git_statistics 0.6.0 → 0.7.0

data/lib/git_statistics/commits.rb

@@ -33,7 +33,7 @@ module GitStatistics
       Dir.entries(path).reject { |file| %w[. ..].include?(file) }
     end
 
-    def flush_commits(force = false)
+    def flush_commits(force=false)
       if size >= limit || force
         file_count = Utilities.number_of_matching_files(path, /\d+\.json/)
         save(File.join(path, "#{file_count}.json"), @pretty)
@@ -54,13 +54,18 @@ module GitStatistics
       # Identify authors and author type
       type = email ? :author_email : :author
 
-      # For all the commit files created
-      Dir.entries(path).each do |file|
-        # Load commit file and extract the commits
-        if file =~ /\d+\.json/
-          load(File.join(path, file))
+      # Process the commits from file or memory
+      files = Dir.entries(path) - [".", ".."]
+      if files.size == 0
           process_commits(type, merge)
-          clear
+      else
+        #Load commit file and extract the commits
+        files.each do |file|
+          if file =~ /\d+\.json/
+            load(File.join(path, file))
+            process_commits(type, merge)
+            clear
+          end
         end
       end
     end
@@ -112,8 +117,9 @@ module GitStatistics
       data[:languages][file[:language].to_sym][:additions] += file[:additions]
       data[:languages][file[:language].to_sym][:deletions] += file[:deletions]
 
+      # Keep count of languages status (i.e., added, deleted) and keep keys consistent (i.e., added_files, deleted_files)
       if file[:status] != nil
-        data[:languages][file[:language].to_sym][file[:status].to_sym] += 1
+        data[:languages][file[:language].to_sym][(file[:status]+'_files').to_sym] += 1
       end
 
       return data
@@ -125,10 +131,8 @@ module GitStatistics
       data[:commits] += 1
       data[:additions] += commit[:additions]
       data[:deletions] += commit[:deletions]
-      data[:create] += commit[:create] if commit[:create] != nil
-      data[:delete] += commit[:delete] if commit[:delete] != nil
-      data[:rename] += commit[:rename] if commit[:rename] != nil
-      data[:copy] += commit[:copy] if commit[:copy] != nil
+      data[:added_files] += commit[:added_files] if !commit[:added_files].nil?
+      data[:deleted_files] += commit[:deleted_files] if !commit[:deleted_files].nil?
       return data
     end
 
@@ -142,9 +146,9 @@ module GitStatistics
         # Ensure the path to the file exists
         FileUtils.mkdir_p(File.dirname(file))
         # Save file in a simple or pretty format
-        File.open(file, 'w') do |file|
+        File.open(file, 'w') do |f|
           json_content = pretty ? JSON.pretty_generate(self) : self.to_json
-          file.write(json_content)
+          f.write(json_content)
         end
       end
     end

data/lib/git_statistics/diff_summary.rb

@@ -0,0 +1,79 @@
+module GitStatistics
+  class DiffSummary < SimpleDelegator
+    def initialize(repo, patch)
+      @repo = repo
+      super(patch)
+    end
+
+    # We flip these around since we are diffing in the opposite direction -- new.diff(old)
+    def additions
+      __getobj__.deletions
+    end
+
+    # We flip these around since we are diffing in the opposite direction -- new.diff(old)
+    def deletions
+      __getobj__.additions
+    end
+
+    def net
+      additions - deletions
+    end
+
+    # We flip these around since we are diffing in the opposite direction -- new.diff(old)
+    def status
+      if delta.status == :deleted
+        return :added
+      elsif delta.status == :added
+        return :deleted
+      else
+        return delta.status
+      end
+    end
+
+    def similarity
+      delta.similarity
+    end
+
+    def filename
+      if (status == :deleted)
+        delta.old_file[:path]
+      else
+        delta.new_file[:path]
+      end
+    end
+
+    # We flip these around since we are diffing in the opposite direction -- new.diff(old)
+    def blob
+      begin
+        if (status == :deleted)
+          blob = @repo.lookup(delta.new_file[:oid])  # Look at new instead of old
+        else
+          blob = @repo.lookup(delta.old_file[:oid])  # Look at old instead of new
+        end
+      rescue Rugged::OdbError
+        Log.warn "Could not find object (most likely a submodule)"
+        blob = nil
+      end
+    end
+
+    def inspect
+      %Q{<GitStatistics::FileStat @filename=#{filename} @status=#{status} @similarity=#{similarity} @language=#{language} @additions=#{additions}, @deletions=#{deletions}, @net=#{net}>}
+    end
+
+    def to_json
+      { filename: filename, status: status, similarity: similarity, language: language, additions: additions, deletions: deletions, net: net}
+    end
+
+    # Determine the language of the file from the blob
+    def language
+      language = "Unknown"
+      unless blob.nil?
+        detected_language = LanguageSniffer.detect(filename, :content => blob.content).language
+        unless detected_language.nil?
+          language = detected_language.name
+        end
+      end
+      language
+    end
+  end
+end

