Refactor

lib/html/proofer.rb

@@ -1,4 +1,6 @@
 require 'nokogiri'
+require 'find'
+require 'html/proofer/checkable'
 require 'html/proofer/checks'
 
 module HTML
@@ -16,15 +18,18 @@ def run
         Find.find(@srcDir) do |path|
           if File.extname(path) == @options[:ext]
             html = HTML::Proofer.create_nokogiri(path)
-            check = klass.new(path, html, @options)
+            check = klass.new(@srcDir, path, html, @options)
             check.run
             check.hydra.run
             self.print_issues(klass, check.issues)
           end
         end
       end
 
-      if [email protected]?
+      if @failedTests.empty?
+        puts "Tests executed sucesfully.".green
+        exit 0
+      else
         # make the hash default to 0 so that += will work correctly
         count = Hash.new(0)
 
@@ -41,7 +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

@@ -10,9 +10,10 @@ class HTML::Proofer::Checks
 
   class Check
 
-    attr_reader :issues, :hydra
+    attr_reader :issues, :hydra, :src, :path, :options, :additional_href_ignores
 
-    def initialize(path, html, opts={})
+    def initialize(src, path, html, opts={})
+      @src    = src
       @path   = path
       @html   = html
       @options = opts
@@ -27,31 +28,13 @@ def run
     end
 
     def add_issue(desc)
-      @issues << desc
+      @issues << "#{@path.blue}: #{desc}"
     end
 
     def output_filenames
       Dir[@site.config[:output_dir] + '/**/*'].select{ |f| File.file?(f) }
     end
 
-    def external_href?(href)
-      uri = URI.parse(href)
-      %w( http https ).include?(uri.scheme)
-    rescue URI::BadURIError
-      false
-    rescue URI::InvalidURIError
-      false
-    end
-
-    def ignore_href?(href)
-      uri = URI.parse(href)
-      %w( mailto ).include?(uri.scheme) || @additional_href_ignores.include?(href)
-    rescue URI::BadURIError
-      false
-    rescue URI::InvalidURIError
-      false
-    end
-
     def validate_url(href, issue_text)
       request = Typhoeus::Request.new(href, {:followlocation => true})
       request.on_complete do |response|
@@ -91,5 +74,6 @@ def self.subclasses
 
       classes
     end
+
   end
 end

lib/html/proofer/checkable.rb

@@ -0,0 +1,106 @@
+module HTML
+  class Proofer
+    class Checkable
+
+      def initialize(obj, check)
+        @src = obj['src']
+        @href = obj['href']
+        @alt = obj['alt']
+        @name = obj['name']
+        @id = obj['id']
+        @check = check
+
+        if @href && @check.options[:href_swap]
+          @options[:href_swap].each do |link, replace|
+            @href = @href.gsub(link, replace)
+          end
+        end
+
+      end
+
+      def url
+        @src || @href || ""
+      end
+
+      def valid?
+        begin
+          URI.parse url
+        rescue
+          false
+        end
+      end
+
+      def parts
+         URI.parse url
+      end
+
+      def path
+        parts.path
+      end
+
+      def hash
+        parts.fragment
+      end
+
+      # path is to an external server
+      def remote?
+        uri = URI.parse url
+        %w( http https ).include?(uri.scheme)
+      rescue URI::BadURIError
+        false
+      rescue URI::InvalidURIError
+        false
+      end
+
+      def ignore?
+        uri = URI.parse url
+        %w( mailto ).include?(uri.scheme) || @check.additional_href_ignores.include?(href)
+      rescue URI::BadURIError
+        false
+      rescue URI::InvalidURIError
+        false
+      end
+
+      # path is external to the file
+      def external?
+        !internal?
+      end
+
+      # path is an anchor
+      def internal?
+        url[0] == "#"
+      end
+
+      def file_path
+
+        return if path.nil?
+
+        if path =~ /^\// #path relative to root
+          base = @check.src
+        elsif File.exist? File.expand_path path, @check.src #relative links, path is a file
+          base = File.dirname @check.path
+        else #relative link, path is a directory
+          base = @check.path
+        end
+
+        file = File.join base, path
+
+        # implicit /index.html support, with support for tailing slashes
+        file = File.join path, "index.html" if File.directory? File.expand_path file, @check.src
+
+        file
+      end
+
+      # checks if a file exists relative to the current pwd
+      def exists?
+        File.exist? absolute_path
+      end
+
+      def absolute_path
+        path = file_path || @check.path
+        File.expand_path path, Dir.pwd
+      end
+
+    end
+  end
+end

lib/html/proofer/checks/images.rb

@@ -1,30 +1,50 @@
 # encoding: utf-8
 
+class Image < ::HTML::Proofer::Checkable
+
+  SCREEN_SHOT_REGEX = /Screen(?: |%20)Shot(?: |%20)\d+-\d+-\d+(?: |%20)at(?: |%20)\d+.\d+.\d+/
+
+  def valid_alt_tag?
+    @alt and [email protected]?
+  end
+
+  def terrible_filename?
+    @src =~ SCREEN_SHOT_REGEX
+  end
+
+  def src
+    @src unless @src.nil? || @src.empty?
+  end
+
+  def missing_src?
+    !src
+  end
+
+end
+
 class Images < ::HTML::Proofer::Checks::Check
 
   def run
     @html.css('img').each do |img|
-      src = img['src']
-
-      # check image sources
-      if src && src.length > 0
-        if !external_href?(src)
-          self.add_issue("#{@path}".blue + ": internal image #{src} does not exist") unless src[0] != "/" and File.exist?(File.join(File.dirname(@path), src))
-        else
-          validate_url(src, "#{@path}".blue + ": external image #{src} does not exist")
-        end
+
+      img = Image.new img, self
+
+      # screenshot filenames, return because invalid URL
+      return self.add_issue "image has a terrible filename (#{img.src})" if img.terrible_filename?
+
+      # does the image exist?
+      if img.missing_src?
+        self.add_issue "image has no src attribute"
+      elsif img.remote?
+        validate_url img.src, "external image #{img.src} does not exist"
       else
-        self.add_issue("#{@path}".blue + ": image has no src attribute")
+        self.add_issue("internal image #{img.src} does not exist") unless img.exists?
       end
 
