module EhcProbe

# File lib/ehc_probe.rb, line 21
def analyze(body, url:)
  html = body.to_s
  doc = Nokogiri::HTML(html)
  metadata = UrlPreview.parse_metadata(html)

  {
    url: url.to_s,
    title: metadata[:title].to_s.strip,
    description: metadata[:description].to_s.strip,
    forms: extract_forms(doc, base_url: url),
    comments: extract_comments(doc),
    path_hints: extract_path_hints(doc),
    external_hosts: extract_external_hosts(doc, base_url: url)
  }
end

# File lib/ehc_probe.rb, line 122
def comment_line(comment)
  return nil if comment.to_s.strip.empty?

  "HTML comment: #{comment}"
end

# File lib/ehc_probe.rb, line 132
def external_host_lines(hosts)
  Array(hosts).first(2).map { |host| "External host: #{host}" }
end

# File lib/ehc_probe.rb, line 85
def extract_comments(doc)
  doc.xpath("//comment()").map { |node| node.text.to_s.strip.gsub(/\s+/, " ") }
     .reject(&:empty?)
     .map { |text| TextUtil.truncate(text, COMMENT_LIMIT) }
end

# File lib/ehc_probe.rb, line 102
def extract_external_hosts(doc, base_url:)
  base_host = host_from_url(base_url)
  hosts = doc.css("script[src], link[href]").filter_map do |node|
    src = node["src"] || node["href"]
    next unless src.to_s.match?(%r{\Ahttps?://}i)

    host_from_url(src)
  end
  hosts.reject { |host| host.empty? || host.casecmp?(base_host) }.uniq
end

# File lib/ehc_probe.rb, line 70
def extract_forms(doc, base_url:)
  doc.css("form").filter_map do |form|
    method = form["method"].to_s.strip.upcase
    method = "GET" if method.empty?
    action = resolve_path(form["action"].to_s, base_url: base_url)
    inputs = form.css("input, textarea, select").filter_map do |node|
      name = node["name"].to_s.strip
      name unless name.empty?
    end.uniq
    next if inputs.empty?

    { method: method, action: action, inputs: inputs }
  end
end

# File lib/ehc_probe.rb, line 91
def extract_path_hints(doc)
  hints = []
  doc.css("[href], [src]").each do |node|
    value = node["href"] || node["src"]
    next unless value.to_s.match?(%r{\A/[^/]})

    hints << value.to_s.split(/[?#]/).first
  end
  hints.uniq.first(PATH_HINT_LIMIT)
end

# File lib/ehc_probe.rb, line 113
def form_line(form)
  return nil unless form

  inputs = Array(form[:inputs]).join(", ")
  hidden = Array(form[:inputs]).grep(/user|id|token|csrf/i)
  detail = hidden.any? ? " (#{hidden.join(', ')})" : ""
  "#{form[:method]} form #{form[:action]}#{detail}"
end

# File lib/ehc_probe.rb, line 150
def host_from_url(url)
  URI.parse(url.to_s).host.to_s.sub(/\Awww\./, "")
rescue URI::InvalidURIError
  ""
end

# File lib/ehc_probe.rb, line 128
def path_hint_lines(hints)
  Array(hints).first(2).map { |path| "Path hint: #{path}" }
end

# File lib/ehc_probe.rb, line 136
def resolve_path(action, base_url:)
  action = action.to_s.strip
  return "/" if action.empty?

  if action.match?(%r{\Ahttps?://}i)
    path = URI.parse(action).path.to_s
    return path.empty? ? "/" : path
  end

  action.start_with?("/") ? action.split(/[?#]/).first : "/#{action.split(/[?#]/).first}"
rescue URI::InvalidURIError
  action
end

# File lib/ehc_probe.rb, line 60
def robots_signals(body)
  text = body.to_s.strip
  return [] if text.empty?

  interesting = text.lines.map(&:strip).grep(/\A(?:Disallow|Allow):/i).first(3)
  return [] if interesting.empty?

  ["robots.txt: #{interesting.join(' \u00b7 ')}"]
end

# File lib/ehc_probe.rb, line 37
def signal_lines(probe, headers: [], robots_body: nil)
  header_list = Array(headers)
  lines = []
  lines << header_list.first if header_list.any?
  title_line = title_signal_line(probe[:title])
  lines << title_line if title_line
  form = form_line(probe[:forms].first)
  lines << form if form
  lines.concat(header_list.drop(1))
  lines << comment_line(probe[:comments].first) if probe[:comments].any?
  lines.concat(path_hint_lines(probe[:path_hints]))
  lines.concat(robots_signals(robots_body))
  lines.concat(external_host_lines(probe[:external_hosts]))
  lines.map(&:strip).reject(&:empty?).first(MAX_SIGNALS)
end

# File lib/ehc_probe.rb, line 53
def title_signal_line(title)
  text = title.to_s.strip
  return nil if text.empty?

  "Title: #{text}"
end