devnull wordfreq();
348 c15ce40c 2005-08-10 devnull * call individual classify routines
350 c15ce40c 2005-08-10 devnull for(i=0; call[i]; i++)
351 c15ce40c 2005-08-10 devnull if((*call[i])())
355 c15ce40c 2005-08-10 devnull * if all else fails,
356 c15ce40c 2005-08-10 devnull * print out gross classification
358 c15ce40c 2005-08-10 devnull if (nbuf < 100 && !mime)
359 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "short ");
360 c15ce40c 2005-08-10 devnull if (guess == Fascii)
361 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "Ascii\n");
362 c15ce40c 2005-08-10 devnull else if (guess == Feascii)
363 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "extended ascii\n");
364 c15ce40c 2005-08-10 devnull else if (guess == Flatin)
365 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "latin ascii\n");
366 c15ce40c 2005-08-10 devnull else if (guess == Futf && utf_count() < 4)
367 c15ce40c 2005-08-10 devnull print_utf();
368 c15ce40c 2005-08-10 devnull else print(mime ? OCTET : "binary\n");
372 c15ce40c 2005-08-10 devnull bump_utf_count(Rune r)
374 c15ce40c 2005-08-10 devnull int low, high, mid;
376 c15ce40c 2005-08-10 devnull high = sizeof(language)/sizeof(language[0])-1;
377 c15ce40c 2005-08-10 devnull for (low = 0; low < high;) {
378 c15ce40c 2005-08-10 devnull mid = (low+high)/2;
379 c15ce40c 2005-08-10 devnull if (r >=language[mid].low) {
380 c15ce40c 2005-08-10 devnull if (r <= language[mid].high) {
381 c15ce40c 2005-08-10 devnull language[mid].count++;
383 c15ce40c 2005-08-10 devnull } else low = mid+1;
384 c15ce40c 2005-08-10 devnull } else high = mid;
389 c15ce40c 2005-08-10 devnull utf_count(void)
391 c15ce40c 2005-08-10 devnull int i, count;
393 c15ce40c 2005-08-10 devnull count = 0;
394 c15ce40c 2005-08-10 devnull for (i = 0; language[i].name; i++)
395 c15ce40c 2005-08-10 devnull if (language[i].count > 0)
396 c15ce40c 2005-08-10 devnull switch (language[i].mode) {
397 c15ce40c 2005-08-10 devnull case Normal:
398 c15ce40c 2005-08-10 devnull case First:
399 c15ce40c 2005-08-10 devnull count++;
401 c15ce40c 2005-08-10 devnull default:
404 c15ce40c 2005-08-10 devnull return count;
408 c15ce40c 2005-08-10 devnull chkascii(void)
412 c15ce40c 2005-08-10 devnull for (i = 'a'; i < 'z'; i++)
413 c15ce40c 2005-08-10 devnull if (cfreq[i])
414 c15ce40c 2005-08-10 devnull return 1;
415 c15ce40c 2005-08-10 devnull for (i = 'A'; i < 'Z'; i++)
416 c15ce40c 2005-08-10 devnull if (cfreq[i])
417 c15ce40c 2005-08-10 devnull return 1;
418 c15ce40c 2005-08-10 devnull return 0;
422 c15ce40c 2005-08-10 devnull find_first(char *name)
426 c15ce40c 2005-08-10 devnull for (i = 0; language[i].name != 0; i++)
427 c15ce40c 2005-08-10 devnull if (language[i].mode == First
428 c15ce40c 2005-08-10 devnull && strcmp(language[i].name, name) == 0)
429 c15ce40c 2005-08-10 devnull return i;
430 c15ce40c 2005-08-10 devnull return -1;
434 c15ce40c 2005-08-10 devnull print_utf(void)
436 c15ce40c 2005-08-10 devnull int i, printed, j;
438 c15ce40c 2005-08-10 devnull if(mime){
439 c15ce40c 2005-08-10 devnull print(PLAIN);
442 c15ce40c 2005-08-10 devnull if (chkascii()) {
443 c15ce40c 2005-08-10 devnull printed = 1;
444 c15ce40c 2005-08-10 devnull print("Ascii");
446 c15ce40c 2005-08-10 devnull printed = 0;
447 c15ce40c 2005-08-10 devnull for (i = 0; language[i].name; i++)
448 c15ce40c 2005-08-10 devnull if (language[i].count) {
449 c15ce40c 2005-08-10 devnull switch(language[i].mode) {
450 c15ce40c 2005-08-10 devnull case Multi:
451 c15ce40c 2005-08-10 devnull j = find_first(language[i].name);
452 c15ce40c 2005-08-10 devnull if (j < 0)
454 c15ce40c 2005-08-10 devnull if (language[j].count > 0)
456 c15ce40c 2005-08-10 devnull /* Fall through */
457 c15ce40c 2005-08-10 devnull case Normal:
458 c15ce40c 2005-08-10 devnull case First:
459 c15ce40c 2005-08-10 devnull if (printed)
460 c15ce40c 2005-08-10 devnull print(" & ");
461 c15ce40c 2005-08-10 devnull else printed = 1;
462 c15ce40c 2005-08-10 devnull print("%s", language[i].name);
464 c15ce40c 2005-08-10 devnull case Shared:
465 c15ce40c 2005-08-10 devnull default:
469 c15ce40c 2005-08-10 devnull if(!printed)
470 c15ce40c 2005-08-10 devnull print("UTF");
471 c15ce40c 2005-08-10 devnull print(" text\n");
