git_statistics 0.3.0 → 0.4.0

data/.gitignore

@@ -16,3 +16,9 @@ tmp
 .yardoc
 _yardoc
 doc/
+
+# Mac artifacts
+.DS_Store
+
+# Application artifacts
+.git_statistics/

data/CHANGELOG.md

@@ -1,5 +1,6 @@
 # CHANGELOG
 
+* [0.4.0 - September 25, 2012](https://github.com/kevinjalbert/git_statistics/compare/v0.3.0...v0.4.0)
 * [0.3.0 - September 12, 2012](https://github.com/kevinjalbert/git_statistics/compare/v0.2.0...v0.3.0)
 * [0.2.0 - September 6, 2012](https://github.com/kevinjalbert/git_statistics/compare/v0.1.2...v0.2.0)
 * [0.1.2 - April 12, 2012](https://github.com/kevinjalbert/git_statistics/compare/v0.1.1...v0.1.2)

data/lib/git_statistics.rb

@@ -6,38 +6,40 @@ module GitStatistics
       @opts = Trollop::options do
         opt :email, "Use author's email instead of name", :default => false
         opt :merges, "Factor in merges when calculating statistics", :default => false
-        opt :save, "Save the commits as commits.json", :default => false
-        opt :pretty, "Save the commits as commits.json in pretty print (larger file size)", :default => false
-        opt :load, "Load commits.json instead of re-collecting data", :default => false
-        opt :update, "Update commits.json with new data (same as save and load together)", :default => false
+        opt :pretty, "Save the commits in git_repo/.git_statistics in pretty print (larger file size)", :default => false
+        opt :update, "Update saved commits with new data", :default => false
         opt :sort, "Sort authors by {commits, additions, deletions, create, delete, rename, copy, merges}", :default => "commits"
         opt :top, "Show the top N authors in results", :default => 0
         opt :branch, "Use current branch for statistics (otherwise all branches)", :default => false
         opt :verbose, "Verbose output (shows progress)", :default => false
+        opt :limit, "The maximum limit of commits to hold in memory at a time", :default => 100
       end
     end
 
     def execute
-      collector = Collector.new(@opts[:verbose])
-
-      # Collect commit data
-      if @opts[:load] || @opts[:update]
-        collector.commits.load("commits.json")
-      else
-        collector.collect(@opts[:branch])
-      end
-
-      # Collect incremental recent data
+      # Collect data (incremental or fresh) based on presence of old data
       if @opts[:update]
-        time = Utilities.get_modified_time("commit.json")
-        collector.collect(@opts[:branch], "--since=\"#{time}\"")
+        # Ensure commit directory is present
+        collector = Collector.new(@opts[:verbose], @opts[:limit], false, @opts[:pretty])
+        commits_directory = collector.repo_path + ".git_statistics" + File::Separator
+        FileUtils.mkdir_p(commits_directory)
+        file_count = Utilities.get_number_of_files(commits_directory, /\d+\.json/) - 1
+
+        # Only use --since if there is data present
+        if file_count >= 0
+          time = Utilities.get_modified_time(commits_directory + "#{file_count}.json")
+          collector.collect(@opts[:branch], "--since=\"#{time}\"")
+          collected = true
+        end
       end
 
-      # Save data
-      if @opts[:save] || @opts[:update] || @opts[:pretty]
-        collector.commits.save("commits.json", @opts[:pretty])
+      # If no data was collected as there was no present data then start fresh
+      if !collected
+        collector = Collector.new(@opts[:verbose], @opts[:limit], true, @opts[:pretty])
+        collector.collect(@opts[:branch])
       end
 
+      # Calculate statistics
       collector.commits.calculate_statistics(@opts[:email], @opts[:merges])
 
       # Print results
@@ -45,4 +47,5 @@ module GitStatistics
       puts results.print_summary(@opts[:sort], @opts[:email], @opts[:top])
     end
   end
+
 end

data/lib/git_statistics/collector.rb

@@ -1,10 +1,9 @@
 module GitStatistics
   class Collector
 
-    attr_accessor :commits, :verbose
+    attr_accessor :repo, :repo_path, :commits_path, :commits, :verbose
 
