如何在Perl中读取二进制文件

Question

I'm having an issue with writing a Perl script to read a binary file.

My code is as the following whereby the $file are files in binary format. I tried to search through the web and apply in my code, tried to print it out, but it seems it doesn't work well.

Currently it only prints the '&&&&&&&&&&&" and ""ppppppppppp", but what I really want is it can print out each of the $line , so that I can do some other post processing later. Also, I'm not quite sure what the $data is as I see it is part of the code from sample in article, stating suppose to be a scalar. I need somebody who can pin point me where the error goes wrong in my code. Below is what I did.

my $tmp = "$basedir/$key";
opendir (TEMP1, "$tmp");
my @dirs = readdir(TEMP1);
closedir(TEMP1);

foreach my $dirs (@dirs) {
    next if ($dirs eq "." || $dirs eq "..");
    print "---->$dirs\n";
    my $d = "$basedir/$key/$dirs";
    if (-d "$d") {
        opendir (TEMP2, $d) || die $!;
        my @files = readdir (TEMP2); # This should read binary files
        closedir (TEMP2);

        #my $buffer = "";
        #opendir (FILE, $d) || die $!;
        #binmode (FILE);
        #my @files =  readdir (FILE, $buffer, 169108570);
        #closedir (FILE);

        foreach my $file (@files) {
            next if ($file eq "." || $file eq "..");
            my $f = "$d/$file";
            print "==>$file\n";
            open FILE, $file || die $!;
            binmode FILE;
            foreach ($line = read (FILE, $data, 169108570)) {
                print "&&&&&&&&&&&$line\n";
                print "ppppppppppp$data\n";
            }
            close FILE;
        }
    }
}

I have altered my code so that it goes like as below. Now I can read the $data. Thanks J-16 SDiZ for pointing out that. I'm trying to push the info I got from the binary file to an array called "@array", thinkking to grep data from the array for string whichever match "p04" but fail. Can someone point out where is the error?

my $tmp = "$basedir/$key";
opendir (TEMP1, "$tmp");
my @dirs = readdir (TEMP1);
closedir (TEMP1);

foreach my $dirs (@dirs) {
    next if ($dirs eq "." || $dirs eq "..");
    print "---->$dirs\n";
    my $d = "$basedir/$key/$dirs";
    if (-d "$d") {
        opendir (TEMP2, $d) || die $!;
        my @files = readdir (TEMP2); #This should read binary files
        closedir (TEMP2);

        foreach my $file (@files) {
            next if ($file eq "." || $file eq "..");
            my $f = "$d/$file";
            print "==>$file\n";
            open FILE, $file || die $!;
            binmode FILE;
            foreach ($line = read (FILE, $data, 169108570)) {
                print "&&&&&&&&&&&$line\n";
                print "ppppppppppp$data\n";
                push @array, $data;
            }
            close FILE;
        }
    }
}

foreach $item (@array) {
    #print "==>$item<==\n"; # It prints out content of binary file without the ==> and <== if I uncomment this.. weird!
    if ($item =~ /p04(.*)/) {
        print "=>$item<===============\n"; # It prints "=><===============" according to the number of binary file I have.  This is wrong that I aspect it to print the content of each binary file instead :(
        next if ($item !~ /^w+/);
        open (LOG, ">log") or die $!;
        #print LOG $item;
        close LOG;
    }
}

Again, I changed my code as following, but it still doesn't work as it do not able to grep the "p04" correctly by checking on the "log" file. It did grep the whole file including binary like this "@^@^@^@^G^D^@^@^@^^@p04bbhi06^@^^@^@^@^@^@^@^@^@hh^R^@^@^@^^@^@^@p04lohhj09^@^@^@^^@@" . What I'm aspecting is it do grep the anything with p04 only such as grepping p04bbhi06 and p04lohhj09. Here is how my code goes:-

foreach my $file (@files) {
    next if ($file eq "." || $file eq "..");
    my $f = "$d/$file";
    print "==>$file\n";
    open FILE, $f || die $!;
    binmode FILE;
    my @lines = <FILE>;
    close FILE;
    foreach $cell (@lines) {
        if ($cell =~ /b12/) {
            push @array, $cell;
        }
    }
}