475 c15ce40c 2005-08-10 devnull wordfreq(void)
477 c15ce40c 2005-08-10 devnull int low, high, mid, r;
478 c15ce40c 2005-08-10 devnull uchar *p, *p2, c;
480 c15ce40c 2005-08-10 devnull p = buf;
481 c15ce40c 2005-08-10 devnull for(;;) {
482 c15ce40c 2005-08-10 devnull while (p < buf+nbuf && !isalpha(*p))
484 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
487 c15ce40c 2005-08-10 devnull while(p < buf+nbuf && isalpha(*p))
491 c15ce40c 2005-08-10 devnull high = sizeof(dict)/sizeof(dict[0]);
492 c15ce40c 2005-08-10 devnull for(low = 0;low < high;) {
493 c15ce40c 2005-08-10 devnull mid = (low+high)/2;
494 c15ce40c 2005-08-10 devnull r = strcmp(dict[mid].word, (char*)p2);
495 c15ce40c 2005-08-10 devnull if(r == 0) {
496 c15ce40c 2005-08-10 devnull wfreq[dict[mid].class]++;
499 c15ce40c 2005-08-10 devnull if(r < 0)
500 c15ce40c 2005-08-10 devnull low = mid+1;
502 c15ce40c 2005-08-10 devnull high = mid;
504 c15ce40c 2005-08-10 devnull *p++ = c;
508 c15ce40c 2005-08-10 devnull typedef struct Filemagic Filemagic;
509 c15ce40c 2005-08-10 devnull struct Filemagic {
510 c15ce40c 2005-08-10 devnull ulong x;
511 c15ce40c 2005-08-10 devnull ulong mask;
512 c15ce40c 2005-08-10 devnull char *desc;
513 c15ce40c 2005-08-10 devnull char *mime;
516 c15ce40c 2005-08-10 devnull Filemagic long0tab[] = {
517 c15ce40c 2005-08-10 devnull 0xF16DF16D, 0xFFFFFFFF, "pac1 audio file\n", OCTET,
518 c15ce40c 2005-08-10 devnull 0x31636170, 0xFFFFFFFF, "pac3 audio file\n", OCTET,
519 c15ce40c 2005-08-10 devnull 0x32636170, 0xFFFF00FF, "pac4 audio file\n", OCTET,
520 c15ce40c 2005-08-10 devnull 0xBA010000, 0xFFFFFFFF, "mpeg system stream\n", OCTET,
521 c15ce40c 2005-08-10 devnull 0x30800CC0, 0xFFFFFFFF, "inferno .dis executable\n", OCTET,
522 c15ce40c 2005-08-10 devnull 0x04034B50, 0xFFFFFFFF, "zip archive\n", "application/zip",
523 c15ce40c 2005-08-10 devnull 070707, 0xFFFF, "cpio archive\n", OCTET,
524 c15ce40c 2005-08-10 devnull 0x2F7, 0xFFFF, "tex dvi\n", "application/dvi",
525 c15ce40c 2005-08-10 devnull 0xfffa0000, 0xfffe0000, "mp3 audio", "audio/mpeg",
529 c15ce40c 2005-08-10 devnull filemagic(Filemagic *tab, int ntab, ulong x)
533 c15ce40c 2005-08-10 devnull for(i=0; i<ntab; i++)
534 c15ce40c 2005-08-10 devnull if((x&tab[i].mask) == tab[i].x){
535 c15ce40c 2005-08-10 devnull print(mime ? tab[i].mime : tab[i].desc);
536 c15ce40c 2005-08-10 devnull return 1;
538 c15ce40c 2005-08-10 devnull return 0;
542 c15ce40c 2005-08-10 devnull long0(void)
544 cbeb0b26 2006-04-01 devnull /* Fhdr *f; */
547 c15ce40c 2005-08-10 devnull seek(fd, 0, 0); /* reposition to start of file */
549 c15ce40c 2005-08-10 devnull if(crackhdr(fd, &f)) {
550 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "%s\n", f.name);
551 c15ce40c 2005-08-10 devnull return 1;
554 c15ce40c 2005-08-10 devnull x = LENDIAN(buf);
555 c15ce40c 2005-08-10 devnull if(filemagic(long0tab, nelem(long0tab), x))
556 c15ce40c 2005-08-10 devnull return 1;
557 c15ce40c 2005-08-10 devnull return 0;
560 c15ce40c 2005-08-10 devnull /* from tar.c */
561 c15ce40c 2005-08-10 devnull enum { NAMSIZ = 100, TBLOCK = 512 };
563 c15ce40c 2005-08-10 devnull union hblock
565 c15ce40c 2005-08-10 devnull char dummy[TBLOCK];
566 c15ce40c 2005-08-10 devnull struct header
568 c15ce40c 2005-08-10 devnull char name[NAMSIZ];
569 c15ce40c 2005-08-10 devnull char mode[8];
570 c15ce40c 2005-08-10 devnull char uid[8];
571 c15ce40c 2005-08-10 devnull char gid[8];
572 c15ce40c 2005-08-10 devnull char size[12];
573 c15ce40c 2005-08-10 devnull char mtime[12];
574 c15ce40c 2005-08-10 devnull char chksum[8];
575 c15ce40c 2005-08-10 devnull char linkflag;
576 c15ce40c 2005-08-10 devnull char linkname[NAMSIZ];
577 c15ce40c 2005-08-10 devnull /* rest are defined by POSIX's ustar format; see p1003.2b */
578 c15ce40c 2005-08-10 devnull char magic[6]; /* "ustar" */
579 c15ce40c 2005-08-10 devnull char version[2];
580 c15ce40c 2005-08-10 devnull char uname[32];
581 c15ce40c 2005-08-10 devnull char gname[32];
582 c15ce40c 2005-08-10 devnull char devmajor[8];
583 c15ce40c 2005-08-10 devnull char devminor[8];
584 c15ce40c 2005-08-10 devnull char prefix[155]; /* if non-null, path = prefix "/" name */
589 c15ce40c 2005-08-10 devnull checksum(union hblock *hp)
