raakt 0.3 → 0.4

data/lib/raakt (kopia).rb

@@ -0,0 +1,495 @@
+# :title: Ruby Accessibility Analysis Kit
+# =Ruby Accessibility Analysis Kit
+#
+# See README for a full explanation of this library.
+
+module Raakt
+  require 'rubyful_soup'
+  
+  MESSAGES = {
+    "missingtitle"    => "The title element is missing. Provide a descriptive title for your document.",
+    "emptytitle"      => "The title element is empty. Provide a descriptive title for your document.",  
+    "missingalt"      => "Missing alt attribute for image (with src '%s').",
+    "missingheading"  => "Missing first level heading (h1). Provide at least one first level heading describing document content.",
+    "wronghstructure" => "Document heading structure is wrong.",
+    "firsthnoth1"     => "The first heading is not h1.",
+    "hasnestedtables" => "You have one or more nested tables.",
+    "missingsemantics"=> "You have used <font>, <b> or <i> for visual formatting. Use CSS instead.",
+    "hasflicker"      => "You have used <blink> or <marquee>. These may create accessibility issues and should be avoided.",
+    "missinglanginfo" => "Document language information is missing. Use the lang attribute on the html element.",
+    "missingth"       => "Missing table headings (th) for table #%s.",
+    "ambiguouslinktext" => "One or more links have the same link text ('%s'). Make sure each link is unambiguous.",
+    "fieldmissinglabel" => "A field (with id/name '%s') is missing a corresponding label element. Make sure a label exists for all visible fields.",
+    "missingframetitle" => "Missing title attribute for frame with url %s",
+    "hasmetarefresh"  => "Client side redirect (meta refresh) detected. Use server side redirection instead."
+  } 
+  
+  VERSION = "0.31"
+  
+  class ErrorMessage
+  
+    attr_reader :eid, :text, :note
+  
+    def initialize(eid, note=nil)
+      @eid = eid
+      if note
+        @text = MESSAGES[eid].sub(/%s/, note)
+      else
+        @text = MESSAGES[eid]
+      end
+      @note = note
+    end
+  
+    def to_s
+      @eid + ": " + @text
+    end
+  end
+
+
+
+
+
+  class Test
+
+    attr_accessor :soup, :html, :user_agent
+  
+    def initialize(html=nil)
+      @html = html
+      @soup = BeautifulSoup.new(@html) if html
+      @user_agent = "Mozilla/5.0 (RAAKT v#{VERSION}; http://raakt.rubyforge.org; The Ruby Accessibility Analysis Kit)"
+    end
+    
+    def feed(html)
+      @html = html || ""
+      if @html.length > 0
+        @soup = BeautifulSoup.new(@html)
+      else
+        raise "You called feed with no data. There is nothing to check."
+      end
+    end
+    
+    
+    
+    def feedurl(url)
+      if url.length == 0
+        raise "You called feedurl with a blank url. There is nothing to check."
+      end
+      
+      #Clean the url and make sure protocol and trailing slash is available
+      url = "http://" + url unless url[0..3] == "http"
+      
+      require 'open-uri'
+
+      open(url, "User-Agent" => @user_agent) { |f|
+        @html = f.read || ""
+      }
+      
+      if @html.length == 0
+        raise "Could not fetch html from the url #{url}. There is nothing to check."
+      else
+        @soup = BeautifulSoup.new(@html)
+      end      
+    
+    end
+    
+    
+    
+    def all
+      #Call all check methods
+      messages = []
+      
+      self.methods.each do |method|
+        if method[0..5] == "check_"
+          messages += self.send(method)
+        end
+      end
+      
+      return messages
+    end
+
+  
+    def check_images
+      #soup = BeautifulSoup.new(html)
+      images = @soup.find_all("img")
+      messages = []
+    
+      for image in images:
+          if image["alt"] == nil:
+            img_src = image["src"] || ""
+            messages << ErrorMessage.new("missingalt", img_src)
+          end
+      end
+      
+      return messages  
+    end
+
+  
+    def check_title
+      title = @soup.find("title")
+      messages = []
+    
+      if title
+        titletext = normalize_text(title.string)
+        if titletext.length == 0
+          messages << ErrorMessage.new("emptytitle")
+        end
+      else
+        messages << ErrorMessage.new("missingtitle")
