#!/usr/bin/ruby

# require File.expand_path('../../config/boot.rb', __FILE__)
# require File.expand_path('../../config/environment.rb', __FILE__)

PATH = File.dirname(__FILE__) + '/..'

PRAGMAS = {
  'nologin'  => true,   # Doesn't require login.
  'root'     => true,   # Requires root (for docs only).
  'norobots' => true,   # Off-limit to web crawlers.
  'prefetch' => true,   # Allow browser to prefetch.
}

# ------------------------------------
#  Get list of views from views dir.
# ------------------------------------

def get_list_of_views(controller)
  # results = []
  # for file in Dir.glob("#{PATH}/app/views/#{controller}/[a-z]*.{rhtml,rxml}")
  #   file.match(/(\w+)\.\w+$/)
  #   results << $1
  # end
  # return results.sort
end

# ---------------------------------------------------------------------
#  Extract existing docs of views from comments at top of controller.
# ---------------------------------------------------------------------

def read_headers(file)
  headers = {}
  File.open(file) do |fh|
    fh.each_line do |line|
      if line.match(/^ *# *:(all_\w+):/)
        headers[$1.to_sym] = true
      # elsif ...
      end
    end
  end
  return headers
end

# -----------------------------------
#  Extract actions from controller.
# -----------------------------------

def extract_actions(file)
  actions = []
  section = nil
  comment = nil
  globals = []
  if file.match(/(\w+)_controller/)
    controller = $1
  end
  File.open(file) do |fh|
    num = 0
    fh.each_line do |line|
      num += 1
      if line.match(/^ *# *:section: *(\S.*\S)/)
        section = $1
      elsif line.match(/^ *# *:all_(\w+):/)
        if !PRAGMAS[pragma = $1]
          $stderr.puts("Unrecognized pragma :all_#{pragma}: on " +
                       "line #{num} of #{file}.")
        else
          globals << pragma
        end
      elsif line.match(/^  # (\S[^\.]*\S)/)
        comment ||= $1
      elsif line.match(/^  def (\w+)( .*|$)/)
        name  = $1
        notes = $2
        action = {
          :controller => controller,
          :section => section,
          :name    => name,
          :comment => comment,
          :notes   => (hash = {}),
        }
        for pragma in globals
          hash[pragma.to_sym] = true
        end
        while notes.sub!(/:(\w+):/, '')
          if !PRAGMAS[pragma = $1]
            $stderr.puts("Unrecognized pragma :#{pragma}: on " +
                         "line #{num} of #{file}.")
          else
            hash[pragma.to_sym] = true
          end
        end
        actions << action
      elsif !line.match(/^  #/)
        comment = nil
      end
    end
  end
  return actions
end

# --------------------------------------------------------
#  Write new controller, updating docs and filter specs.
# --------------------------------------------------------

def redo_headers(file, headers, actions, views)
  # save_copy(file)
  # File.open(file, 'w') do |fh|
  # end
end

# -----------------------------------------------------------------------
#  Make sure we don't have any unused redirects in observer controller.
# -----------------------------------------------------------------------

def check_for_unused_redirects(actions, old_actions)
  norobots = []
  for path in old_actions
    controller, name = path.split('/')[1..2]
    used = false
    for action in actions
      if (action[:controller] == controller) and
         (action[:name] == name)
        used = true
        if action[:notes][:norobots]
          norobots << "/observer/#{name}"
        end
      end
    end
    if !used
      $stderr.puts("No longer need redirect to #{controller} #{name} from observer!")
    end
  end
  return norobots
end

# ------------------------------------------------
#  Add actions marked :norobots: to master list.
# ------------------------------------------------

def save_norobots(master, controller, actions)
  last_one = nil
  for action in actions.sort_by {|x| x[:name]}
    name = action[:name]
    if action[:notes][:norobots]
      master << "/#{controller}/#{name}"
      last_one = name
    elsif last_one and
          (name[0,last_one.length] == last_one)
      $stderr.puts("Accidentally blocked /#{controller}/#{name}!!")
    end
  end
end

# ----------------------------
#  Write robots.txt.
# ----------------------------

def write_robots(data)
  file = "#{PATH}/public/robots.txt"
  save_copy(file)
  File.open(file, 'w') do |fh|
    fh.puts <<EOB
# See http://www.robotstxt.org/wc/norobots.html for documentation on how to use
# the robots.txt file.

User-agent: *
EOB
    for file in data.sort.uniq
      fh.puts "Disallow: #{file}"
    end
  end
end

# ----------------------------
#  Save copy of old file.
# ----------------------------

def save_copy(new_file)
  old_file = "#{new_file}.old"
  if File.exists?(new_file)
    if File.exists?(old_file)
      File.delete(old_file)
    end
    File.rename(new_file, old_file)
  end
end

# ----------------------------
#  Main program.
# ----------------------------

norobots = [
  '/ajax/',
  '/api/',
]

# These are all being redirected from observer.
old_actions = [
  '/comment/add_comment',
  '/comment/destroy_comment',
  '/comment/edit_comment',
  '/comment/list_comments',
  '/comment/show_comment',
  '/comment/show_comments_for_user',
  '/image/add_image',
  '/image/destroy_image',
  '/image/edit_image',
  '/image/license_updater',
  '/image/list_images',
  '/image/next_image',
  '/image/prev_image',
  '/image/remove_images',
  '/image/reuse_image',
  '/image/show_image',
  '/name/approve_name',
  '/name/bulk_name_edit',
  '/name/change_synonyms',
  '/name/deprecate_name',
  '/name/edit_name',
  '/name/map',
  '/name/observation_index',
  '/name/show_name',
  '/name/show_past_name',
  '/species_list/add_observation_to_species_list',
  '/species_list/create_species_list',
  '/species_list/destroy_species_list',
  '/species_list/edit_species_list',
  '/species_list/list_species_lists',
  '/species_list/manage_species_lists',
  '/species_list/remove_observation_from_species_list',
  '/species_list/show_species_list',
  '/species_list/species_lists_by_title',
  '/species_list/upload_species_list',
]

all_actions = []
for file in Dir.glob("#{PATH}/app/controllers/*_controller.rb").sort
  file.match(/(\w+)_controller.rb/)
  controller = $1
  if controller != 'api'
    headers = read_headers(file)
    actions = extract_actions(file)
    views   = get_list_of_views(controller)
    redo_headers(file, headers, actions, views)
    if headers[:all_norobots]
      norobots << "/#{controller}/"
    else
      save_norobots(norobots, controller, actions)
    end
    all_actions += actions
  end
end

norobots += check_for_unused_redirects(all_actions, old_actions)
write_robots(norobots)

exit 0