indent-stat.rb

#!/usr/bin/env ruby -W
# encoding: utf-8

# This script takes a glob pattern as input (eg. "*.[hm]"), and reports
# indentation statistics for files matching the pattern in the current
# directory, recursively.

require 'set'

pattern = ARGV.shift
if pattern.nil?
  puts "No pattern given"
  exit 1
end

root_pattern = File.join(Dir.getwd, '**', pattern)

stats = {}

Dir.glob(root_pattern) do |file|
  stats[file] ||= {
    lines: 0,
    tabs: 0,
    spaces: 0,
  }
  IO.foreach(file) do |line|
    stats[file][:lines] += 1
    begin
      if /^	/.match(line) then
        stats[file][:tabs] += 1
      end
      if /^ /.match(line) then
        stats[file][:spaces] += 1
      end
    rescue
      puts "Failed to match #{file}:#{line}"
    end
  end
end

line_count = stats.collect {|file, stat| stat[:lines]}.reduce :+
tabs_count = stats.collect {|file, stat| stat[:tabs]}.reduce :+
spaces_count = stats.collect {|file, stat| stat[:spaces]}.reduce :+

file_tabs_count = stats.select {|file, stat| stat[:tabs] != 0 && stat[:spaces] == 0}.collect { 1 }.reduce :+
file_spaces_count = stats.select {|file, stat| stat[:tabs] == 0 && stat[:spaces] != 0}.collect { 1 }.reduce :+

mismatched_files = stats.select {|file, stat| stat[:tabs] != 0 and stat[:spaces] != 0 }

puts "Processed #{line_count} lines in #{stats.count} files, #{tabs_count} tab lines, #{spaces_count} spaces lines"
puts "#{file_tabs_count} tab files, #{file_spaces_count} space files"

mismatch_stats = {tabs: 0, spaces: 0}
puts "#{mismatched_files.count} files have mismatching indent :"
mismatched_files.each do |file, stat|
  file = file.sub Dir.getwd+'', "."
  msg = case
    when stat[:tabs] > stat[:spaces] then
      mismatch_stats[:tabs] += 1
      "has more tabs"
    when stat[:tabs] < stat[:spaces] then
      mismatch_stats[:spaces] += 1
      "has more spaces"
    else "has same number"
  end
  puts "#{file} #{msg}"
end
puts "#{mismatch_stats[:tabs]} mismatching files have more tabs, #{mismatch_stats[:spaces]} have more spaces"