+      end
+      
+      return messages 
+    end
+
+  
+    def check_has_heading
+      messages = []  
+      
+      if @soup.find_all("h1").length == 0
+        messages << ErrorMessage.new("missingheading")
+      end
+      
+      return messages
+    end
+
+
+    def headings
+      headings = []
+      headings.push(@soup.find_all("h1")) if @soup.find_all("h1").length > 0
+      headings.push(@soup.find_all("h2")) if @soup.find_all("h2").length > 0
+      headings.push(@soup.find_all("h3")) if @soup.find_all("h3").length > 0
+      headings.push(@soup.find_all("h4")) if @soup.find_all("h4").length > 0
+      headings.push(@soup.find_all("h5")) if @soup.find_all("h5").length > 0
+      headings.push(@soup.find_all("h6")) if @soup.find_all("h6").length > 0
+      
+      return headings.flatten
+    end
+
+    
+    def level(heading)
+      Integer(heading[1,1])
+    end
+
+    
+    def check_document_structure
+      messages = []
+      currentitem = 0
+      docheadings = headings
+      
+      for heading in docheadings
+        if currentitem == 0
+          if level(heading.name) != 1
+            messages << ErrorMessage.new("firsthnoth1", "h" + heading.name[1,1])
+          end
+        else
+          if level(heading.name) - level(docheadings[currentitem - 1].name) > 1
+            messages << ErrorMessage.new("wronghstructure")
+            break
+          end  
+        end
+        
+        currentitem += 1
+        
+      end
+      
+      return messages
+    end
+
+    
+    def check_for_nested_tables
+      
+      messages = []  
+      tables = @soup.find_all("table")
+      
+      for table in tables
+        if table.find_all("table").length > 0
+          messages << ErrorMessage.new("hasnestedtables")
+          break
+        end
+      end
+      
+      return messages
+    end
+
+    
+    def check_tables
+      
+      messages = []  
+      tables = @soup.find_all("table")   
+      hasth = false
+      currenttable = 1
+      
+      for table in tables     
+        if table.thead
+          if table.thead.tr
+            if table.thead.tr.th   
+              hasth = true
+            end
+          end
+        end
+        
+        if table.tr
+          if table.tr.th
+            hasth = true
+          end
+        end
+        
+        unless hasth
+          messages << ErrorMessage.new("missingth", currenttable.to_s)
+        end
+        
+        currenttable += 1
+      end
+      
+      return messages
+    end
+    
+    
+    def check_for_formatting_elements
+      
+      messages = []  
+      formatting_items = @soup.find_all(["font", "b", "i"])
+      flicker_items = @soup.find_all(["blink", "marquee"])
+      
+      if formatting_items.length > 0
+          messages << ErrorMessage.new("missingsemantics")  
+      end
+
+      if flicker_items.length > 0
+          messages << ErrorMessage.new("hasflicker")  
+      end
+            
+      return messages   
+    end
+    
+    
+    def check_for_language_info
+      messages = []
+      
+      htmlelement = @soup.find("html")
+  
+      lang = langinfo(htmlelement) || ""
+      
+      unless lang.length > 1
+        messages << ErrorMessage.new("missinglanginfo")
+      end
+      
+      return messages
+    end
+    
+    
+    def check_link_text
+      messages = []
+      links = get_links
+      linktexts = links.collect { |el| el[3] }
+      
+      for link_a in links
+        #compare to other links in collection
+        for link_b in links
+          if link_a[0] != link_b[0]
+            if is_ambiguous_link(link_a, link_b)
+              #add message if not added already for link text
+              unless find_errormsg_with_text(messages, link_a[3])
+                messages << ErrorMessage.new("ambiguouslinktext", link_a[3])
+              end
+            end
+          end
+        end
+      end
+      
+      return messages      
+    end
+        
+    
+    def check_form
+      messages = []
+      labels = get_labels
+      fields = get_editable_fields
+      
+      #make sure all fields have associated labels
+      
+      label_for_ids = []
+      for label in labels
+        if label["for"]
