Blame


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>
6 c15ce40c 2005-08-10 devnull
7 c15ce40c 2005-08-10 devnull /*
8 c15ce40c 2005-08-10 devnull * file - determine type of file
9 c15ce40c 2005-08-10 devnull */
10 c15ce40c 2005-08-10 devnull #define LENDIAN(p) ((p)[0] | ((p)[1]<<8) | ((p)[2]<<16) | ((p)[3]<<24))
11 c15ce40c 2005-08-10 devnull
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;
17 c15ce40c 2005-08-10 devnull int fd;
18 c15ce40c 2005-08-10 devnull char *fname;
19 c15ce40c 2005-08-10 devnull char *slash;
20 c15ce40c 2005-08-10 devnull
21 c15ce40c 2005-08-10 devnull enum
22 c15ce40c 2005-08-10 devnull {
23 c15ce40c 2005-08-10 devnull Cword,
24 c15ce40c 2005-08-10 devnull Fword,
25 c15ce40c 2005-08-10 devnull Aword,
26 c15ce40c 2005-08-10 devnull Alword,
27 c15ce40c 2005-08-10 devnull Lword,
28 c15ce40c 2005-08-10 devnull I1,
29 c15ce40c 2005-08-10 devnull I2,
30 c15ce40c 2005-08-10 devnull I3,
31 c15ce40c 2005-08-10 devnull Clatin = 128,
32 c15ce40c 2005-08-10 devnull Cbinary,
33 c15ce40c 2005-08-10 devnull Cnull,
34 c15ce40c 2005-08-10 devnull Ceascii,
35 c15ce40c 2005-08-10 devnull Cutf,
36 c15ce40c 2005-08-10 devnull };
37 c15ce40c 2005-08-10 devnull struct
38 c15ce40c 2005-08-10 devnull {
39 c15ce40c 2005-08-10 devnull char* word;
40 c15ce40c 2005-08-10 devnull int class;
41 c15ce40c 2005-08-10 devnull } dict[] =
42 c15ce40c 2005-08-10 devnull {
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 fa325e9b 2020-01-10 cross "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,
62 c15ce40c 2005-08-10 devnull "h", I3,
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,
81 c15ce40c 2005-08-10 devnull "u", I2,
82 c15ce40c 2005-08-10 devnull "void", Cword,
83 c15ce40c 2005-08-10 devnull };
84 c15ce40c 2005-08-10 devnull
85 c15ce40c 2005-08-10 devnull /* codes for 'mode' field in language structure */
86 c15ce40c 2005-08-10 devnull enum {
87 c15ce40c 2005-08-10 devnull Normal = 0,
88 c15ce40c 2005-08-10 devnull First, /* first entry for language spanning several ranges */
89 fa325e9b 2020-01-10 cross Multi, /* later entries " " " ... */
90 c15ce40c 2005-08-10 devnull Shared, /* codes used in several languages */
91 c15ce40c 2005-08-10 devnull };
92 c15ce40c 2005-08-10 devnull
93 c15ce40c 2005-08-10 devnull struct
94 c15ce40c 2005-08-10 devnull {
95 c15ce40c 2005-08-10 devnull int mode; /* see enum above */
96 c15ce40c 2005-08-10 devnull int count;
97 c15ce40c 2005-08-10 devnull int low;
98 c15ce40c 2005-08-10 devnull int high;
99 c15ce40c 2005-08-10 devnull char *name;
100 fa325e9b 2020-01-10 cross
101 c15ce40c 2005-08-10 devnull } language[] =
102 c15ce40c 2005-08-10 devnull {
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 */
129 c15ce40c 2005-08-10 devnull };
130 fa325e9b 2020-01-10 cross
131 fa325e9b 2020-01-10 cross
132 c15ce40c 2005-08-10 devnull enum
133 c15ce40c 2005-08-10 devnull {
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;
141 c15ce40c 2005-08-10 devnull
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);
170 c15ce40c 2005-08-10 devnull
171 c15ce40c 2005-08-10 devnull int (*call[])(void) =
172 c15ce40c 2005-08-10 devnull {
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 */
191 c15ce40c 2005-08-10 devnull 0
192 c15ce40c 2005-08-10 devnull };
193 c15ce40c 2005-08-10 devnull
194 c15ce40c 2005-08-10 devnull int mime;
195 c15ce40c 2005-08-10 devnull
196 c15ce40c 2005-08-10 devnull #define OCTET "application/octet-stream\n"
197 c15ce40c 2005-08-10 devnull #define PLAIN "text/plain\n"
198 c15ce40c 2005-08-10 devnull
199 c15ce40c 2005-08-10 devnull void
200 c15ce40c 2005-08-10 devnull main(int argc, char *argv[])
201 c15ce40c 2005-08-10 devnull {
202 c15ce40c 2005-08-10 devnull int i, j, maxlen;
203 c15ce40c 2005-08-10 devnull char *cp;
204 c15ce40c 2005-08-10 devnull Rune r;
205 c15ce40c 2005-08-10 devnull
206 c15ce40c 2005-08-10 devnull ARGBEGIN{
207 c15ce40c 2005-08-10 devnull case 'm':
208 c15ce40c 2005-08-10 devnull mime = 1;
209 c15ce40c 2005-08-10 devnull break;
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;
214 c15ce40c 2005-08-10 devnull
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))
219 c15ce40c 2005-08-10 devnull ;
220 c15ce40c 2005-08-10 devnull if(j > maxlen)
221 c15ce40c 2005-08-10 devnull maxlen = j;
222 c15ce40c 2005-08-10 devnull }
223 c15ce40c 2005-08-10 devnull }
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);
228 c15ce40c 2005-08-10 devnull }
229 c15ce40c 2005-08-10 devnull else {
230 c15ce40c 2005-08-10 devnull for(i = 0; i < argc; i++)
231 c15ce40c 2005-08-10 devnull type(argv[i], maxlen);
232 c15ce40c 2005-08-10 devnull }
233 c15ce40c 2005-08-10 devnull exits(0);
234 c15ce40c 2005-08-10 devnull }
235 c15ce40c 2005-08-10 devnull
236 c15ce40c 2005-08-10 devnull void
237 c15ce40c 2005-08-10 devnull type(char *file, int nlen)
238 c15ce40c 2005-08-10 devnull {
239 c15ce40c 2005-08-10 devnull Rune r;
240 c15ce40c 2005-08-10 devnull int i;
241 c15ce40c 2005-08-10 devnull char *p;
242 c15ce40c 2005-08-10 devnull
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 */
249 c15ce40c 2005-08-10 devnull }
250 c15ce40c 2005-08-10 devnull print("%s:%*s",file, nlen-i+1, "");
251 c15ce40c 2005-08-10 devnull }
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");
255 c15ce40c 2005-08-10 devnull return;
256 c15ce40c 2005-08-10 devnull }
257 c15ce40c 2005-08-10 devnull filetype(fd);
258 c15ce40c 2005-08-10 devnull close(fd);
259 c15ce40c 2005-08-10 devnull }
260 c15ce40c 2005-08-10 devnull
261 c15ce40c 2005-08-10 devnull void
262 c15ce40c 2005-08-10 devnull filetype(int fd)
263 c15ce40c 2005-08-10 devnull {
264 c15ce40c 2005-08-10 devnull Rune r;
265 c15ce40c 2005-08-10 devnull int i, f, n;
266 c15ce40c 2005-08-10 devnull char *p, *eob;
267 c15ce40c 2005-08-10 devnull
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");
272 c15ce40c 2005-08-10 devnull return;
273 c15ce40c 2005-08-10 devnull }
274 c15ce40c 2005-08-10 devnull if(mbuf->mode & DMDIR) {
275 c15ce40c 2005-08-10 devnull print(mime ? "text/directory\n" : "directory\n");
276 c15ce40c 2005-08-10 devnull return;
277 c15ce40c 2005-08-10 devnull }
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);
281 c15ce40c 2005-08-10 devnull return;
282 c15ce40c 2005-08-10 devnull }
283 c15ce40c 2005-08-10 devnull nbuf = read(fd, buf, sizeof(buf)-1);
