radare-common 1:1.5.2-4 / usr / lib / radare / bin / bin2xml

#!/usr/bin/env perl
#
# Reverse Engineering tool that parses a binary
# supported by objdump and prints out a report
# analysis.
#
# A txt2bin tool should be written, or at least
# a context from the assembly code
#
# Usage:
#   $ SYNTAX=intel rsc bin2txt /bin/ls > ls.txt
#
# author: pancake
#
# TODO: Support for arm and powerpc
# TODO: Find lost calls and unaligned ones (follow them)
#
my $VERSION = "0.1";

use warnings;
use strict;

my $syntax = $ENV{"SYNTAX"};
$syntax="att" unless ($syntax);
my $objdump = $ENV{"OBJDUMP"};
$objdump="objdump" unless ($objdump);
my $target = $ARGV[0];
$target = "-h" unless ($target);
if ($target eq "-h") {
	print "Usage: bin2txt [a.out] > a.out.txt\n";
	print "Environment: OBJDUMP SYNTAX\n";
	exit 1;
}
unless (-f $target) {
	print "Cannot find $target.\n";
	exit 1;
}

# TODO USE HASHER
# rahash -a md5 -qvf $file
my $md5bin="md5sum";
$md5bin="md5" if (`md5 -x 2>/dev/null`);
chomp (my $md5 = qx($md5bin $target));
$md5=~/([a-z0-9]*)/; $md5 = $1;
chomp(my $date = qx(date));
my $size=(lstat($target))[7];

print "<radarexml>\n";

print " <generator name=\"bin2xml\" version=\"$VERSION\" />\n";
print " <metadata>\n";
print "  <filename>$target</filename>\n";
print "  <size>$size</size>\n";
print "  <md5>$md5</md5>\n";
print "  <date>$date</date>\n"; # XXX (`date` is br0ken)

###########
# base address
###########

my $str = qx(readelf -e $target |grep PHDR | sed -e 's,  , ,g');
while($str=~/\ \ /) { $str=~s/\ \ /\ /g; }
my ($magic, $one, $two, $tri)=split(/ /,$str);
my $baddr;
eval("\$baddr=$tri-$two;");
#printf("BASE ADDRESS: _%08x\n", $baddr);
print "  <base address=\"$baddr\" />\n";
print " </metadata>\n";

###########
# elf header
###########
my $entry;
print "<header type=\"elf\">\n";
print " <readelf>\n";
#system("readelf -h $target");
open(PD, "readelf -h $target|") or die("brbrb");
while(<PD>) {
	my $arg = $_;
	my ($key,$value)=split(/:/);
	unless ($value) {
		$arg=~/_(^\ )* <(.*)>/;
		$key = $1;
		$value = $2;
	}
	
	if ($key=~/Entry point/) {
		$entry = $value;
		chomp($entry);
	}
	if ($value=~/0x/) {
		$value=~s/0x/_0/;
#		$value=~tr/[a-z]/[A-Z]/;
	}
	print "$key: $value";
}
close(PD);
print " </readelf>\n";

my $entry_seek;
eval("\$entry_seek=$entry-$baddr");
print " <entry>$entry_seek</entry>\n";
print "</header>\n\n";


###########
# elf sections
###########

my $od = sprintf ".bin2txt.%08x", rand 0xffffffff;
my $od2 = sprintf ".bin2txt.%08x", rand 0xffffffff;
system("$objdump -M $syntax -wd $target > $od");

if (0) {
my @sections_offset = split(' ', qx(cat $od | grep '<.' | grep '>:' | sed -e 's,:,,g' -e 's,^\ ,0,g' -e 's,\t, ,g'));

print "\nELF sections: {{{\n";
my $i;
for($i=0;$i<$#sections_offset;$i+=2) {
	#$sections_offset[$i]=~tr/[a-z]/[A-Z]/;
	my $section_offset = $sections_offset[$i];
	$sections_offset[$i+1]=~s/[\<\>]//g;
	my $section_name   = $sections_offset[$i+1];
	next if ($section_name=~/plt/);
	printf(" _$section_offset $section_name\n", $section_offset);
}
print "}}}\n";

###########
# PLT
###########
print "ELF PLT: {{{\n";
for($i=0;$i<$#sections_offset;$i+=2) {
	my $section_offset = $sections_offset[$i];
	my $section_name   = $sections_offset[$i+1];
	next unless ($section_name=~/plt/);
	printf(" _$section_offset $section_name\n", $section_offset);
}
print "}}}\n";


###########
# code entries
###########

my @entries=split(' ',`cat $od | grep -e jmp -e call | grep -v '>:' | sed -e 's,:,,g' -e 's,^\ ,0,g' -e 's,\t, ,g' -e 's,^,0x,g' | cut -d ' ' -f 1 | xargs printf "%08x\n"|uniq`);

# add program syms
my @entries2=split("\n",`cat $od | grep '>:'`);
foreach $entry (@entries2) {
	push @entries, (split('<',$entry))[0];
}

print "<labels>\n";
foreach $entry (@entries) {
	my $ref = qx(cat $od | grep $entry);
	$ref=~s/$entry//;
	$ref=~s/<//;
	$ref=~s/>://;
	chomp($ref);
	print " <label name=\"_$entry\">$ref</label>\n";
	#print " _$entry : $ref\n";
}
print "</labels>\n\n";

} # if(0)
###############
# disassembly #
###############

print "<disassembly>\n";

# XXX x86 only here
my $jmpout = "perl -ne 'print; print \";--\\n\" if (/>:/||/jmp/||/call/||/ret/||/je /||/jne /||/jz /||/jnz /||/jbe /||/ja /||/jb /||/jae /||/jle /);'";

system("cat $od | sed -e 's,^\ ,_0,g' -e 's,^0,_0,g' -e 's, 804, _0804,g' | $jmpout | rsc ffftoneg > $od2");
my $log = 0;
open FD, "<$od2";
while(<FD>) {
	if (/^;--/) {
		$log^=1;
		if ($log) {
			print "<block>\n";
		} else {
			print "</block>\n";
		}
		next;
	}
	s/</&lt;/g;
	s/>/&gt;/g;
	print if ($log);
}
close FD;
print "</disassembly>\n";
print "</radarexml>\n";

unlink ($od);