#my @matches = grep /p04/, @lines;
#foreach $item (@matches) {
foreach $item (@array) {
    #print "-->$item<--";
    open (LOG, ">log") or die $!;
    print LOG $item;
    close LOG;
}

Answer 1

Use:

$line = read (FILE, $data, 169108570);

The data is in $data ; and $line is the number of bytes read.

       my $f = "$d/$file" ;
       print "==>$file\n" ;
       open FILE, $file || die $! ;

I guess the full path is in $f , but you are opening $file . (In my testing -- even $f is not the full path, but I guess you may have some other glue code...)

If you just want to walk all the files in a directory, try File::DirWalk or File::Find .

Answer 2

I am not sure if I understood you right.

If you need to read a binary file, you can do the same as for a text file:

open F, "/bin/bash";
my $file = do { local $/; <F> };
close F;

Under Windows you may need to add binmode F; under *nix it works without it.

If you need to find which lines in an array contains some word, you can use grep function:

my @matches = grep /something/, @array_to_grep;

You will get all matched lines in the new array @matches .

BTW: I don't think it's a good idea to read tons of binary files into memory at once. You can search them 1 by 1...

If you need to find where the match occurs you can use another standard function, index :

my $offset = index('myword', $file);

Answer 3

I'm not sure I'll be able to answer the OP question exactly, but here are some notes that may be related. (edit: this is the same approach as answer by @Dimanoid, but with more detail)

Say you have a file, which is a mix of ASCII data, and binary. Here is an example in a bash terminal:

$ echo -e "aa aa\x00\x0abb bb" | tee tester.txt
aa aa
bb bb
$ du -b tester.txt 
13  tester.txt
$ hexdump -C tester.txt 
00000000  61 61 20 61 61 00 0a 62  62 20 62 62 0a           |aa aa..bb bb.|
0000000d

Note that byte 00 (specified as \\x00 ) is a non-printable character, (and in C , it also means "end of a string") - thereby, its presence makes tester.txt a binary file. The file has size of 13 bytes as seen by du , because of the trailing \\n added by the echo (as it can be seen from hexdump ).

Now, let's see what happens when we try to read it with perl 's <> diamond operator (see also What's the use of <> in perl? ):

$ perl -e '
open IN, "<./tester.txt";
binmode(IN);
$data = <IN>; # does this slurp entire file in one go?
close(IN);
print "length is: " . length($data) . "\n";
print "data is: --$data--\n";
'

length is: 7
data is: --aa aa
--

Clearly, the entire file didn't get slurped - it broke at the line end \\n (and not at the binary \\x00 ). That is because the diamond filehandle <FH> operator is actually shortcut for readline (see Perl Cookbook: Chapter 8, File Contents )

The same link tells that one should undef the input record separator, \\$ (which by default is set to \\n ), in order to slurp the entire file. You may want to have this change be only local, which is why the braces and local are used instead of undef (see Perl Idioms Explained - my $string = do { local $/; }; ); so we have:

$ perl -e '
open IN, "<./tester.txt";
print "_$/_\n"; # check if $/ is \n
binmode(IN);
{
local $/; # undef $/; is global
$data = <IN>; # this should slurp one go now
};
print "_$/_\n"; # check again if $/ is \n
close(IN);
print "length is: " . length($data) . "\n";
print "data is: --$data--\n";
'

_
_
_
_
length is: 13
data is: --aa aa
bb bb
--

... and now we can see the file is slurped in its entirety.

Since binary data implies unprintable characters, you may want to inspect the actual contents of $data by printing via sprintf or pack / unpack instead.

Hope this helps someone,
Cheers!