592 c15ce40c 2005-08-10 devnull char *cp;
593 c15ce40c 2005-08-10 devnull struct header *hdr = &hp->dbuf;
595 c15ce40c 2005-08-10 devnull for (cp = hdr->chksum; cp < &hdr->chksum[sizeof hdr->chksum]; cp++)
596 c15ce40c 2005-08-10 devnull *cp = ' ';
598 c15ce40c 2005-08-10 devnull for (cp = hp->dummy; cp < &hp->dummy[TBLOCK]; cp++)
599 c15ce40c 2005-08-10 devnull i += *cp & 0xff;
600 c15ce40c 2005-08-10 devnull return i;
604 c15ce40c 2005-08-10 devnull istar(void)
606 c15ce40c 2005-08-10 devnull int chksum;
607 c15ce40c 2005-08-10 devnull char tblock[TBLOCK];
608 c15ce40c 2005-08-10 devnull union hblock *hp = (union hblock *)tblock;
609 c15ce40c 2005-08-10 devnull struct header *hdr = &hp->dbuf;
611 c15ce40c 2005-08-10 devnull seek(fd, 0, 0); /* reposition to start of file */
612 c15ce40c 2005-08-10 devnull if (readn(fd, tblock, sizeof tblock) != sizeof tblock)
613 c15ce40c 2005-08-10 devnull return 0;
614 c15ce40c 2005-08-10 devnull chksum = strtol(hdr->chksum, 0, 8);
615 c15ce40c 2005-08-10 devnull if (hdr->name[0] != '\0' && checksum(hp) == chksum) {
616 c15ce40c 2005-08-10 devnull if (strcmp(hdr->magic, "ustar") == 0)
617 c15ce40c 2005-08-10 devnull print(mime? "application/x-ustar\n":
618 c15ce40c 2005-08-10 devnull "posix tar archive\n");
620 c15ce40c 2005-08-10 devnull print(mime? "application/x-tar\n": "tar archive\n");
621 c15ce40c 2005-08-10 devnull return 1;
623 c15ce40c 2005-08-10 devnull return 0;
627 c15ce40c 2005-08-10 devnull * initial words to classify file
629 c15ce40c 2005-08-10 devnull struct FILE_STRING
631 c15ce40c 2005-08-10 devnull char *key;
632 c15ce40c 2005-08-10 devnull char *filetype;
633 c15ce40c 2005-08-10 devnull int length;
634 c15ce40c 2005-08-10 devnull char *mime;
635 c15ce40c 2005-08-10 devnull } file_string[] =
637 c15ce40c 2005-08-10 devnull "!<arch>\n__.SYMDEF", "archive random library", 16, "application/octet-stream",
638 c15ce40c 2005-08-10 devnull "!<arch>\n", "archive", 8, "application/octet-stream",
639 c15ce40c 2005-08-10 devnull "070707", "cpio archive - ascii header", 6, "application/octet-stream",
640 c15ce40c 2005-08-10 devnull "%!", "postscript", 2, "application/postscript",
641 c15ce40c 2005-08-10 devnull "\004%!", "postscript", 3, "application/postscript",
642 c15ce40c 2005-08-10 devnull "x T post", "troff output for post", 8, "application/troff",
643 c15ce40c 2005-08-10 devnull "x T Latin1", "troff output for Latin1", 10, "application/troff",
644 c15ce40c 2005-08-10 devnull "x T utf", "troff output for UTF", 7, "application/troff",
645 c15ce40c 2005-08-10 devnull "x T 202", "troff output for 202", 7, "application/troff",
646 c15ce40c 2005-08-10 devnull "x T aps", "troff output for aps", 7, "application/troff",
647 c15ce40c 2005-08-10 devnull "GIF", "GIF image", 3, "image/gif",
648 c15ce40c 2005-08-10 devnull "\0PC Research, Inc\0", "ghostscript fax file", 18, "application/ghostscript",
649 c15ce40c 2005-08-10 devnull "%PDF", "PDF", 4, "application/pdf",
650 c15ce40c 2005-08-10 devnull "<html>\n", "HTML file", 7, "text/html",
651 c15ce40c 2005-08-10 devnull "<HTML>\n", "HTML file", 7, "text/html",
652 c15ce40c 2005-08-10 devnull "compressed\n", "Compressed image or subfont", 11, "application/octet-stream",
653 c15ce40c 2005-08-10 devnull "\111\111\052\000", "tiff", 4, "image/tiff",
654 c15ce40c 2005-08-10 devnull "\115\115\000\052", "tiff", 4, "image/tiff",
655 c15ce40c 2005-08-10 devnull "\377\330\377\340", "jpeg", 4, "image/jpeg",
656 c15ce40c 2005-08-10 devnull "\377\330\377\341", "jpeg", 4, "image/jpeg",
657 c15ce40c 2005-08-10 devnull "\377\330\377\333", "jpeg", 4, "image/jpeg",
658 c15ce40c 2005-08-10 devnull "BM", "bmp", 2, "image/bmp",
659 c15ce40c 2005-08-10 devnull "\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1", "microsoft office document", 8, "application/octet-stream",
660 c15ce40c 2005-08-10 devnull "<MakerFile ", "FrameMaker file", 11, "application/framemaker",
661 c15ce40c 2005-08-10 devnull "\033%-12345X", "HPJCL file", 9, "application/hpjcl",
662 c15ce40c 2005-08-10 devnull "ID3", "mp3 audio with id3", 3, "audio/mpeg",
667 c15ce40c 2005-08-10 devnull istring(void)
669 adee1676 2005-08-31 devnull int i, j;
670 c15ce40c 2005-08-10 devnull struct FILE_STRING *p;
