filey-diff 1.0.2 → 1.1.0

data/Gemfile

@@ -1,2 +1,2 @@
-source :rubygems
+source 'https://rubygems.org'
 gemspec

data/changelog.md

@@ -1,5 +1,13 @@
 # Changelog
 
+This project is [Semantically Versioned](http://semver.org).
+
+## 1.1.0
+
+* Add support for gzipped S3 objects
+
+  Thanks to Alex Marchant for implementing this!
+
 ## 1.0.2
 
 * Remove the -number suffix from S3 ETags ([thanks

data/filey-diff.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name              = 'filey-diff'
-  s.version           = '1.0.2'
+  s.version           = '1.1.0'
 
   s.summary     = "Compare two data sources that contain file-like objects"
   s.description =

data/lib/filey-diff/data-sources/aws_sdk_s3.rb

@@ -16,13 +16,34 @@ module Filey
             path = ''
             name = s3_object.key
           end
+
+          if (s3_object.head[:content_encoding] == "gzip")
+            last_modified, md5 =
+              last_modified_and_md5_from_gzipped(s3_object, path)
+          else
+            last_modified = s3_object.last_modified
+            md5 = s3_object.etag.gsub(/"/, '').split('-',2).first
+          end
+
           normalised_path = "./#{path}"
           Filey.new(normalised_path,
                     name,
-                    s3_object.last_modified,
-                    s3_object.etag.gsub(/"/, '').split('-',2).first)
+                    last_modified,
+                    md5)
         }
       end
+
+      def last_modified_and_md5_from_gzipped(s3_object, path)
+        tempfile = Tempfile.new(File.basename(path))
+        tempfile.write s3_object.read
+        tempfile.close
+
+        gz = Zlib::GzipReader.open(tempfile.path)
+        last_modified = gz.mtime
+        md5 = Digest::MD5.hexdigest(gz.read)
+        gz.close
+        [last_modified, md5]
+      end
     end
   end
 end

data/spec/comparison_spec.rb

@@ -73,5 +73,4 @@ describe Filey::Comparison do
       end
     end
   end
-
 end

data/spec/data_sources_spec.rb

@@ -1,4 +1,6 @@
 require File.dirname(__FILE__) + '/spec_helper'
+require 'tempfile'
+require 'zlib'
 
 shared_examples "a data source" do |source|
   let(:data_source) { described_class.new(source) }
@@ -57,6 +59,42 @@ describe Filey::DataSources::AwsSdkS3 do
     }
   )
   it_should_behave_like "a data source", s3_bucket
+
+  it 'provides the original md5/mtime of a gzipped file' do
+    original_object = objects.first
+    tempfile = gzip_into_tmp_file(original_object)
+    data_source = s3_data_source_from_file(tempfile, original_object[:path])
+    filey = data_source.get_fileys[0]
+
+    filey.md5.should eq(Digest::MD5.hexdigest(original_object[:content]))
+    # GzipWriter seems to cut off fractions of a second,
+    # to_i adjusts the original file to match
+    filey.last_modified.to_i.should eq(original_object[:mtime].to_i)
+  end
+
+  def s3_data_source_from_file(file, path)
+    file.open
+    data_source = Filey::DataSources::AwsSdkS3.new(S3Bucket.new([S3Object.new(
+      path,
+      file.mtime,
+      file.read,
+      { :content_encoding => 'gzip' }
+    )]))
+  end
+
+  def gzip_into_tmp_file(s3_mock_object)
+    tempfile = Tempfile.new("temp")
+
+    gz = Zlib::GzipWriter.open(tempfile.path, Zlib::BEST_COMPRESSION, Zlib::DEFAULT_STRATEGY)
+    gz.mtime = s3_mock_object[:mtime]
+    gz.write s3_mock_object[:content]
+
+    gz.flush
+    tempfile.flush
+
+    gz.close
+    tempfile
+  end
 end
 
 describe Filey::DataSources::FileSystem do

data/spec/spec_helper.rb

@@ -4,17 +4,22 @@ require 'digest/md5'
 require File.dirname(__FILE__) + '/../lib/filey-diff'
 
 class S3Object
-  attr_reader :key, :last_modified
+  attr_reader :key, :last_modified, :head
 
-  def initialize(key, last_modified, content)
+  def initialize(key, last_modified, content, head = {})
     @key = key
     @last_modified = last_modified
     @content = content
+    @head = head
   end
 
   def etag
     Digest::MD5.hexdigest(@content)
   end
+
+  def read
+    @content
+  end
 end
 
 class S3Bucket

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: filey-diff
 version: !ruby/object:Gem::Version
-  version: 1.0.2
+  version: 1.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-01-13 00:00:00.000000000 Z
+date: 2013-03-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake