Use sed to read file into middle of line - perl

Question

File A contains this text (assume that "alpha" and "bravo" are arbitrarily long chunks on a single line):
alpha {FOO} bravo
File B contains an arbitrary amount of text, including all sorts of wacky characters.
I want to replace the string "{FOO}" in file A with the contents of file B. Using sed's 'r' command as follows doesn't work because it inserts the content of file B after that line:
cat A | sed -e "/{FOO}/r B"
Is there any way, using sed, to end up with a file that consists of:
alpha [the contents of B] bravo
? If it would be easier to do this with, say, perl, that's fine too. But I know even less about perl than I do about sed. ;)

Answer 1

Short perl solution:
FOO="$( cat replacement.txt )" perl -pe's/\{FOO\}/$ENV{FOO}/g'
This will work with any character except 00 NUL. If you have to deal with binary files, you can use:
perl -pe'
BEGIN {
open(my $fh, "<", shift(#ARGV)) or die $!;
local $/;
$FOO = <$fh>;
}
s/\{FOO\}/$FOO/g
' replacement.txt
Usage:
perl -i~ -pe'...' file # In-place edit with backup
perl -i -pe'...' file # In-place edit without backup
perl -pe'...' file.in >file.out # Read from named file(s)
perl -pe'...' <file.in >file.out # Read from STDIN

Answer 2

If this is Bash on Linux, this seems to work:
sed -i "s/{FOO}/$(cat B.txt)/g" A.txt
This will directly edit the file A.txt - they don't have to be .txt files, I just added those in to make it more obvious.
As #ikegami points out, this will have problems if there any / in the file - also any \ will probably be ignored.
So in an attempt to solve that, you should be able to use:
sed -i "s/\//%2F/g" B.txt
sed -i "s/{FOO}/$(cat B.txt)/g" A.txt
sed -i "s/%2F/\//g" B.txt
You won't have to use %2F though.

Answer 3

#!/usr/bin/perl
use strict;
use warnings;
open my $fh_a, '<', $ARGV[0] or die "Failed to open $ARGV[0] for reading";
open my $fh_b, '<', $ARGV[1] or die "Failed to open $ARGV[1] for reading";
my $a;
my $b;
{
local $/;
$a = <$fh_a>;
$b = <$fh_b>;
}
close $fh_a;
close $fh_b;
$a =~ s/{FOO}/$b/;
print $a;
As long as your files both fit in memory twice, this should be fine. The local $/; puts the I/O system into 'slurp' mode, reading the whole file in a single operation.
Usage:
perl replace_foo.pl fileA fileB