-    def initialize(verbose)
-      @commits = Commits.new
+    def initialize(verbose, limit, fresh, pretty)
       @verbose = verbose
       @repo = Utilities.get_repository
 
@@ -12,6 +11,10 @@ module GitStatistics
       if @repo == nil
         raise ("No git Repository Found")
       end
+
+      @repo_path = File.expand_path("..", @repo.path) + File::Separator
+      @commits_path = @repo_path + ".git_statistics" + File::Separator
+      @commits = Commits.new(@commits_path, fresh, limit, pretty)
     end
 
     def collect(branch, time_since="", time_until="")
@@ -42,6 +45,10 @@ module GitStatistics
 
           extract_commit(buffer) if not buffer.empty?
           buffer = []
+
+          # Save commits to file if size exceeds limit or forced
+          @commits.flush_commits
+          @repo = Utilities.get_repository
         end
 
         buffer << line
@@ -49,10 +56,10 @@ module GitStatistics
 
       # Extract the last commit
       extract_commit(buffer) if not buffer.empty?
+      @commits.flush_commits(true)
     end
 
     def fall_back_collect_commit(sha)
-
       # Create pipe for the git log to acquire commits
       pipe = open("|git --no-pager show #{sha} --date=iso --reverse"\
                   " --no-color --find-copies-harder --numstat --encoding=utf-8 "\

data/lib/git_statistics/commits.rb

@@ -1,17 +1,43 @@
 module GitStatistics
   class Commits < Hash
 
-    attr_accessor :stats, :author_list, :language_list, :totals
+    attr_accessor :stats, :totals, :path, :fresh, :limit, :pretty
+
+    def initialize(path, fresh, limit, pretty)
+      super()
+      @path = path
+      @fresh = fresh
+      @limit = limit
+      @pretty = pretty
+      clean
+    end
+
+    def clean
+      # Ensure the path exists
+      FileUtils.mkdir_p(@path)
+
+      # Remove all files within path if saving
+      if @fresh
+        Dir.entries(@path).each do |file|
+          next if file == "." || file == ".."
+          File.delete(path + File::Separator + file)
+        end
+      end
 
-    def initialize
-      super
-      @stats = Hash.new(0)
-      @author_list = []
-      @language_list = []
+      # Initilize/resets stats and totals
+      @stats = Hash.new
       @totals = Hash.new(0)
       @totals[:languages] = {}
     end
 
+    def flush_commits(force=false)
+      if self.size >= limit || force
+        file_count = Utilities.get_number_of_files(path, /\d+\.json/)
+        save(path + File::Separator + file_count.to_s + ".json", @pretty)
+        self.clear
+      end
+    end
+
     def author_top_n_type(type, top_n=0)
       top_n = 0 if top_n < 0
       return nil if @stats.size == 0
@@ -20,34 +46,42 @@ module GitStatistics
     end
 
     def calculate_statistics(email, merge)
-
-      # Check that there are statistics to calculate
-      if self.size == 0
-        puts "Cannot calculate statistics as there is no data"
-        return
-      end
-
       # Identify authors and author type
       if email
         type = :author_email
       else
         type = :author
       end
-      @author_list = Utilities.unique_data_in_hash(self, type)
 
-      # Initialize the stats hash
-      @author_list.each do |author|
-        @stats[author] = Hash.new(0)
-        @stats[author][:languages] = {}
+      # For all the commit files created
+      Dir.entries(path).each do |file|
+        # Load commit file and extract the commits
+        if file =~ /\d+\.json/
+          load(path + File::Separator + file)
+          process_commits(type, merge)
+          self.clear
+        end
       end
+    end
 
+    def process_commits(type, merge)
       # Collect the stats from each commit
       self.each do |key,value|
         if !merge && value[:merge]
           next
         else
 
-          author = (@stats[value[type]] ||= Hash.new(0))
+          # Acquire author (make if not seen before)
+          author = @stats[value[type]]
+
+          if author == nil
+            @stats[value[type]] = Hash.new(0)
+            author = @stats[value[type]]
+            author[:languages] = {}
+          end
+
+          # If there are no changed files move to next commit
+          next if value[:files].size == 0
 