284 c15ce40c 2005-08-10 devnull
285 c15ce40c 2005-08-10 devnull if(nbuf < 0) {
286 c15ce40c 2005-08-10 devnull print("cannot read\n");
287 c15ce40c 2005-08-10 devnull return;
288 c15ce40c 2005-08-10 devnull }
289 c15ce40c 2005-08-10 devnull if(nbuf == 0) {
290 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "empty file\n");
291 c15ce40c 2005-08-10 devnull return;
292 c15ce40c 2005-08-10 devnull }
293 c15ce40c 2005-08-10 devnull buf[nbuf] = 0;
294 c15ce40c 2005-08-10 devnull
295 c15ce40c 2005-08-10 devnull /*
296 c15ce40c 2005-08-10 devnull * build histogram table
297 c15ce40c 2005-08-10 devnull */
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)
304 c15ce40c 2005-08-10 devnull break;
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 */
319 c15ce40c 2005-08-10 devnull else {
320 c15ce40c 2005-08-10 devnull bump_utf_count(r);
321 c15ce40c 2005-08-10 devnull f = Cutf; /* UTF extension */
322 c15ce40c 2005-08-10 devnull }
323 c15ce40c 2005-08-10 devnull cfreq[f]++; /* ASCII chars peg directly */
324 c15ce40c 2005-08-10 devnull }
325 c15ce40c 2005-08-10 devnull /*
326 c15ce40c 2005-08-10 devnull * gross classify
327 c15ce40c 2005-08-10 devnull */
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");
338 c15ce40c 2005-08-10 devnull return;
339 c15ce40c 2005-08-10 devnull }
340 c15ce40c 2005-08-10 devnull else guess = Fascii;
341 c15ce40c 2005-08-10 devnull /*
342 c15ce40c 2005-08-10 devnull * lookup dictionary words
343 c15ce40c 2005-08-10 devnull */
344 c15ce40c 2005-08-10 devnull memset(wfreq, 0, sizeof(wfreq));
345 fa325e9b 2020-01-10 cross if(guess == Fascii || guess == Flatin || guess == Futf)
346 c15ce40c 2005-08-10 devnull wordfreq();
347 c15ce40c 2005-08-10 devnull /*
348 c15ce40c 2005-08-10 devnull * call individual classify routines
349 c15ce40c 2005-08-10 devnull */
350 c15ce40c 2005-08-10 devnull for(i=0; call[i]; i++)
351 c15ce40c 2005-08-10 devnull if((*call[i])())
352 c15ce40c 2005-08-10 devnull return;
353 c15ce40c 2005-08-10 devnull
354 c15ce40c 2005-08-10 devnull /*
355 c15ce40c 2005-08-10 devnull * if all else fails,
356 c15ce40c 2005-08-10 devnull * print out gross classification
357 c15ce40c 2005-08-10 devnull */
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");
369 c15ce40c 2005-08-10 devnull }
370 c15ce40c 2005-08-10 devnull
371 c15ce40c 2005-08-10 devnull void
372 c15ce40c 2005-08-10 devnull bump_utf_count(Rune r)
373 c15ce40c 2005-08-10 devnull {
374 c15ce40c 2005-08-10 devnull int low, high, mid;
375 c15ce40c 2005-08-10 devnull
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++;
382 c15ce40c 2005-08-10 devnull break;
383 c15ce40c 2005-08-10 devnull } else low = mid+1;
384 c15ce40c 2005-08-10 devnull } else high = mid;
385 c15ce40c 2005-08-10 devnull }
386 c15ce40c 2005-08-10 devnull }
387 c15ce40c 2005-08-10 devnull
388 c15ce40c 2005-08-10 devnull int
389 c15ce40c 2005-08-10 devnull utf_count(void)
390 c15ce40c 2005-08-10 devnull {
391 c15ce40c 2005-08-10 devnull int i, count;
392 c15ce40c 2005-08-10 devnull
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++;
400 c15ce40c 2005-08-10 devnull break;
401 c15ce40c 2005-08-10 devnull default:
402 c15ce40c 2005-08-10 devnull break;
403 c15ce40c 2005-08-10 devnull }
404 c15ce40c 2005-08-10 devnull return count;
405 c15ce40c 2005-08-10 devnull }
406 c15ce40c 2005-08-10 devnull
407 c15ce40c 2005-08-10 devnull int
408 c15ce40c 2005-08-10 devnull chkascii(void)
409 c15ce40c 2005-08-10 devnull {
410 c15ce40c 2005-08-10 devnull int i;
411 c15ce40c 2005-08-10 devnull
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;
419 c15ce40c 2005-08-10 devnull }
420 c15ce40c 2005-08-10 devnull
421 c15ce40c 2005-08-10 devnull int
422 c15ce40c 2005-08-10 devnull find_first(char *name)
423 c15ce40c 2005-08-10 devnull {
424 c15ce40c 2005-08-10 devnull int i;
425 c15ce40c 2005-08-10 devnull
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;
431 c15ce40c 2005-08-10 devnull }
432 c15ce40c 2005-08-10 devnull
433 c15ce40c 2005-08-10 devnull void
434 c15ce40c 2005-08-10 devnull print_utf(void)
435 c15ce40c 2005-08-10 devnull {
436 c15ce40c 2005-08-10 devnull int i, printed, j;
437 c15ce40c 2005-08-10 devnull
438 c15ce40c 2005-08-10 devnull if(mime){
439 c15ce40c 2005-08-10 devnull print(PLAIN);
440 c15ce40c 2005-08-10 devnull return;
441 c15ce40c 2005-08-10 devnull }
442 c15ce40c 2005-08-10 devnull if (chkascii()) {
443 c15ce40c 2005-08-10 devnull printed = 1;
444 c15ce40c 2005-08-10 devnull print("Ascii");
445 c15ce40c 2005-08-10 devnull } else
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)
453 c15ce40c 2005-08-10 devnull break;
454 c15ce40c 2005-08-10 devnull if (language[j].count > 0)
455 c15ce40c 2005-08-10 devnull break;
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);
463 c15ce40c 2005-08-10 devnull break;
464 c15ce40c 2005-08-10 devnull case Shared:
465 c15ce40c 2005-08-10 devnull default:
466 c15ce40c 2005-08-10 devnull break;
467 c15ce40c 2005-08-10 devnull }
468 c15ce40c 2005-08-10 devnull }
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");
472 c15ce40c 2005-08-10 devnull }
473 c15ce40c 2005-08-10 devnull
474 c15ce40c 2005-08-10 devnull void
475 c15ce40c 2005-08-10 devnull wordfreq(void)
476 c15ce40c 2005-08-10 devnull {
477 c15ce40c 2005-08-10 devnull int low, high, mid, r;
478 c15ce40c 2005-08-10 devnull uchar *p, *p2, c;
479 c15ce40c 2005-08-10 devnull
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))
483 c15ce40c 2005-08-10 devnull p++;
484 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
485 c15ce40c 2005-08-10 devnull return;
486 c15ce40c 2005-08-10 devnull p2 = p;
487 c15ce40c 2005-08-10 devnull while(p < buf+nbuf && isalpha(*p))
488 c15ce40c 2005-08-10 devnull p++;
489 c15ce40c 2005-08-10 devnull c = *p;
490 c15ce40c 2005-08-10 devnull *p = 0;
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]++;
497 c15ce40c 2005-08-10 devnull break;
498 c15ce40c 2005-08-10 devnull }
499 c15ce40c 2005-08-10 devnull if(r < 0)
500 c15ce40c 2005-08-10 devnull low = mid+1;
501 c15ce40c 2005-08-10 devnull else
502 c15ce40c 2005-08-10 devnull high = mid;
503 c15ce40c 2005-08-10 devnull }
504 c15ce40c 2005-08-10 devnull *p++ = c;
505 c15ce40c 2005-08-10 devnull }
506 c15ce40c 2005-08-10 devnull }
507 c15ce40c 2005-08-10 devnull
