A Rakefile for simple web site creation and maintenance

simple_webtasks_rakefile.rb

# -*- mode: ruby; coding: utf-8 -*-

#
# Require:
#  * xoxo gem
#  * lftp binary
#  * linkchecker binary
#  * html tidy binary
#  * svn or hg or git
#  * htpasswd binary
#

require 'uri'
require 'yaml'
require 'open3'
require 'erb'
require 'pp'
require 'fileutils'
include FileUtils::Verbose

Dir.chdir( File.dirname( __FILE__ ) )

module SCM
  %w( svn hg git ).each { |scm|
    define_method( "scm_#{scm}?".to_sym ) {
      File.exist?( ".#{scm}" ) and which( scm )
    }
  }
end
include SCM

class Transfer
  class InvalidConfiguration < StandardError; end

  def initialize( conf = nil )
    if ( conf.has_key?( 'bin' ) and conf.has_key?( 'uri' ) )
      begin
        @conf = conf
        @bin  = conf['bin']
        @uri  = URI( conf['uri'] )
      rescue => e
        bad_configuration( e, conf )
      end
    else
      bad_configuration( InvalidConfiguration.new, conf )
    end
  end
  attr_reader :bin, :uri

  def bad_configuration( e, conf )
    puts "\n" + conf.to_yaml + "\n"
    raise e
  end

  def target
    case  @bin
    when 'lftp'
      @uri.path
    when 'rsync'
      @uri
    end
  end

  def opts( c )
    case @bin
    when 'lftp'
      "-u #{c.user},#{c.password} -e 'set ftp:list-options -a; mirror --delete -v <%= reverse %>-X #{ignores.join( ' -X ' )} <%= from %> <%= to %>; bye' #{c.host}"
    when 'rsync'
      "-rvt --exclude=#{ignores.join( ' --exclude=' )} --delete <%= from %> <%= to%>"
    end
  end

  def up
    prepare_htaccess( @conf )
    reverse = '-R '
    from    = '.'
    to      = target
    opts    = ERB.new( opts( @uri ) ).result( binding )
    system( "#{@bin} #{opts}" )
    remove_htaccess
  end

  def down
    reverse = ''
    from = target
    to   = '.'
    opts = ERB.new( opts( @uri ) ).result( binding )
    system( "#{@bin} #{opts}" )
  end

  def self.minimum
    {'production' =>
      { 'bin' => 'lftp',
        'uri' => 'ftp://user:pass@host/path'},
      'default' => 'production'
    }.to_yaml
  end
end

class BasicAuth
  def self.minimum
    return <<EOD
AuthType Basic
AuthName The restriction area for <%= host %>
AuthUserFile <%%= conf['htpasswd'] %>
Require valid-user
EOD
  end
end

HTACCESS      = '.htaccess'
HTACCESS_SEED = '_htaccess.txt'
SITEMAP       = '_sitemap.html'

def rake_self
  File.basename( __FILE__ )
end

def win?
  return ( RUBY_PLATFORM =~ /mingw/ or
           (RUBY_PLATFORM =~ /win/ and
            RUBY_PLATFORM !~ /darwin/ and RUBY_PLATFORM !~ /cygwin/ ) )
end

def linebreak
  if ( !@linebreak )
    @linebreak = ( win? ) ? "\r\n" : "\n"
  end

  return @linebreak
end

def which( cmd )
  path = ENV['PATH']
  path += File::PATH_SEPARATOR + '/Program Files' if win?
  cmd  += '.exe' if win?

  paths = path.split( File::PATH_SEPARATOR ).select { |path|
    File.exist?( path ) and File.exist?( File.join( path, cmd ) )
  }
  if ( paths.size > 0 )
    paths.first + File::SEPARATOR + cmd
  else
    nil
  end
end

def cut( cmd, chr = '?' )
  `#{cmd}`.map { |line|
    c = line.split( /\s+/ )
    if ( c[0] =~ Regexp.new( "[#{chr}]" ) )
      c[c.size-1]
    else
      nil
    end
  }.compact
end

def htmlfiles
  Dir.glob( "**/*.html\0**/*.htm" )
end

def trans_conf
  '_trans.yml'
end

def load_trans_conf( env = nil )
  if ( !@conf and File.exist?( trans_conf ) )
    @conf = YAML.load_file( trans_conf )
  end

  return ( @conf and env and @conf.has_key?( env ) ) ? @conf[env] : @conf
end

def trans_envs
  conf = load_trans_conf
  env  = conf.keys if ( conf.is_a?( Hash ) )