           # Collect language stats
           value[:files].each do |file|
@@ -57,11 +91,6 @@ module GitStatistics
 
             # Add to repository's languages
             add_language_stats(@totals, file)
-
-            # Add language to language list if not encountered before
-            if not @language_list.include?(file[:language])
-              @language_list << file[:language]
-            end
           end
 
           # Add commit stats to author
@@ -69,6 +98,10 @@ module GitStatistics
 
           # Add commit stats to repository
           add_commit_stats(@totals, value)
+
+          # Save new changes back to stats
+          @stats[value[type]] = author
+          author = nil
         end
       end
     end
@@ -107,6 +140,13 @@ module GitStatistics
     end
 
     def save(file, pretty)
+      # Don't save if there is no information (i.e., using updates)
+      return if self.size == 0
+
+      # Ensure the path to the file exists
+      FileUtils.mkdir_p(File.dirname(file))
+
+      # Save file in a simple or pretty format
       if pretty
         File.open(file, 'w') {|file| file.write(JSON.pretty_generate(self))}
       else

data/lib/git_statistics/results.rb

@@ -11,7 +11,7 @@ module GitStatistics
       # Default to a 0 if given a negative number to display
       top_n = 0 if top_n < 0
 
-      # Acquire data based on sorty type and top # to show
+      # Acquire data based on sort type and top # to show
       data = @commits.author_top_n_type(sort.to_sym, top_n)
       if data == nil
         raise "Parameter for --sort is not valid"
@@ -19,8 +19,8 @@ module GitStatistics
 
       # Create config
       config = {:data => data,
-                :author_length => Utilities.find_longest_length(data, 17),
-                :language_length => Utilities.find_longest_length(@commits.language_list, 8),
+                :author_length => Utilities.find_longest_length(data.keys, 17),
+                :language_length => Utilities.find_longest_length(@commits.totals[:languages].keys, 8),
                 :sort => sort,
                 :email => email,
                 :top_n => top_n}
@@ -72,7 +72,7 @@ module GitStatistics
     end
 
     def print_header(config)
-      total_authors = @commits.author_list.length
+      total_authors = @commits.stats.size
 
       output = ""
       # Print summary information of displayed results

data/lib/git_statistics/utilities.rb

@@ -97,12 +97,23 @@ module GitStatistics
 
     def self.get_modified_time(file)
       if OS.mac?
-        Time.at(`stat -f %m commits.json`.to_i)
+        Time.at(`stat -f %m #{file}`.to_i)
       elsif OS.linux?
-        Time.at(`stat -c %Y commits.json`.to_i)
+        Time.at(`stat -c %Y #{file}`.to_i)
       else
         raise "Update on the Windows operating system is not supported"
       end
     end
+
+    def self.get_number_of_files(directory, pattern)
+      count = 0
+      files = Dir.entries(directory)
+
+      files.each do |file|
+        count += 1 if file =~ pattern
+      end
+
+      return count
+    end
   end
 end

data/lib/git_statistics/version.rb

@@ -1,3 +1,3 @@
 module GitStatistics
-  VERSION = "0.3.0"
+  VERSION = "0.4.0"
 end

data/spec/collector_spec.rb

@@ -2,36 +2,41 @@ require File.dirname(__FILE__) + '/spec_helper'
 include GitStatistics
 
 describe Collector do
-
-  collector = Collector.new(false)
+  let(:verbose) {false}
+  let(:limit) {100}
+  let(:fresh) {true}
+  let(:pretty) {false}
+  let(:collector) {Collector.new(verbose, limit, fresh, pretty)}
 
   describe "#collect_branches" do
+    let(:branches) {collector.collect_branches(fixture(fixture_file))}
+
     context "with many branches" do
-      branches = collector.collect_branches(fixture("git_many_branches.txt"))
+    let(:fixture_file) {"git_many_branches.txt"}
       it {branches.size.should == 2}
       it {branches[0].should == "issue_2"}
       it {branches[1].should == "master"}
     end
 
     context "with zero branches" do
-      branches = collector.collect_branches(fixture("git_zero_branches.txt"))
+      let(:fixture_file) {"git_zero_branches.txt"}
       it {branches.size.should == 1}
       it {branches[0].should == "master"}
     end
   end
 