672 c15ce40c 2005-08-10 devnull for(p = file_string; p->key; p++) {
673 c15ce40c 2005-08-10 devnull if(nbuf >= p->length && !memcmp(buf, p->key, p->length)) {
674 c15ce40c 2005-08-10 devnull if(mime)
675 c15ce40c 2005-08-10 devnull print("%s\n", p->mime);
677 c15ce40c 2005-08-10 devnull print("%s\n", p->filetype);
678 c15ce40c 2005-08-10 devnull return 1;
681 c15ce40c 2005-08-10 devnull if(strncmp((char*)buf, "TYPE=", 5) == 0) { /* td */
682 c15ce40c 2005-08-10 devnull for(i = 5; i < nbuf; i++)
683 c15ce40c 2005-08-10 devnull if(buf[i] == '\n')
685 c15ce40c 2005-08-10 devnull if(mime)
686 c15ce40c 2005-08-10 devnull print(OCTET);
688 c15ce40c 2005-08-10 devnull print("%.*s picture\n", utfnlen((char*)buf+5, i-5), (char*)buf+5);
689 c15ce40c 2005-08-10 devnull return 1;
691 adee1676 2005-08-31 devnull if(buf[0]=='#' && buf[1]=='!'){
693 adee1676 2005-08-31 devnull for(j=2; j < nbuf && buf[j] != ' ' && buf[j] != '\n' && buf[j] != '\r'; j++)
694 adee1676 2005-08-31 devnull if(buf[j] == '/')
695 adee1676 2005-08-31 devnull i = j+1;
696 adee1676 2005-08-31 devnull if(mime)
697 adee1676 2005-08-31 devnull print(PLAIN);
699 adee1676 2005-08-31 devnull print("%.*s executable file script\n", utfnlen((char*)buf+i, j-i), (char*)buf+i);
700 adee1676 2005-08-31 devnull return 1;
702 c15ce40c 2005-08-10 devnull return 0;
706 c15ce40c 2005-08-10 devnull iff(void)
708 c15ce40c 2005-08-10 devnull if (strncmp((char*)buf, "FORM", 4) == 0 &&
709 c15ce40c 2005-08-10 devnull strncmp((char*)buf+8, "AIFF", 4) == 0) {
710 c15ce40c 2005-08-10 devnull print("%s\n", mime? "audio/x-aiff": "aiff audio");
711 c15ce40c 2005-08-10 devnull return 1;
713 c15ce40c 2005-08-10 devnull return 0;
716 c15ce40c 2005-08-10 devnull char* html_string[] =
718 c15ce40c 2005-08-10 devnull "title",
721 c15ce40c 2005-08-10 devnull "strong",
737 c15ce40c 2005-08-10 devnull ishtml(void)
739 c15ce40c 2005-08-10 devnull uchar *p, *q;
740 c15ce40c 2005-08-10 devnull int i, count;
742 c15ce40c 2005-08-10 devnull /* compare strings between '<' and '>' to html table */
743 c15ce40c 2005-08-10 devnull count = 0;
744 c15ce40c 2005-08-10 devnull p = buf;
745 c15ce40c 2005-08-10 devnull for(;;) {
746 c15ce40c 2005-08-10 devnull while (p < buf+nbuf && *p != '<')
749 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
751 c15ce40c 2005-08-10 devnull if(*p == '/')
754 c15ce40c 2005-08-10 devnull while(p < buf+nbuf && *p != '>')
756 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
758 c15ce40c 2005-08-10 devnull for(i = 0; html_string[i]; i++) {
759 c15ce40c 2005-08-10 devnull if(cistrncmp(html_string[i], (char*)q, p-q) == 0) {
760 c15ce40c 2005-08-10 devnull if(count++ > 4) {
761 c15ce40c 2005-08-10 devnull print(mime ? "text/html\n" : "HTML file\n");
762 c15ce40c 2005-08-10 devnull return 1;
769 c15ce40c 2005-08-10 devnull return 0;
772 c15ce40c 2005-08-10 devnull char* rfc822_string[] =
774 c15ce40c 2005-08-10 devnull "from:",
775 c15ce40c 2005-08-10 devnull "date:",
777 c15ce40c 2005-08-10 devnull "subject:",
778 c15ce40c 2005-08-10 devnull "received:",
779 c15ce40c 2005-08-10 devnull "reply to:",
780 c15ce40c 2005-08-10 devnull "sender:",
785 c15ce40c 2005-08-10 devnull isrfc822(void)
788 c15ce40c 2005-08-10 devnull char *p, *q, *r;
789 c15ce40c 2005-08-10 devnull int i, count;
791 c15ce40c 2005-08-10 devnull count = 0;
792 c15ce40c 2005-08-10 devnull p = (char*)buf;
793 c15ce40c 2005-08-10 devnull for(;;) {
794 c15ce40c 2005-08-10 devnull q = strchr(p, '\n');
795 c15ce40c 2005-08-10 devnull if(q == nil)
798 c15ce40c 2005-08-10 devnull if(p == (char*)buf && strncmp(p, "From ", 5) == 0 && strstr(p, " remote from ")){
799 c15ce40c 2005-08-10 devnull count++;
800 c15ce40c 2005-08-10 devnull *q = '\n';
801 c15ce40c 2005-08-10 devnull p = q+1;
802 c15ce40c 2005-08-10 devnull continue;
804 c15ce40c 2005-08-10 devnull *q = '\n';
805 c15ce40c 2005-08-10 devnull if(*p != '\t' && *p != ' '){
806 c15ce40c 2005-08-10 devnull r = strchr(p, ':');
807 c15ce40c 2005-08-10 devnull if(r == 0 || r > q)
809 c15ce40c 2005-08-10 devnull for(i = 0; rfc822_string[i]; i++) {
810 c15ce40c 2005-08-10 devnull if(cistrncmp(p, rfc822_string[i], strlen(rfc822_string[i])) == 0){
811 c15ce40c 2005-08-10 devnull count++;