data/lib/git_statistics/formatters/console.rb

@@ -2,10 +2,13 @@ module GitStatistics
   module Formatters
     class Console
 
+      attr_reader :output
       attr_accessor :commits, :config
 
       def initialize(commits)
         @commits = commits
+        @config = {}
+        @output = []
       end
 
       def prepare_result_summary(sort, email, top_n = 0)
@@ -26,7 +29,7 @@ module GitStatistics
 
         # Acquire formatting pattern for output
         @pattern = "| %-#{config[:author_length]}s | %-#{config[:language_length]}s | %7s | %9s | %9s | %7s | %7s | %7s | %6s | %6s |"
-        config
+        @config
       end
 
       def print_summary(sort, email, top_n = 0)
@@ -35,47 +38,45 @@ module GitStatistics
         config = prepare_result_summary(sort, email, top_n)
 
         # Print query/header information
-        output = print_header
-
+        print_header
         # Print per author information
         config[:data].each do |name, commit_data|
-          output << print_row(name, commit_data)
-          output << print_language_data(commit_data)
+          print_row(name, commit_data)
+          print_language_data(commit_data)
         end
-        output << separator
-        output << print_row("Repository Totals", commit_totals)
-        output << print_language_data(commit_totals)
-        output.flatten.join("\n")
+        add_row separator
+        print_row("Repository Totals", commit_totals)
+        print_language_data(commit_totals)
+        display!
+      end
+
+      def display!
+        output.join("\n")
       end
 
       def print_language_data(data)
-        output = []
         # Print information of each language for the data
         data[:languages].each do |language, commit_data|
-          output << print_row("", commit_data, language)
+          print_row("", commit_data, language)
         end
         output
       end
 
       def print_row(name, commit_info, language = '')
-        format_for_row(name, language, commit_info[:commits],
-                        commit_info[:additions], commit_info[:deletions], commit_info[:create],
-                        commit_info[:delete], commit_info[:rename], commit_info[:copy], commit_info[:merges])
+        add_row format_for_row(name, language, commit_info[:commits],
+                        commit_info[:additions], commit_info[:deletions], commit_info[:added_files],
+                        commit_info[:deleted_files], commit_info[:renamed_files], commit_info[:copied_files], commit_info[:merges])
       end
 
       def print_header
-        output = []
-        output << get_author_info(@commits.stats.size)
-        output << get_header_info
-        output
+        get_author_info(@commits.stats.size)
+        get_header_info
       end
 
       def get_header_info
-        headers = []
-        headers << separator
-        headers << format_for_row('Name/Email', 'Language', 'Commits', 'Additions', 'Deletions', 'Creates', 'Deletes', 'Renames', 'Copies', 'Merges')
-        headers << separator
-        headers
+        add_row separator
+        add_row format_for_row('Name/Email', 'Language', 'Commits', 'Additions', 'Deletions', 'Creates', 'Deletes', 'Renames', 'Copies', 'Merges')
+        add_row separator
       end
 
       def format_for_row(*columns)
@@ -83,15 +84,19 @@ module GitStatistics
       end
 
       def separator
-        "-" * 89 + "-"*config[:author_length] + "-"*config[:language_length]
+        ("-" * 89) + ("-" * config[:author_length]) + ("-" * config[:language_length])
       end
 
       def get_author_info(total_authors)
         if config[:top_n] > 0 && config[:top_n] < total_authors
-          return "Top #{config[:top_n]} authors(#{total_authors}) sorted by #{config[:sort]}"
+          add_row "Top #{config[:top_n]} authors(#{total_authors}) sorted by #{config[:sort]}"
+        else
+          add_row "All authors(#{total_authors}) sorted by #{config[:sort]}"
         end
+      end
 