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;
514 c15ce40c 2005-08-10 devnull };
515 c15ce40c 2005-08-10 devnull
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 3ae09bee 2018-11-16 rsc 0x04034B50, 0xFFFFFFFF, "zip archive\n", "application/zip\n",
523 c15ce40c 2005-08-10 devnull 070707, 0xFFFF, "cpio archive\n", OCTET,
524 3ae09bee 2018-11-16 rsc 0x2F7, 0xFFFF, "tex dvi\n", "application/dvi\n",
525 3ae09bee 2018-11-16 rsc 0xfffa0000, 0xfffe0000, "mp3 audio\n", "audio/mpeg\n",
526 3ebbb99c 2018-11-16 rsc 0xcafebabe, 0xFFFFFFFF, "Mach-O fat executable\n", "application/x-mach-binary\n",
527 3ebbb99c 2018-11-16 rsc 0xfeedface, 0xFFFFFFFE, "Mach-O executable\n", "application/x-mach-binary\n",
528 3ebbb99c 2018-11-16 rsc 0xbebafeca, 0xFFFFFFFF, "Java class\n", "application/x-java-applet\n",
529 c15ce40c 2005-08-10 devnull };
530 c15ce40c 2005-08-10 devnull
531 c15ce40c 2005-08-10 devnull int
532 c15ce40c 2005-08-10 devnull filemagic(Filemagic *tab, int ntab, ulong x)
533 c15ce40c 2005-08-10 devnull {
534 c15ce40c 2005-08-10 devnull int i;
535 c15ce40c 2005-08-10 devnull
536 c15ce40c 2005-08-10 devnull for(i=0; i<ntab; i++)
537 c15ce40c 2005-08-10 devnull if((x&tab[i].mask) == tab[i].x){
538 c15ce40c 2005-08-10 devnull print(mime ? tab[i].mime : tab[i].desc);
539 c15ce40c 2005-08-10 devnull return 1;
540 c15ce40c 2005-08-10 devnull }
541 c15ce40c 2005-08-10 devnull return 0;
542 c15ce40c 2005-08-10 devnull }
543 fa325e9b 2020-01-10 cross
544 c15ce40c 2005-08-10 devnull int
545 c15ce40c 2005-08-10 devnull long0(void)
546 c15ce40c 2005-08-10 devnull {
547 cbeb0b26 2006-04-01 devnull /* Fhdr *f; */
548 c15ce40c 2005-08-10 devnull long x;
549 c15ce40c 2005-08-10 devnull
550 c15ce40c 2005-08-10 devnull seek(fd, 0, 0); /* reposition to start of file */
551 c15ce40c 2005-08-10 devnull /*
552 c15ce40c 2005-08-10 devnull if(crackhdr(fd, &f)) {
553 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "%s\n", f.name);
554 c15ce40c 2005-08-10 devnull return 1;
555 c15ce40c 2005-08-10 devnull }
556 c15ce40c 2005-08-10 devnull */
557 c15ce40c 2005-08-10 devnull x = LENDIAN(buf);
558 c15ce40c 2005-08-10 devnull if(filemagic(long0tab, nelem(long0tab), x))
559 c15ce40c 2005-08-10 devnull return 1;
560 c15ce40c 2005-08-10 devnull return 0;
561 c15ce40c 2005-08-10 devnull }
562 c15ce40c 2005-08-10 devnull
563 c15ce40c 2005-08-10 devnull /* from tar.c */
564 c15ce40c 2005-08-10 devnull enum { NAMSIZ = 100, TBLOCK = 512 };
565 c15ce40c 2005-08-10 devnull
566 c15ce40c 2005-08-10 devnull union hblock
567 c15ce40c 2005-08-10 devnull {
568 c15ce40c 2005-08-10 devnull char dummy[TBLOCK];
569 c15ce40c 2005-08-10 devnull struct header
570 c15ce40c 2005-08-10 devnull {
571 c15ce40c 2005-08-10 devnull char name[NAMSIZ];
572 c15ce40c 2005-08-10 devnull char mode[8];
573 c15ce40c 2005-08-10 devnull char uid[8];
574 c15ce40c 2005-08-10 devnull char gid[8];
575 c15ce40c 2005-08-10 devnull char size[12];
576 c15ce40c 2005-08-10 devnull char mtime[12];
577 c15ce40c 2005-08-10 devnull char chksum[8];
578 c15ce40c 2005-08-10 devnull char linkflag;
579 c15ce40c 2005-08-10 devnull char linkname[NAMSIZ];
580 c15ce40c 2005-08-10 devnull /* rest are defined by POSIX's ustar format; see p1003.2b */
581 c15ce40c 2005-08-10 devnull char magic[6]; /* "ustar" */
582 c15ce40c 2005-08-10 devnull char version[2];
583 c15ce40c 2005-08-10 devnull char uname[32];
584 c15ce40c 2005-08-10 devnull char gname[32];
585 c15ce40c 2005-08-10 devnull char devmajor[8];
586 c15ce40c 2005-08-10 devnull char devminor[8];
587 c15ce40c 2005-08-10 devnull char prefix[155]; /* if non-null, path = prefix "/" name */
588 c15ce40c 2005-08-10 devnull } dbuf;
589 c15ce40c 2005-08-10 devnull };
590 c15ce40c 2005-08-10 devnull
591 c15ce40c 2005-08-10 devnull int
592 c15ce40c 2005-08-10 devnull checksum(union hblock *hp)
593 c15ce40c 2005-08-10 devnull {
594 c15ce40c 2005-08-10 devnull int i;
595 c15ce40c 2005-08-10 devnull char *cp;
596 c15ce40c 2005-08-10 devnull struct header *hdr = &hp->dbuf;
597 c15ce40c 2005-08-10 devnull
598 c15ce40c 2005-08-10 devnull for (cp = hdr->chksum; cp < &hdr->chksum[sizeof hdr->chksum]; cp++)
599 c15ce40c 2005-08-10 devnull *cp = ' ';
600 c15ce40c 2005-08-10 devnull i = 0;
601 c15ce40c 2005-08-10 devnull for (cp = hp->dummy; cp < &hp->dummy[TBLOCK]; cp++)
602 c15ce40c 2005-08-10 devnull i += *cp & 0xff;
603 c15ce40c 2005-08-10 devnull return i;
604 c15ce40c 2005-08-10 devnull }
605 c15ce40c 2005-08-10 devnull
606 c15ce40c 2005-08-10 devnull int
607 c15ce40c 2005-08-10 devnull istar(void)
608 c15ce40c 2005-08-10 devnull {
609 c15ce40c 2005-08-10 devnull int chksum;
610 c15ce40c 2005-08-10 devnull char tblock[TBLOCK];
611 c15ce40c 2005-08-10 devnull union hblock *hp = (union hblock *)tblock;
612 c15ce40c 2005-08-10 devnull struct header *hdr = &hp->dbuf;
613 c15ce40c 2005-08-10 devnull
614 c15ce40c 2005-08-10 devnull seek(fd, 0, 0); /* reposition to start of file */
615 c15ce40c 2005-08-10 devnull if (readn(fd, tblock, sizeof tblock) != sizeof tblock)
616 c15ce40c 2005-08-10 devnull return 0;
617 c15ce40c 2005-08-10 devnull chksum = strtol(hdr->chksum, 0, 8);
618 c15ce40c 2005-08-10 devnull if (hdr->name[0] != '\0' && checksum(hp) == chksum) {
619 c15ce40c 2005-08-10 devnull if (strcmp(hdr->magic, "ustar") == 0)
620 c15ce40c 2005-08-10 devnull print(mime? "application/x-ustar\n":
621 c15ce40c 2005-08-10 devnull "posix tar archive\n");
622 c15ce40c 2005-08-10 devnull else
623 c15ce40c 2005-08-10 devnull print(mime? "application/x-tar\n": "tar archive\n");
624 c15ce40c 2005-08-10 devnull return 1;
625 c15ce40c 2005-08-10 devnull }
626 c15ce40c 2005-08-10 devnull return 0;
627 c15ce40c 2005-08-10 devnull }
628 c15ce40c 2005-08-10 devnull
629 c15ce40c 2005-08-10 devnull /*
630 c15ce40c 2005-08-10 devnull * initial words to classify file
631 c15ce40c 2005-08-10 devnull */
632 c15ce40c 2005-08-10 devnull struct FILE_STRING
633 c15ce40c 2005-08-10 devnull {
634 c15ce40c 2005-08-10 devnull char *key;
635 c15ce40c 2005-08-10 devnull char *filetype;
636 c15ce40c 2005-08-10 devnull int length;
637 c15ce40c 2005-08-10 devnull char *mime;
638 c15ce40c 2005-08-10 devnull } file_string[] =
639 c15ce40c 2005-08-10 devnull {
640 c15ce40c 2005-08-10 devnull "!<arch>\n__.SYMDEF", "archive random library", 16, "application/octet-stream",