  ( env ) ? env - ['default'] : []
end

def sitemap_yaml
  yaml = nil
  
  %w( yaml yml ).each { |suffix|
    %w( .sitemap _sitemap sitemap ).each { |basename|
      file = "#{basename}.#{suffix}"
      if ( File.exist?( file ) )
        yaml = file
        break
      end
    }
  }
  
  return yaml
end

def load_sitemap
  ( sitemap_yaml ) ? YAML.load_file( sitemap_yaml ) : nil
end

def ignores
  %w(*~ *.bak .*~ .*.bak .*ignore CVS/ .svn/ .hg/ .git/ .DS_Store Thumbs.db) + [rake_self, trans_conf, HTACCESS_SEED, SITEMAP, sitemap_yaml]
end

def prepare_htaccess( conf = nil )
  if ( File.exist?( HTACCESS_SEED ) )
    cp( HTACCESS_SEED, HTACCESS )
    open( HTACCESS, 'r+b' ) { |f|
      erb = ERB.new( f.read )
      f.truncate( 0 )
      f.rewind
      f.puts erb.result( binding )
    }
  end
end

def remove_htaccess
  if ( File.exist?( HTACCESS_SEED ) )
    rm( HTACCESS )
  end
end

LINEBREAK   = linebreak
HTMLFILES   = htmlfiles
TIDY        = 'tidy -qe'
LINKCHECKER = 'linkchecker --stdin --no-status -a --ignore-url=^mailto:'

#

namespace :transfer do
  trans_envs.each { |env|
    namespace env do
      trans = Transfer.new( load_trans_conf( env ) )
      if ( which( trans.bin ) )
        desc "upload contents from cwd to #{trans.uri.host}"
        task :up => :clean do
          trans.up
        end

        desc "download contents into cwd from #{trans.uri.host}"
        task :down do
          trans.down
        end
      end
    end
  }

  c = load_trans_conf
  if ( c )
    if ( c.has_key?( 'default' ) )
      trans = Transfer.new( c[c['default']] )
      if ( which( trans.bin ) )
        desc "upload contents from cwd to #{trans.uri.host}"
        task :up => :clean do
          trans.up
        end
        
        desc "download contents into cwd from #{trans.uri.host}"
        task :down do
          trans.down
        end
      end
    end

    namespace :conf do
      desc "show transfer config"
      task :get do
        puts c.to_yaml
      end
    end
  end

  namespace :conf do
    desc "set transfer config"
    task :set do
      if ( !File.exist?( trans_conf ) )
        puts "Hit any key and edit transfer setting with your favorite EDITOR"
        ARGV.replace( [] )
        gets
        File.open( trans_conf, 'wb' ) { |f|
          f.puts( Transfer.minimum )
        }
      end
      system( ENV['EDITOR'], trans_conf )
    end
  end
end

if ( which( 'tidy' ) and HTMLFILES.size > 0 )
  desc 'check syntax of htmls'
  task :tidy do
    check = HTMLFILES.map { |html|
      Open3.popen3( TIDY + " #{html}" ) { |stdin, stdout, stder|
        e = stder.read
        html + LINEBREAK + e if e.size > 0
      }
    }.compact.join

    if ( check.size > 0 )
      puts check
    else
      puts "All html file's syntax OK."
    end
  end
end

if ( which( 'linkchecker' ) and HTMLFILES.size > 0 )
  desc "check links in htmls"
  task :linkcheck do
    Open3.popen3( LINKCHECKER ) { |stdin, stdout, stderr|
      stdin.puts HTMLFILES.join( ' ' )
      stdin.close