+          label_for_ids << label["for"]
+        end
+      end
+      
+      field_id = nil
+      
+      for field in fields
+        field_id = (field["id"] || "")
+        field_identifier = (field["id"] || field["name"] || "unknown")
+        if not label_for_ids.include?(field_id)
+          messages << ErrorMessage.new("fieldmissinglabel", field_identifier)
+        end
+      end   
+      
+      return messages
+    end
+    
+    
+    def check_frames
+      #Verify frame titles
+      
+      messages = []  
+      if is_frameset
+        frames = @soup.find_all("frame")
+        frame_title = ""
+      
+        for frame in frames
+          frame_title = frame["title"] || ""
+          if normalize_text(frame_title).length == 0
+            messages << ErrorMessage.new("missingframetitle", frame["src"])
+          end
+        end
+      end
+      
+      return messages            
+    end
+    
+    
+    def check_refresh
+      
+      messages = []
+      meta_elements = @soup.find_all("meta")
+      
+      for element in meta_elements
+        if element["http-equiv"] == "refresh"
+          messages << ErrorMessage.new("hasmetarefresh")
+        end
+      end
+      
+      return messages
+    end
+    
+    
+    #Utility methods    
+    
+    def is_ambiguous_link(link_a, link_b)
+      #Link A and B are ambiguous if:
+      #1. The url differs
+      #2. The link text is identical
+      #3. The title text is identical (if present)
+      if link_a[1] != link_b[1] and 
+         normalize_text(link_a[2]) == normalize_text(link_b[2]) and
+         normalize_text(link_a[3]) == normalize_text(link_b[3]) then
+         return true
+      end
+      
+      return false
+    end
+    
+    
+    def find_errormsg_with_text(messages, text)
+      for errormessage in messages
+        if errormessage.note == text
+          return errormessage
+        end
+      end
+      
+      return nil
+    end
+    
+    
+    def get_links      
+      linkelements = @soup.find_all("a")
+      links = []
+      currentlink = 0
+      
+      for element in linkelements
+        title = normalize_text((element['title'] || "").strip)
+        linktext = normalize_text((elements_to_text(element) || "").strip)
+        url = element['href']
+        links << [currentlink, url, title, linktext]
+        currentlink += 1
+      end      
+      
+      return links
+    end
+
+
+    def langinfo(element)
+      langval = ""
+      
+      if element.class.to_s == 'Tag'      
+        if element['lang']
+          langval = element['lang']
+        end      
+      else
+        return nil
+      end
+      
+      return langval
+    end    
+    
+    
+    def img_to_text(imgtag)
+      return (imgtag['alt'] || "")
+    end
+    
+    
+    def elements_to_text(element)
+      retval = ""
+      
+      for el in element.contents
+        if el.class.to_s == 'NavigableString'
+          retval += el
+        else
+          if el.name == "img"
+            retval += img_to_text(el)
+          else
+            retval += elements_to_text(el)
+          end
+        end        
+      end
+      
+      return retval
+    end
+    
+    
+    def normalize_text(text)
+      text = (text || "")
+      retval = text.gsub(/&nbsp;/, " ")
+      retval = retval.gsub(/&#160;/, " ")
+      retval = retval.gsub(/\n/, "")
+      retval = retval.gsub(/\r/, "")
+      retval = retval.gsub(/\t/, "")
+      while /  /.match(retval) do
+        retval = retval.gsub(/  /, " ")
+      end
+      
+      retval = retval.strip
+      
+      return retval
+    end
+    
+    
+    def get_labels
+      return @soup.find_all("label")
+    end
+
+
+    def get_editable_fields
+      allfields = @soup.find_all(["textarea", "select", "input"])
+      fields = []
+      field_type = ""
+      
+      for field in allfields do
+        field_type = field["type"] || ""
+        unless ["button", "submit", "hidden", "image"].include?(field_type)
+          fields << field
+        end
+        
+      end
+      
+      return fields
+    end
+        
+        
+    def is_frameset
+      return (@soup.find("frameset") != nil)
+    end    
+    
+  end
+
+end

