validate-pp

#!/usr/bin/env ruby
require 'thread'

# Valid command line flags
VALID_ARGS = %w(-a -g -f -d -h --help)

def usage
  puts "usage: validate-pp [-a] [-g] [-f <file-1>..<file-n>]"
  puts "       [-d <directory-1>..<directory-n>] [-h | --help]"
end

# Parse ARGV[0], make sure the flag passed is valid.
unless VALID_ARGS.include? ARGV[0]
  puts "validate-pp needs a command line option." unless ARGV[0]
  puts "Invalid option: '#{ARGV[0]}'" if ARGV[0]
  usage
  exit!
end

# Help output.
if %w(-h --help).include? ARGV[0]
  usage
  puts
  puts "validate-pp uses the `puppet parser validate` command to validate"
  puts "files with the '.pp' extension in bulk. This makes it easy to validate"
  puts "an entire puppet directory before committing to a git repo, for example."
  puts
  puts "Valid options:"
  puts
  puts "-a"
  puts "    Validate every '.pp' file in the current directory and all subdirectories.\n\n"
  puts "-d <directory-1>..<directory-n>"
  puts "    Validate every '.pp' file in each given directory, and in all of their subdirectories.\n\n"
  puts "-f <file-1>..<file-n>"
  puts "    Validate each file listed on the command line.\n\n"
  puts "-g"
  puts "    If the current directory is a valid git repository, validate any '.pp' file in the repo"
  puts "    that has been changed, copied, moved or renamed since the last commit. This can be"
  puts "    significantly faster than using -a to parse a large puppet repo unnecessarily if"
  puts "    most of the files in the repo are unchanged.\n\n"
  puts "-h"
  puts "    Display this help message."
  puts
  exit!
end


# Class to represent .pp file objects.
class PuppetFile

  # This will be the work queue our thread pool pulls from
  # to validate the .pp files.
  @@all_files = Queue.new
  # Array to keep track of files that fail validation.
  @@invalid_files = []

  # return the queue of puppet files.
  def self.all_files
    @@all_files
  end
  # return the list of invalid files.
  def self.invalid_files
    @@invalid_files
  end

  attr_accessor :filename

  # Create a puppet file. All we need is the filename,
  # and to validate that it's actually a .pp file.
  def initialize(filename)
    unless File.extname(filename) == '.pp'
      puts "Error, #{filename} does not appear to be a valid puppet file."
      exit!
    end
    @filename = File.expand_path(filename).sub(' ', '\ ')
    # Add the new object to the queue.
    @@all_files.push(self)
  end
  
  # Use puppet parser to validate the puppet file.
  # if it fails validation, add it to the @@invalid_files array.
  def validate
    puts "Validating #{self.filename}..."
    output = `puppet parser validate #{self.filename} 2>&1`
    if $?.exitstatus != 0
      @@invalid_files << { filename: self.filename, error: output }
    end
  end
end



# parse_args method figures out what we're supposed to do
# based on the command line arguments.
def parse_args

  case ARGV[0]
  when '-a'
    # check that no arg was passed with the -a flag.
    if ARGV[1]
      puts "Error, '-a' flag cannot accept additional arguments."
      puts "Run `validate-pp -h` for more info."
      exit!
    end
    parse_directory(File.expand_path(Dir.pwd))

  when '-d'
    ARGV.shift
    ARGV.each do |d|
      unless File.directory?(d)
        puts "Error, #{d} is not a directory."
        puts "Run `validate-pp -h` for more info."
        exit!
      end
    end
    dirs = ARGV.map { |d| File.expand_path(d) }
    dirs.each { |d| parse_directory(d) }

  when "-f"
    ARGV.shift
    ARGV.each do |f|
      unless File.extname(f) == '.pp'
        puts "Error, #{f} is not a '.pp' file"
        puts "Run `validate-pp -h` for more info."
        exit!
      end  
    end
    ARGV.each { |f| PuppetFile.new(File.expand_path(f)) }

  when '-g'
    if ARGV[1]
      puts "Error, '-g' flag cannot accept additional arguments."
      puts "Run `validate-pp -h` for more info."
      exit!
    end
    parse_git_repo
  end
      
end


# Parse a given directory for .pp files and make new 
# PuppetFile instances with them.
def parse_directory(dir)
  search = File.join("**", "*.pp")
  Dir.chdir dir
  ppfiles = Dir.glob(search)
  
  ppfiles.each { |f| PuppetFile.new(File.expand_path(f)) }

end


# If the current directory is a valid git repo, get the list
# of new or modified files. Make PuppetFile instances out of the
# ones that are .pp files.
def parse_git_repo
  git_output = `git status --porcelain 2>&1`
  unless $?.exitstatus == 0
    puts "Git error: make sure the current directory is a valid repo and that git is working."
    exit!
  end
  list = git_output.split("\n")
  # Get just what was added, modified, copied or moved.
  # Skip deleted files.
  files = list.reduce([]) do |a, f|
    file = /[\?MARC ]{1,2} (.*\.pp)/.match f
    a << file[1] if file
    a
  end
  files.uniq! # remove dupes, just in case.
  files.each { |f| PuppetFile.new(File.expand_path(f)) }
end




# Validate the files stored in our PuppetFile files queue.
def validate_files
  # Thread pool with 5 workers. Adjust as desired.
  workers = (0..4).map do
    Thread.new do
      begin
        while puppetfile = PuppetFile.all_files.pop(true)
          puppetfile.validate
        end
      rescue ThreadError
      end
    end
  end
  workers.map(&:join)
end


# How'd it go?
def output_results
  if PuppetFile.invalid_files.any?
    puts "\nThe following files had errors:\n\n"
    PuppetFile.invalid_files.each do |file|
      puts "File: #{file[:filename]}"
      puts file[:error]
      puts
    end
  else
    puts "No errors found, all parsed files appear valid."
  end

end

# -------------------------- Begin Script ---------------------------

parse_args

if PuppetFile.all_files.empty?
  puts "No files found to validate."
  exit!
else
  validate_files
end

output_results

exit 1 if PuppetFile.invalid_files.any?

exit 0