816 c15ce40c 2005-08-10 devnull p = q+1;
818 c15ce40c 2005-08-10 devnull if(count >= 3){
819 c15ce40c 2005-08-10 devnull print(mime ? "message/rfc822\n" : "email file\n");
820 c15ce40c 2005-08-10 devnull return 1;
822 c15ce40c 2005-08-10 devnull return 0;
826 c15ce40c 2005-08-10 devnull ismbox(void)
828 c15ce40c 2005-08-10 devnull char *p, *q;
830 c15ce40c 2005-08-10 devnull p = (char*)buf;
831 c15ce40c 2005-08-10 devnull q = strchr(p, '\n');
832 c15ce40c 2005-08-10 devnull if(q == nil)
833 c15ce40c 2005-08-10 devnull return 0;
835 c15ce40c 2005-08-10 devnull if(strncmp(p, "From ", 5) == 0 && strstr(p, " remote from ") == nil){
836 c15ce40c 2005-08-10 devnull print(mime ? "text/plain\n" : "mail box\n");
837 c15ce40c 2005-08-10 devnull return 1;
839 c15ce40c 2005-08-10 devnull *q = '\n';
840 c15ce40c 2005-08-10 devnull return 0;
844 c15ce40c 2005-08-10 devnull isc(void)
848 c15ce40c 2005-08-10 devnull n = wfreq[I1];
850 c15ce40c 2005-08-10 devnull * includes
852 c15ce40c 2005-08-10 devnull if(n >= 2 && wfreq[I2] >= n && wfreq[I3] >= n && cfreq['.'] >= n)
853 c15ce40c 2005-08-10 devnull goto yes;
854 c15ce40c 2005-08-10 devnull if(n >= 1 && wfreq[Alword] >= n && wfreq[I3] >= n && cfreq['.'] >= n)
855 c15ce40c 2005-08-10 devnull goto yes;
857 c15ce40c 2005-08-10 devnull * declarations
859 c15ce40c 2005-08-10 devnull if(wfreq[Cword] >= 5 && cfreq[';'] >= 5)
860 c15ce40c 2005-08-10 devnull goto yes;
862 c15ce40c 2005-08-10 devnull * assignments
864 c15ce40c 2005-08-10 devnull if(cfreq[';'] >= 10 && cfreq['='] >= 10 && wfreq[Cword] >= 1)
865 c15ce40c 2005-08-10 devnull goto yes;
866 c15ce40c 2005-08-10 devnull return 0;
869 c15ce40c 2005-08-10 devnull if(mime){
870 c15ce40c 2005-08-10 devnull print(PLAIN);
871 c15ce40c 2005-08-10 devnull return 1;
873 c15ce40c 2005-08-10 devnull if(wfreq[Alword] > 0)
874 c15ce40c 2005-08-10 devnull print("alef program\n");
876 c15ce40c 2005-08-10 devnull print("c program\n");
877 c15ce40c 2005-08-10 devnull return 1;
881 c15ce40c 2005-08-10 devnull islimbo(void)
885 c15ce40c 2005-08-10 devnull * includes
887 c15ce40c 2005-08-10 devnull if(wfreq[Lword] < 4)
888 c15ce40c 2005-08-10 devnull return 0;
889 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "limbo program\n");
890 c15ce40c 2005-08-10 devnull return 1;
894 c15ce40c 2005-08-10 devnull isas(void)
898 c15ce40c 2005-08-10 devnull * includes
900 c15ce40c 2005-08-10 devnull if(wfreq[Aword] < 2)
901 c15ce40c 2005-08-10 devnull return 0;
902 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "as program\n");
903 c15ce40c 2005-08-10 devnull return 1;
907 c15ce40c 2005-08-10 devnull * low entropy means encrypted
910 c15ce40c 2005-08-10 devnull ismung(void)
912 c15ce40c 2005-08-10 devnull int i, bucket[8];
913 c15ce40c 2005-08-10 devnull float cs;
915 c15ce40c 2005-08-10 devnull if(nbuf < 64)
916 c15ce40c 2005-08-10 devnull return 0;
917 c15ce40c 2005-08-10 devnull memset(bucket, 0, sizeof(bucket));
918 c15ce40c 2005-08-10 devnull for(i=0; i<64; i++)
919 c15ce40c 2005-08-10 devnull bucket[(buf[i]>>5)&07] += 1;
921 c15ce40c 2005-08-10 devnull cs = 0.;
922 c15ce40c 2005-08-10 devnull for(i=0; i<8; i++)
923 c15ce40c 2005-08-10 devnull cs += (bucket[i]-8)*(bucket[i]-8);
924 c15ce40c 2005-08-10 devnull cs /= 8.;
925 c15ce40c 2005-08-10 devnull if(cs <= 24.322) {
926 c15ce40c 2005-08-10 devnull if(buf[0]==0x1f && (buf[1]==0x8b || buf[1]==0x9d))
927 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "compressed\n");
929 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "encrypted\n");
930 c15ce40c 2005-08-10 devnull return 1;
932 c15ce40c 2005-08-10 devnull return 0;
936 c15ce40c 2005-08-10 devnull * english by punctuation and frequencies
939 c15ce40c 2005-08-10 devnull isenglish(void)
941 c15ce40c 2005-08-10 devnull int vow, comm, rare, badpun, punct;
942 c15ce40c 2005-08-10 devnull char *p;
944 c15ce40c 2005-08-10 devnull if(guess != Fascii && guess != Feascii)
945 c15ce40c 2005-08-10 devnull return 0;
946 c15ce40c 2005-08-10 devnull badpun = 0;
947 c15ce40c 2005-08-10 devnull punct = 0;
948 c15ce40c 2005-08-10 devnull for(p = (char *)buf; p < (char *)buf+nbuf-1; p++)
949 c15ce40c 2005-08-10 devnull switch(*p) {