   describe "#extract_change_file" do
+    let(:file) {collector.extract_change_file(line)}
+
     context "with a simple changed file" do
-      line =  "37	30	lib/file.rb"
-      file = collector.extract_change_file(line)
+      let(:line) {"37	30	lib/file.rb"}
       it {file[:additions].should == 37}
       it {file[:deletions].should == 30}
       it {file[:file].should == "lib/file.rb"}
     end
 
     context "with a simple rename/copy changed file" do
-      line = "11	3	old_file.rb => lib/file.rb"
-      file = collector.extract_change_file(line)
+      let(:line) {"11	3	old_file.rb => lib/file.rb"}
       it {file[:additions].should == 11}
       it {file[:deletions].should == 3}
       it {file[:file].should == "lib/file.rb"}
@@ -39,8 +44,7 @@ describe Collector do
     end
 
     context "with a complex rename/copy changed file" do
-      line = "-	-	lib/{old_dir => new_dir}/file.rb"
-      file = collector.extract_change_file(line)
+      let(:line) {"-	-	lib/{old_dir => new_dir}/file.rb"}
       it {file[:additions].should == 0}
       it {file[:deletions].should == 0}
       it {file[:file].should == "lib/new_dir/file.rb"}
@@ -49,25 +53,26 @@ describe Collector do
   end
 
   describe "#extract_create_delete_file" do
+    let(:file) {collector.extract_create_delete_file(line)}
+
     context "with a create changed file" do
-      line = "create mode 100644 lib/dir/file.rb"
-      file = collector.extract_create_delete_file(line)
+      let(:line) {"create mode 100644 lib/dir/file.rb"}
       it {file[:status].should == "create"}
       it {file[:file].should == "lib/dir/file.rb"}
     end
 
     context "with a delete changed file" do
-      line = "delete mode 100644 lib/file.rb"
-      file = collector.extract_create_delete_file(line)
+      let(:line) {"delete mode 100644 lib/file.rb"}
       it {file[:status].should == "delete"}
       it {file[:file].should == "lib/file.rb"}
     end
   end
 
   describe "#extract_rename_copy_file" do
+    let(:file) {collector.extract_rename_copy_file(line)}
+
     context "with a rename changed file" do
-      line = "rename lib/{old_dir => new_dir}/file.rb (100%)"
-      file = collector.extract_rename_copy_file(line)
+      let(:line) {"rename lib/{old_dir => new_dir}/file.rb (100%)"}
       it {file[:status].should == "rename"}
       it {file[:old_file].should == "lib/old_dir/file.rb"}
       it {file[:new_file].should == "lib/new_dir/file.rb"}
@@ -75,8 +80,7 @@ describe Collector do
     end
 
     context "with a copy changed file" do
-      line = "copy lib/dir/{old_file.rb => new_file.rb} (75%)"
-      file = collector.extract_rename_copy_file(line)
+      let(:line) {"copy lib/dir/{old_file.rb => new_file.rb} (75%)"}
       it {file[:status].should == "copy"}
       it {file[:old_file].should == "lib/dir/old_file.rb"}
       it {file[:new_file].should == "lib/dir/new_file.rb"}
@@ -85,56 +89,56 @@ describe Collector do
   end
 
   describe "#acquire_commit_data" do
+    let(:input) {fixture(fixture_file).read}
+    let(:data) {collector.acquire_commit_data(input)}
+
     context "no parent, first commit" do
-      collector.commits.clear
-      input = fixture("commit_buffer_information_first.txt").read
-      commit_data = collector.acquire_commit_data(input)
-      it {commit_data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
-      it {commit_data[:data][:author].should == "Test Author"}
-      it {commit_data[:data][:author_email].should == "author@test.com"}
-      it {commit_data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
-      it {commit_data[:data][:merge].should be_false}
+      let(:fixture_file) {"commit_buffer_information_first.txt"}
+      it {data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
+      it {data[:data][:author].should == "Test Author"}
+      it {data[:data][:author_email].should == "author@test.com"}
+      it {data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
+      it {data[:data][:merge].should be_false}
     end
 
