Improved parser support for erb blocks inside html tags

BDQ · May 24, 2011 · a8f14db · a8f14db
1 parent 1b271d8
commit a8f14db
Show file tree

Hide file tree

Showing 2 changed files with 72 additions and 24 deletions.
diff --git a/lib/deface/parser.rb b/lib/deface/parser.rb
@@ -1,20 +1,48 @@
 require 'nokogiri'
-require 'erubis'
 require 'cgi'
 
 module Deface
   class Parser
     # converts erb to markup
     #
     def self.erb_markup!(source)
+
+      #all opening html tags that contain <% %> blocks
+      source.scan(/<\w+[^<>]+(?:<%.*?%>[^<>]*)+/m).each do |line|
+
+        #regexs to catch <% %> inside attributes id="<% something %>" - with double, single or no quotes
+        erb_attrs_regexs = [/([\w-]+)(\s?=\s?)(")([^"]*<%.*?%>[^"]*)/m,
+          /([\w-]+)(\s?=\s?)(')([^']*<%.*?%>[^']*)'/m,
+          /([\w-]+)(\s?=\s?)()(<%.*?%>)(?:\s|>|\z)/m]
+
+        replace_line = erb_attrs_regexs.inject(line.clone) do |replace_line, regex|
+
+          replace_line = line.scan(regex).inject(replace_line) do |replace_line, match|
+            replace_line.sub("#{match[0]}#{match[1]}#{match[2]}#{match[3]}#{match[2]}") { |m| m = " data-erb-#{match[0]}=\"#{CGI.escapeHTML(match[3])}\"" }
+          end
+
+          replace_line
+        end
+
+
+        i = -1
+        #catch all <% %> inside tags id <p <%= test %>> , not inside attrs
+        replace_line.scan(/(<%.*?%>)/m).each do |match|
+          replace_line.sub!(match[0]) { |m| m = " data-erb-#{i += 1}=\"#{CGI.escapeHTML(match[0])}\"" }
+        end
+
+        source.sub!(line) { |m| m = replace_line }
+      end
+
+      #replaces all <% %> not inside opening html tags
       replacements = [ {"<%=" => "<code erb-loud>"},
                        {"<%"  => "<code erb-silent>"},
                        {"%>"  => "</code>"} ]
 
       replacements.each{ |h| h.each { |replace, with| source.gsub! replace, with } }
 
       source.scan(/(<code.*?>)((?:(?!<\/code>)[\s\S])*)(<\/code>)/).each do |match|
-        source.gsub!("#{match[0]}#{match[1]}#{match[2]}", "#{match[0]}#{CGI.escapeHTML(match[1])}#{match[2]}")
+        source.sub!("#{match[0]}#{match[1]}#{match[2]}") { |m| m = "#{match[0]}#{CGI.escapeHTML(match[1])}#{match[2]}" }
       end
 
       source
@@ -25,18 +53,21 @@ def self.erb_markup!(source)
     def self.undo_erb_markup!(source)
       replacements = [ {"<code erb-silent>" => '<%'},
                        {"<code erb-loud>"   => '<%='},
-                       {"</code>"           => '%>'},
-                       {/(<|&lt;)code(\s|%20)erb-silent(&gt;|>)/ => '<%'},
-                       {/(<|&lt;)code(\s|%20)erb-loud(&gt;|>)/   => '<%='},
-                       {/(<|&lt;)\/code(&gt;|>)/                 => '%>'} ]
+                       {"</code>"           => '%>'}]
 
       replacements.each{ |h| h.each { |replace, with| source.gsub! replace, with } }
 
+      source.scan(/data-erb-(\d+)+=(['"])(.*?)\2/m).each do |match|
+        source.gsub!("data-erb-#{match[0]}=#{match[1]}#{match[2]}#{match[1]}") { |m| m = CGI.unescapeHTML(match[2]) }
+      end
+
+      source.scan(/data-erb-([\w-]+)+=(["'])(.*?)\2/m).each do |match|
+        source.gsub!("data-erb-#{match[0]}=#{match[1]}#{match[2]}#{match[1]}") { |m| "#{match[0]}=#{match[1]}#{CGI.unescapeHTML(match[2])}#{match[1]}" }
+      end
+
       #un-escape changes from Nokogiri and erb-markup!
       source.scan(/(<%.*?)((?:(?!%>)[\s\S])*)(%>)/).each do |match|
-        escaped = URI.unescape match[1]
-        escaped = CGI.unescapeHTML CGI.unescapeHTML(escaped)
-        source.gsub!("#{match[0]}#{match[1]}#{match[2]}", "#{match[0]}#{escaped}#{match[2]}")
+        source.gsub!("#{match[0]}#{match[1]}#{match[2]}") { |m| m = "#{match[0]}#{ CGI.unescapeHTML match[1] }#{match[2]}" }
       end
 
       source

diff --git a/spec/deface/parser_spec.rb b/spec/deface/parser_spec.rb
@@ -37,12 +37,34 @@ module Deface
         Deface::Parser.convert("<%= method_name %>").to_s.should == "<code erb-loud> method_name </code>"
       end
 
-      it "should convert nested <% ... %>" do
-        Deface::Parser.convert("<p id=\"<% method_name %>\"></p>").to_s.should == "<p id=\"&lt;code erb-silent&gt; method_name &lt;/code&gt;\"></p>"
+      it "should convert first <% ... %> inside html tag" do
+        Deface::Parser.convert("<p <% method_name %>></p>").to_s.should == "<p data-erb-0=\"&lt;% method_name %&gt;\"></p>"
       end
 
