|
|
1.1 root 1: #ifndef lint
2: static char sccsid[] = "@(#)file.c 4.12 (Berkeley) 11/17/85";
3: #endif
4: /*
5: * file - determine type of file
6: */
7:
8: #include <sys/param.h>
9: #include <sys/stat.h>
10: #include <stdio.h>
11: #include <ctype.h>
12: #include <a.out.h>
13: int errno;
14: int sys_nerr;
15: char *sys_errlist[];
16: int in;
17: int i = 0;
18: char buf[BUFSIZ];
19: char *troff[] = { /* new troff intermediate lang */
20: "x","T","res","init","font","202","V0","p1",0};
21: char *fort[] = {
22: "function","subroutine","common","dimension","block","integer",
23: "real","data","double",0};
24: char *asc[] = {
25: "chmk","mov","tst","clr","jmp",0};
26: char *c[] = {
27: "int","char","float","double","struct","extern",0};
28: char *as[] = {
29: "globl","byte","align","text","data","comm",0};
30: char *sh[] = {
31: "fi", "elif", "esac", "done", "export",
32: "readonly", "trap", "PATH", "HOME", 0 };
33: char *csh[] = {
34: "alias", "breaksw", "endsw", "foreach", "limit", "onintr",
35: "repeat", "setenv", "source", "path", "home", 0 };
36: int ifile;
37:
38: main(argc, argv)
39: char **argv;
40: {
41: FILE *fl;
42: register char *p;
43: char ap[MAXPATHLEN + 1];
44: extern char _sobuf[];
45:
46: if (argc < 2) {
47: fprintf(stderr, "usage: %s file ...\n", argv[0]);
48: exit(3);
49: }
50:
51: if (argc>1 && argv[1][0]=='-' && argv[1][1]=='f') {
52: if ((fl = fopen(argv[2], "r")) == NULL) {
53: perror(argv[2]);
54: exit(2);
55: }
56: while ((p = fgets(ap, sizeof ap, fl)) != NULL) {
57: int l = strlen(p);
58: if (l>0)
59: p[l-1] = '\0';
60: printf("%s: ", p);
61: type(p);
62: if (ifile>=0)
63: close(ifile);
64: }
65: exit(1);
66: }
67: while(argc > 1) {
68: printf("%s: ", argv[1]);
69: type(argv[1]);
70: fflush(stdout);
71: argc--;
72: argv++;
73: if (ifile >= 0)
74: close(ifile);
75: }
76: exit(0);
77: }
78:
79: type(file)
80: char *file;
81: {
82: int j,nl;
83: char ch;
84: struct stat mbuf;
85: char slink[MAXPATHLEN + 1];
86:
87: ifile = -1;
88: if (lstat(file, &mbuf) < 0) {
89: printf("%s\n",
90: (unsigned)errno < sys_nerr? sys_errlist[errno]: "Cannot stat");
91: return;
92: }
93: switch (mbuf.st_mode & S_IFMT) {
94:
95: case S_IFLNK:
96: printf("symbolic link");
97: j = readlink(file, slink, sizeof slink - 1);
98: if (j >= 0) {
99: slink[j] = '\0';
100: printf(" to %s", slink);
101: }
102: printf("\n");
103: return;
104:
105: case S_IFDIR:
106: if (mbuf.st_mode & S_ISVTX)
107: printf("append-only ");
108: printf("directory\n");
109: return;
110:
111: case S_IFCHR:
112: case S_IFBLK:
113: printf("%s special (%d/%d)\n",
114: (mbuf.st_mode&S_IFMT) == S_IFCHR ? "character" : "block",
115: major(mbuf.st_rdev), minor(mbuf.st_rdev));
116: return;
117:
118: case S_IFSOCK:
119: printf("socket\n");
120: return;
121: }
122:
123: ifile = open(file, 0);
124: if(ifile < 0) {
125: printf("%s\n",
126: (unsigned)errno < sys_nerr? sys_errlist[errno]: "Cannot read");
127: return;
128: }
129: in = read(ifile, buf, BUFSIZ);
130: if(in == 0){
131: printf("empty\n");
132: return;
133: }
134: switch(*(int *)buf) {
135:
136: case 0413:
137: printf("demand paged ");
138:
139: case 0410:
140: printf("pure ");
141: goto exec;
142:
143: case 0411:
144: printf("jfr or pdp-11 unix 411 executable\n");
145: return;
146:
147: case 0407:
148: exec:
149: if (mbuf.st_mode & S_ISUID)
150: printf("set-uid ");
151: if (mbuf.st_mode & S_ISGID)
152: printf("set-gid ");
153: if (mbuf.st_mode & S_ISVTX)
154: printf("sticky ");
155: printf("executable");
156: if(((int *)buf)[4] != 0) {
157: printf(" not stripped");
158: if(oldo(buf))
159: printf(" (old format symbol table)");
160: }
161: printf("\n");
162: return;
163:
164: case 0177555:
165: printf("very old archive\n");
166: return;
167:
168: case 0177545:
169: printf("old archive\n");
170: return;
171:
172: case 070707:
173: printf("cpio data\n");