-        "All authors(#{total_authors}) sorted by #{config[:sort]}"
+      def add_row(string_or_array)
+        output.concat Array(string_or_array)
       end
     end
   end

data/lib/git_statistics/initialize.rb

@@ -1,10 +1,11 @@
 require 'json'
-require 'grit'
-require 'linguist'
 require 'pathname'
+require 'ostruct'
+require 'optparse'
+require 'delegate'
+require 'fileutils'
 
-# Must be required before all other files
-require 'git_statistics/blob'
-require 'git_statistics/regex_matcher'
+require 'rugged'
+require 'language_sniffer'
 
-Dir.glob(File.dirname(__FILE__) + '/**/*.rb') {|file| require file}
+Dir.glob(File.dirname(__FILE__) + '/**/*.rb') { |file| require file }

data/lib/git_statistics/pipe.rb

@@ -3,7 +3,7 @@ module GitStatistics
     include Enumerable
 
     def initialize(command)
-      @command  = command
+      @command = command
     end
 
     def command

data/lib/git_statistics/utilities.rb

@@ -3,103 +3,35 @@ require 'rbconfig'
 module GitStatistics
   module Utilities
 
-    class NotInRepository < StandardError; end
-
-    def self.get_repository(path = Dir.pwd)
-      ascender = Pathname.new(path).to_enum(:ascend)
-      repo_path = ascender.detect { |path| (path + '.git').exist? }
-      raise NotInRepository unless repo_path
-      Grit::Repo.new(repo_path.to_s)
-    rescue NotInRepository
-      Log.error "You must be within a Git project to run git-statistics."
-      exit 0
-    end
-
-    def self.max_length_in_list(list, max = nil)
-      return nil if list.nil?
-      list.each do |key,value|
-        max = key.length if max.nil? || key.length > max
-      end
-      max
+    def self.max_length_in_list(list, min_length = nil)
+      list ||= []
+      min_length = min_length.to_i
+      list_max = list.map { |k,_| k.length }.max || 0
+      list_max >= min_length ? list_max : min_length
     end
 
-    def self.split_old_new_file(old, new)
-      # Split the old and new chunks up (separted by the =>)
-      split_old = old.split('{')
-      split_new = new.split('}')
-
-      # Handle recombine the file splits into their whole paths)
-      if split_old.one? && split_new.one?
-        old_file = split_old[0]
-        new_file = split_new[0]
-      elsif split_new.one?
-        old_file = split_old[0] + split_old[1]
-        new_file = split_old[0] + split_new[0]
-      elsif split_old.one?
-        old_file = split_old[0] + split_new[1]
-        new_file = split_old[0] + split_new[0] + split_new[1]
-      else
-        old_file = split_old[0] + split_old[1] + split_new[1]
-        new_file = split_old[0] + split_new[0] + split_new[1]
-      end
-
-      # Return files, yet remove the '//' if present from combining splits
-      return {:old_file => old_file.gsub('//', '/'),
-              :new_file => new_file.gsub('//', '/')}
-    end
-
-    def self.find_blob_in_tree(tree, file)
-      # Check If cannot find tree in commit or if we found a submodule as the changed file
-      if tree.nil? || file.nil?
-        return nil
-      elsif tree.instance_of?(Grit::Submodule)
-        return tree
-      end
-
-      # If the blob is within the current directory (tree)
-      if file.one?
-        blob = tree / file.first
-
-        # Check if blob is nil (could not find changed file in tree)
-        if blob.nil?
-
-          # Try looking for submodules as they cannot be found using tree / file notation
-          tree.contents.each do |content|
-            if file.first == content.name
-              return tree
-            end
-          end
-
-          # Exit through recusion with the base case of a nil tree/blob
-          return find_blob_in_tree(blob, file)
-        end
-        return blob
-      else
-        # Explore deeper in the tree to find the blob of the changed file
-        return find_blob_in_tree(tree / file.first, file[1..-1])
-      end
-    end
+    COMMANDS = {
+      :windows => ->{ raise "`stat` is not supported on Windows" },
+      :mac =>     ->{ "-f %m" }
+    }
+    COMMANDS.default = ->{ "-c %Y" }
 
     def self.get_modified_time(file)
-      if os == :windows
-        raise "`stat` is not supported on the Windows operating system"
-      end
-      flags = os == :mac ? "-f %m" : "-c %Y"
+      flags = COMMANDS[os].()
       time_at("stat #{flags} #{file}")
     end
 