-      it "should convert nested <%= ... %> including href attribute" do
-        Deface::Parser.convert(%(<a href="<%= x 'y' + "z" %>">A Link</a>)).to_s.should == "<a href=\"&lt;code%20erb-loud&gt;%20x%20'y'%20+%20%22z%22%20&lt;/code&gt;\">A Link</a>"
+      it "should convert second <% ... %> inside html tag" do
+        Deface::Parser.convert("<p <% method_name %> <% x = y %>></p>").to_s.should == "<p data-erb-0=\"&lt;% method_name %&gt;\" data-erb-1=\"&lt;% x = y %&gt;\"></p>"
+      end
+
+      it "should convert <% ... %> inside double quoted attr value" do
+        Deface::Parser.convert("<p id=\"<% method_name %>\"></p>").to_s.should == "<p data-erb-id=\"&lt;% method_name %&gt;\"></p>"
+      end
+
+      it "should convert <% ... %> inside single quoted attr value" do
+        Deface::Parser.convert("<p id='<% method_name %>'></p>").to_s.should == "<p data-erb-id=\"&lt;% method_name %&gt;\"></p>"
+      end
+
+      it "should convert <% ... %> inside non-quoted attr value" do
+        Deface::Parser.convert("<p id=<% method_name %>></p>").to_s.should == "<p data-erb-id=\"&lt;% method_name %&gt;\"></p>"
+        Deface::Parser.convert("<p id=<% method_name %> alt=\"test\"></p>").to_s.should == "<p data-erb-id=\"&lt;% method_name %&gt;\" alt=\"test\"></p>"
+      end
+
+      it "should convert multiple <% ... %> inside html tag" do
+        Deface::Parser.convert(%q{<p <%= method_name %> alt="<% x = 'y' + 
+                               \"2\" %>" title='<% method_name %>' <%= other_method %></p>}).to_s.should == "<p data-erb-0=\"&lt;%= method_name %&gt;\" data-erb-alt=\"&lt;% x = 'y' + \n                               \\&quot;2\\&quot; %&gt;\" data-erb-title=\"&lt;% method_name %&gt;\" data-erb-1=\"&lt;%= other_method %&gt;\"></p>"
+      end
+
+      it "should convert <%= ... %> including href attribute" do
+        Deface::Parser.convert(%(<a href="<%= x 'y' + "z" %>">A Link</a>)).to_s.should == "<a data-erb-href=\"&lt;%= x 'y' + &quot;z&quot; %&gt;\">A Link</a>"
       end
 
       it "should escape contents code tags" do
@@ -59,21 +81,16 @@ module Deface
         Deface::Parser.undo_erb_markup!("<code erb-loud> method_name </code>").should == "<%= method_name %>"
       end
 
-      it "should revert nested <code erb-silent>" do
-        Deface::Parser.undo_erb_markup!("<p id=\"&lt;code erb-silent&gt; method_name > 1 &lt;/code&gt;\"></p>").should == "<p id=\"<% method_name > 1 %>\"></p>"
-      end
-
-      it "should revert nested <code erb-silent> including href attribute" do
-        Deface::Parser.undo_erb_markup!("<a href=\"&lt;code%20erb-silent&gt;%20method_name%20&lt;/code&gt;\">A Link</a>").should == "<a href=\"<% method_name %>\">A Link</a>"
+      it "should revert data-erb-x attrs inside html tag" do
+        Deface::Parser.undo_erb_markup!("<p data-erb-0=\"&lt;% method_name %&gt;\" data-erb-1=\"&lt;% x = y %&gt;\"></p>").should == "<p <% method_name %> <% x = y %>></p>"
       end
 
-      it "should revert nested <code erb-loud>" do
-        Deface::Parser.undo_erb_markup!("<p id=\"&lt;code erb-loud&gt; method_name < 2 &lt;/code&gt;\"></p>").should == "<p id=\"<%= method_name < 2 %>\"></p>"
+      it "should revert data-erb-id attr inside html tag" do
+        Deface::Parser.undo_erb_markup!("<p data-erb-id=\"&lt;% method_name &gt; 1 %&gt;\"></p>").should == "<p id=\"<% method_name > 1 %>\"></p>"
       end
 
-      it "should revert nested <code erb-loud> including href attribute" do
-        Deface::Parser.undo_erb_markup!("<a href=\"&lt;code%20erb-loud&gt;%20x%20'y'%20+%20'z'%20&lt;/code&gt;\">A Link</a>").should == %(<a href="<%= x 'y' + 'z' %>">A Link</a>)
-        Deface::Parser.undo_erb_markup!("<a href=\"&lt;code%20erb-loud&gt;%20x%20'y'%20+%20%22z%22%20&lt;/code&gt;\">A Link</a>").should == %(<a href="<%= x 'y' + "z" %>">A Link</a>)
+      it "should revert data-erb-href attr inside html tag" do
+        Deface::Parser.undo_erb_markup!("<a data-erb-href=\"&lt;%= x 'y' + &quot;z&quot; %&gt;\">A Link</a>").should == %(<a href="<%= x 'y' + \"z\" %>">A Link</a>)
       end
 
       it "should unescape contents of code tags" do