950 c15ce40c 2005-08-10 devnull case '.':
951 c15ce40c 2005-08-10 devnull case ',':
952 c15ce40c 2005-08-10 devnull case ')':
953 c15ce40c 2005-08-10 devnull case '%':
954 c15ce40c 2005-08-10 devnull case ';':
955 c15ce40c 2005-08-10 devnull case ':':
956 c15ce40c 2005-08-10 devnull case '?':
957 c15ce40c 2005-08-10 devnull punct++;
958 c15ce40c 2005-08-10 devnull if(p[1] != ' ' && p[1] != '\n')
959 c15ce40c 2005-08-10 devnull badpun++;
961 c15ce40c 2005-08-10 devnull if(badpun*5 > punct)
962 c15ce40c 2005-08-10 devnull return 0;
963 c15ce40c 2005-08-10 devnull if(cfreq['>']+cfreq['<']+cfreq['/'] > cfreq['e']) /* shell file test */
964 c15ce40c 2005-08-10 devnull return 0;
965 c15ce40c 2005-08-10 devnull if(2*cfreq[';'] > cfreq['e'])
966 c15ce40c 2005-08-10 devnull return 0;
968 c15ce40c 2005-08-10 devnull vow = 0;
969 c15ce40c 2005-08-10 devnull for(p="AEIOU"; *p; p++) {
970 bce0902a 2005-08-11 devnull vow += cfreq[(uchar)*p];
971 bce0902a 2005-08-11 devnull vow += cfreq[tolower((uchar)*p)];
973 c15ce40c 2005-08-10 devnull comm = 0;
974 c15ce40c 2005-08-10 devnull for(p="ETAION"; *p; p++) {
975 bce0902a 2005-08-11 devnull comm += cfreq[(uchar)*p];
976 bce0902a 2005-08-11 devnull comm += cfreq[tolower((uchar)*p)];
978 c15ce40c 2005-08-10 devnull rare = 0;
979 c15ce40c 2005-08-10 devnull for(p="VJKQXZ"; *p; p++) {
980 bce0902a 2005-08-11 devnull rare += cfreq[(uchar)*p];
981 bce0902a 2005-08-11 devnull rare += cfreq[tolower((uchar)*p)];
983 c15ce40c 2005-08-10 devnull if(vow*5 >= nbuf-cfreq[' '] && comm >= 10*rare) {
984 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "English text\n");
985 c15ce40c 2005-08-10 devnull return 1;
987 c15ce40c 2005-08-10 devnull return 0;
991 c15ce40c 2005-08-10 devnull * pick up a number with
992 c15ce40c 2005-08-10 devnull * syntax _*[0-9]+_
994 c15ce40c 2005-08-10 devnull #define P9BITLEN 12
996 c15ce40c 2005-08-10 devnull p9bitnum(uchar *bp)
998 c15ce40c 2005-08-10 devnull int n, c, len;
1000 c15ce40c 2005-08-10 devnull len = P9BITLEN;
1001 c15ce40c 2005-08-10 devnull while(*bp == ' ') {
1004 c15ce40c 2005-08-10 devnull if(len <= 0)
1005 c15ce40c 2005-08-10 devnull return -1;
1008 c15ce40c 2005-08-10 devnull while(len > 1) {
1009 c15ce40c 2005-08-10 devnull c = *bp++;
1010 c15ce40c 2005-08-10 devnull if(!isdigit(c))
1011 c15ce40c 2005-08-10 devnull return -1;
1012 c15ce40c 2005-08-10 devnull n = n*10 + c-'0';
1015 c15ce40c 2005-08-10 devnull if(*bp != ' ')
1016 c15ce40c 2005-08-10 devnull return -1;
1017 c15ce40c 2005-08-10 devnull return n;
1021 c15ce40c 2005-08-10 devnull depthof(char *s, int *newp)
1023 c15ce40c 2005-08-10 devnull char *es;
1026 c15ce40c 2005-08-10 devnull *newp = 0;
1027 c15ce40c 2005-08-10 devnull es = s+12;
1028 c15ce40c 2005-08-10 devnull while(s<es && *s==' ')
1030 c15ce40c 2005-08-10 devnull if(s == es)
1031 c15ce40c 2005-08-10 devnull return -1;
1032 c15ce40c 2005-08-10 devnull if('0'<=*s && *s<='9')
1033 c15ce40c 2005-08-10 devnull return 1<<atoi(s);
1035 c15ce40c 2005-08-10 devnull *newp = 1;
1037 c15ce40c 2005-08-10 devnull while(s<es && *s!=' '){
1038 c15ce40c 2005-08-10 devnull s++; /* skip letter */
1039 c15ce40c 2005-08-10 devnull d += strtoul(s, &s, 10);
1042 c15ce40c 2005-08-10 devnull switch(d){
1043 c15ce40c 2005-08-10 devnull case 32:
1044 c15ce40c 2005-08-10 devnull case 24:
1045 c15ce40c 2005-08-10 devnull case 16:
1046 c15ce40c 2005-08-10 devnull case 8:
1047 c15ce40c 2005-08-10 devnull return d;
1049 c15ce40c 2005-08-10 devnull return -1;
1053 c15ce40c 2005-08-10 devnull isp9bit(void)
1055 c15ce40c 2005-08-10 devnull int dep, lox, loy, hix, hiy, px, new;
1056 c15ce40c 2005-08-10 devnull ulong t;
1057 c15ce40c 2005-08-10 devnull long len;
1058 c15ce40c 2005-08-10 devnull char *newlabel;
1060 c15ce40c 2005-08-10 devnull newlabel = "old ";
1062 c15ce40c 2005-08-10 devnull dep = depthof((char*)buf + 0*P9BITLEN, &new);
1063 c15ce40c 2005-08-10 devnull if(new)
1064 c15ce40c 2005-08-10 devnull newlabel = "";
1065 c15ce40c 2005-08-10 devnull lox = p9bitnum(buf + 1*P9BITLEN);
1066 c15ce40c 2005-08-10 devnull loy = p9bitnum(buf + 2*P9BITLEN);
1067 c15ce40c 2005-08-10 devnull hix = p9bitnum(buf + 3*P9BITLEN);
1068 c15ce40c 2005-08-10 devnull hiy = p9bitnum(buf + 4*P9BITLEN);
