What is the best way to parse a web page in Ruby?

Question

I have been looking at XML and HTML libraries on rubyforge for a simple way to pull data out of a web page. For example if I want to parse a user page on stackoverflow how can I get the data into a usable format?

Say I want to parse my own user page for my current reputation score and badge listing. I tried to convert the source retrieved from my user page into xml but the conversion failed due to a missing div. I know I could do a string compare and find the text I'm looking for, but there has to be a much better way of doing this.

I want to incorporate this into a simple script that spits out my user data at the command line, and possibly expand it into a GUI application.

Answer 1

Hpricot is over !

Use Nokogiri now.

Answer 2

Unfortunately stackoverflow is claiming to be XML but actually isn't. Hpricot however can parse this tag soup into a tree of elements for you.

require 'hpricot'
require 'open-uri'

doc = Hpricot(open("http://stackoverflow.com/users/19990/armin-ronacher"))
reputation = (doc / "td.summaryinfo div.summarycount").text.gsub(/[^\d]+/, "").to_i

And so forth.

Answer 3

try hpricot, its well... awesome

I've used it several times for screen scraping.

Answer 4

I always really like what Ilya Grigorik writes, and he wrote up a nice post about using hpricot.

I also read this post a while back and it looks like it would be useful for you.

Haven't done either myself, so YMMV but these seem pretty useful.

Answer 5

Something I ran into trying to do this before is that few web pages are well-formed XML documents. Hpricot may be able to deal with that (I haven't used it) but when I was doing a similar project in the past (using Python and its library's built in parsing functions) it helped to have a pre-processor to clean up the HTML. I used the python bindings for HTML Tidy as this and it made life a lot easier. Ruby bindings are here but I haven't tried them.

Good luck!

Answer 6

it seems to be an old topic but here is a new one. Example getting reputation:

#!/usr/bin/env ruby

require 'rubygems'
require 'hpricot'
require 'open-uri'

user = "619673/100kg"
html = "http://stackoverflow.com/users/%s?tab=reputation"

page = html % user
puts page

doc = Hpricot(open(page))
pars = Array.new
doc.search("div[@class='subheader user-full-tab-header']/h1/span[@class='count']").text.each do |p|
  pars << p
end

puts "reputation " + pars[0]