Imagine this scenario: you've got a directory full of jar files, and your build is failing because it's missing some classes that are in one of them. You're not sure which; only that the missing classes are in here somewhere.

I've been running into this problem a lot lately, as I try to convert a series of old Ant builds over to Maven. At times, I've found that the jar name doesn't even come close to indicating whether it contains the key classes I need to make my build start working. To address this problem, I've started a short Ruby script that I'm tentatively calling jargrep.rb.

Before I show you this, please remember that I'm a little green when it comes to Ruby, so there may be some elegant one-liner that I'm missing here or there...and it may look decidedly Java-ish to the "native" Ruby programmer. Oh, and I'm not looking to win any architectural or OO-design awards here; I only needed a script to scratch a very persistent itch. Finally, this script is lacking a lot of polish when it comes to options I might want to include, etc. For this, I'd use OptionParser, but I couldn't be bothered to lookup the syntax, etc. today when I needed this script...

jargrep.rb

#!/usr/bin/ruby

if ( ARGV.length < 2 )
  STDOUT.puts "Searches jar archives for files within that match a particular pattern.\n"
  STDOUT.puts "Usage: #{$0} <pattern> -"
  STDOUT.puts "\tRead jar list from standard in, searching them for the specified pattern."
  STDOUT.puts "\t\t-OR-"
  STDOUT.puts "Usage: #{$0} <pattern> <file> [<file>]*"
  STDOUT.puts "\tSearch the jars in the file list for the specified pattern."
  STDOUT.puts "\n\nNOTES:"
  STDOUT.puts "\n\t1. Does NOT use egrep (yet)."
  STDOUT.puts "\n\t2. Currently, lists matches in the following format:"
  STDOUT.puts "\n\t\t<jar-file-name[0]>:\n"
  STDOUT.puts "\n\t\t<matching-path-within-jar[0]>"
  STDOUT.puts "\n\t\t<matching-path-within-jar[1]>"
  STDOUT.puts "\n\t\t<matching-path-within-jar[2]>"
  STDOUT.puts "\n\n\t\t<jar-file-name[1]>:\n"
  STDOUT.puts "\n\t\t<matching-path-within-jar[0]>"
  STDOUT.puts "\n\t\t..."
  exit 1
end

pattern = ARGV[0]

if ( ARGV[1] == '-' )

  files = STDIN.read
  files.each_line do |file|
  
    file.chomp!
    matches = `jar tvf #{file} | grep #{pattern}`.chomp
    
#    status = $?.exitstatus
#    STDOUT.puts "jar returned with status: #{status} for file: #{file}"
    
    if matches.index( 'java.util.zip.ZipException' ) == 0
      STDOUT.puts "\n\n!!!\nFailed to search jar: #{file}\nError:\n\n#{match}!!!\n\n"
    elsif ( !matches.empty? )
      STDOUT.puts "\n#{file}:\n\n#{matches}"
    end
    
  end
  
else

  ARGV.shift
  ARGV.each do |file|
  
    file.chomp!
    matches = `jar tvf #{file} | grep #{pattern}`.chomp
    
    if ( !matches.empty? )
      STDOUT.puts "\n#{file}:\n\n#{matches}"
    end
    
  end
  
end

Have you ever run into a situation like this? How did you solve it? Remember, each time I run across a problem like this, I fully expect to see it come up again, so I'm not looking for the ideal linux command line to execute one time then lose once it rolls out of my .bash_history file... Anyway, for what it's worth, enjoy.

Responses[7]

Posted on July 4, 2007 8:28:14 PM EDT #