Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!decvax!ucbvax!CITHEX.CALTECH.EDU!carl
From: carl@CITHEX.CALTECH.EDU.UUCP
Newsgroups: mod.computers.vax
Subject: Re: File attributes - binary?
Message-ID: <870120220430.0a4@CitHex.Caltech.Edu>
Date: Wed, 21-Jan-87 01:04:43 EST
Article-I.D.: CitHex.870120220430.0a4
Posted: Wed Jan 21 01:04:43 1987
Date-Received: Wed, 21-Jan-87 18:50:52 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 28
Approved: info-vax@sri-kl.arpa

The routine "stat" in the C run time library fills a structure also called
"stat" that contains, among other things, the elements:

stat.st_mode:		the file "mode", part of which lets you tell if the
			file's a directory, a terminal, etc.
stat.st_fab_rfm:	the file's record format (e.g. fixed variable, or
		        stream records, and indexed, relative, or sequential
			file organization.)
stat.st_fab_rat:	record attributes (what sort of carriage control
			is associated with the records of this file, and can
		       	records cross block boundaries?

Now, "text" files tend to:
	1) Not be directories;
	2) Have a record format of either variable length or some sort of
	   stream records;
	3) Belong to files organized sequentially;
	4) Have PRINT, FORTRAN, or IMPLIED (CARRIAGE-RETURN) carriage control
	   attributes, and the blocks are generally allowed to cross block
	   boundaries.
Checking these fields in the stat structure will come about as close as
you can come to deciding whether you're dealing with a "text" file or a 
binary file without resorting to something like:
	1) Checking a few of the leading records of the file to discover
	   whether they conform to the formats of, say executable, object,
	   or library files; or
	2) Running the thing through a dictionary so see if it has may
	   recognizable words in it.