1069 c15ce40c 2005-08-10 devnull if(dep < 0 || lox < 0 || loy < 0 || hix < 0 || hiy < 0)
1070 c15ce40c 2005-08-10 devnull return 0;
1072 c15ce40c 2005-08-10 devnull if(dep < 8){
1073 c15ce40c 2005-08-10 devnull px = 8/dep; /* pixels per byte */
1074 c15ce40c 2005-08-10 devnull /* set l to number of bytes of data per scan line */
1075 c15ce40c 2005-08-10 devnull if(lox >= 0)
1076 c15ce40c 2005-08-10 devnull len = (hix+px-1)/px - lox/px;
1077 c15ce40c 2005-08-10 devnull else{ /* make positive before divide */
1078 c15ce40c 2005-08-10 devnull t = (-lox)+px-1;
1079 c15ce40c 2005-08-10 devnull t = (t/px)*px;
1080 c15ce40c 2005-08-10 devnull len = (t+hix+px-1)/px;
1083 c15ce40c 2005-08-10 devnull len = (hix-lox)*dep/8;
1084 c15ce40c 2005-08-10 devnull len *= (hiy-loy); /* col length */
1085 c15ce40c 2005-08-10 devnull len += 5*P9BITLEN; /* size of initial ascii */
1088 c15ce40c 2005-08-10 devnull * for image file, length is non-zero and must match calculation above
1089 c15ce40c 2005-08-10 devnull * for /dev/window and /dev/screen the length is always zero
1090 c15ce40c 2005-08-10 devnull * for subfont, the subfont header should follow immediately.
1092 c15ce40c 2005-08-10 devnull if (len != 0 && mbuf->length == 0) {
1093 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1094 c15ce40c 2005-08-10 devnull return 1;
1096 c15ce40c 2005-08-10 devnull if (mbuf->length == len) {
1097 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1098 c15ce40c 2005-08-10 devnull return 1;
1100 c15ce40c 2005-08-10 devnull /* Ghostscript sometimes produces a little extra on the end */
1101 c15ce40c 2005-08-10 devnull if (mbuf->length < len+P9BITLEN) {
1102 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1103 c15ce40c 2005-08-10 devnull return 1;
1105 c15ce40c 2005-08-10 devnull if (p9subfont(buf+len)) {
1106 c15ce40c 2005-08-10 devnull print("%ssubfont file\n", newlabel);
1107 c15ce40c 2005-08-10 devnull return 1;
1109 c15ce40c 2005-08-10 devnull return 0;
1113 c15ce40c 2005-08-10 devnull p9subfont(uchar *p)
1115 c15ce40c 2005-08-10 devnull int n, h, a;
1117 c15ce40c 2005-08-10 devnull /* if image too big, assume it's a subfont */
1118 c15ce40c 2005-08-10 devnull if (p+3*P9BITLEN > buf+sizeof(buf))
1119 c15ce40c 2005-08-10 devnull return 1;
1121 c15ce40c 2005-08-10 devnull n = p9bitnum(p + 0*P9BITLEN); /* char count */
1122 c15ce40c 2005-08-10 devnull if (n < 0)
1123 c15ce40c 2005-08-10 devnull return 0;
1124 c15ce40c 2005-08-10 devnull h = p9bitnum(p + 1*P9BITLEN); /* height */
1125 c15ce40c 2005-08-10 devnull if (h < 0)
1126 c15ce40c 2005-08-10 devnull return 0;
1127 c15ce40c 2005-08-10 devnull a = p9bitnum(p + 2*P9BITLEN); /* ascent */
1128 c15ce40c 2005-08-10 devnull if (a < 0)
1129 c15ce40c 2005-08-10 devnull return 0;
1130 c15ce40c 2005-08-10 devnull return 1;
1133 c15ce40c 2005-08-10 devnull #define WHITESPACE(c) ((c) == ' ' || (c) == '\t' || (c) == '\n')
1136 c15ce40c 2005-08-10 devnull isp9font(void)
1138 c15ce40c 2005-08-10 devnull uchar *cp, *p;
1139 c15ce40c 2005-08-10 devnull int i, n;
1140 c15ce40c 2005-08-10 devnull char pathname[1024];
1142 c15ce40c 2005-08-10 devnull cp = buf;
1143 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* height */
1144 c15ce40c 2005-08-10 devnull return 0;
1145 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* ascent */
1146 c15ce40c 2005-08-10 devnull return 0;
1147 c15ce40c 2005-08-10 devnull for (i = 0; 1; i++) {
1148 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* min */
1150 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* max */
1151 c15ce40c 2005-08-10 devnull return 0;
1152 c15ce40c 2005-08-10 devnull while (WHITESPACE(*cp))
1154 c15ce40c 2005-08-10 devnull for (p = cp; *cp && !WHITESPACE(*cp); cp++)
1156 c15ce40c 2005-08-10 devnull /* construct a path name, if needed */
1158 c15ce40c 2005-08-10 devnull if (*p != '/' && slash) {
1159 c15ce40c 2005-08-10 devnull n = slash-fname+1;
1160 c15ce40c 2005-08-10 devnull if (n < sizeof(pathname))
1161 c15ce40c 2005-08-10 devnull memcpy(pathname, fname, n);
1162 c15ce40c 2005-08-10 devnull else n = 0;
1164 c15ce40c 2005-08-10 devnull if (n+cp-p < sizeof(pathname)) {
1165 c15ce40c 2005-08-10 devnull memcpy(pathname+n, p, cp-p);