       # check alt tag
-      self.add_issue("#{@path}".blue + ": image #{src} does not have an alt attribute") unless img['alt'] and !img['alt'].empty?
-
-      screenShotRegExp = /Screen(?: |%20)Shot(?: |%20)\d+-\d+-\d+(?: |%20)at(?: |%20)\d+.\d+.\d+/
+      self.add_issue "image #{img.src} does not have an alt attribute" unless img.valid_alt_tag?
+
 
-      if src =~ screenShotRegExp
-        self.add_issue("#{@path}".blue + ": image has a terrible filename (#{src})")
-      end
     end
   end
 end

lib/html/proofer/checks/links.rb

@@ -1,57 +1,48 @@
 # encoding: utf-8
 
+class Link < ::HTML::Proofer::Checkable
+
+  def href
+    @href unless @href.nil? || @href.empty?
+  end
+
+  def missing_href?
+    href.nil? and @name.nil? and @id.nil?
+  end
+
+end
+
 class Links < ::HTML::Proofer::Checks::Check
 
   def run
-    @html.css('a').each do |a|
-      href = a['href']
+    @html.css('a').each do |link|
 
-      if href && href.length > 0
-        if @options[:href_swap]
-          @options[:href_swap].each do |link, replace|
-            href = href.gsub(link, replace)
-          end
-        end
+      link = Link.new link, self
+      return if link.ignore?
 
-        return if ignore_href?(href)
+      # is there even a href?
+      return self.add_issue("link has no href attribute") if link.missing_href?
 
-        if href.include? '#'
-          href_split = href.split('#')
-        end
-        if !external_href?(href)
-          # an internal link, with a hash
-          if href_split && !href_split.empty?
-            href_file = href_split[0]
-            href_hash = href_split[1]
+      # is it even a valid URL?
+      return self.add_issue "#{link.href} is an invalid URL" unless link.valid?
 
-            # it's not an internal hash; it's pointing to some other file
-            if href_file.length > 0
-              href_location = resolve_path File.join(File.dirname(@path), href_file)
-              if !File.exist?(href_location)
-                self.add_issue("#{@path}".blue + ": internal link #{href_location} does not exist")
-              else
-                href_html = HTML::Proofer.create_nokogiri(href_location)
-                found_hash_match = false
-                unless hash_check(href_html, href_hash)
-                  self.add_issue("#{@path}".blue + ": linking to #{href}, but #{href_hash} does not exist")
-                end
-              end
-            # it is an internal link, with an internal hash
-            else
-              unless hash_check(@html, href_hash)
-                self.add_issue("#{@path}".blue + ": linking to an internal hash called #{href_hash} that does not exist")
-              end
-            end
-          # internal link, no hash
-          else
-            href_location = resolve_path File.join(File.dirname(@path), href)
-            self.add_issue("#{@path}".blue + ": internally linking to #{href_location}, which does not exist") unless File.exist?(href_location)
-          end
-        else
-          validate_url(href, "#{@path}".blue + ": externally linking to #{href}, which does not exist")
-        end
+      # does the file even exist?
+      if link.remote?
+        validate_url link.href, "externally linking to #{link.href}, which does not exist"
       else
-        self.add_issue("#{@path}".blue + ": link has no href attribute") unless a['name'] || a['id']
+        self.add_issue "internally linking to #{link.href}, which does not exist" unless link.exists?
+      end
+
+      # verify the target hash
+      if link.hash
+        if link.remote?
+          #not yet checked
+        elsif link.internal?
+          self.add_issue "linking to internal hash ##{link.hash} that does not exist" unless hash_check @html, link.hash
+        elsif link.external?
+          target_html = HTML::Proofer.create_nokogiri link.absolute_path
+          self.add_issue "linking to #{link.href}, but #{link.hash} does not exist" unless hash_check target_html, link.hash
+        end
       end
     end
   end
@@ -60,9 +51,4 @@ def hash_check(html, href_hash)
     html.xpath("//*[@id='#{href_hash}']", "//*[@name='#{href_hash}']").length > 0
   end
 
-  #support for implicit /index.html in URLs
-  def resolve_path(path)
-    path << "index.html" if File.directory? path
-    path
-  end
 end

spec/html/proofer/fixtures/brokenLinkWithNumber.html

@@ -2,7 +2,7 @@
 
 <body>
 
-	<p>Blah blah blah. <a name="25-method-not-allowed" class="anchor" href="#25-method-not-allowed"><span class="mini-icon mini-icon-link"></span></a>25 Method not allowed</h4></p>
+	<p>Blah blah blah. <a class="anchor" href="#25-method-not-allowed"><span class="mini-icon mini-icon-link"></span></a>25 Method not allowed</h4></p>
 </body>
 
-</html>
+</html>

spec/html/proofer/fixtures/index.html

@@ -1 +1 @@
-root.
+<h1 id="anchor">root.</a>

spec/html/proofer/fixtures/relativeLinks.html

@@ -0,0 +1,4 @@
+<a href="/index.html">Relative to root</a>
+<a href="index.html">Relative to self</a>
+<a href="folder/">Folder relative to self</a>
+<a href="index.html#anchor">Anchor relative to self</a>