-    def os
-      case RbConfig::CONFIG['host_os']
-      when /mswin|msys|mingw|cygwin|bccwin|wince|emc/
-        :windows
-      when /darwin|mac os/
-        :mac
-      when /linux/
-        :linux
-      when /solaris|bsd/
-        :unix
-      else
-        :unknown
+    class OperatingSystem
+      OPERATING_SYSTEMS = {
+        /mswin|msys|mingw|cygwin|bccwin|wince|emc/ => :windows,
+        /darwin|mac os/ => :mac,
+        /linux/ => :linux,
+        /solaris|bsd/ => :unix
+      }
+      OPERATING_SYSTEMS.default = :unknown
+
+      def self.determine(os_name)
+        OPERATING_SYSTEMS.select { |k,_| k =~ os_name }.first
       end
     end
 
@@ -107,11 +39,16 @@ module GitStatistics
       Time.at(%x{#{cmd}}.to_i)
     end
 
+    def self.os
+      OperatingSystem.determine(RbConfig::CONFIG['host_os'])
+    end
+
     def self.number_of_matching_files(directory, pattern)
       Dir.entries(directory).grep(pattern).size
     rescue SystemCallError
-      warn "No such directory #{File.expand_path(directory)}"
+      Log.error "No such directory #{File.expand_path(directory)}"
       0
     end
+
   end
 end

data/lib/git_statistics/version.rb

@@ -1,3 +1,3 @@
 module GitStatistics
-  VERSION = "0.6.0"
+  VERSION = "0.7.0"
 end

data/spec/collector_spec.rb

@@ -5,33 +5,29 @@ describe Collector do
   let(:limit) {100}
   let(:fresh) {true}
   let(:pretty) {false}
-  let(:collector) {Collector.new(limit, fresh, pretty)}
-
-  # Create buffer which is an array of cleaned lines
-  let(:buffer) {
-    fixture(fixture_file).lines
-  }
+  let(:repo) { GIT_REPO }
+  let(:collector) {Collector.new(repo, limit, fresh, pretty)}
 
   describe "#collect" do
-    let(:branch) {""}
+    let(:branch) {CLI::DEFAULT_BRANCH}
     let(:email) {false}
     let(:merge) {true}
-    let(:time_since) {"--since \"Tue Sep 25 14:15:44 2012 -0400\""}
-    let(:time_until) {"--until \"Tue Sep 25 14:45:05 2012 -0400\""}
+    let(:time_since) {"Tue Sep 24 14:15:44 2012 -0400"}
+    let(:time_until) {"Tue Sep 26 14:45:05 2012 -0400"}
     let(:author) {"Kevin Jalbert"}
 
     let(:setup) {
-      collector.collect(branch, time_since, time_until)
+      collector.collect({:branch => branch, :time_since => time_since, :time_until => time_until})
       collector.commits.calculate_statistics(email, merge)
       @subject = collector.commits.stats[author]
     }
 
     context "with no merge commits" do
       let(:merge) {false}
-      let(:time_since) {"--since \"Tue Sep 10 14:15:44 2012 -0400\""}
-      let(:time_until) {"--until \"Tue Sep 11 14:45:05 2012 -0400\""}
+      let(:time_since) {"Tue Sep 10 14:15:44 2012 -0400"}
+      let(:time_until) {"Tue Sep 11 14:45:05 2012 -0400"}
 
-      before(:all) {setup}
+      before(:all) { setup }
 
       it{@subject[:additions].should == 276}
       it{@subject[:deletions].should == 99}
@@ -40,279 +36,89 @@ describe Collector do
 
       it{@subject[:languages][:Ruby][:additions].should == 270}
       it{@subject[:languages][:Ruby][:deletions].should == 99}
-      it{@subject[:languages][:Ruby][:create].should == 2}
-      it{@subject[:languages][:Unknown][:additions].should == 6}
-      it{@subject[:languages][:Unknown][:deletions].should == 0}
-      it{@subject[:languages][:Unknown][:create].should == 1}
+      it{@subject[:languages][:Ruby][:added_files].should == 2}
+      it{@subject[:languages][:Text][:additions].should == 6}
+      it{@subject[:languages][:Text][:deletions].should == 0}
+      it{@subject[:languages][:Text][:added_files].should == 1}
     end
 
     context "with merge commits and merge option" do
-      before(:all) {setup}
+      before(:all) { setup }
 
