file.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

#include <u.h>
#include <libc.h>
#include <bio.h>
#include <ctype.h>
#include <mach.h>

/*
 * file - determine type of file
 */
#define	LENDIAN(p)	((p)[0] | ((p)[1]<<8) | ((p)[2]<<16) | ((p)[3]<<24))

uchar	buf[6001];
short	cfreq[140];
short	wfreq[50];
int	nbuf;
Dir*	mbuf;
int	fd;
char 	*fname;
char	*slash;

enum
{
	Cword,
	Fword,
	Aword,
	Alword,
	Lword,
	I1,
	I2,
	I3,
	Clatin	= 128,
	Cbinary,
	Cnull,
	Ceascii,
	Cutf,
};
struct
{
	char*	word;
	int	class;
} dict[] =
{
	"PATH",		Lword,
	"TEXT",		Aword,
	"adt",		Alword,
	"aggr",		Alword,
	"alef",		Alword,
	"array",	Lword,
	"block",	Fword,
	"chan",		Alword,
	"char",		Cword,
	"common",	Fword,
	"con",		Lword,
	"data",		Fword,
	"dimension",	Fword,
	"double",	Cword,
	"extern",	Cword,
	"bio",		I2,
	"float",	Cword,
	"fn",		Lword,
	"function",	Fword,
	"h",		I3,
	"implement",	Lword,
	"import",	Lword,
	"include",	I1,
	"int",		Cword,
	"integer",	Fword,
	"iota",		Lword,
	"libc",		I2,
	"long",		Cword,
	"module",	Lword,
	"real",		Fword,
	"ref",		Lword,
	"register",	Cword,
	"self",		Lword,
	"short",	Cword,
	"static",	Cword,
	"stdio",	I2,
	"struct",	Cword,
	"subroutine",	Fword,
	"u",		I2,
	"void",		Cword,
};

/* codes for 'mode' field in language structure */
enum	{
		Normal	= 0,
		First,		/* first entry for language spanning several ranges */
		Multi,		/* later entries "   "       "  ... */
		Shared,		/* codes used in several languages */
	};

struct
{
	int	mode;		/* see enum above */
	int 	count;
	int	low;
	int	high;
	char	*name;

} language[] =
{
	Normal,	0,	0x0100,	0x01FF,	"Extended Latin",
	Normal,	0,	0x0370,	0x03FF,	"Greek",
	Normal,	0,	0x0400,	0x04FF,	"Cyrillic",
	Normal,	0,	0x0530,	0x058F,	"Armenian",
	Normal,	0,	0x0590,	0x05FF,	"Hebrew",
	Normal,	0,	0x0600,	0x06FF,	"Arabic",
	Normal,	0,	0x0900,	0x097F,	"Devanagari",
	Normal,	0,	0x0980,	0x09FF,	"Bengali",
	Normal,	0,	0x0A00,	0x0A7F,	"Gurmukhi",
	Normal,	0,	0x0A80,	0x0AFF,	"Gujarati",
	Normal,	0,	0x0B00,	0x0B7F,	"Oriya",
	Normal,	0,	0x0B80,	0x0BFF,	"Tamil",
	Normal,	0,	0x0C00,	0x0C7F,	"Telugu",
	Normal,	0,	0x0C80,	0x0CFF,	"Kannada",
	Normal,	0,	0x0D00,	0x0D7F,	"Malayalam",
	Normal,	0,	0x0E00,	0x0E7F,	"Thai",
	Normal,	0,	0x0E80,	0x0EFF,	"Lao",
	Normal,	0,	0x1000,	0x105F,	"Tibetan",
	Normal,	0,	0x10A0,	0x10FF,	"Georgian",
	Normal,	0,	0x3040,	0x30FF,	"Japanese",
	Normal,	0,	0x3100,	0x312F,	"Chinese",
	First,	0,	0x3130,	0x318F,	"Korean",
	Multi,	0,	0x3400,	0x3D2F,	"Korean",
	Shared,	0,	0x4e00,	0x9fff,	"CJK",
	Normal,	0,	0,	0,	0,		/* terminal entry */
};