1166 c15ce40c 2005-08-10 devnull n += cp-p;
1167 c15ce40c 2005-08-10 devnull pathname[n] = 0;
1168 c15ce40c 2005-08-10 devnull if (access(pathname, AEXIST) < 0)
1169 c15ce40c 2005-08-10 devnull return 0;
1172 c15ce40c 2005-08-10 devnull if (i) {
1173 c15ce40c 2005-08-10 devnull print(mime ? "text/plain\n" : "font file\n");
1174 c15ce40c 2005-08-10 devnull return 1;
1176 c15ce40c 2005-08-10 devnull return 0;
1180 c15ce40c 2005-08-10 devnull getfontnum(uchar *cp, uchar **rp)
1182 c15ce40c 2005-08-10 devnull while (WHITESPACE(*cp)) /* extract ulong delimited by whitespace */
1184 c15ce40c 2005-08-10 devnull if (*cp < '0' || *cp > '9')
1185 c15ce40c 2005-08-10 devnull return 0;
1186 c15ce40c 2005-08-10 devnull strtoul((char *)cp, (char **)rp, 0);
1187 c15ce40c 2005-08-10 devnull if (!WHITESPACE(**rp))
1188 c15ce40c 2005-08-10 devnull return 0;
1189 c15ce40c 2005-08-10 devnull return 1;
1193 c15ce40c 2005-08-10 devnull isrtf(void)
1195 c15ce40c 2005-08-10 devnull if(strstr((char *)buf, "\\rtf1")){
1196 c15ce40c 2005-08-10 devnull print(mime ? "application/rtf\n" : "rich text format\n");
1197 c15ce40c 2005-08-10 devnull return 1;
1199 c15ce40c 2005-08-10 devnull return 0;
1203 c15ce40c 2005-08-10 devnull ismsdos(void)
1205 c15ce40c 2005-08-10 devnull if (buf[0] == 0x4d && buf[1] == 0x5a){
1206 c15ce40c 2005-08-10 devnull print(mime ? "application/x-msdownload\n" : "MSDOS executable\n");
1207 c15ce40c 2005-08-10 devnull return 1;
1209 c15ce40c 2005-08-10 devnull return 0;
1213 c15ce40c 2005-08-10 devnull iself(void)
1215 bce0902a 2005-08-11 devnull static char *cpu[] = { /* NB: incomplete and arbitary list */
1217 bce0902a 2005-08-11 devnull /*1*/ "WE32100",
1218 bce0902a 2005-08-11 devnull /*2*/ "SPARC",
1219 bce0902a 2005-08-11 devnull /*3*/ "i386",
1220 bce0902a 2005-08-11 devnull /*4*/ "M68000",
1221 bce0902a 2005-08-11 devnull /*5*/ "M88000",
1222 bce0902a 2005-08-11 devnull /*6*/ "i486",
1223 bce0902a 2005-08-11 devnull /*7*/ "i860",
1224 bce0902a 2005-08-11 devnull /*8*/ "R3000",
1225 bce0902a 2005-08-11 devnull /*9*/ "S370",
1226 bce0902a 2005-08-11 devnull /*10*/ "R4000",
1227 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1228 bce0902a 2005-08-11 devnull /*15*/ "HP-PA",
1231 bce0902a 2005-08-11 devnull /*18*/ "sparc v8+",
1232 bce0902a 2005-08-11 devnull /*19*/ "i960",
1233 bce0902a 2005-08-11 devnull /*20*/ "PPC-32",
1234 bce0902a 2005-08-11 devnull /*21*/ "PPC-64",
1235 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1236 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1237 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1238 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1239 bce0902a 2005-08-11 devnull /*40*/ "ARM",
1240 bce0902a 2005-08-11 devnull /*41*/ "Alpha",
1242 bce0902a 2005-08-11 devnull /*43*/ "sparc v9",
1243 bce0902a 2005-08-11 devnull nil, nil,
1244 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1245 b2a905b7 2005-08-31 devnull /*50*/ "IA-64",
1246 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1247 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1249 bce0902a 2005-08-11 devnull /*62*/ "AMD64",
1250 bce0902a 2005-08-11 devnull nil, nil, nil,
1251 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1252 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1253 bce0902a 2005-08-11 devnull /*75*/ "VAX",
1257 adee1676 2005-08-31 devnull if (memcmp(buf, "\177ELF", 4) == 0){
1258 bce0902a 2005-08-11 devnull /* gcc misparses \x7FELF as \x7FE L F */
1259 c15ce40c 2005-08-10 devnull if (!mime){
1260 c15ce40c 2005-08-10 devnull int n = (buf[19] << 8) | buf[18];
1261 c15ce40c 2005-08-10 devnull char *p = "unknown";
1263 c15ce40c 2005-08-10 devnull if (n > 0 && n < nelem(cpu) && cpu[n])
1264 c15ce40c 2005-08-10 devnull p = cpu[n];
1266 c15ce40c 2005-08-10 devnull /* try the other byte order */
1267 c15ce40c 2005-08-10 devnull n = (buf[18] << 8) | buf[19];
1268 c15ce40c 2005-08-10 devnull if (n > 0 && n < nelem(cpu) && cpu[n])
1269 c15ce40c 2005-08-10 devnull p = cpu[n];
1271 c15ce40c 2005-08-10 devnull print("%s ELF executable\n", p);
1274 c15ce40c 2005-08-10 devnull print("application/x-elf-executable");
1275 c15ce40c 2005-08-10 devnull return 1;
1278 c15ce40c 2005-08-10 devnull return 0;