module AiUrlContext

# File lib/ai/url_context.rb, line 117
def build_context_lines(question, fetch: nil)
  url = extract_url(question)
  return [] unless url
  return [] unless safe_url?(url)
  return [] if skip_url?(url)

  result = fetch_link(url, fetch: fetch)
  return [] unless result[:found]

  lines = ["URL: #{result[:url]}"]
  lines << "Title: #{result[:title]}" if result[:title]
  lines << "Content: #{result[:text]}"
  lines
end

# File lib/ai/url_context.rb, line 85
def clean_plain_text(body)
  body.to_s.strip.gsub(/\r\n?/, "\n").squeeze(" ")
end

# File lib/ai/url_context.rb, line 132
def default_fetch
  ->(url) { RabbotHttp.fetch(url, user_agent: USER_AGENT) }
end

# File lib/ai/url_context.rb, line 63
def extract_content(body, url: nil)
  _ = url
  if html_body?(body)
    metadata = UrlPreview.parse_metadata(body)
    paragraphs = UrlPreview.extract_paragraphs(body, limit: 10, offset: 0)
    if paragraphs.empty?
      paragraphs = UrlPreview.fallback_paragraphs(body, metadata: metadata, offset: 0, limit: 3)
    end

    parts = []
    title = metadata[:title].to_s.strip
    parts << title unless title.empty?
    paragraphs.each do |paragraph|
      line = paragraph.to_s.strip
      parts << line unless line.empty?
    end
    parts.join("\n\n")
  else
    clean_plain_text(body)
  end
end

# File lib/ai/url_context.rb, line 24
def extract_url(question)
  UrlLink.extract_first_url(question)
end

# File lib/ai/url_context.rb, line 96
def fetch_link(url, fetch: nil)
  fetch ||= default_fetch
  resolved = resolve_fetch_url(url)
  body = fetch.call(resolved)
  return { found: false, url: url, title: nil, text: nil, error: "empty" } if body.to_s.strip.empty?

  title = nil
  if html_body?(body)
    metadata = UrlPreview.parse_metadata(body)
    title = metadata[:title].to_s.strip
    title = nil if title.empty?
  end

  text = truncate_content(extract_content(body, url: resolved))
  return { found: false, url: url, title: title, text: nil, error: "no content" } if text.empty?

  { found: true, url: resolved, title: title, text: text, error: nil }
rescue StandardError => e
  { found: false, url: url, title: nil, text: nil, error: e.message }
end

# File lib/ai/url_context.rb, line 58
def html_body?(body)
  text = body.to_s.lstrip
  text.start_with?("<") || text.match?(/\A(?:<!DOCTYPE|<!--)/i)
end

# File lib/ai/url_context.rb, line 42
def localhost_hostname?(host)
  LOCALHOST_HOSTNAMES.include?(host.to_s.strip.downcase)
end

# File lib/ai/url_context.rb, line 50
def resolve_fetch_url(url)
  text = url.to_s.strip
  match = text.match(PASTEBIN_PAGE_PATTERN)
  return text.sub(%r{pastebin\.com/\K[a-zA-Z0-9]+}i, "raw/#{match[1]}") if match

  text
end

# File lib/ai/url_context.rb, line 28
def safe_url?(url)
  uri = URI.parse(url.to_s)
  return false unless %w[http https].include?(uri.scheme)
  return false if uri.host.to_s.strip.empty?
  return false if localhost_hostname?(uri.host)

  ip = GeoLookup.extract_ip(uri.host)
  return false if ip && GeoLookup.private_ip?(ip)

  true
rescue URI::InvalidURIError, ArgumentError
  false
end

# File lib/ai/url_context.rb, line 46
def skip_url?(url)
  UrlLink.reserved_url?(url)
end

# File lib/ai/url_context.rb, line 89
def truncate_content(text)
  line = text.to_s.strip
  return line if line.length <= MAX_CONTENT_CHARS

  "#{line[0, MAX_CONTENT_CHARS - 3]}..."
end