enum
{
	Fascii,		/* printable ascii */
	Flatin,		/* latin 1*/
	Futf,		/* UTF character set */
	Fbinary,	/* binary */
	Feascii,	/* ASCII with control chars */
	Fnull,		/* NULL in file */
} guess;

void	bump_utf_count(Rune);
int	cistrncmp(char*, char*, int);
void	filetype(int);
int	getfontnum(uchar*, uchar**);
int	isas(void);
int	isc(void);
int	iscint(void);
int	isenglish(void);
int	ishp(void);
int	ishtml(void);
int	isrfc822(void);
int	ismbox(void);
int	islimbo(void);
int	ismung(void);
int	isp9bit(void);
int	isp9font(void);
int	isrtf(void);
int	ismsdos(void);
int	iself(void);
int	istring(void);
int	isoffstr(void);
int	iff(void);
int	long0(void);
int	longoff(void);
int	istar(void);
int	isface(void);
int	isexec(void);
int	p9bitnum(uchar*);
int	p9subfont(uchar*);
void	print_utf(void);
void	type(char*, int);
int	utf_count(void);
void	wordfreq(void);

int	(*call[])(void) =
{
	long0,		/* recognizable by first 4 bytes */
	istring,	/* recognizable by first string */
	iself,		/* ELF (foreign) executable */
	isexec,		/* native executables */
	iff,		/* interchange file format (strings) */
	longoff,	/* recognizable by 4 bytes at some offset */
	isoffstr,	/* recognizable by string at some offset */
	isrfc822,	/* email file */
	ismbox,		/* mail box */
	istar,		/* recognizable by tar checksum */
	ishtml,		/* html keywords */
	iscint,		/* compiler/assembler intermediate */
	islimbo,	/* limbo source */
	isc,		/* c & alef compiler key words */
	isas,		/* assembler key words */
	ismung,		/* entropy compressed/encrypted */
	isp9font,	/* plan 9 font */
	isp9bit,	/* plan 9 image (as from /dev/window) */
	isenglish,	/* char frequency English */
	isrtf,		/* rich text format */
	ismsdos,	/* msdos exe (virus file attachement) */
	isface,		/* ascii face file */
	0
};

int mime;

#define OCTET	"application/octet-stream\n"
#define PLAIN	"text/plain\n"

void
main(int argc, char *argv[])
{
	int i, j, maxlen;
	char *cp;
	Rune r;

	ARGBEGIN{
	case 'm':
		mime = 1;
		break;
	default:
		fprint(2, "usage: file [-m] [file...]\n");
		exits("usage");
	}ARGEND;

	maxlen = 0;
	if(mime == 0 || argc > 1){
		for(i = 0; i < argc; i++) {
			for (j = 0, cp = argv[i]; *cp; j++, cp += chartorune(&r, cp))
					;
			if(j > maxlen)
				maxlen = j;
		}
	}
	if (argc <= 0) {
		if(!mime)
			print ("stdin: ");
		filetype(0);
	}
	else {
		for(i = 0; i < argc; i++)
			type(argv[i], maxlen);
	}
	exits(0);
}

void
type(char *file, int nlen)
{
	Rune r;
	int i;
	char *p;

	if(nlen > 0){
		slash = 0;
		for (i = 0, p = file; *p; i++) {
			if (*p == '/')			/* find rightmost slash */
				slash = p;
			p += chartorune(&r, p);		/* count runes */
		}
		print("%s:%*s",file, nlen-i+1, "");
	}
	fname = file;
	if ((fd = open(file, OREAD)) < 0) {
		print("cannot open\n");
		return;
	}
	filetype(fd);
	close(fd);
}