      started = false
      result  = {}
      check = stdout.read.split( LINEBREAK ).map { |line|
        if ( !started )
          if ( line =~ /\AStart checking/ )
            started = true
          end
          nil
        else
          if ( line =~ /\AThat's it/ )
            result['warning'] = $1.to_i if ( line =~ /([0-9]+) warnings/ )
            result['error']   = $1.to_i if ( line =~ /([0-9]+) error/ )
          end
          line
        end
      }.compact.join( LINEBREAK )

      if ( (result.has_key?( 'warning' ) and result['warning'] > 0) or
           (result.has_key?( 'error' ) and  result['error'] > 0) )
        puts check
      else
        puts "All html file's link OK."
      end
    }
  end
end

if ( sitemap_yaml )
  desc "create sitemap html from #{sitemap_yaml}"
  task :sitemap_html do
    require 'xoxo'
    open( SITEMAP, 'w' ) { |f|
      f.puts load_sitemap.to_xoxo
    }
  end
end

if ( methods.grep( /\Ascm_/ ).size > 0 )
  desc "list untracked files"
  task :untracked do
    if ( scm_hg? )
      puts cut( 'hg stat -A', 'I?' )
    elsif ( scm_git? )
      puts `git ls-files -o`
    elsif ( scm_svn? )
      puts cut( 'svn stat -v', '?' )
    end
  end
end

if ( methods.grep( /\Ascm_/ ).size > 0 and
     !( scm_git? and methods.grep( /\Ascm_/ ).size == 1 ) )
  desc "list missing files"
  task :missing do
    if ( scm_hg? )
      puts cut( 'hg stat -A', '!' )
    elsif ( scm_svn? )
      puts cut( 'svn stat -v', '!' )
    else
      puts "noncompliant scm tool only"
    end
  end
end

if ( methods.grep( /\Ascm_/ ).size > 0 )
  desc "diffs of all tracking files"
  task :diff do
    if ( STDOUT.tty? )
      color = which( 'colordiff' ) ? ' | colordiff'       : ''
      pager = ENV['PAGER']         ? " | #{ENV['PAGER']}" : ''
    end
    if ( scm_hg? )
      system( "hg diff#{color}#{pager}" )
    elsif ( scm_git? )
      system( "git diff#{color}#{pager}" )
    elsif ( scm_svn? )
      system( "svn diff#{color}#{pager}" )
    end
  end
end

if ( HTMLFILES.size > 0 )
  desc "s/{WAVEDASH}/{FULLWIDTH_TILDE}/ (utf-8)"
  task :fix_wavedash4oldwin do
    system( <<EOD )
find . -name '*.html' -o -name '*.htm' -o -name '*.css' -o -name '*.js' | xargs grep -l 〜 | xargs perl -i.bak -pe 's/〜/～/'
EOD
  end
end

if ( !File.exist?( HTACCESS_SEED ) and
     c     = load_trans_conf( 'production' ) and
     trans = Transfer.new( c ) )
  desc 'create BASIC-Auth setting'
  task :create_basic_auth do
    open( HTACCESS_SEED, 'wb' ) { |f|
      host = trans.uri.host
      f.puts( ERB.new( BasicAuth.minimum ).result( binding ) )
    }
    touch( '.htpasswd' )
    # same as ftp setting
    system( "htpasswd -b .htpasswd #{trans.uri.host} #{trans.uri.password} > /dev/null 2>&1" )
  end
end

desc "test html files"
task :test => [:tidy, :linkcheck] do; end

desc "remove backup and auto creation file"
task :clean do
  Dir.glob( "**/*~\0**/.*~\0**/.DS_Store\0**/Thumbs.db\0**/._*" ).map { |e|
    rm( e )
  }
end

task :default do
  if ( HTMLFILES.size == 0 and !trans_conf and !sitemap_yaml )
    puts 'No HTML file, No Task.'
  else
    app = Rake.application
    app.options.show_task_pattern = Regexp.new('')
    app.display_tasks_and_comments
  end
end