flushing perl STDIN buffer?

Daravon · 05-21-2012, 11:52 AM

Perl is not doing what I expect.

I have a script that accepts 2 commandline arguments. Both of them will happen to be filenames, but according to my concept of standard input, the input to the script itself will just be the strings entered as arguments. If the commandline arguments happen to be filenames, the shell will not actually expand the commandline arguments to the contents of the files they are filenames for.

The first thing I do in the script is assign the values of $ARGV[0] and $ARGV[1] to variables. I don't need to access the commandline arguments after that.

Later in the script, I want to prompt the user for some input. The usual way I do this is

Code:

$userinput = <STDIN>;

However, when I do this, $userinput gets loaded with not just the first commandline argument I supplied, but since Perl is so eager to help, my variable actually gets loaded with the CONTENT of the file! Since my first argument is an image file, this is what I get. I guess there must be a carriage return embedded in the jpeg file after a few characters:

Code:

ÿØÿà^@^PJFIF^@^A^A^A^Aà^Aà^@^@ÿá£vExif^@^@II*^@^H^@^@^@

Anyway, is there some way I can flush the STDIN buffer after I assign the commandline arguments to variables, so that I can prompt for user input from the keyboard later?

markush · 05-21-2012, 12:26 PM

Hello Daravon,

please post your code completely. From what I have understood $userinput should only be the userinput, there must be something wrong in another place in your code.

Markus

Daravon · 05-21-2012, 01:00 PM

I don't have it finished, but this is what I have now

Code:

# USAGE: texpic <picture.jpg> <mydoc.tex>

# a script to jam pictures into \LaTeX documents. 
# it converts the picture to encapsulated postscript (EPS)
# format, then inserts a \figure{} element into the 
# \LaTeX document 

$printflag = 'N';

if ( !(2 == scalar(@ARGV)) ){
    die "expecting exactly 2 arguments. \n USAGE: texpic <picture.jpg> <mydoc.tex>";
}

$pic = $ARGV[0];
$file = $ARGV[1];
$pic =~ /(^.+)(\..*)/;
$picbase = $1;
#$suffix = $2;

#print "pic=$pic; \$1 = $1; \$2 = $2\n";
#print "file is $file";

qx( 
convert -scale 1000 $pic scaled_$pic
convert scaled_$pic $picbase.eps
cp $file .$file.tmp
);

print "enter caption:\n";

$caption = <>;   #this is where I need to read user input, but I get the contents of the filed specified by the first commandline arg!

I think this is just perl being normal. Normally you can do "while (<>)" or whatever and perl just starts reading in the first file supplied as an argument. That might be handy behavior normally but how do I stop it?

markush · 05-21-2012, 02:05 PM

As far as I understood what your script should do, it works here.

Markus

Daravon · 05-21-2012, 04:13 PM

It works if I use <STDIN>. If I use the <> operator by itself, it tries to read in the files.

Birei · 05-21-2012, 04:31 PM

Hi Daravon,

As I understand, use splice to remove everything from your argument list, and later use <> to ask for input.

Code:

my ($pic, $file) = splice @ARGV;
...
my $caption = <>;

markush · 05-21-2012, 04:41 PM

I don't understand why the OP would not use the <STDIN> instead of <>

Markus

Birei · 05-21-2012, 04:47 PM

Yes, markush. It should work using <STDIN> too.

markush · 05-21-2012, 05:06 PM

The Perldocs say that <> is only used if there are no commandline-arguments and <STDIN> in the other cases.

In my opinion the code is much more readable without <>. I would also make it a little more readable, for example:

Code:

my $usage = "expecting exactly 2 arguments. \n USAGE: texpic <picture.jpg> <mydoc.tex>" ;
die  "$usage" if @ARGV != 2 ;

my ( $pic, $file ) = @ARGV ;
$pic =~ /(^.+)(\..*)/;
my ( $picbase, $suffix ) = ( $1, $2 ) ;

# and so on

Markus