     context "without merge, one parent" do
-      collector.commits.clear
-      input = fixture("commit_buffer_information.txt").read
-      commit_data = collector.acquire_commit_data(input)
-      it {commit_data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
-      it {commit_data[:data][:author].should == "Test Author"}
-      it {commit_data[:data][:author_email].should == "author@test.com"}
-      it {commit_data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
-      it {commit_data[:data][:merge].should be_false}
+      let(:fixture_file) {"commit_buffer_information.txt"}
+      it {data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
+      it {data[:data][:author].should == "Test Author"}
+      it {data[:data][:author_email].should == "author@test.com"}
+      it {data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
+      it {data[:data][:merge].should be_false}
     end
 
     context "with merge, two parents" do
-      collector.commits.clear
-      input = fixture("commit_buffer_information_with_merge.txt").read
-      commit_data = collector.acquire_commit_data(input)
-      it {commit_data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
-      it {commit_data[:data][:author].should == "Test Author"}
-      it {commit_data[:data][:author_email].should == "author@test.com"}
-      it {commit_data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
-      it {commit_data[:data][:merge].should be_true}
+      let(:fixture_file) {"commit_buffer_information_with_merge.txt"}
+      it {data[:sha].should == "111111aa111a11111a11aa11aaaa11a111111a11"}
+      it {data[:data][:author].should == "Test Author"}
+      it {data[:data][:author_email].should == "author@test.com"}
+      it {data[:data][:time].should == "2011-01-11 11:11:11 +0000"}
+      it {data[:data][:merge].should be_true}
     end
   end
 
   describe "#identify_changed_files" do
+    let(:files) {collector.identify_changed_files(buffer)}
+
     context "with no changes" do
-      buffer = []
-      files = collector.identify_changed_files(buffer)
+      let(:buffer) {[]}
       it {files.size.should == 0}
       it {files[0].should == nil}
     end
 
     context "with all types (create,delete,rename,copy) of files" do
       # Create buffer which is an array of cleaned lines
-      buffer = []
-      fixture("commit_buffer_changes.txt").readlines.each do |line|
-        buffer << Utilities.clean_string(line)
-      end
+      let(:buffer) {
+        buffer = []
+        fixture("commit_buffer_changes.txt").readlines.each do |line|
+          buffer << Utilities.clean_string(line)
+        end
+        buffer
+      }
 
-      files = collector.identify_changed_files(buffer)
       it {files.size.should == 5}
 
       it {files[0][:additions].should == 45}
@@ -166,13 +170,18 @@ describe Collector do
   end
 
   describe "#extract_commit" do
-    context "with valid buffer" do
-      collector.commits.clear
+    # Create buffer which is an array of cleaned lines
+    let(:buffer) {
       buffer = []
-      fixture("commit_buffer_whole.txt").readlines.each do |line|
+      fixture(fixture_file).readlines.each do |line|
         buffer << Utilities.clean_string(line)
       end
-      data = collector.extract_commit(buffer)
+      buffer
+    }
+    let(:data) {collector.extract_commit(buffer)}
+
+    context "with valid buffer" do
+      let(:fixture_file) {"commit_buffer_whole.txt"}
 
       it {data[:author].should == "Kevin Jalbert"}
       it {data[:author_email].should == "kevin.j.jalbert@gmail.com"}
@@ -215,85 +224,75 @@ describe Collector do
     end
 
     context "with buffer that has no file changes" do
-      collector.commits.clear
-      buffer = []
-      fixture("commit_buffer_information.txt").readlines.each do |line|
-        buffer << Utilities.clean_string(line)
-      end
-      data = collector.extract_commit(buffer)
-
+      let(:fixture_file) {"commit_buffer_information.txt"}
       it {data.should == nil}
     end
 
     context "with invalid buffer" do
-      collector.commits.clear
-      buffer = "invalid input"
-      data = collector.extract_commit(buffer)
-
+      let(:buffer) {"invalid input"}
       it {data.should == nil}
     end
   end
 
   describe "#fall_back_collect_commit" do
+    let(:buffer) {collector.fall_back_collect_commit(sha)}
     context "with valid sha" do
-      buffer = collector.fall_back_collect_commit("260bc61e2c42930d91f3503c5849b0a2351275cf")
-
       # Create buffer which is an array of cleaned lines
-      expected = []
-      fixture("commit_buffer_whole.txt").readlines.each do |line|
-        expected << Utilities.clean_string(line)
-      end
-
-     it {buffer.should == expected}
+      let(:expected) {
+        expected = []
+        fixture("commit_buffer_whole.txt").readlines.each do |line|
+          expected << Utilities.clean_string(line)
+        end
+        expected
+      }
+      let(:sha) {"260bc61e2c42930d91f3503c5849b0a2351275cf"}
+
+      it {buffer.should == expected}
     end
 
