From 5203e193226728c3d57525e898a47508be3d75e2 Mon Sep 17 00:00:00 2001
From: Donald Ball <donald.ball@gmail.com>
Date: Thu, 13 Jan 2011 16:29:35 -0500
Subject: [PATCH] Provide explicit constructor to bypass URI.parse

---
 README.rdoc           |  6 +++++-
 data_uri.gemspec      |  2 +-
 lib/data_uri/uri.rb   | 11 ++++++++++-
 test/test_data_uri.rb | 19 +++++++++++++++++++
 4 files changed, 35 insertions(+), 3 deletions(-)

diff --git a/README.rdoc b/README.rdoc
index 6b64473..87b7596 100644
--- a/README.rdoc
+++ b/README.rdoc
@@ -13,7 +13,7 @@ a line.
 
  require 'data_uri'
 
- uri = URI.parse('data:image/gif;base64,...')
+ uri = URI::Data.new('data:image/gif;base64,...')
  uri.content_type # image/gif
  uri.data # Base64 decoded data
 
@@ -32,6 +32,10 @@ add if anyone's interested.
 
 == Features & Limitations
 
+URI.parse knows about URI::Data, but unfortunately, its regexp for splitting
+URIs into components maxes out at 92 characters for an opaque URI, which is
+far too small to be useful for data URIs.
+
 It accepts URIs with charset MIME parameters:
 
  data:text/html;charset=utf-8,...
diff --git a/data_uri.gemspec b/data_uri.gemspec
index 45fabc7..11e6a8f 100644
--- a/data_uri.gemspec
+++ b/data_uri.gemspec
@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name        = "data_uri"
-  s.version     = "0.0.2"
+  s.version     = "0.0.3"
   s.author      = "Donald Ball"
   s.email       = "donald.ball@gmail.com"
   s.homepage    = "http://github.com/dball/data_uri"
diff --git a/lib/data_uri/uri.rb b/lib/data_uri/uri.rb
index d89eded..f513633 100644
--- a/lib/data_uri/uri.rb
+++ b/lib/data_uri/uri.rb
@@ -9,7 +9,16 @@ class Data < Generic
     attr_reader :content_type, :data
 
     def initialize(*args)
-      super(*args)
+      if args.length == 1
+        uri = args.first.to_s
+        unless uri.match(/^data:/)
+          raise 'Invalid Data URI: ' + args.first.inspect
+        end
+        @scheme = 'data'
+        @opaque = uri[5 .. -1]
+      else
+        super(*args)
+      end
       @data = @opaque
       if md = MIME_TYPE_RE.match(@data)
         @content_type = md[1]
diff --git a/test/test_data_uri.rb b/test/test_data_uri.rb
index a6009f0..66faf40 100644
--- a/test/test_data_uri.rb
+++ b/test/test_data_uri.rb
@@ -68,6 +68,25 @@
   
     end
 
+    describe "a big data binary data URI" do
+    
+      before do
+        @data = Array.new(100000) { rand(256) }.pack('c*')
+        @raw = "data:application/octet-stream;base64,#{Base64.encode64(@data).chop}"
+      end
+
+      it "shouldn't be parsed by URI.parse because the ABS_URI regexp is silly" do
+        uri = URI.parse(@raw)
+        assert uri.data != @data
+      end
+
+      it "should be parsed by URI::Data.new" do
+        uri = URI::Data.new(@raw)
+        assert uri.data == @data
+      end
+
+    end
+
   end
 
   describe "building" do