641 c15ce40c 2005-08-10 devnull "!<arch>\n", "archive", 8, "application/octet-stream",
642 c15ce40c 2005-08-10 devnull "070707", "cpio archive - ascii header", 6, "application/octet-stream",
643 c15ce40c 2005-08-10 devnull "%!", "postscript", 2, "application/postscript",
644 c15ce40c 2005-08-10 devnull "\004%!", "postscript", 3, "application/postscript",
645 c15ce40c 2005-08-10 devnull "x T post", "troff output for post", 8, "application/troff",
646 c15ce40c 2005-08-10 devnull "x T Latin1", "troff output for Latin1", 10, "application/troff",
647 c15ce40c 2005-08-10 devnull "x T utf", "troff output for UTF", 7, "application/troff",
648 c15ce40c 2005-08-10 devnull "x T 202", "troff output for 202", 7, "application/troff",
649 c15ce40c 2005-08-10 devnull "x T aps", "troff output for aps", 7, "application/troff",
650 c15ce40c 2005-08-10 devnull "GIF", "GIF image", 3, "image/gif",
651 c15ce40c 2005-08-10 devnull "\0PC Research, Inc\0", "ghostscript fax file", 18, "application/ghostscript",
652 c15ce40c 2005-08-10 devnull "%PDF", "PDF", 4, "application/pdf",
653 c15ce40c 2005-08-10 devnull "<html>\n", "HTML file", 7, "text/html",
654 c15ce40c 2005-08-10 devnull "<HTML>\n", "HTML file", 7, "text/html",
655 c15ce40c 2005-08-10 devnull "compressed\n", "Compressed image or subfont", 11, "application/octet-stream",
656 c15ce40c 2005-08-10 devnull "\111\111\052\000", "tiff", 4, "image/tiff",
657 c15ce40c 2005-08-10 devnull "\115\115\000\052", "tiff", 4, "image/tiff",
658 c15ce40c 2005-08-10 devnull "\377\330\377\340", "jpeg", 4, "image/jpeg",
659 c15ce40c 2005-08-10 devnull "\377\330\377\341", "jpeg", 4, "image/jpeg",
660 c15ce40c 2005-08-10 devnull "\377\330\377\333", "jpeg", 4, "image/jpeg",
661 5f67fe1b 2008-10-09 jas "\106\117\126\142", "x3f", 4, "image/x3f",
662 c15ce40c 2005-08-10 devnull "BM", "bmp", 2, "image/bmp",
663 c15ce40c 2005-08-10 devnull "\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1", "microsoft office document", 8, "application/octet-stream",
664 c15ce40c 2005-08-10 devnull "<MakerFile ", "FrameMaker file", 11, "application/framemaker",
665 c15ce40c 2005-08-10 devnull "\033%-12345X", "HPJCL file", 9, "application/hpjcl",
666 c15ce40c 2005-08-10 devnull "ID3", "mp3 audio with id3", 3, "audio/mpeg",
667 c15ce40c 2005-08-10 devnull 0,0,0,0
668 c15ce40c 2005-08-10 devnull };
669 c15ce40c 2005-08-10 devnull
670 c15ce40c 2005-08-10 devnull int
671 c15ce40c 2005-08-10 devnull istring(void)
672 c15ce40c 2005-08-10 devnull {
673 adee1676 2005-08-31 devnull int i, j;
674 c15ce40c 2005-08-10 devnull struct FILE_STRING *p;
675 c15ce40c 2005-08-10 devnull
676 c15ce40c 2005-08-10 devnull for(p = file_string; p->key; p++) {
677 c15ce40c 2005-08-10 devnull if(nbuf >= p->length && !memcmp(buf, p->key, p->length)) {
678 c15ce40c 2005-08-10 devnull if(mime)
679 c15ce40c 2005-08-10 devnull print("%s\n", p->mime);
680 c15ce40c 2005-08-10 devnull else
681 c15ce40c 2005-08-10 devnull print("%s\n", p->filetype);
682 c15ce40c 2005-08-10 devnull return 1;
683 c15ce40c 2005-08-10 devnull }
684 c15ce40c 2005-08-10 devnull }
685 c15ce40c 2005-08-10 devnull if(strncmp((char*)buf, "TYPE=", 5) == 0) { /* td */
686 c15ce40c 2005-08-10 devnull for(i = 5; i < nbuf; i++)
687 c15ce40c 2005-08-10 devnull if(buf[i] == '\n')
688 c15ce40c 2005-08-10 devnull break;
689 c15ce40c 2005-08-10 devnull if(mime)
690 c15ce40c 2005-08-10 devnull print(OCTET);
691 c15ce40c 2005-08-10 devnull else
692 c15ce40c 2005-08-10 devnull print("%.*s picture\n", utfnlen((char*)buf+5, i-5), (char*)buf+5);
693 c15ce40c 2005-08-10 devnull return 1;
694 c15ce40c 2005-08-10 devnull }
695 adee1676 2005-08-31 devnull if(buf[0]=='#' && buf[1]=='!'){
696 adee1676 2005-08-31 devnull i=2;
697 adee1676 2005-08-31 devnull for(j=2; j < nbuf && buf[j] != ' ' && buf[j] != '\n' && buf[j] != '\r'; j++)
698 adee1676 2005-08-31 devnull if(buf[j] == '/')
699 adee1676 2005-08-31 devnull i = j+1;
700 adee1676 2005-08-31 devnull if(mime)
701 adee1676 2005-08-31 devnull print(PLAIN);
702 adee1676 2005-08-31 devnull else
703 adee1676 2005-08-31 devnull print("%.*s executable file script\n", utfnlen((char*)buf+i, j-i), (char*)buf+i);
704 adee1676 2005-08-31 devnull return 1;
705 adee1676 2005-08-31 devnull }
706 c15ce40c 2005-08-10 devnull return 0;
707 c15ce40c 2005-08-10 devnull }
708 c15ce40c 2005-08-10 devnull
709 c15ce40c 2005-08-10 devnull int
710 c15ce40c 2005-08-10 devnull iff(void)
711 c15ce40c 2005-08-10 devnull {
712 c15ce40c 2005-08-10 devnull if (strncmp((char*)buf, "FORM", 4) == 0 &&
713 c15ce40c 2005-08-10 devnull strncmp((char*)buf+8, "AIFF", 4) == 0) {
714 c15ce40c 2005-08-10 devnull print("%s\n", mime? "audio/x-aiff": "aiff audio");
715 c15ce40c 2005-08-10 devnull return 1;
716 c15ce40c 2005-08-10 devnull }
717 c15ce40c 2005-08-10 devnull return 0;
718 c15ce40c 2005-08-10 devnull }
719 c15ce40c 2005-08-10 devnull
720 c15ce40c 2005-08-10 devnull char* html_string[] =
721 c15ce40c 2005-08-10 devnull {
722 c15ce40c 2005-08-10 devnull "title",
723 c15ce40c 2005-08-10 devnull "body",
724 c15ce40c 2005-08-10 devnull "head",
725 c15ce40c 2005-08-10 devnull "strong",
726 c15ce40c 2005-08-10 devnull "h1",
727 c15ce40c 2005-08-10 devnull "h2",
728 c15ce40c 2005-08-10 devnull "h3",
729 c15ce40c 2005-08-10 devnull "h4",
730 c15ce40c 2005-08-10 devnull "h5",
731 c15ce40c 2005-08-10 devnull "h6",
732 c15ce40c 2005-08-10 devnull "ul",
733 c15ce40c 2005-08-10 devnull "li",
734 c15ce40c 2005-08-10 devnull "dl",
735 c15ce40c 2005-08-10 devnull "br",
736 c15ce40c 2005-08-10 devnull "em",
737 c15ce40c 2005-08-10 devnull 0,
738 c15ce40c 2005-08-10 devnull };
739 c15ce40c 2005-08-10 devnull
740 c15ce40c 2005-08-10 devnull int
741 c15ce40c 2005-08-10 devnull ishtml(void)
742 c15ce40c 2005-08-10 devnull {
743 c15ce40c 2005-08-10 devnull uchar *p, *q;
744 c15ce40c 2005-08-10 devnull int i, count;
745 c15ce40c 2005-08-10 devnull
746 c15ce40c 2005-08-10 devnull /* compare strings between '<' and '>' to html table */
747 c15ce40c 2005-08-10 devnull count = 0;
748 c15ce40c 2005-08-10 devnull p = buf;
749 c15ce40c 2005-08-10 devnull for(;;) {
750 c15ce40c 2005-08-10 devnull while (p < buf+nbuf && *p != '<')
751 c15ce40c 2005-08-10 devnull p++;
752 c15ce40c 2005-08-10 devnull p++;
753 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
754 c15ce40c 2005-08-10 devnull break;
