filetype.c

来自「<B>Digital的Unix操作系统VAX 4.2源码</B>」· C语言 代码 · 共 943 行 · 第 1/2 页

if(lookupe(fort) == 1) {
	strcpy(Tftypes,"for");
	Type = TEXT;	/* Fortran program text */
	goto outa;

}/*E if lookupe fort */

/*
 */
notfort:
i=0;
if (!ascom()) goto notas;

j = i-1;
if (buf[i] == '.') {
	i++;
	if(lookupe(as) == 1) {
		strcpy(Tftypes,"asm");
		Type = TEXT;	/* Assembler program text */
		goto outa;

	}/*T if lookupe as .. */
	else	if(buf[j] == '\n' && isalpha(buf[j+2])) {

			strcpy(Tftypes,"rof");
			Type = TEXT;	/* roff, nroff or eqn
					 * input text.
					 */
			goto outa;
		}
}/*E if buf[i] ..*/

while (!lookupe(asc)) {
	if (!ascom()) goto notas;

	while(buf[i] != '\n' && buf[i++] != ':')
		if(i >= in) goto notas;

	while(buf[i] == '\n' || buf[i] == ' ' || buf[i] == '\t'
		|| buf[i] == '\f' || buf[i] == NL)
		if(i++ >= in) goto notas;

	j = i-1;
	if(buf[i] == '.') {
		i++;
		if(lookupe(as) == 1) {

			strcpy(Tftypes,"asm");
			Type = TEXT; /* ASM program text */
			goto outa;
		}
		else if(buf[j] == '\n' && isalpha(buf[j+2])) {

				strcpy(Tftypes,"rof");
				Type = TEXT;	/* roff, nroff, or
						 * eqn input text.
						 */
				goto outa;
		}
	}/*E if buf[i] == . */
}/* while lookupe(asc) ..*/

strcpy(Tftypes,"asm");
Type = TEXT;	/* Assembler program text */	
goto outa;

/*
 */
notas:

for (i=0; i < in; i++)
	if (buf[i] & 0200) {
		if (((int)buf[0]==(int)'\100') && ((int)buf[1]==(int)'\357')) {
			fclose(ifile);
			strcpy(Tftypes,"rof");
			return(BINARY);	/* troff (CAT) output */
		}

	fclose(ifile);
	strcpy(Tftypes,"bin");
	return(BINARY);	/* Data file of some type */

}/*E for i=0 ..*/

if (mbuf.st_mode&((S_IEXEC)|(S_IEXEC>>3)|(S_IEXEC>>6))) {

	strcpy(Tftypes,"com");
	Type = TEXT;	/* Commands text */
	goto outa;
}
	else if (troffint(in)) {

		strcpy(Tftypes,"rof");
		Type = BINARY; /* troff intermediate output text */
		}
	else if (english(buf, in)) {

		strcpy(Tftypes,"eng");
		Type = ENGLISH;	/* English text */
		}
	else	{
		strcpy(Tftypes,"asc");
		Type = TEXT;	/* Ascii text */
		}
/*
 */
outa:

while(i < in)
	if((!buf[i]) || ((buf[i++] & 0377) > 0176)) {

		/* With garbage, assume binary
		 */
		fclose(ifile);
		strcpy(Tftypes,"bin");
		return(BINARY);
	}

if (((Type == TEXT) || (Type == ENGLISH))) {
	/*
	 *	We don't want to be mislead. If we think this is
	 *	a 'text' file... Make certain of it..
	 */
	while((in = read(fileno(ifile),buf,MBUFSIZ)) > 0) {

		for(i = 0; i < in; i++)
			if((!buf[i]) || ((buf[i] & 0377) > 0176)) {

				/* With garbage, assume binary
				 */
				fclose(ifile);
				strcpy(Tftypes,"bin");
				return(BINARY);
			}
		/* If it doesn't meet the English frequency test,
		 * again assume it is TEXT of some kind.
		 */
		if (Type == ENGLISH)
			if (!(english(buf, in))) {
				/*
				 * If it fails English freq test,
				 * switch over to TEXT and continue
				 * checking for BINARY data.
				 */
				strcpy(Tftypes,"asc");
				Type = TEXT;
			}
	}/*E while in .. */
	if (in < 0) {
	    PERROR "\n%s: %s %s\n", caller, CANTRD, file);
	    perror(caller);
	    exit(FAIL);
	}

}/*E if Type == TEXT */

	/* If the above logic didn't change our mind,
	 * return the tentative file type as the actual.
	 * Convert tentative specifics to generalized types
	 * as/if required.
	 */
if (Type == ENGLISH) {
	fclose(ifile);
	return(TEXT);
}
else {
	fclose(ifile);
	return(Type);
}
}/*E Filetype() */
/**/
/*
 *
 * Function:
 *
 *	ascom
 *
 * Function Description:
 *
 *	
 *
 * Arguments:
 *
 *	none
 *
 * Return values:
 *
 *	Indication of what value(s) are returned.
 *
 *
 * Side Effects:
 *
 */

ascom()
{

while(buf[i] == '/') { 
	i++;

	while(buf[i++] != '\n')if(i >= in)
		return(0);

	while(buf[i] == '\n')if(i++ >= in)
		return(0);

}/*E while buf[i] ..*/

return(1);

