Improve Content-Type parsing

Rakefile

@@ -22,7 +22,7 @@ end
 
 require 'yardstick/rake/verify'
 Yardstick::Rake::Verify.new do |verify|
-  verify.threshold = 56.3
+  verify.threshold = 56.8
 end
 
 task :default => [:spec, :rubocop, :verify_measurements]

lib/http/content_type.rb

@@ -0,0 +1,27 @@
+module HTTP
+  ContentType = Struct.new(:mime_type, :charset) do
+    MIME_TYPE_RE = %r{^([^/]+/[^;]+)(?:$|;)}
+    CHARSET_RE   = /;\s*charset=([^;]+)/i
+
+    class << self
+      # Parse string and return ContentType struct
+      def parse(str)
+        new mime_type(str), charset(str)
+      end
+
+    private
+
+      # :nodoc:
+      def mime_type(str)
+        md = str.to_s.match MIME_TYPE_RE
+        md && md[1].to_s.strip.downcase
+      end
+
+      # :nodoc:
+      def charset(str)
+        md = str.to_s.match CHARSET_RE
+        md && md[1].to_s.strip.gsub(/^"|"$/, '')
+      end
+    end
+  end
+end

lib/http/response.rb

@@ -1,5 +1,6 @@
 require 'delegate'
 require 'http/header'
+require 'http/content_type'
 
 module HTTP
   class Response
@@ -120,10 +121,22 @@ def to_s
     end
     alias_method :to_str, :to_s
 
+    # Parsed Content-Type header
+    # @return [HTTP::ContentType]
+    def content_type
+      @content_type ||= ContentType.parse @headers['Content-Type']
+    end
+
     # MIME type of response (if any)
     # @return [String, nil]
     def mime_type
-      @mime_type ||= @headers['Content-Type'].split(/;\s*/).first
+      @mime_type ||= content_type.mime_type
+    end
+
+    # Charset of response (if any)
+    # @return [String, nil]
+    def charset
+      @mime_type ||= content_type.charset
     end
 
     # Inspect a response

spec/http/content_type_spec.rb

@@ -0,0 +1,47 @@
+require 'spec_helper'
+
+describe HTTP::ContentType do
+  describe '.parse' do
+    context 'with text/plain' do
+      subject { described_class.parse 'text/plain' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should be_nil }
+    end
+
+    context 'with tEXT/plaIN' do
+      subject { described_class.parse 'tEXT/plaIN' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should be_nil }
+    end
+
+    context 'with text/plain; charset=utf-8' do
+      subject { described_class.parse 'text/plain; charset=utf-8' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should eq 'utf-8' }
+    end
+
+    context 'with text/plain; charset="utf-8"' do
+      subject { described_class.parse 'text/plain; charset="utf-8"' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should eq 'utf-8' }
+    end
+
+    context 'with text/plain; charSET=utf-8' do
+      subject { described_class.parse 'text/plain; charSET=utf-8' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should eq 'utf-8' }
+    end
+
+    context 'with text/plain; foo=bar; charset=utf-8' do
+      subject { described_class.parse 'text/plain; foo=bar; charset=utf-8' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should eq 'utf-8' }
+    end
+
+    context 'with text/plain;charset=utf-8;foo=bar' do
+      subject { described_class.parse 'text/plain;charset=utf-8;foo=bar' }
+      its(:mime_type) { should eq 'text/plain' }
+      its(:charset)   { should eq 'utf-8' }
+    end
+  end
+end

spec/http/response_spec.rb

@@ -25,10 +25,40 @@
   end
 
   describe 'mime_type' do
-    it 'strips out charset' do
-      headers  = {'Content-Type' => 'text/html; charset=utf8'}
-      response = HTTP::Response.new 200, '1.1', headers, ''
-      expect(response.mime_type).to eq 'text/html'
+    subject { HTTP::Response.new(200, '1.1', headers, '').mime_type }
+
+    context 'without Content-Type header' do
+      let(:headers) { {} }
+      it { should be_nil }
+    end
+
+    context 'with Content-Type: text/html' do
+      let(:headers) { {'Content-Type' => 'text/html'} }
+      it { should eq 'text/html' }
+    end
+
+    context 'with Content-Type: text/html; charset=utf-8' do
+      let(:headers) { {'Content-Type' => 'text/html; charset=utf-8'} }
+      it { should eq 'text/html' }
+    end
+  end
+
+  describe 'charset' do
+    subject { HTTP::Response.new(200, '1.1', headers, '').charset }
+
+    context 'without Content-Type header' do
+      let(:headers) { {} }
+      it { should be_nil }
+    end
+
+    context 'with Content-Type: text/html' do
+      let(:headers) { {'Content-Type' => 'text/html'} }
+      it { should be_nil }
+    end
+
+    context 'with Content-Type: text/html; charset=utf-8' do
+      let(:headers) { {'Content-Type' => 'text/html; charset=utf-8'} }
+      it { should eq 'utf-8' }
     end
   end
 end