     context "with invalid sha" do
-      buffer = collector.fall_back_collect_commit("111111aa111a11111a11aa11aaaa11a111111a11")
+      let(:sha) {"111111aa111a11111a11aa11aaaa11a111111a11"}
       it {buffer.should == nil}
     end
   end
 
   describe "#get_blob" do
-    sha = "695b487432e8a1ede765b4e3efda088ab87a77f8"  # Commit within repository
+    let(:sha) {"695b487432e8a1ede765b4e3efda088ab87a77f8"}  # Commit within repository
+    let(:blob) {collector.get_blob(sha, file)}
 
     context "with valid blob" do
-      file = {:file => "Gemfile.lock"}
-      blob = collector.get_blob(sha, file)
-
+      let(:file) {{:file => "Gemfile.lock"}}
       it {blob.instance_of?(Grit::Blob).should be_true}
       it {blob.name.should == file[:file].split(File::Separator).last}
     end
 
     context "with invalid blob" do
-      file = {:file => "dir/nothing.rb"}
-      blob = collector.get_blob(sha, file)
-
+      let(:file) {{:file => "dir/nothing.rb"}}
       it {blob.should == nil}
     end
 
     context "with deleted file" do
-      file = {:file => "spec/collector_spec.rb"}
-      blob = collector.get_blob(sha, file)
-
+      let(:file) {{:file => "spec/collector_spec.rb"}}
       it {blob.instance_of?(Grit::Blob).should be_true}
       it {blob.name.should == file[:file].split(File::Separator).last}
     end
   end
 
   describe "#process_blob" do
-    sha = "695b487432e8a1ede765b4e3efda088ab87a77f8"  # Commit within repository
-
-    context "with status (delete) blob" do
-      file = {:file => "spec/collector_spec.rb",
-              :additions => 0,
-              :deletions => 6,
-              :status => "delete"}
-
-      blob = collector.get_blob(sha, file)
+    let(:sha) {"695b487432e8a1ede765b4e3efda088ab87a77f8"}  # Commit within repository
+    let(:blob) {collector.get_blob(sha, file)}
+    let(:data) {
       data = Hash.new(0)
       data[:files] = []
-      data = collector.process_blob(data, blob, file)
-      data_file = data[:files].first
+      collector.process_blob(data, blob, file)
+    }
+    let(:data_file) {data_file = data[:files].first}
+
+    context "with status (delete) blob" do
+      let(:file) {{:file => "spec/collector_spec.rb",
+                   :additions => 0,
+                   :deletions => 6,
+                   :status => "delete"}}
 
       it {data[:additions].should == file[:additions]}
       it {data[:deletions].should == file[:deletions]}
@@ -310,16 +309,10 @@ describe Collector do
     end
 
     context "with invalid language blob" do
-      file = {:file => "Gemfile.lock",
-              :additions => 33,
-              :deletions => 11,
-              :status => nil}
-
-      blob = collector.get_blob(sha, file)
-      data = Hash.new(0)
-      data[:files] = []
-      data = collector.process_blob(data, blob, file)
-      data_file = data[:files].first
+      let(:file) {{:file => "Gemfile.lock",
+                   :additions => 33,
+                   :deletions => 11,
+                   :status => nil}}
 
       it {data[:additions].should == file[:additions]}
       it {data[:deletions].should == file[:deletions]}
@@ -336,16 +329,10 @@ describe Collector do
     end
 
     context "with valid language blob" do
-      file = {:file => "README.md",
-              :additions => 7,
-              :deletions => 3,
-              :status => nil}
-
-      blob = collector.get_blob(sha, file)
-      data = Hash.new(0)
-      data[:files] = []
-      data = collector.process_blob(data, blob, file)
-      data_file = data[:files].first
+      let(:file) {{:file => "README.md",
+                   :additions => 7,
+                   :deletions => 3,
+                   :status => nil}}
 
       it {data[:additions].should == file[:additions]}
       it {data[:deletions].should == file[:deletions]}