755 c15ce40c 2005-08-10 devnull if(*p == '/')
756 c15ce40c 2005-08-10 devnull p++;
757 c15ce40c 2005-08-10 devnull q = p;
758 c15ce40c 2005-08-10 devnull while(p < buf+nbuf && *p != '>')
759 c15ce40c 2005-08-10 devnull p++;
760 c15ce40c 2005-08-10 devnull if (p >= buf+nbuf)
761 c15ce40c 2005-08-10 devnull break;
762 c15ce40c 2005-08-10 devnull for(i = 0; html_string[i]; i++) {
763 c15ce40c 2005-08-10 devnull if(cistrncmp(html_string[i], (char*)q, p-q) == 0) {
764 c15ce40c 2005-08-10 devnull if(count++ > 4) {
765 c15ce40c 2005-08-10 devnull print(mime ? "text/html\n" : "HTML file\n");
766 c15ce40c 2005-08-10 devnull return 1;
767 c15ce40c 2005-08-10 devnull }
768 c15ce40c 2005-08-10 devnull break;
769 c15ce40c 2005-08-10 devnull }
770 c15ce40c 2005-08-10 devnull }
771 c15ce40c 2005-08-10 devnull p++;
772 c15ce40c 2005-08-10 devnull }
773 c15ce40c 2005-08-10 devnull return 0;
774 c15ce40c 2005-08-10 devnull }
775 c15ce40c 2005-08-10 devnull
776 c15ce40c 2005-08-10 devnull char* rfc822_string[] =
777 c15ce40c 2005-08-10 devnull {
778 c15ce40c 2005-08-10 devnull "from:",
779 c15ce40c 2005-08-10 devnull "date:",
780 c15ce40c 2005-08-10 devnull "to:",
781 c15ce40c 2005-08-10 devnull "subject:",
782 c15ce40c 2005-08-10 devnull "received:",
783 c15ce40c 2005-08-10 devnull "reply to:",
784 c15ce40c 2005-08-10 devnull "sender:",
785 c15ce40c 2005-08-10 devnull 0,
786 c15ce40c 2005-08-10 devnull };
787 c15ce40c 2005-08-10 devnull
788 c15ce40c 2005-08-10 devnull int
789 c15ce40c 2005-08-10 devnull isrfc822(void)
790 c15ce40c 2005-08-10 devnull {
791 c15ce40c 2005-08-10 devnull
792 c15ce40c 2005-08-10 devnull char *p, *q, *r;
793 c15ce40c 2005-08-10 devnull int i, count;
794 c15ce40c 2005-08-10 devnull
795 c15ce40c 2005-08-10 devnull count = 0;
796 c15ce40c 2005-08-10 devnull p = (char*)buf;
797 c15ce40c 2005-08-10 devnull for(;;) {
798 c15ce40c 2005-08-10 devnull q = strchr(p, '\n');
799 c15ce40c 2005-08-10 devnull if(q == nil)
800 c15ce40c 2005-08-10 devnull break;
801 c15ce40c 2005-08-10 devnull *q = 0;
802 c15ce40c 2005-08-10 devnull if(p == (char*)buf && strncmp(p, "From ", 5) == 0 && strstr(p, " remote from ")){
803 c15ce40c 2005-08-10 devnull count++;
804 c15ce40c 2005-08-10 devnull *q = '\n';
805 c15ce40c 2005-08-10 devnull p = q+1;
806 c15ce40c 2005-08-10 devnull continue;
807 c15ce40c 2005-08-10 devnull }
808 c15ce40c 2005-08-10 devnull *q = '\n';
809 c15ce40c 2005-08-10 devnull if(*p != '\t' && *p != ' '){
810 c15ce40c 2005-08-10 devnull r = strchr(p, ':');
811 c15ce40c 2005-08-10 devnull if(r == 0 || r > q)
812 c15ce40c 2005-08-10 devnull break;
813 c15ce40c 2005-08-10 devnull for(i = 0; rfc822_string[i]; i++) {
814 c15ce40c 2005-08-10 devnull if(cistrncmp(p, rfc822_string[i], strlen(rfc822_string[i])) == 0){
815 c15ce40c 2005-08-10 devnull count++;
816 c15ce40c 2005-08-10 devnull break;
817 c15ce40c 2005-08-10 devnull }
818 c15ce40c 2005-08-10 devnull }
819 c15ce40c 2005-08-10 devnull }
820 c15ce40c 2005-08-10 devnull p = q+1;
821 c15ce40c 2005-08-10 devnull }
822 c15ce40c 2005-08-10 devnull if(count >= 3){
823 c15ce40c 2005-08-10 devnull print(mime ? "message/rfc822\n" : "email file\n");
824 c15ce40c 2005-08-10 devnull return 1;
825 c15ce40c 2005-08-10 devnull }
826 c15ce40c 2005-08-10 devnull return 0;
827 c15ce40c 2005-08-10 devnull }
828 c15ce40c 2005-08-10 devnull
829 c15ce40c 2005-08-10 devnull int
830 c15ce40c 2005-08-10 devnull ismbox(void)
831 c15ce40c 2005-08-10 devnull {
832 c15ce40c 2005-08-10 devnull char *p, *q;
833 c15ce40c 2005-08-10 devnull
834 c15ce40c 2005-08-10 devnull p = (char*)buf;
835 c15ce40c 2005-08-10 devnull q = strchr(p, '\n');
836 c15ce40c 2005-08-10 devnull if(q == nil)
837 c15ce40c 2005-08-10 devnull return 0;
838 c15ce40c 2005-08-10 devnull *q = 0;
839 c15ce40c 2005-08-10 devnull if(strncmp(p, "From ", 5) == 0 && strstr(p, " remote from ") == nil){
840 c15ce40c 2005-08-10 devnull print(mime ? "text/plain\n" : "mail box\n");
841 c15ce40c 2005-08-10 devnull return 1;
842 c15ce40c 2005-08-10 devnull }
843 c15ce40c 2005-08-10 devnull *q = '\n';
844 c15ce40c 2005-08-10 devnull return 0;
845 c15ce40c 2005-08-10 devnull }
846 c15ce40c 2005-08-10 devnull
847 c15ce40c 2005-08-10 devnull int
848 c15ce40c 2005-08-10 devnull isc(void)
849 c15ce40c 2005-08-10 devnull {
850 c15ce40c 2005-08-10 devnull int n;
851 c15ce40c 2005-08-10 devnull
852 c15ce40c 2005-08-10 devnull n = wfreq[I1];
853 c15ce40c 2005-08-10 devnull /*
854 c15ce40c 2005-08-10 devnull * includes
855 c15ce40c 2005-08-10 devnull */
856 c15ce40c 2005-08-10 devnull if(n >= 2 && wfreq[I2] >= n && wfreq[I3] >= n && cfreq['.'] >= n)
857 c15ce40c 2005-08-10 devnull goto yes;
858 c15ce40c 2005-08-10 devnull if(n >= 1 && wfreq[Alword] >= n && wfreq[I3] >= n && cfreq['.'] >= n)
859 c15ce40c 2005-08-10 devnull goto yes;
860 c15ce40c 2005-08-10 devnull /*
861 c15ce40c 2005-08-10 devnull * declarations
862 c15ce40c 2005-08-10 devnull */
863 c15ce40c 2005-08-10 devnull if(wfreq[Cword] >= 5 && cfreq[';'] >= 5)
864 c15ce40c 2005-08-10 devnull goto yes;
865 c15ce40c 2005-08-10 devnull /*
866 c15ce40c 2005-08-10 devnull * assignments
867 c15ce40c 2005-08-10 devnull */
868 c15ce40c 2005-08-10 devnull if(cfreq[';'] >= 10 && cfreq['='] >= 10 && wfreq[Cword] >= 1)
869 c15ce40c 2005-08-10 devnull goto yes;
870 c15ce40c 2005-08-10 devnull return 0;
871 c15ce40c 2005-08-10 devnull
872 c15ce40c 2005-08-10 devnull yes:
873 c15ce40c 2005-08-10 devnull if(mime){
874 c15ce40c 2005-08-10 devnull print(PLAIN);
875 c15ce40c 2005-08-10 devnull return 1;
876 c15ce40c 2005-08-10 devnull }
877 c15ce40c 2005-08-10 devnull if(wfreq[Alword] > 0)
878 c15ce40c 2005-08-10 devnull print("alef program\n");
879 fa325e9b 2020-01-10 cross else
880 c15ce40c 2005-08-10 devnull print("c program\n");
881 c15ce40c 2005-08-10 devnull return 1;
882 c15ce40c 2005-08-10 devnull }
883 c15ce40c 2005-08-10 devnull
884 c15ce40c 2005-08-10 devnull int
885 c15ce40c 2005-08-10 devnull islimbo(void)
886 c15ce40c 2005-08-10 devnull {
887 c15ce40c 2005-08-10 devnull
888 c15ce40c 2005-08-10 devnull /*
889 c15ce40c 2005-08-10 devnull * includes
890 c15ce40c 2005-08-10 devnull */
891 c15ce40c 2005-08-10 devnull if(wfreq[Lword] < 4)
892 c15ce40c 2005-08-10 devnull return 0;
893 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "limbo program\n");