/*
 * Unicode 4.0 4-byte runes.
 */
typedef int Rune1;

enum {
	UTFmax1 = 4,
};

int
fullrune1(char *p, int n)
{
	int c;
	
	if(n >= 1) {
		c = *(uchar*)p;
		if(c < 0x80)
			return 1;
		if(n >= 2 && c < 0xE0)
			return 1;
		if(n >= 3 && c < 0xF0)
			return 1;
		if(n >= 4)
			return 1;
	}
	return 0;
}

int
chartorune1(Rune1 *rune, char *str)
{
	int c, c1, c2, c3, n;
	Rune r;
	
	c = *(uchar*)str;
	if(c < 0xF0){
		r = 0;
		n = chartorune(&r, str);
		*rune = r;
		return n;
	}
	c &= ~0xF0;
	c1 = *(uchar*)(str+1) & ~0x80;
	c2 = *(uchar*)(str+2) & ~0x80;
	c3 = *(uchar*)(str+3) & ~0x80;
	n = (c<<18) | (c1<<12) | (c2<<6) | c3;
	if(n < 0x10000 || n > 0x10FFFF){
		*rune = Runeerror;
		return 1;
	}
	*rune = n;
	return 4;
}

void
filetype(int fd)
{
	Rune1 r;
	int i, f, n;
	char *p, *eob;

	free(mbuf);
	mbuf = dirfstat(fd);
	if(mbuf == nil){
		print("cannot stat: %r\n");
		return;
	}
	if(mbuf->mode & DMDIR) {
		print(mime ? "text/directory\n" : "directory\n");
		return;
	}
	if(mbuf->type != 'M' && mbuf->type != '|') {
		print(mime ? OCTET : "special file #%c/%s\n",
			mbuf->type, mbuf->name);
		return;
	}
	nbuf = read(fd, buf, sizeof(buf)-1);

	if(nbuf < 0) {
		print("cannot read\n");
		return;
	}
	if(nbuf == 0) {
		print(mime ? PLAIN : "empty file\n");
		return;
	}
	buf[nbuf] = 0;

	/*
	 * build histogram table
	 */
	memset(cfreq, 0, sizeof(cfreq));
	for (i = 0; language[i].name; i++)
		language[i].count = 0;
	eob = (char *)buf+nbuf;
	for(n = 0, p = (char *)buf; p < eob; n++) {
		if (!fullrune1(p, eob-p) && eob-p < UTFmax1)
			break;
		p += chartorune1(&r, p);
		if (r == 0)
			f = Cnull;
		else if (r <= 0x7f) {
			if (!isprint(r) && !isspace(r))
				f = Ceascii;	/* ASCII control char */
			else f = r;
		} else if (r == 0x80) {
			bump_utf_count(r);
			f = Cutf;
		} else if (r < 0xA0)
			f = Cbinary;	/* Invalid Runes */
		else if (r <= 0xff)
			f = Clatin;	/* Latin 1 */
		else {
			bump_utf_count(r);
			f = Cutf;		/* UTF extension */
		}
		cfreq[f]++;			/* ASCII chars peg directly */
	}
	/*
	 * gross classify
	 */
	if (cfreq[Cbinary])
		guess = Fbinary;
	else if (cfreq[Cutf])
		guess = Futf;
	else if (cfreq[Clatin])
		guess = Flatin;
	else if (cfreq[Ceascii])
		guess = Feascii;
	else if (cfreq[Cnull])
		guess = Fbinary;
	else
		guess = Fascii;
	/*
	 * lookup dictionary words
	 */
	memset(wfreq, 0, sizeof(wfreq));
	if(guess == Fascii || guess == Flatin || guess == Futf)
		wordfreq();
	/*
	 * call individual classify routines
	 */
	for(i=0; call[i]; i++)
		if((*call[i])())
			return;

	/*
	 * if all else fails,
	 * print out gross classification
	 */
	if (nbuf < 100 && !mime)
		print(mime ? PLAIN : "short ");
	if (guess == Fascii)
		print(mime ? PLAIN : "Ascii\n");
	else if (guess == Feascii)
		print(mime ? PLAIN : "extended ascii\n");
	else if (guess == Flatin)
		print(mime ? PLAIN : "latin ascii\n");
	else if (guess == Futf && utf_count() < 4)
		print_utf();
	else print(mime ? OCTET : "binary\n");
}

void
bump_utf_count(Rune r)
{
	int low, high, mid;

