added parsing for %2f and default to return passed in value when can'…

…t be parsed

README.textile

@@ -1,5 +1,5 @@
 
-h1. Absolutize v0.0.9
+h1. Absolutize v0.0.11
 
 h2. Intro 
 
@@ -23,6 +23,7 @@ Creates a new absolutize object based on a base_url
     ** :remove_anchors - removes any anchor tags (false)
     ** :force_escaping - forces escaping of urls if it hasn't been done already (true)
     ** :output_debug - puts debug info while processing the url (false)
+    ** :raise_exceptions - raises exceptions when it can't parse a url, otherwise returns passed in value (false)
 
 bq. absolutizer = Absolutize.new("http://www.vamosa.com/mainfolder/", :remove_anchors => true)  
 

absolutize.gemspec

@@ -1,6 +1,6 @@
 spec = Gem::Specification.new do |s|
   s.name              = "absolutize"
-  s.version           = "0.0.9"
+  s.version           = "0.0.11"
   s.author            = "Stewart McKee"
   s.email             = "stewart.mckee@vamosa.com"
   s.homepage          = "http://github.com/vamosa/absolutize"

changelog.txt

@@ -1,3 +1,11 @@
+v0.0.11
+
+added raise_exceptions option and defaulted to not raising exceptions, passed in value will be returned when cant be parsed unless option set to true
+
+v0.0.10
+
+fixed parsing for %2f to allow it to work as a / when joining url
+
 v0.0.9
 
 removed the removal of // in the uri path

lib/absolutize.rb

@@ -11,6 +11,8 @@ def initialize(base_url, options = {})
     @options[:remove_anchors] = false if @options[:remove_anchors].nil? 
     @options[:force_escaping] = true if @options[:force_escaping].nil?
     @options[:output_debug] = false if @options[:output_debug].nil?
+    @options[:raise_exceptions] = false if @options[:output_debug].nil?
+
   end
 
   def url(relative_url)
@@ -25,23 +27,29 @@ def url(relative_url)
     begin
       absolute_url = URI.join(@base_url, relative_url)
     rescue URI::InvalidURIError => uri
-      puts "Unable to use URI.join attempting manually" if @options[:output_debug]
-      if @base_url =~ /\Ahttp/ and relative_url =~ /\A\//
-        puts "base url starts with http and relative_url is relative to root" if @options[:output_debug]
-        uri = URI.parse(@base_url)
-        if uri.port
-          absolute_url = URI.parse("#{uri.scheme}://#{uri.host}:#{uri.port}#{relative_url}")
+      begin
+        puts "Unable to use URI.join attempting manually" if @options[:output_debug]
+        if @base_url =~ /\Ahttp/ and relative_url =~ /\A\//
+          puts "base url starts with http and relative_url is relative to root" if @options[:output_debug]
+          uri = URI.parse(@base_url)
+          if uri.port
+            absolute_url = URI.parse("#{uri.scheme}://#{uri.host}:#{uri.port}#{relative_url}")
+          else
+            absolute_url = URI.parse("#{uri.scheme}://#{uri.host}#{relative_url}")
+          end
+        elsif relative_url =~ /\Ahttp/
+          #new url is absolute anyway
+          absolute_url = URI.parse(relative_url)
         else
-          absolute_url = URI.parse("#{uri.scheme}://#{uri.host}#{relative_url}")
+          raise "Unable to absolutize #{@base_url} and #{relative_url}" if @options[:raise_exceptions]
+          absolute_url = relative_url
         end
-      elsif relative_url =~ /\Ahttp/
-        #new url is absolute anyway
-        absolute_url = URI.parse(relative_url)
-      else
-        raise "Unable to absolutize #{@base_url} and #{relative_url}"
+      rescue URI::InvalidURIError => uri_2
+        # ok, givin it a fair go trying to get this url, raise an exception if the option is set otherwise give up and return original relative url
+        raise "Unable to absolutize #{@base_url} and #{relative_url}" if @options[:raise_exceptions]
+        absolute_url = relative_url        
       end
     end
-
     absolute_url
   end
 end

spec/absolutize/absolutize_spec.rb

@@ -49,6 +49,9 @@
     it "should encode with pipes" do
       @absolutize.url("/root%20folder/asdf.html?lang=en|us").to_s.should == "http://www.baseurl.com/root%20folder/asdf.html?lang=en%7Cus"         
     end
+    it "should not raise an exception with really bad url" do
+      @absolutize.url("http://.asdf.com/").to_s.should == "http://.asdf.com/"
+    end
   end
 
   describe "with remove_anchors true" do
@@ -65,15 +68,27 @@
       @absolutize = Absolutize.new(@base_url, :force_escaping => false)
     end
     it "should not escape invalid characters" do
-      lambda { @absolutize.url("/root folder/asdf.html#anchor")}.should raise_error
-
       ab = Absolutize.new("http://www.baseurl.com/top_folder/index.html", :force_escaping => false)
-      #this is actually wrong, but we have not forced escaping
+
+      #this example should be decoded first, but we are checking that when force_escaping is off it does not
       ab.url("%2ftop_folder%2fsub_folder%2findex.html").to_s.should == "http://www.baseurl.com/top_folder/%2ftop_folder%2fsub_folder%2findex.html"
-      #should work fine
       ab.url("/top_folder/sub_folder/index.html").to_s.should == "http://www.baseurl.com/top_folder/sub_folder/index.html"
-
     end
-
+  end
+
+  describe "with raise_exceptions true" do
+    before(:each) do
+      @absolutize = Absolutize.new(@base_url, :raise_exceptions => true)
+    end
+    it "should raise an exception when absolutize can't fix a url" do
+      lambda { @absolutize.url("http://.asdf.com/root folder/asdf.html#anchor")}.should raise_error
+    end
+    it "should not raise an exception when a valid url is given" do
+      lambda { @absolutize.url("http://www.asdf.com/rootfolder/asdf.html#anchor")}.should_not raise_error
+    end
+
+    it "should not raise an exception when a fixable url is given" do
+      lambda { @absolutize.url("http://www.asdf.com/root folder[1]/asdf.html#anchor")}.should_not raise_error    
+    end 
   end
 end