#!/usr/bin/perl -w
#$Header: sunfingerprint,v0.9 2000/12 afb$
#$! sunfingerprint: checks file list against Suns md5 database
# reads input file/stdin, of the format:
# /usr/bin/ls
# /usr/bin/cat
# that is, one file per line, md5's the files and fetches the results 
+of
# a search against Sun's md5 executable db.
# Relies on some formatting of Sun's results page, which may vary some
+day.
# Always tests the md5 executable, also.
# fixed 20/0 matches oops - multiple matches appear for same exe in mu
+ltiple
# Sol versions
# added -l libs flag so you can check lib files (not executables)

require LWP::UserAgent;
use Getopt::Std;
use strict;

my $debug = 0;
my $md5 = "/usr2/local/bin/md5";
my $request;
my $output_all = 0;
my $errors_only = 0;
my $un_html = 1;
# Sun's fingerprint site
my $request_http = "http://sunsolve.Sun.COM/pub-cgi/fileFingerprints.p
+l";
my $Usage = "Usage: $0 [-d X|-l|-e|-h|-o X] file_list
   file_list: a text file w/ full path names for the files to be viewe
+d
      use - to use stdin, e.g.: find /bin -print | $0
       -d X debug level X
       -h   leave html tags in output
       -e   errors only
       -l   libs, ignore executable status
       -o X output level X (0 matches only, 1 Sun results,
                            2 full text, 3 full html page)
    
";
my %opts;
# must have a file list
die "$Usage" unless @ARGV;
getopts('lho:d:e',\%opts) or die "$Usage";
$debug = $opts{d} ? $opts{d} : 0;
$errors_only = 1 if $opts{e};
$un_html = 0 if $opts{h};
$output_all = $opts{o} ? $opts{o} : 0;
$output_all = 0 if $errors_only;

my $mdlist;
my %mdlist;
my %matches;
my $mdhash;

# read input, check file's existence/executable
while (<>) {
  chomp;
  next unless /^\s+\//;
  $mdlist{$_} = "Not Found!";
  if ( -s $_ ) {
    $mdlist .= "$_ " ;
    $mdlist{$_} = "Found";
    if ( ! -x $_ ) {
      $mdlist{$_} .= ", Not Executable" unless $opts{l};
    }    # if ! -x
  }      # if -s
}        # while <>

# add md5 exe to list
$mdlist .= "$md5";
# ugly hack
$mdlist{$md5} = "Found";

# md5 the files
open(MD5, "$md5 $mdlist | ") or die "can't open md5 $md5: $!";
while (<MD5>) {
  $mdhash .= "$_";
}
close MD5;

# fetch the cgi results from sun
my $ua = LWP::UserAgent->new;
print STDERR "POSTing $mdhash to $request_http\n" if $debug > 20;
if ( $mdhash ) {
  #$request = new HTTP::Request POST => $request_http;
  $request = HTTP::Request->new('POST', $request_http);
  $request->content_type('application/x-www-form-urlencoded');
  $request->content("md5list=$mdhash");
} else {
  die "$Usage";
}

print STDERR "Request: ", $request->content, "\n" if $debug > 20;
my $response = $ua->request($request); #  actually fetch the doc

if ( $response->is_success ) {
# split up the html return and process the lines between the
# "Results ..." line and the end of the list </dl>
  my $pr = 0;
  my $file_path;
  print STDERR "Success\n" if $debug > 10;
  foreach ( split(/\n/, $response->content) ) {
    next if /^\s*$/;
    if ( $pr ) {
      last if /<\/DL>/;
      s/<[ =#\/\w]+>//g if $un_html;
# line w/ sun's response:
# 36e7b0c2c6b2b391722c2471688a060b - (/usr/bin/ls) - 1 match(es)
      if ( /match/ ) {
        print STDERR "match: $_\n" if $debug > 2;
# get file name and matches
        /\((\S+)\) - (.*)/;
        $file_path = $1;
# add Sun's response
        $mdlist{$file_path} .= ", $2 w/ Sun";
        $matches{$file_path} = "$_\n" if  $output_all;
      } else {
        $matches{$file_path} .= "$_\n" if  $output_all > 1 and $file_p
+ath;
      }
    }           # if pr
# start printing
    $pr++ if /Results of Last Search/i;
  }             # foreach content

# output the final data, input files and any results
  if ( $output_all != 3 ) {
    foreach my $file ( sort keys %mdlist ) {
      if ( $errors_only ) {
        $_ = $mdlist{$file};
        print "$file: $mdlist{$file}\n" if /Not/ or / 0 match/;
      } else {
        if ( $matches{$file} ) {
          print "$file: $matches{$file}\n";
        } else {
          print "$file: $mdlist{$file}\n\n";
        }    # if matches
      }      # if errors_only
    }        # foreach file
  } else {
# dump the whole page, if wanted
    print "Content: ", $response->content, "\n";
  }          # if output_all == 3
} else {
  print STDERR "Result: cgi request failed! $request_http\n";
}