#!/usr/bin/env ruby
# frozen_string_literal: true

require 'optparse'
require 'httpspell/spider'
require 'httpspell/spellchecker'
require 'httpspell/version'

personal_dictionary_path = nil
force_language = nil
tracing = nil
verbose = nil
whitelist = nil
blacklist = []

begin
  OptionParser.new do |parser|
    parser.banner.prepend <<~BANNER
      Spellchecks a website via HTTP.

    BANNER
    parser.version = HttpSpell::VERSION

    parser.on('-p', '--personal-dictionary=FILE', 'path to the personal dictionary file') do |p|
      personal_dictionary_path = p
    end

    parser.on('-l', '--language=LANGUAGE', 'override LANGUAGE of content') do |l|
      force_language = l
    end

    parser.on('-w', '--whitelist=EXPRESSION', 'when recursively retrieving URLs, allow only those matching the given regular EXPRESSION') do |w|
      whitelist ||= []
      whitelist << Regexp.new(w)
    end

    parser.on('-t', '--trace', 'enable error tracing') do
      tracing = true
    end

    parser.on('-V', '--verbose', "explain what's happening") do
      verbose = true
    end

    parser.on('-b', '--blacklist=EXPRESSION', 'blacklist (ignore) URLs matching the given regular EXPRESSION') do |b|
      blacklist << Regexp.new(b)
    end

    # TODO: --recursive, defaults to false
    # TODO wget has some additional options for recursive behavior that should be reviewed
  end.parse!
rescue StandardError
  warn "Error - #{$ERROR_INFO}"
  exit 1
end

if ARGV.size != 1
  warn "Expected exactly one argument, but received #{ARGV.size}."
  exit 1
end

spell_checker = HttpSpell::SpellChecker.new(personal_dictionary_path, tracing: tracing)
has_unknown_words = false

spider_success = HttpSpell::Spider.new(ARGV.first, whitelist: whitelist, blacklist: blacklist, tracing: tracing).start do |url, doc|
  lang = force_language || doc.root['lang'] || ENV['LANGUAGE']

  # Remove sections that are not to be spellchecked
  doc.css('pre').each(&:unlink)
  doc.css('code').each(&:unlink)
  doc.css('[spellcheck=false]').each(&:unlink)

  # TODO: Find sections with a lang attribute and handle them separately
  unknown_words = spell_checker.check(doc.to_s, lang)

  if unknown_words.empty?
    warn "No unknown words (language is #{lang}) at #{url}." if verbose
  else
    warn "#{unknown_words.size} unknown words (language is #{lang}) at #{url}:" if verbose
    puts unknown_words
    has_unknown_words = true
  end
end

exit 2 unless spider_success
exit 1 if has_unknown_words