	high = sizeof(language)/sizeof(language[0])-1;
	for (low = 0; low < high;) {
		mid = (low+high)/2;
		if (r >= language[mid].low) {
			if (r <= language[mid].high) {
				language[mid].count++;
				break;
			} else low = mid+1;
		} else high = mid;
	}
}

int
utf_count(void)
{
	int i, count;

	count = 0;
	for (i = 0; language[i].name; i++)
		if (language[i].count > 0)
			switch (language[i].mode) {
			case Normal:
			case First:
				count++;
				break;
			default:
				break;
			}
	return count;
}

int
chkascii(void)
{
	int i;

	for (i = 'a'; i < 'z'; i++)
		if (cfreq[i])
			return 1;
	for (i = 'A'; i < 'Z'; i++)
		if (cfreq[i])
			return 1;
	return 0;
}

int
find_first(char *name)
{
	int i;

	for (i = 0; language[i].name != 0; i++)
		if (language[i].mode == First
			&& strcmp(language[i].name, name) == 0)
			return i;
	return -1;
}

void
print_utf(void)
{
	int i, printed, j;

	if(mime){
		print(PLAIN);
		return;
	}
	if (chkascii()) {
		printed = 1;
		print("Ascii");
	} else
		printed = 0;
	for (i = 0; language[i].name; i++)
		if (language[i].count) {
			switch(language[i].mode) {
			case Multi:
				j = find_first(language[i].name);
				if (j < 0)
					break;
				if (language[j].count > 0)
					break;
				/* Fall through */
			case Normal:
			case First:
				if (printed)
					print(" & ");
				else printed = 1;
				print("%s", language[i].name);
				break;
			case Shared:
			default:
				break;
			}
		}
	if(!printed)
		print("UTF");
	print(" text\n");
}

void
wordfreq(void)
{
	int low, high, mid, r;
	uchar *p, *p2, c;

	p = buf;
	for(;;) {
		while (p < buf+nbuf && !isalpha(*p))
			p++;
		if (p >= buf+nbuf)
			return;
		p2 = p;
		while(p < buf+nbuf && isalpha(*p))
			p++;
		c = *p;
		*p = 0;
		high = sizeof(dict)/sizeof(dict[0]);
		for(low = 0;low < high;) {
			mid = (low+high)/2;
			r = strcmp(dict[mid].word, (char*)p2);
			if(r == 0) {
				wfreq[dict[mid].class]++;
				break;
			}
			if(r < 0)
				low = mid+1;
			else
				high = mid;
		}
		*p++ = c;
	}
}

typedef struct Filemagic Filemagic;
struct Filemagic {
	ulong x;
	ulong mask;
	char *desc;
	char *mime;
};

/*
 * integers in this table must be as seen on a little-endian machine
 * when read from a file.
 */
Filemagic long0tab[] = {
	0xF16DF16D,	0xFFFFFFFF,	"pac1 audio file\n",	OCTET,
	/* "pac1" */
	0x31636170,	0xFFFFFFFF,	"pac3 audio file\n",	OCTET,
	/* "pXc2 */
	0x32630070,	0xFFFF00FF,	"pac4 audio file\n",	OCTET,
	0xBA010000,	0xFFFFFFFF,	"mpeg system stream\n",	OCTET,
	0x30800CC0,	0xFFFFFFFF,	"inferno .dis executable\n", OCTET,
	0x04034B50,	0xFFFFFFFF,	"zip archive\n", "application/zip",
	070707,		0xFFFF,		"cpio archive\n", OCTET,
	0x2F7,		0xFFFF,		"tex dvi\n", "application/dvi",
	0xfaff,		0xfeff,		"mp3 audio\n",	"audio/mpeg",
	0xfeff0000,	0xffffffff,	"utf-32be\n",	"text/plain charset=utf-32be",
	0xfffe,		0xffffffff,	"utf-32le\n",	"text/plain charset=utf-32le",
	0xfeff,		0xffff,		"utf-16be\n",	"text/plain charset=utf-16be",
	0xfffe,		0xffff,		"utf-16le\n",	"text/plain charset=utf-16le",
	/*
	 * venti & fossil magic numbers are stored big-endian on disk,
	 * thus the numbers appear reversed in this table.
	 */
	0xad4e5cd1,	0xFFFFFFFF,	"venti arena\n", OCTET,
};