data/lib/raakt.rb

@@ -14,7 +14,7 @@ module Raakt
     "wronghstructure" => "Document heading structure is wrong.",
     "firsthnoth1"     => "The first heading is not h1.",
     "hasnestedtables" => "You have one or more nested tables.",
-    "missingsemantics"=> "You have used <font>, <b> or <i> for visual formatting. Use CSS instead.",
+    "missingsemantics"=> "You have used %s for visual formatting. Use CSS instead.",
     "hasflicker"      => "You have used <blink> or <marquee>. These may create accessibility issues and should be avoided.",
     "missinglanginfo" => "Document language information is missing. Use the lang attribute on the html element.",
     "missingth"       => "Missing table headings (th) for table #%s.",
@@ -24,7 +24,7 @@ module Raakt
     "hasmetarefresh"  => "Client side redirect (meta refresh) detected. Use server side redirection instead."
   } 
   
-  VERSION = "0.3"
+  VERSION = "0.4"
   
   class ErrorMessage
   
@@ -51,11 +51,12 @@ module Raakt
 
   class Test
 
-    attr_accessor :soup, :html
+    attr_accessor :soup, :html, :user_agent
   
     def initialize(html=nil)
       @html = html
       @soup = BeautifulSoup.new(@html) if html
+      @user_agent = "Mozilla/5.0 (RAAKT v#{VERSION}; http://raakt.rubyforge.org; The Ruby Accessibility Analysis Kit)"
     end
     
     def feed(html)
@@ -67,8 +68,34 @@ module Raakt
       end
     end
     
+    
+    
+    def feedurl(url)
+      if url.length == 0
+        raise "You called feedurl with a blank url. There is nothing to check."
+      end
+      
+      #Clean the url and make sure protocol and trailing slash is available
+      url = "http://" + url unless url[0..3] == "http"
+      
+      require 'open-uri'
+
+      open(url, "User-Agent" => @user_agent) { |f|
+        @html = f.read || ""
+      }
+      
+      if @html.length == 0
+        raise "Could not fetch html from the url #{url}. There is nothing to check."
+      else
+        @soup = BeautifulSoup.new(@html)
+      end      
+    
+    end
+    
+    
+    
     def all
-      #Do all checks
+      #Call all check methods
       messages = []
       
       self.methods.each do |method|
@@ -220,11 +247,11 @@ module Raakt
     def check_for_formatting_elements
       
       messages = []  
-      formatting_items = @soup.find_all(["font", "b", "i"])
+      formatting_items = @soup.find_all(%w(font b i u tt small big strike s))
       flicker_items = @soup.find_all(["blink", "marquee"])
       
-      if formatting_items.length > 0
-          messages << ErrorMessage.new("missingsemantics")  
+      formatting_items.each do |element|
+        messages << ErrorMessage.new("missingsemantics", "<#{element.name}>")  
       end
 
       if flicker_items.length > 0

data/tests/invalidelements1.htm

@@ -14,5 +14,6 @@
     <table><tr><td>Test table</td></tr></table>
     <H1>Third h1 heading</H1>
 	<marquee>Marquees are so nineties!</marquee>
+	<center>Centering should be done with CSS</center>
   </body>
 </html>

data/tests/langinfodoc1.htm

@@ -0,0 +1,17 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
+  "http://www.w3.org/TR/html4/strict.dtd">
+<html lang="zh-tw" dir="ltr">
+  <head>
+    <?xml version="1.0" encoding="UTF-8"?>
+    <title>This is the title</title>
+    <link rel="schema.DC" href="http://purl.org/DC/elements/1.0/" />
+  </head>
+  <body>
+      <TABLE><tr><td>
+    <p>This is a minimal <a href="http://www.w3.org/TR/html/">HTML 4</a> 
+    document.</p>
+    </td>
+    </tr>
+    </table>
+  </body>
+</html>

data/tests/raakt_test.rb

@@ -6,13 +6,32 @@ class RaaktTest < Test::Unit::TestCase
   
   def setup
     @raakt = Raakt::Test.new
+    
+    #Use includeremote option to run unit tests that fetch test data from http://raakt.rubyforge.org
+    @remote_tests = false
+    if ARGV[0] == "includeremote"
+      @remote_tests = true
+    end
   end
   
   def test_all
     @raakt.feed(data_full_google)
-    assert_equal 7, @raakt.all.length
+    assert_equal 15, @raakt.all.length
+  end
+  
+  
+  def test_feedurl
+    if @remote_tests
+      @raakt.feedurl("http://raakt.rubyforge.org/tests/langdoc2-iso88591.htm")
+      assert_equal 477, @raakt.html.length
+      
+      #a redirect
+    else
+      assert true
+    end
   end
   