}/*E ascom */
/**/
/*
 *
 * Function:
 *
 *	ccom
 *
 * Function Description:
 *
 *	This section will provide a description of the function.
 *
 * Arguments:
 *
 *	none
 *
 * Return values:
 *
 *	Indication of what value(s) are returned.
 *
 *
 * Side Effects:
 *
 *	
 */

ccom()
{

/*------*\
   Code
\*------*/

while((ch = buf[i]) == ' ' || ch == '\t' || ch == '\n'
	|| ch == '\f' || ch == NL)
	if(i++ >= in)
	{
	return(0);
}

if(buf[i] == '/' && buf[i+1] == '*') {
	i += 2;
	while(buf[i] != '*' || buf[i+1] != '/') {
		if(buf[i] == '\\')
			i += 2;
		else
			i++;

		if(i >= in) {
			return(0);
		}

	}/* while buf[i] ..*/

	if((i += 2) >= in) {
		return(0);
	}
}/*E if buf[i] ..*/

if(buf[i] == '\n')
	if(ccom() == 0) {
		return(0);
	}
return(1);

}/*E ccom() */
/**/
/*
 *
 * Function:
 *
 *	english
 *
 * Function Description:
 *
 *	This routine attempts to determine if the file contains
 *	english text based on the frequency (or lack thereof)
 *	of key letter usage in the English language.	
 *
 * Arguments:
 *
 *	char	*bp	Pointer to the buffer of text
 *	int	n	Number of characters in the buffer.
 *
 * Return values:
 *
 *	Zero	if the buffer doesn't look like english text.
 *	Non-zero if buffer appears to be English.
 *
 *
 * Side Effects:
 *
 *	none	
 */

english(bp, n)
	char	*bp;
	int	n;
{
/* Local variables
 */

int	ct[NASC], freq, rare, vow;
int	badpun = 0, punct = 0;

/*-*/

if (n<50)
	return(0); /* no point in statistics on squibs */

for(j=0; j<NASC; j++)
	ct[j]=0;

for(j=0; j<n; j++) {
	if (bp[j]<NASC)
		ct[bp[j]|040]++;

	switch (bp[j]) {

		case '.': 
		case ',': 
		case ')': 
		case '%':
		case ';': 
		case ':': 
		case '?':
			punct++;
			if ( j < n-1 &&
			    bp[j+1] != ' '  &&
			    bp[j+1] != '\f' &&
			    bp[j+1] != '\t' &&
			    bp[j+1] != NL   &&
			    bp[j+1] != '\n')
				badpun++;

	}/*E switch bp[j] */
}/*E for j=0 ..*/

if (badpun*5 > punct)
	return(0);

vow = ct['a'] + ct['e'] + ct['i'] + ct['o'] + ct['u'];
freq = ct['e'] + ct['t'] + ct['a'] + ct['i'] + ct['o'] + ct['n'];
rare = ct['v'] + ct['j'] + ct['k'] + ct['q'] + ct['x'] + ct['z'];

if (2*ct[';'] > ct['e'])
	return(0);

if ( (ct['>']+ct['<']+ct['/'])>ct['e'])
	return(0); /* shell file test */

return (vow*5 >= n-ct[' '] && freq >= 10*rare);

}/*E english() */
/**/
/*
 *
 * Function:
 *
 *	lookupe
 *
 * Function Description:
 *
 *	This section will provide a description of the function.
 *
 * Arguments:
 *
 *	char	*tab	??
 *
 * Return values:
 *
 *	Indication of what value(s) are returned.
 *
 *
 * Side Effects:
 *
 *	
 */

lookupe(tab)
	char	*tab[];
{
/* Local variables
 */

int	k, l;

/*-*/

while(buf[i] == ' ' || buf[i] == '\t' || buf[i] == '\n'
	|| buf[i] == '\f' || buf[i] == NL)
	i++;

for(j=0; tab[j] != 0; j++) {
	l=0;
	for(k=i; ((ch=tab[j][l++]) == buf[k] && ch != '\0');k++)
		;
	if(ch == '\0')
		if(buf[k] == ' ' || buf[k] == '\n' || buf[k] == '\t'
		    || buf[k] == '\f' || buf[k] == '{' ||
			 buf[k] == NL || buf[k] == '/') {
			i=k;
			return(1);

		}/*E if buf[k] ..*/
}/*E for j=0 ..*/

return(0);

}/*E lookupe() */
/**/
/*
 *
 * Function:
 *
 *	troffint
 *
 * Function Description:
 *
 *	See if the file appears to be  troff  intermediate text
 *
 * Arguments:
 *
 *	int	n	?
 *
 * Return values:
 *
 *	Indication of what value(s) are returned.
 *
 *
 * Side Effects:
 *
 *	
 */

troffint(n)
	int	n;
{

i = 0;
for (j = 0; j < 6; j++) {
	if (lookupe(troff) == 0)
		return(0);
	if (lookupe(troff) == 0)
		return(0);
	while (i < n && buf[i] != '\n')
		i++;
	if (i++ >= n)
		return(0);

}/*E for j=0 .. */

return(1);

}/*E troffint() */


/**\\**\\**\\**\\**\\**  EOM  filetype.c  **\\**\\**\\**\\**\\*/
/**\\**\\**\\**\\**\\**  EOM  filetype.c  **\\**\\**\\**\\**\\*/