int
filemagic(Filemagic *tab, int ntab, ulong x)
{
	int i;

	for(i=0; i<ntab; i++)
		if((x&tab[i].mask) == tab[i].x){
			print(mime ? tab[i].mime : tab[i].desc);
			return 1;
		}
	return 0;
}

int
long0(void)
{
	return filemagic(long0tab, nelem(long0tab), LENDIAN(buf));
}

typedef struct Fileoffmag Fileoffmag;
struct Fileoffmag {
	ulong	off;
	Filemagic;
};

/*
 * integers in this table must be as seen on a little-endian machine
 * when read from a file.
 */
Fileoffmag longofftab[] = {
	/*
	 * venti & fossil magic numbers are stored big-endian on disk,
	 * thus the numbers appear reversed in this table.
	 */
	256*1024, 0xe7a5e4a9, 0xFFFFFFFF, "venti arenas partition\n", OCTET,
	256*1024, 0xc75e5cd1, 0xFFFFFFFF, "venti index section\n", OCTET,
	128*1024, 0x89ae7637, 0xFFFFFFFF, "fossil write buffer\n", OCTET,
};

int
fileoffmagic(Fileoffmag *tab, int ntab)
{
	int i;
	ulong x;
	Fileoffmag *tp;
	uchar buf[sizeof(long)];

	for(i=0; i<ntab; i++) {
		tp = tab + i;
		seek(fd, tp->off, 0);
		if (read(fd, buf, sizeof buf) != sizeof buf)
			continue;
		x = LENDIAN(buf);
		if((x&tp->mask) == tp->x){
			print(mime? tp->mime: tp->desc);
			return 1;
		}
	}
	return 0;
}

int
longoff(void)
{
	return fileoffmagic(longofftab, nelem(longofftab));
}

int
isexec(void)
{
	Fhdr f;

	seek(fd, 0, 0);		/* reposition to start of file */
	if(crackhdr(fd, &f)) {
		print(mime ? OCTET : "%s\n", f.name);
		return 1;
	}
	return 0;
}


/* from tar.c */
enum { NAMSIZ = 100, TBLOCK = 512 };

union	hblock
{
	char	dummy[TBLOCK];
	struct	header
	{
		char	name[NAMSIZ];
		char	mode[8];
		char	uid[8];
		char	gid[8];
		char	size[12];
		char	mtime[12];
		char	chksum[8];
		char	linkflag;
		char	linkname[NAMSIZ];
		/* rest are defined by POSIX's ustar format; see p1003.2b */
		char	magic[6];	/* "ustar" */
		char	version[2];
		char	uname[32];
		char	gname[32];
		char	devmajor[8];
		char	devminor[8];
		char	prefix[155];  /* if non-null, path = prefix "/" name */
	} dbuf;
};

int
checksum(union hblock *hp)
{
	int i;
	char *cp;
	struct header *hdr = &hp->dbuf;

	for (cp = hdr->chksum; cp < &hdr->chksum[sizeof hdr->chksum]; cp++)
		*cp = ' ';
	i = 0;
	for (cp = hp->dummy; cp < &hp->dummy[TBLOCK]; cp++)
		i += *cp & 0xff;
	return i;
}

int
istar(void)
{
	int chksum;
	char tblock[TBLOCK];
	union hblock *hp = (union hblock *)tblock;
	struct header *hdr = &hp->dbuf;

	seek(fd, 0, 0);		/* reposition to start of file */
	if (readn(fd, tblock, sizeof tblock) != sizeof tblock)
		return 0;
	chksum = strtol(hdr->chksum, 0, 8);
	if (hdr->name[0] != '\0' && checksum(hp) == chksum) {
		if (strcmp(hdr->magic, "ustar") == 0)
			print(mime? "application/x-ustar\n":
				"posix tar archive\n");
		else
			print(mime? "application/x-tar\n": "tar archive\n");
		return 1;
	}
	return 0;
}

/*
 * initial words to classify file
 */
struct	FILE_STRING
{