894 c15ce40c 2005-08-10 devnull return 1;
895 c15ce40c 2005-08-10 devnull }
896 c15ce40c 2005-08-10 devnull
897 c15ce40c 2005-08-10 devnull int
898 c15ce40c 2005-08-10 devnull isas(void)
899 c15ce40c 2005-08-10 devnull {
900 c15ce40c 2005-08-10 devnull
901 c15ce40c 2005-08-10 devnull /*
902 c15ce40c 2005-08-10 devnull * includes
903 c15ce40c 2005-08-10 devnull */
904 c15ce40c 2005-08-10 devnull if(wfreq[Aword] < 2)
905 c15ce40c 2005-08-10 devnull return 0;
906 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "as program\n");
907 c15ce40c 2005-08-10 devnull return 1;
908 c15ce40c 2005-08-10 devnull }
909 c15ce40c 2005-08-10 devnull
910 c15ce40c 2005-08-10 devnull /*
911 c15ce40c 2005-08-10 devnull * low entropy means encrypted
912 c15ce40c 2005-08-10 devnull */
913 c15ce40c 2005-08-10 devnull int
914 c15ce40c 2005-08-10 devnull ismung(void)
915 c15ce40c 2005-08-10 devnull {
916 c15ce40c 2005-08-10 devnull int i, bucket[8];
917 c15ce40c 2005-08-10 devnull float cs;
918 c15ce40c 2005-08-10 devnull
919 c15ce40c 2005-08-10 devnull if(nbuf < 64)
920 c15ce40c 2005-08-10 devnull return 0;
921 c15ce40c 2005-08-10 devnull memset(bucket, 0, sizeof(bucket));
922 c15ce40c 2005-08-10 devnull for(i=0; i<64; i++)
923 c15ce40c 2005-08-10 devnull bucket[(buf[i]>>5)&07] += 1;
924 c15ce40c 2005-08-10 devnull
925 c15ce40c 2005-08-10 devnull cs = 0.;
926 c15ce40c 2005-08-10 devnull for(i=0; i<8; i++)
927 c15ce40c 2005-08-10 devnull cs += (bucket[i]-8)*(bucket[i]-8);
928 c15ce40c 2005-08-10 devnull cs /= 8.;
929 c15ce40c 2005-08-10 devnull if(cs <= 24.322) {
930 c15ce40c 2005-08-10 devnull if(buf[0]==0x1f && (buf[1]==0x8b || buf[1]==0x9d))
931 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "compressed\n");
932 c15ce40c 2005-08-10 devnull else
933 c15ce40c 2005-08-10 devnull print(mime ? OCTET : "encrypted\n");
934 c15ce40c 2005-08-10 devnull return 1;
935 c15ce40c 2005-08-10 devnull }
936 c15ce40c 2005-08-10 devnull return 0;
937 c15ce40c 2005-08-10 devnull }
938 c15ce40c 2005-08-10 devnull
939 c15ce40c 2005-08-10 devnull /*
940 c15ce40c 2005-08-10 devnull * english by punctuation and frequencies
941 c15ce40c 2005-08-10 devnull */
942 c15ce40c 2005-08-10 devnull int
943 c15ce40c 2005-08-10 devnull isenglish(void)
944 c15ce40c 2005-08-10 devnull {
945 c15ce40c 2005-08-10 devnull int vow, comm, rare, badpun, punct;
946 c15ce40c 2005-08-10 devnull char *p;
947 c15ce40c 2005-08-10 devnull
948 c15ce40c 2005-08-10 devnull if(guess != Fascii && guess != Feascii)
949 c15ce40c 2005-08-10 devnull return 0;
950 c15ce40c 2005-08-10 devnull badpun = 0;
951 c15ce40c 2005-08-10 devnull punct = 0;
952 c15ce40c 2005-08-10 devnull for(p = (char *)buf; p < (char *)buf+nbuf-1; p++)
953 c15ce40c 2005-08-10 devnull switch(*p) {
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 case '%':
958 c15ce40c 2005-08-10 devnull case ';':
959 c15ce40c 2005-08-10 devnull case ':':
960 c15ce40c 2005-08-10 devnull case '?':
961 c15ce40c 2005-08-10 devnull punct++;
962 c15ce40c 2005-08-10 devnull if(p[1] != ' ' && p[1] != '\n')
963 c15ce40c 2005-08-10 devnull badpun++;
964 c15ce40c 2005-08-10 devnull }
965 c15ce40c 2005-08-10 devnull if(badpun*5 > punct)
966 c15ce40c 2005-08-10 devnull return 0;
967 c15ce40c 2005-08-10 devnull if(cfreq['>']+cfreq['<']+cfreq['/'] > cfreq['e']) /* shell file test */
968 c15ce40c 2005-08-10 devnull return 0;
969 c15ce40c 2005-08-10 devnull if(2*cfreq[';'] > cfreq['e'])
970 c15ce40c 2005-08-10 devnull return 0;
971 c15ce40c 2005-08-10 devnull
972 c15ce40c 2005-08-10 devnull vow = 0;
973 c15ce40c 2005-08-10 devnull for(p="AEIOU"; *p; p++) {
974 bce0902a 2005-08-11 devnull vow += cfreq[(uchar)*p];
975 bce0902a 2005-08-11 devnull vow += cfreq[tolower((uchar)*p)];
976 c15ce40c 2005-08-10 devnull }
977 c15ce40c 2005-08-10 devnull comm = 0;
978 c15ce40c 2005-08-10 devnull for(p="ETAION"; *p; p++) {
979 bce0902a 2005-08-11 devnull comm += cfreq[(uchar)*p];
980 bce0902a 2005-08-11 devnull comm += cfreq[tolower((uchar)*p)];
981 c15ce40c 2005-08-10 devnull }
982 c15ce40c 2005-08-10 devnull rare = 0;
983 c15ce40c 2005-08-10 devnull for(p="VJKQXZ"; *p; p++) {
984 bce0902a 2005-08-11 devnull rare += cfreq[(uchar)*p];
985 bce0902a 2005-08-11 devnull rare += cfreq[tolower((uchar)*p)];
986 c15ce40c 2005-08-10 devnull }
987 c15ce40c 2005-08-10 devnull if(vow*5 >= nbuf-cfreq[' '] && comm >= 10*rare) {
988 c15ce40c 2005-08-10 devnull print(mime ? PLAIN : "English text\n");
989 c15ce40c 2005-08-10 devnull return 1;
990 c15ce40c 2005-08-10 devnull }
991 c15ce40c 2005-08-10 devnull return 0;
992 c15ce40c 2005-08-10 devnull }
993 c15ce40c 2005-08-10 devnull
994 c15ce40c 2005-08-10 devnull /*
995 c15ce40c 2005-08-10 devnull * pick up a number with
996 c15ce40c 2005-08-10 devnull * syntax _*[0-9]+_
997 c15ce40c 2005-08-10 devnull */
998 c15ce40c 2005-08-10 devnull #define P9BITLEN 12
999 c15ce40c 2005-08-10 devnull int
1000 c15ce40c 2005-08-10 devnull p9bitnum(uchar *bp)
1001 c15ce40c 2005-08-10 devnull {
1002 c15ce40c 2005-08-10 devnull int n, c, len;
1003 c15ce40c 2005-08-10 devnull
1004 c15ce40c 2005-08-10 devnull len = P9BITLEN;
1005 c15ce40c 2005-08-10 devnull while(*bp == ' ') {
1006 c15ce40c 2005-08-10 devnull bp++;
1007 c15ce40c 2005-08-10 devnull len--;
1008 c15ce40c 2005-08-10 devnull if(len <= 0)
1009 c15ce40c 2005-08-10 devnull return -1;
1010 c15ce40c 2005-08-10 devnull }
1011 c15ce40c 2005-08-10 devnull n = 0;
1012 c15ce40c 2005-08-10 devnull while(len > 1) {
1013 c15ce40c 2005-08-10 devnull c = *bp++;
1014 c15ce40c 2005-08-10 devnull if(!isdigit(c))
1015 c15ce40c 2005-08-10 devnull return -1;
1016 c15ce40c 2005-08-10 devnull n = n*10 + c-'0';
1017 c15ce40c 2005-08-10 devnull len--;
1018 c15ce40c 2005-08-10 devnull }
1019 c15ce40c 2005-08-10 devnull if(*bp != ' ')
1020 c15ce40c 2005-08-10 devnull return -1;
1021 c15ce40c 2005-08-10 devnull return n;
1022 c15ce40c 2005-08-10 devnull }
1023 c15ce40c 2005-08-10 devnull
1024 c15ce40c 2005-08-10 devnull int
1025 c15ce40c 2005-08-10 devnull depthof(char *s, int *newp)
1026 c15ce40c 2005-08-10 devnull {
1027 c15ce40c 2005-08-10 devnull char *es;
1028 c15ce40c 2005-08-10 devnull int d;
1029 c15ce40c 2005-08-10 devnull
1030 c15ce40c 2005-08-10 devnull *newp = 0;
1031 c15ce40c 2005-08-10 devnull es = s+12;
1032 c15ce40c 2005-08-10 devnull while(s<es && *s==' ')