174: return;
175: }
176:
177: if (buf[0] == '#' && buf[1] == '!' && shellscript(buf+2, &mbuf))
178: return;
179: if (buf[0] == '\037' && buf[1] == '\235') {
180: if (buf[2]&0x80)
181: printf("block ");
182: printf("compressed %d bit code data\n", buf[2]&0x1f);
183: return;
184: }
185: if(strncmp(buf, "!<arch>\n__.SYMDEF", 17) == 0 ) {
186: printf("archive random library\n");
187: return;
188: }
189: if (strncmp(buf, "!<arch>\n", 8)==0) {
190: printf("archive\n");
191: return;
192: }
193: if (mbuf.st_size % 512 == 0) { /* it may be a PRESS file */
194: lseek(ifile, -512L, 2); /* last block */
195: if (read(ifile, buf, BUFSIZ) > 0 && *(short *)buf == 12138) {
196: printf("PRESS file\n");
197: return;
198: }
199: }
200: i = 0;
201: if(ccom() == 0)goto notc;
202: while(buf[i] == '#'){
203: j = i;
204: while(buf[i++] != '\n'){
205: if(i - j > 255){
206: printf("data\n");
207: return;
208: }
209: if(i >= in)goto notc;
210: }
211: if(ccom() == 0)goto notc;
212: }
213: check:
214: if(lookup(c) == 1){
215: while((ch = buf[i++]) != ';' && ch != '{')if(i >= in)goto notc;
216: printf("c program text");
217: goto outa;
218: }
219: nl = 0;
220: while(buf[i] != '('){
221: if(buf[i] <= 0)
222: goto notas;
223: if(buf[i] == ';'){
224: i++;
225: goto check;
226: }
227: if(buf[i++] == '\n')
228: if(nl++ > 6)goto notc;
229: if(i >= in)goto notc;
230: }
231: while(buf[i] != ')'){
232: if(buf[i++] == '\n')
233: if(nl++ > 6)goto notc;
234: if(i >= in)goto notc;
235: }
236: while(buf[i] != '{'){
237: if(buf[i++] == '\n')
238: if(nl++ > 6)goto notc;
239: if(i >= in)goto notc;
240: }
241: printf("c program text");
242: goto outa;
243: notc:
244: i = 0;
245: while(buf[i] == 'c' || buf[i] == '#'){
246: while(buf[i++] != '\n')if(i >= in)goto notfort;
247: }
248: if(lookup(fort) == 1){
249: printf("fortran program text");
250: goto outa;
251: }
252: notfort:
253: i=0;
254: if(ascom() == 0)goto notas;
255: j = i-1;
256: if(buf[i] == '.'){
257: i++;
258: if(lookup(as) == 1){
259: printf("assembler program text");
260: goto outa;
261: }
262: else if(buf[j] == '\n' && isalpha(buf[j+2])){
263: printf("roff, nroff, or eqn input text");
264: goto outa;
265: }
266: }
267: while(lookup(asc) == 0){
268: if(ascom() == 0)goto notas;
269: while(buf[i] != '\n' && buf[i++] != ':')
270: if(i >= in)goto notas;
271: while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t')if(i++ >= in)goto notas;
272: j = i-1;
273: if(buf[i] == '.'){
274: i++;
275: if(lookup(as) == 1){
276: printf("assembler program text");
277: goto outa;
278: }
279: else if(buf[j] == '\n' && isalpha(buf[j+2])){
280: printf("roff, nroff, or eqn input text");
281: goto outa;
282: }
283: }
284: }
285: printf("assembler program text");
286: goto outa;
287: notas:
288: for(i=0; i < in; i++)if(buf[i]&0200){
289: if (buf[0]=='\100' && buf[1]=='\357')
290: printf("troff (CAT) output\n");
291: else
292: printf("data\n");
293: return;
294: }
295: if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) {
296: if (mbuf.st_mode & S_ISUID)
297: printf("set-uid ");
298: if (mbuf.st_mode & S_ISGID)
299: printf("set-gid ");
300: if (mbuf.st_mode & S_ISVTX)
301: printf("sticky ");
302: if (shell(buf, in, sh))
303: printf("shell script");
304: else if (shell(buf, in, csh))
305: printf("c-shell script");
306: else
307: printf("commands text");
308: } else if (troffint(buf, in))
309: printf("troff intermediate output text");
310: else if (shell(buf, in, sh))
311: printf("shell commands");
312: else if (shell(buf, in, csh))
313: printf("c-shell commands");
314: else if (english(buf, in))
315: printf("English text");
316: else
317: printf("ascii text");
318: outa:
319: while(i < in)
320: if((buf[i++]&0377) > 127){
321: printf(" with garbage\n");
322: return;
323: }
324: /* if next few lines in then read whole file looking for nulls ...
