如何让perl解释代表地址的字符串变量

Question

I want to feed input to a C program with a perl script like this ./cprogram $(perl -e 'print "\\xab\\xcd\\xef";') .

However, the string must be read from a file. So I get something like this: ./cprogram $(perl -e 'open FILE, "<myfile.txt"; $file_contents = do { local $/; <FILE> }; print $file_contents' . However, now perl interprets the string as the string "\\xab\\xcd\\xef" , and I want it to interpret it as the byte sequence as in the first example.

How can this be achieved? It has to be ran on a server without File::Slurp.

Answer 1

In the first case, you pass the three bytes AB CD EF (produced by the string literal "\\xAB\\xCD\\xEF" ) to print .

In the second case, you must be passing something other than those three bytes to print . I suspect you are passing the twelve character string \\xAB\\xCD\\xEF to print .

So your question becomes: How does one convert the twelve-character string \\xAB\\xCD\\xEF into the three bytes AB CD EF . Well, you'd require some kind of parser such as

s/\\x([0-9a-fA-F][0-9a-fA-F])|\\([^x])|([^\\]+)/
   $1 ? chr(hex($1)) : $2 ? $2 : $3
/eg

And here it is at work:

$ perl -e'print "\\xAB\\xCD\\xEF";' >file

$ echo -n "$( perl -0777pe'
     s{\\x([0-9a-fA-F][0-9a-fA-F])|\\([^x])|([^\\]+)}{
        $1 ? chr(hex($1)) : $2 // $3
     }eg;
  ' file )" | od -t x1
0000000 ab cd ef
0000003

Answer 2

using a bash trick:

perl -e "$(echo "print \"$(cat input)"\")"

which for your example becomes:

./cprogram "$(perl -e "$(echo "print \"$(cat myfile.txt)"\")")"

Answer 3

Is Perl's eval too evil? If not, end in print eval("\\"$file_contents\\"");

Or can you prepare the file in advance using Perl? EG print FILE "\\xAB\\xCD\\xED"; then read the resulting file with your existing code.