1033 c15ce40c 2005-08-10 devnull s++;
1034 c15ce40c 2005-08-10 devnull if(s == es)
1035 c15ce40c 2005-08-10 devnull return -1;
1036 c15ce40c 2005-08-10 devnull if('0'<=*s && *s<='9')
1037 c15ce40c 2005-08-10 devnull return 1<<atoi(s);
1038 c15ce40c 2005-08-10 devnull
1039 c15ce40c 2005-08-10 devnull *newp = 1;
1040 c15ce40c 2005-08-10 devnull d = 0;
1041 c15ce40c 2005-08-10 devnull while(s<es && *s!=' '){
1042 c15ce40c 2005-08-10 devnull s++; /* skip letter */
1043 c15ce40c 2005-08-10 devnull d += strtoul(s, &s, 10);
1044 c15ce40c 2005-08-10 devnull }
1045 fa325e9b 2020-01-10 cross
1046 c15ce40c 2005-08-10 devnull switch(d){
1047 c15ce40c 2005-08-10 devnull case 32:
1048 c15ce40c 2005-08-10 devnull case 24:
1049 c15ce40c 2005-08-10 devnull case 16:
1050 c15ce40c 2005-08-10 devnull case 8:
1051 c15ce40c 2005-08-10 devnull return d;
1052 c15ce40c 2005-08-10 devnull }
1053 c15ce40c 2005-08-10 devnull return -1;
1054 c15ce40c 2005-08-10 devnull }
1055 c15ce40c 2005-08-10 devnull
1056 c15ce40c 2005-08-10 devnull int
1057 c15ce40c 2005-08-10 devnull isp9bit(void)
1058 c15ce40c 2005-08-10 devnull {
1059 c15ce40c 2005-08-10 devnull int dep, lox, loy, hix, hiy, px, new;
1060 c15ce40c 2005-08-10 devnull ulong t;
1061 c15ce40c 2005-08-10 devnull long len;
1062 c15ce40c 2005-08-10 devnull char *newlabel;
1063 c15ce40c 2005-08-10 devnull
1064 c15ce40c 2005-08-10 devnull newlabel = "old ";
1065 c15ce40c 2005-08-10 devnull
1066 c15ce40c 2005-08-10 devnull dep = depthof((char*)buf + 0*P9BITLEN, &new);
1067 c15ce40c 2005-08-10 devnull if(new)
1068 c15ce40c 2005-08-10 devnull newlabel = "";
1069 c15ce40c 2005-08-10 devnull lox = p9bitnum(buf + 1*P9BITLEN);
1070 c15ce40c 2005-08-10 devnull loy = p9bitnum(buf + 2*P9BITLEN);
1071 c15ce40c 2005-08-10 devnull hix = p9bitnum(buf + 3*P9BITLEN);
1072 c15ce40c 2005-08-10 devnull hiy = p9bitnum(buf + 4*P9BITLEN);
1073 c15ce40c 2005-08-10 devnull if(dep < 0 || lox < 0 || loy < 0 || hix < 0 || hiy < 0)
1074 c15ce40c 2005-08-10 devnull return 0;
1075 c15ce40c 2005-08-10 devnull
1076 c15ce40c 2005-08-10 devnull if(dep < 8){
1077 c15ce40c 2005-08-10 devnull px = 8/dep; /* pixels per byte */
1078 c15ce40c 2005-08-10 devnull /* set l to number of bytes of data per scan line */
1079 c15ce40c 2005-08-10 devnull if(lox >= 0)
1080 c15ce40c 2005-08-10 devnull len = (hix+px-1)/px - lox/px;
1081 c15ce40c 2005-08-10 devnull else{ /* make positive before divide */
1082 c15ce40c 2005-08-10 devnull t = (-lox)+px-1;
1083 c15ce40c 2005-08-10 devnull t = (t/px)*px;
1084 c15ce40c 2005-08-10 devnull len = (t+hix+px-1)/px;
1085 c15ce40c 2005-08-10 devnull }
1086 c15ce40c 2005-08-10 devnull }else
1087 c15ce40c 2005-08-10 devnull len = (hix-lox)*dep/8;
1088 c15ce40c 2005-08-10 devnull len *= (hiy-loy); /* col length */
1089 c15ce40c 2005-08-10 devnull len += 5*P9BITLEN; /* size of initial ascii */
1090 c15ce40c 2005-08-10 devnull
1091 c15ce40c 2005-08-10 devnull /*
1092 c15ce40c 2005-08-10 devnull * for image file, length is non-zero and must match calculation above
1093 c15ce40c 2005-08-10 devnull * for /dev/window and /dev/screen the length is always zero
1094 c15ce40c 2005-08-10 devnull * for subfont, the subfont header should follow immediately.
1095 c15ce40c 2005-08-10 devnull */
1096 c15ce40c 2005-08-10 devnull if (len != 0 && mbuf->length == 0) {
1097 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1098 c15ce40c 2005-08-10 devnull return 1;
1099 c15ce40c 2005-08-10 devnull }
1100 c15ce40c 2005-08-10 devnull if (mbuf->length == len) {
1101 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1102 c15ce40c 2005-08-10 devnull return 1;
1103 c15ce40c 2005-08-10 devnull }
1104 c15ce40c 2005-08-10 devnull /* Ghostscript sometimes produces a little extra on the end */
1105 c15ce40c 2005-08-10 devnull if (mbuf->length < len+P9BITLEN) {
1106 c15ce40c 2005-08-10 devnull print("%splan 9 image\n", newlabel);
1107 c15ce40c 2005-08-10 devnull return 1;
1108 c15ce40c 2005-08-10 devnull }
1109 c15ce40c 2005-08-10 devnull if (p9subfont(buf+len)) {
1110 c15ce40c 2005-08-10 devnull print("%ssubfont file\n", newlabel);
1111 c15ce40c 2005-08-10 devnull return 1;
1112 c15ce40c 2005-08-10 devnull }
1113 c15ce40c 2005-08-10 devnull return 0;
1114 c15ce40c 2005-08-10 devnull }
1115 c15ce40c 2005-08-10 devnull
1116 c15ce40c 2005-08-10 devnull int
1117 c15ce40c 2005-08-10 devnull p9subfont(uchar *p)
1118 c15ce40c 2005-08-10 devnull {
1119 c15ce40c 2005-08-10 devnull int n, h, a;
1120 c15ce40c 2005-08-10 devnull
1121 c15ce40c 2005-08-10 devnull /* if image too big, assume it's a subfont */
1122 c15ce40c 2005-08-10 devnull if (p+3*P9BITLEN > buf+sizeof(buf))
1123 c15ce40c 2005-08-10 devnull return 1;
1124 c15ce40c 2005-08-10 devnull
1125 c15ce40c 2005-08-10 devnull n = p9bitnum(p + 0*P9BITLEN); /* char count */
1126 c15ce40c 2005-08-10 devnull if (n < 0)
1127 c15ce40c 2005-08-10 devnull return 0;
1128 c15ce40c 2005-08-10 devnull h = p9bitnum(p + 1*P9BITLEN); /* height */
1129 c15ce40c 2005-08-10 devnull if (h < 0)
1130 c15ce40c 2005-08-10 devnull return 0;
1131 c15ce40c 2005-08-10 devnull a = p9bitnum(p + 2*P9BITLEN); /* ascent */
1132 c15ce40c 2005-08-10 devnull if (a < 0)
1133 c15ce40c 2005-08-10 devnull return 0;
1134 c15ce40c 2005-08-10 devnull return 1;
1135 c15ce40c 2005-08-10 devnull }
1136 c15ce40c 2005-08-10 devnull
1137 c15ce40c 2005-08-10 devnull #define WHITESPACE(c) ((c) == ' ' || (c) == '\t' || (c) == '\n')
1138 c15ce40c 2005-08-10 devnull
1139 c15ce40c 2005-08-10 devnull int
1140 c15ce40c 2005-08-10 devnull isp9font(void)
1141 c15ce40c 2005-08-10 devnull {
1142 c15ce40c 2005-08-10 devnull uchar *cp, *p;
1143 c15ce40c 2005-08-10 devnull int i, n;
1144 c15ce40c 2005-08-10 devnull char pathname[1024];
1145 c15ce40c 2005-08-10 devnull
1146 c15ce40c 2005-08-10 devnull cp = buf;
1147 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* height */
1148 c15ce40c 2005-08-10 devnull return 0;
1149 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* ascent */
1150 c15ce40c 2005-08-10 devnull return 0;
1151 c15ce40c 2005-08-10 devnull for (i = 0; 1; i++) {
1152 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* min */
1153 c15ce40c 2005-08-10 devnull break;
1154 c15ce40c 2005-08-10 devnull if (!getfontnum(cp, &cp)) /* max */
1155 c15ce40c 2005-08-10 devnull return 0;
1156 c15ce40c 2005-08-10 devnull while (WHITESPACE(*cp))
