discourse/lib/retrieve_title.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

84 lines
2.1 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
module RetrieveTitle
CRAWL_TIMEOUT = 1
def self.crawl(url)
fetch_title(url)
rescue Exception
# If there was a connection error, do nothing
end
def self.extract_title(html, encoding = nil)
title = nil
if doc = Nokogiri::HTML5(html, nil, encoding)
title = doc.at('title')&.inner_text
# A horrible hack - YouTube uses `document.title` to populate the title
# for some reason. For any other site than YouTube this wouldn't be worth it.
if title == "YouTube" && html =~ /document\.title *= *"(.*)";/
title = Regexp.last_match[1].sub(/ - YouTube$/, '')
end
if !title && node = doc.at('meta[property="og:title"]')
title = node['content']
end
end
if title.present?
title.gsub!(/\n/, ' ')
title.gsub!(/ +/, ' ')
title.strip!
return title
end
nil
end
private
def self.max_chunk_size(uri)
# Amazon and YouTube leave the title until very late. Exceptions are bad
# but these are large sites.
return 500 if uri.host =~ /amazon\.(com|ca|co\.uk|es|fr|de|it|com\.au|com\.br|cn|in|co\.jp|com\.mx)$/
return 300 if uri.host =~ /youtube\.com$/ || uri.host =~ /youtu.be/
# default is 10k
10
end
# Fetch the beginning of a HTML document at a url
def self.fetch_title(url)
fd = FinalDestination.new(url, timeout: CRAWL_TIMEOUT)
current = nil
title = nil
encoding = nil
fd.get do |_response, chunk, uri|
unless Net::HTTPRedirection === _response
if current
current << chunk
else
current = chunk
end
if !encoding && content_type = _response['content-type']&.strip&.downcase
if content_type =~ /charset="?([a-z0-9_-]+)"?/
encoding = Regexp.last_match(1)
if !Encoding.list.map(&:name).map(&:downcase).include?(encoding)
encoding = nil
end
end
end
2018-06-07 13:28:18 +08:00
max_size = max_chunk_size(uri) * 1024
title = extract_title(current, encoding)
throw :done if title || max_size < current.length
end
end
title
2018-06-07 13:28:18 +08:00
end
end