325: while((in = read(ifile,buf,BUFSIZ)) > 0)
326: for(i = 0; i < in; i++)
327: if((buf[i]&0377) > 127){
328: printf(" with garbage\n");
329: return;
330: }
331: /*.... */
332: printf("\n");
333: }
334:
335: oldo(cp)
336: char *cp;
337: {
338: struct exec ex;
339: struct stat stb;
340:
341: ex = *(struct exec *)cp;
342: if (fstat(ifile, &stb) < 0)
343: return(0);
344: if (N_STROFF(ex)+sizeof(off_t) > stb.st_size)
345: return (1);
346: return (0);
347: }
348:
349:
350:
351: troffint(bp, n)
352: char *bp;
353: int n;
354: {
355: int k;
356:
357: i = 0;
358: for (k = 0; k < 6; k++) {
359: if (lookup(troff) == 0)
360: return(0);
361: if (lookup(troff) == 0)
362: return(0);
363: while (i < n && buf[i] != '\n')
364: i++;
365: if (i++ >= n)
366: return(0);
367: }
368: return(1);
369: }
370: lookup(tab)
371: char *tab[];
372: {
373: char r;
374: int k,j,l;
375: while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n')i++;
376: for(j=0; tab[j] != 0; j++){
377: l=0;
378: for(k=i; ((r=tab[j][l++]) == buf[k] && r != '\0');k++);
379: if(r == '\0')
380: if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
381: || buf[k] == '{' || buf[k] == '/'){
382: i=k;
383: return(1);
384: }
385: }
386: return(0);
387: }
388: ccom(){
389: char cc;
390: while((cc = buf[i]) == ' ' || cc == '\t' || cc == '\n')if(i++ >= in)return(0);
391: if(buf[i] == '/' && buf[i+1] == '*'){
392: i += 2;
393: while(buf[i] != '*' || buf[i+1] != '/'){
394: if(buf[i] == '\\')i += 2;
395: else i++;
396: if(i >= in)return(0);
397: }
398: if((i += 2) >= in)return(0);
399: }
400: if(buf[i] == '\n')if(ccom() == 0)return(0);
401: return(1);
402: }
403: ascom(){
404: while(buf[i] == '/'){
405: i++;
406: while(buf[i++] != '\n')if(i >= in)return(0);
407: while(buf[i] == '\n')if(i++ >= in)return(0);
408: }
409: return(1);
410: }
411:
412: english (bp, n)
413: char *bp;
414: {
415: # define NASC 128
416: int ct[NASC], j, vow, freq, rare;
417: int badpun = 0, punct = 0;
418: if (n<50) return(0); /* no point in statistics on squibs */
419: for(j=0; j<NASC; j++)
420: ct[j]=0;
421: for(j=0; j<n; j++)
422: {
423: if (bp[j]<NASC)
424: ct[bp[j]|040]++;
425: switch (bp[j])
426: {
427: case '.':
428: case ',':
429: case ')':
430: case '%':
431: case ';':
432: case ':':
433: case '?':
434: punct++;
435: if ( j < n-1 &&
436: bp[j+1] != ' ' &&
437: bp[j+1] != '\n')
438: badpun++;
439: }
440: }
441: if (badpun*5 > punct)
442: return(0);
443: vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
444: freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
445: rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];
446: if (2*ct[';'] > ct['e']) return(0);
447: if ( (ct['>']+ct['<']+ct['/'])>ct['e']) return(0); /* shell file test */
448: return (vow*5 >= n-ct[' '] && freq >= 10*rare);
449: }
450:
451: shellscript(buf, sb)
452: char buf[];
453: struct stat *sb;
454: {
455: register char *tp;
456: char *cp, *xp, *index();
457:
458: cp = index(buf, '\n');
459: if (cp == 0 || cp - buf > in)
460: return (0);
461: for (tp = buf; tp != cp && isspace(*tp); tp++)
462: if (!isascii(*tp))
463: return (0);
464: for (xp = tp; tp != cp && !isspace(*tp); tp++)
465: if (!isascii(*tp))
466: return (0);
467: if (tp == xp)
468: return (0);
469: if (sb->st_mode & S_ISUID)
470: printf("set-uid ");
471: if (sb->st_mode & S_ISGID)
472: printf("set-gid ");
473: if (strncmp(xp, "/bin/sh", tp-xp) == 0)
474: xp = "shell";
475: else if (strncmp(xp, "/bin/csh", tp-xp) == 0)
476: xp = "c-shell";
477: else
478: *tp = '\0';
479: printf("executable %s script\n", xp);
480: return (1);
481: }
482:
483: shell(bp, n, tab)
484: char *bp;
485: int n;
486: char *tab[];
487: {
488:
489: i = 0;
490: do {
491: if (buf[i] == '#' || buf[i] == ':')
492: while (i < n && buf[i] != '\n')
493: i++;
494: if (++i >= n)
495: break;
496: if (lookup(tab) == 1)
497: return (1);
498: } while (i < n);
499: return (0);
500: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.