Question

I have a Ruby program that takes as input an HTML document. The structure of this HTML document is unknown and can really be anything.

Also it is guaranteed that the document uses CSS for formatting.

The output needs to be a hash where each key is a CSS class used in the document and the value is the number of times that CSS class appears in the body of the document.

Answer 1

require 'open-uri'
require 'nokogiri'

doc = Nokogiri.HTML(open('http://stackoverflow.com'))
classes = doc.xpath('//@class').map{ |attr| attr.value.split(/\s+/) }.flatten
counts = Hash[
  classes.group_by(&:to_s).map{ |n,ary| [n,ary.count] }.sort_by{ |n,i| [-i,n] }
]

require 'pp'
pp counts
#=>{"post-tag"=>320,
#=> "mini-counts"=>270,
#=> "cp"=>90,
#=> "narrow"=>90,
#=> "question-hyperlink"=>90,
#=> ...etc...

Described:

1. Use xpath to find all class="..." attributes, e.g. class="a b", class="c", class="b"
2. split on whitespace, e.g. class="a b" ⇒ ["a","b"]
3. flatten into a single list, e.g. [["a","b"],["c"],["b"]] ⇒ ["a","b","c","b"]
4. group_by to hash from string to hits, e.g. ⇒ {"a"=>["a"],"b"=>["b","b"],"c"=>["c"]}
5. map to pair name with length, e.g. ⇒ [["a",1],["b",2],["c",1]]
6. For fun, sort_by occurrences descending, e.g. ⇒ [["b",2],["a",1],["c",1]]
7. Hash.[] to turn into a hash again, e.g. ⇒ {"b"=>2,"a"=>1",c"=>1}

Alternatively (not functional-style, more straightforward) after step 3 just loop through and add counts to a hash. For bonus points, initialize the hash with default ‘0’ values:

counts = Hash.new(0)
classes.each{ |name| counts[name] += 1 }