1157 c15ce40c 2005-08-10 devnull cp++;
1158 c15ce40c 2005-08-10 devnull for (p = cp; *cp && !WHITESPACE(*cp); cp++)
1159 c15ce40c 2005-08-10 devnull ;
1160 c15ce40c 2005-08-10 devnull /* construct a path name, if needed */
1161 c15ce40c 2005-08-10 devnull n = 0;
1162 c15ce40c 2005-08-10 devnull if (*p != '/' && slash) {
1163 c15ce40c 2005-08-10 devnull n = slash-fname+1;
1164 c15ce40c 2005-08-10 devnull if (n < sizeof(pathname))
1165 c15ce40c 2005-08-10 devnull memcpy(pathname, fname, n);
1166 c15ce40c 2005-08-10 devnull else n = 0;
1167 c15ce40c 2005-08-10 devnull }
1168 c15ce40c 2005-08-10 devnull if (n+cp-p < sizeof(pathname)) {
1169 c15ce40c 2005-08-10 devnull memcpy(pathname+n, p, cp-p);
1170 c15ce40c 2005-08-10 devnull n += cp-p;
1171 c15ce40c 2005-08-10 devnull pathname[n] = 0;
1172 c15ce40c 2005-08-10 devnull if (access(pathname, AEXIST) < 0)
1173 c15ce40c 2005-08-10 devnull return 0;
1174 c15ce40c 2005-08-10 devnull }
1175 c15ce40c 2005-08-10 devnull }
1176 c15ce40c 2005-08-10 devnull if (i) {
1177 c15ce40c 2005-08-10 devnull print(mime ? "text/plain\n" : "font file\n");
1178 c15ce40c 2005-08-10 devnull return 1;
1179 c15ce40c 2005-08-10 devnull }
1180 c15ce40c 2005-08-10 devnull return 0;
1181 c15ce40c 2005-08-10 devnull }
1182 c15ce40c 2005-08-10 devnull
1183 c15ce40c 2005-08-10 devnull int
1184 c15ce40c 2005-08-10 devnull getfontnum(uchar *cp, uchar **rp)
1185 c15ce40c 2005-08-10 devnull {
1186 c15ce40c 2005-08-10 devnull while (WHITESPACE(*cp)) /* extract ulong delimited by whitespace */
1187 c15ce40c 2005-08-10 devnull cp++;
1188 c15ce40c 2005-08-10 devnull if (*cp < '0' || *cp > '9')
1189 c15ce40c 2005-08-10 devnull return 0;
1190 c15ce40c 2005-08-10 devnull strtoul((char *)cp, (char **)rp, 0);
1191 c15ce40c 2005-08-10 devnull if (!WHITESPACE(**rp))
1192 c15ce40c 2005-08-10 devnull return 0;
1193 c15ce40c 2005-08-10 devnull return 1;
1194 c15ce40c 2005-08-10 devnull }
1195 c15ce40c 2005-08-10 devnull
1196 c15ce40c 2005-08-10 devnull int
1197 c15ce40c 2005-08-10 devnull isrtf(void)
1198 c15ce40c 2005-08-10 devnull {
1199 c15ce40c 2005-08-10 devnull if(strstr((char *)buf, "\\rtf1")){
1200 c15ce40c 2005-08-10 devnull print(mime ? "application/rtf\n" : "rich text format\n");
1201 c15ce40c 2005-08-10 devnull return 1;
1202 c15ce40c 2005-08-10 devnull }
1203 c15ce40c 2005-08-10 devnull return 0;
1204 c15ce40c 2005-08-10 devnull }
1205 c15ce40c 2005-08-10 devnull
1206 c15ce40c 2005-08-10 devnull int
1207 c15ce40c 2005-08-10 devnull ismsdos(void)
1208 c15ce40c 2005-08-10 devnull {
1209 c15ce40c 2005-08-10 devnull if (buf[0] == 0x4d && buf[1] == 0x5a){
1210 c15ce40c 2005-08-10 devnull print(mime ? "application/x-msdownload\n" : "MSDOS executable\n");
1211 c15ce40c 2005-08-10 devnull return 1;
1212 c15ce40c 2005-08-10 devnull }
1213 c15ce40c 2005-08-10 devnull return 0;
1214 c15ce40c 2005-08-10 devnull }
1215 c15ce40c 2005-08-10 devnull
1216 c15ce40c 2005-08-10 devnull int
1217 c15ce40c 2005-08-10 devnull iself(void)
1218 c15ce40c 2005-08-10 devnull {
1219 bce0902a 2005-08-11 devnull static char *cpu[] = { /* NB: incomplete and arbitary list */
1220 bce0902a 2005-08-11 devnull nil,
1221 bce0902a 2005-08-11 devnull /*1*/ "WE32100",
1222 bce0902a 2005-08-11 devnull /*2*/ "SPARC",
1223 bce0902a 2005-08-11 devnull /*3*/ "i386",
1224 bce0902a 2005-08-11 devnull /*4*/ "M68000",
1225 bce0902a 2005-08-11 devnull /*5*/ "M88000",
1226 bce0902a 2005-08-11 devnull /*6*/ "i486",
1227 bce0902a 2005-08-11 devnull /*7*/ "i860",
1228 bce0902a 2005-08-11 devnull /*8*/ "R3000",
1229 bce0902a 2005-08-11 devnull /*9*/ "S370",
1230 bce0902a 2005-08-11 devnull /*10*/ "R4000",
1231 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1232 bce0902a 2005-08-11 devnull /*15*/ "HP-PA",
1233 bce0902a 2005-08-11 devnull nil,
1234 bce0902a 2005-08-11 devnull nil,
1235 bce0902a 2005-08-11 devnull /*18*/ "sparc v8+",
1236 bce0902a 2005-08-11 devnull /*19*/ "i960",
1237 bce0902a 2005-08-11 devnull /*20*/ "PPC-32",
1238 bce0902a 2005-08-11 devnull /*21*/ "PPC-64",
1239 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1240 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1241 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1242 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1243 bce0902a 2005-08-11 devnull /*40*/ "ARM",
1244 bce0902a 2005-08-11 devnull /*41*/ "Alpha",
1245 bce0902a 2005-08-11 devnull nil,
1246 bce0902a 2005-08-11 devnull /*43*/ "sparc v9",
1247 bce0902a 2005-08-11 devnull nil, nil,
1248 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1249 b2a905b7 2005-08-31 devnull /*50*/ "IA-64",
1250 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1251 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1252 bce0902a 2005-08-11 devnull nil,
1253 bce0902a 2005-08-11 devnull /*62*/ "AMD64",
1254 bce0902a 2005-08-11 devnull nil, nil, nil,
1255 bce0902a 2005-08-11 devnull nil, nil, nil, nil, nil,
1256 bce0902a 2005-08-11 devnull nil, nil, nil, nil,
1257 bce0902a 2005-08-11 devnull /*75*/ "VAX",
1258 c15ce40c 2005-08-10 devnull };
1259 c15ce40c 2005-08-10 devnull
1260 c15ce40c 2005-08-10 devnull
1261 adee1676 2005-08-31 devnull if (memcmp(buf, "\177ELF", 4) == 0){
1262 bce0902a 2005-08-11 devnull /* gcc misparses \x7FELF as \x7FE L F */
1263 c15ce40c 2005-08-10 devnull if (!mime){
1264 c15ce40c 2005-08-10 devnull int n = (buf[19] << 8) | buf[18];
1265 c15ce40c 2005-08-10 devnull char *p = "unknown";
1266 c15ce40c 2005-08-10 devnull
1267 c15ce40c 2005-08-10 devnull if (n > 0 && n < nelem(cpu) && cpu[n])
1268 c15ce40c 2005-08-10 devnull p = cpu[n];
1269 c15ce40c 2005-08-10 devnull else {
1270 c15ce40c 2005-08-10 devnull /* try the other byte order */
1271 c15ce40c 2005-08-10 devnull n = (buf[18] << 8) | buf[19];
1272 c15ce40c 2005-08-10 devnull if (n > 0 && n < nelem(cpu) && cpu[n])
1273 c15ce40c 2005-08-10 devnull p = cpu[n];
1274 c15ce40c 2005-08-10 devnull }
1275 c15ce40c 2005-08-10 devnull print("%s ELF executable\n", p);
1276 c15ce40c 2005-08-10 devnull }
1277 c15ce40c 2005-08-10 devnull else
1278 c15ce40c 2005-08-10 devnull print("application/x-elf-executable");
1279 c15ce40c 2005-08-10 devnull return 1;
1280 c15ce40c 2005-08-10 devnull }
1281 c15ce40c 2005-08-10 devnull
1282 c15ce40c 2005-08-10 devnull return 0;
1283 c15ce40c 2005-08-10 devnull }