How to parse og meta tags using httparty for rails 3

Question

I am trying to parse og meta tags using the HTTParty gem using this code:

link = http://www.usatoday.com/story/gameon/2013/01/08/nfl-jets-tony-sparano-fired/1817037/
# link = http://news.yahoo.com/chicago-lottery-winners-death-ruled-homicide-181627271.html
resp = HTTParty.get(link)
ret_body = resp.body

# title
  og_title = ret_body.match(/\<[Mm][Ee][Tt][Aa] property\=\"og:title\"\ content\=\"(.*?)\"\/\>/)
  og_title = og_title[1].to_s

The problem is that it worked on some sites (yahoo!) but not others (usa today)

Answer 1

Don't parse HTML with regular expressions, because they're too fragile for anything but the simplest problems. A tiny change to the HTML can break the pattern, causing you to begin a slow battle of maintaining an ever expanding pattern. It's a war you won't win.

Instead, use a HTML parser. Ruby has Nokogiri, which is excellent. Here's how I'd do what you want:

require 'nokogiri'
require 'httparty'

%w[
  http://www.usatoday.com/story/gameon/2013/01/08/nfl-jets-tony-sparano-fired/1817037/
  http://news.yahoo.com/chicago-lottery-winners-death-ruled-homicide-181627271.html
].each do |link|
  resp = HTTParty.get(link)

  doc = Nokogiri::HTML(resp.body)
  puts doc.at('meta[property="og:title"]')['content']
end

Which outputs:

Jets fire offensive coordinator Tony Sparano
Chicago lottery winner's death ruled a homicide

Answer 2

Perhaps I can offer an easier solution? Check out the OpenGraph gem.

It's a simple library for parsing Open Graph protocol information from web sites and should solve your problem.

Answer 3

Solution:

og_title = ret_body.match(/\<[Mm][Ee][Tt][Aa] property\=\"og:title\"\ content\=\"(.*?)\"[\s\/\>|\/\>]/)
og_title = og_title[1].to_s

Trailing whitespace messed up the parsing so make sure to check for that. I added an OR clause to the regex to allow for both trailing and non trailing whitespace.