grepdocx

#!/usr/bin/env ruby

$stderr.sync = true

begin
  require "docopt"
rescue LoadError
  STDERR.puts "The docopt gem could not be loaded, is it installed?"
  STDERR.puts "-> gem install docopt"
  exit 1
end
begin
  require 'docx'
rescue LoadError
  STDERR.puts "The docx gem could not be loaded, is it installed?"
  STDERR.puts "-> gem install docx"
  exit 1
end

doc = <<DOCOPT
Search any given Microsoft Word documents, selecting paragraphs that match a pattern.

Note that tables, headings etc are not searched, only paragraphs. 

Usage:
  #{__FILE__} [-i] <pattern> <docx>...
  #{__FILE__} -h | --help

Options:
  -i, --ignore-case        Perform case insensitive matching.
  -h, --help               Show this output.

DOCOPT

begin
  options = Docopt::docopt(doc)
rescue Docopt::Exit => e
  STDERR.puts e.message
  exit 1
end

# check arguments

options['<docx>'].each do |file|
  if not File.exists?(file)
    STDERR.puts "[!] #{file} does not exist!"
    exit 1
  end
end

options['<docx>'].each do |file|
  begin
    docx = Docx::Document.open(file)
  rescue
    STDERR.puts "[!] Error opening #{file}"
    next
  end
  docx.each_paragraph do |p|
    if options['--ignore-case']
      if p.to_s.match(/#{options['<pattern>']}/i)
        if options['<docx>'].length > 1
          puts "#{file}: #{p.to_s}"
        else
          puts p
        end
      end
    else
      if p.to_s.match(/#{options['<pattern>']}/)
        if options['<docx>'].length > 1
          puts "#{file}: #{p.to_s}"
        else
          puts p
        end
      end
    end
  end
end