+  
   def test_check_images
     @raakt.feed(data_imagedoc1)
     assert_equal 1, @raakt.check_images.length
@@ -62,12 +81,15 @@ class RaaktTest < Test::Unit::TestCase
     assert_equal 0, @raakt.headings.length
   end
   
+  
   def test_level
     assert_equal 1, @raakt.level("h1")
     assert_equal 2, @raakt.level("h2")
     assert_equal 6, @raakt.level("h6")
   end
   
+  
+  
   def test_check_has_heading
     @raakt.feed(data_empty)
     assert_equal 1, @raakt.check_has_heading.length
@@ -84,6 +106,7 @@ class RaaktTest < Test::Unit::TestCase
     assert_equal "missingheading", @raakt.check_has_heading[0].eid
   end
   
+  
   def test_check_document_structure
     
     @raakt.feed(data_headingsdoc1)
@@ -126,6 +149,7 @@ class RaaktTest < Test::Unit::TestCase
     assert_equal 1, @raakt.check_for_nested_tables.length
   end
   
+  
   def test_check_tables
     @raakt.feed(data_tabledoc4)
     assert_equal 0, @raakt.check_tables.length
@@ -155,6 +179,16 @@ class RaaktTest < Test::Unit::TestCase
     @raakt.feed(data_tablelayoutdoc)
     assert_equal 1, @raakt.check_for_language_info.length   
     
+    @raakt.feed(data_langinfodoc1)
+    assert_equal 0, @raakt.check_for_language_info.length
+    
+    if @remote_tests
+      @raakt.feedurl("http://raakt.rubyforge.org/tests/langdoc2-iso88591.htm")
+      assert_equal 1, @raakt.check_for_language_info.length
+    
+      @raakt.feedurl("http://raakt.rubyforge.org/tests/langdoc1-utf8.htm")
+      assert_equal 0, @raakt.check_for_language_info.length
+    end
   end
   
   
@@ -295,9 +329,10 @@ class RaaktTest < Test::Unit::TestCase
   def test_check_for_formatting_elements
     @raakt.feed(data_invalidelements1)
     invaliderrs = @raakt.check_for_formatting_elements
-    assert_equal 2, invaliderrs.length
+    assert_equal 3, invaliderrs.length
     assert_equal "missingsemantics", invaliderrs[0].eid
-    assert_equal "hasflicker", invaliderrs[1].eid
+    assert_equal "missingsemantics", invaliderrs[1].eid
+    assert_equal "hasflicker", invaliderrs[2].eid
     
     @raakt.feed(data_xhtmldoc1)
     assert_equal 0, @raakt.check_for_formatting_elements.length

metadata

@@ -3,13 +3,13 @@ rubygems_version: 0.8.11
 specification_version: 1
 name: raakt
 version: !ruby/object:Gem::Version 
-  version: "0.3"
-date: 2006-07-13 00:00:00 +02:00
+  version: "0.4"
+date: 2006-08-17 00:00:00 +02:00
 summary: A toolkit to find accessibility issues in HTML documents.
 require_paths: 
 - lib
 email: peter.krantzNODAMNSPAM@gmail.com
-homepage: http://peterkrantz.com/projects/raakt
+homepage: http://raakt.rubyforge.org
 rubyforge_project: raakt
 description: 
 autorequire: raakt
@@ -28,6 +28,7 @@ cert_chain:
 authors: 
 - Peter Krantz
 files: 
+- lib/raakt (kopia).rb
 - lib/raakt.rb
 - tests/empty.htm
 - tests/emptytitledoc.htm
@@ -55,6 +56,7 @@ files:
 - tests/invalidhtmldoc1.htm
 - tests/invalidhtmldoc2.htm
 - tests/invalidxhtmldoc1.htm
+- tests/langinfodoc1.htm
 - tests/linkdoc1.htm
 - tests/linkdoc2.htm
 - tests/linkdoc3.htm