Merge pull request #18 from gjtorikian/cache-matched-urls

Cache successful paths internally

Gemfile.lock

@@ -22,7 +22,7 @@ GEM
     ethon (0.6.1)
       ffi (>= 1.3.0)
       mime-types (~> 1.18)
-    ffi (1.9.0)
+    ffi (1.9.3)
     gemoji (1.4.0)
     github-markdown (0.5.5)
     html-pipeline (0.0.14)
@@ -34,7 +34,7 @@ GEM
       rinku (~> 1.7)
       sanitize (~> 2.0)
     i18n (0.6.5)
-    mime-types (1.25)
+    mime-types (1.25.1)
     mini_portile (0.5.1)
     minitest (4.7.5)
     multi_json (1.8.1)

Rakefile

@@ -5,4 +5,4 @@ require 'rspec/core/rake_task'
 
 RSpec::Core::RakeTask.new(:spec)
 
-task :default => :spec
+task :default => :spec

lib/html/proofer.rb

@@ -46,7 +46,7 @@ def run
     end
 
     def self.create_nokogiri(path)
-      path << "/index.html" if File.directory? path #support for Jekyll-style links
+      path << "/index.html" if File.directory? path # support for Jekyll-style links
       Nokogiri::HTML(File.read(path))
     end
 

lib/html/proofer/check.rb

@@ -18,6 +18,7 @@ def initialize(src, path, html, opts={})
       @html   = html
       @options = opts
       @issues = []
+      @checked_urls = {}
 
       @hydra = Typhoeus::Hydra.hydra
       @additional_href_ignores = @options[:href_ignore] || []
@@ -36,10 +37,11 @@ def output_filenames
     end
 
     def validate_url(href, issue_text)
+      return @checked_urls[href] if @checked_urls.has_key? href
       request = Typhoeus::Request.new(href, {:followlocation => true})
       request.on_complete do |response|
         if response.success?
-          # no op
+          @checked_urls[href] = true
         elsif response.timed_out?
           self.add_issue(issue_text + " got a time out")
         elsif response.code == 0
@@ -57,6 +59,8 @@ def validate_url(href, issue_text)
             self.add_issue("#{issue_text} HTTP request failed: #{response_code}")
           end
         end
+
+        @checked_urls[href] = false unless response.success?
       end
       hydra.queue(request)
     end

lib/html/proofer/checkable.rb

@@ -10,6 +10,7 @@ def initialize(obj, check)
         @id = obj['id']
         @data_ignore_proofer = obj['data-proofer-ignore']
         @check = check
+        @checked_paths = {}
 
         if @href && @check.options[:href_swap]
           @check.options[:href_swap].each do |link, replace|
@@ -73,7 +74,6 @@ def internal?
       end
 
       def file_path
-
         return if path.nil?
 
         if path =~ /^\// #path relative to root
@@ -94,7 +94,8 @@ def file_path
 
       # checks if a file exists relative to the current pwd
       def exists?
-        File.exist? absolute_path
+        return @checked_paths[absolute_path] if @checked_paths.has_key? absolute_path
+        @checked_paths[absolute_path] = File.exist? absolute_path
       end
 
       def absolute_path