-      it{@subject[:additions].should == 667}
-      it{@subject[:deletions].should == 483}
-      it{@subject[:commits].should == 3}
+      it{@subject[:additions].should == 1240}
+      it{@subject[:deletions].should == 934}
+      it{@subject[:commits].should == 9}
       it{@subject[:merges].should == 1}
 
       it{@subject[:languages][:Markdown][:additions].should == 1}
       it{@subject[:languages][:Markdown][:deletions].should == 0}
-      it{@subject[:languages][:Ruby][:additions].should == 654}
-      it{@subject[:languages][:Ruby][:deletions].should == 483}
+      it{@subject[:languages][:Ruby][:additions].should == 1227}
+      it{@subject[:languages][:Ruby][:deletions].should == 934}
       it{@subject[:languages][:Unknown][:additions].should == 12}
       it{@subject[:languages][:Unknown][:deletions].should == 0}
     end
 
     context "with merge commits and no merge option" do
       let(:merge) {false}
-      before(:all) {setup}
+      before(:all) { setup }
 
-      it{@subject[:additions].should == 8}
-      it{@subject[:deletions].should == 1}
-      it{@subject[:commits].should == 2}
+      it{@subject[:additions].should == 581}
+      it{@subject[:deletions].should == 452}
+      it{@subject[:commits].should == 8}
       it{@subject[:merges].should == 0}
 
       it{@subject[:languages][:Markdown][:additions].should == 1}
       it{@subject[:languages][:Markdown][:deletions].should == 0}
-      it{@subject[:languages][:Ruby][:additions].should == 1}
-      it{@subject[:languages][:Ruby][:deletions].should == 1}
+      it{@subject[:languages][:Ruby][:additions].should == 574}
+      it{@subject[:languages][:Ruby][:deletions].should == 452}
       it{@subject[:languages][:Unknown][:additions].should == 6}
       it{@subject[:languages][:Unknown][:deletions].should == 0}
     end
   end
 
-  describe "#acquire_commit_data" do
-    let(:data) {collector.acquire_commit_data(buffer.first)}
-
-    context "no parent, first commit" do
-      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
-      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
-      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)}
-    let(:fixture_file) {"commit_buffer_changes.txt"}
-
-    context "with no changes" do
-      let(:buffer) {[]}
-      it {files.size.should == 0}
-      it {files[0].should.nil?}
-    end
-
-    context "with all types (create,delete,rename,copy) of files" do
-      it {files.size.should == 5}
-
-      it {files[0][:additions].should == 45}
-      it {files[0][:deletions].should == 1}
-      it {files[0][:file].should == "dir/README"}
-
-      it {files[1][:additions].should == 6}
-      it {files[1][:deletions].should == 5}
-      it {files[1][:old_file].should == "dir/lib/old_dir/copy_file.rb"}
-      it {files[1][:file].should == "dir/lib/new_dir/copy_file.rb"}
-      it {files[1][:status].should == "copy"}
-
-      it {files[2][:additions].should == 0}
-      it {files[2][:deletions].should == 0}
-      it {files[2][:old_file].should == "dir/lib/rename/old_file.rb"}
-      it {files[2][:file].should == "dir/lib/rename/new_file.rb"}
-      it {files[2][:status].should == "rename"}
-
-      it {files[3][:additions].should == 0}
-      it {files[3][:deletions].should == 127}
-      it {files[3][:file].should == "dir/lib/delete_file.rb"}
-      it {files[3][:status].should == "delete"}
-
-      it {files[4][:additions].should == 60}
-      it {files[4][:deletions].should == 0}
-      it {files[4][:file].should == "dir/lib/create_file.rb"}
-      it {files[4][:status].should == "create"}
-    end
-  end
-
   describe "#extract_commit" do
-    let(:data) {collector.extract_commit(buffer)}
-
-    context "with valid buffer" do
-      let(:fixture_file) {"commit_buffer_whole.txt"}
+    let(:commit) {repo.lookup(sha)}
+    let(:data) {collector.extract_commit(commit, nil)}
 
+    context "with valid commit" do
+      let(:sha) {"260bc61e2c42930d91f3503c5849b0a2351275cf"}
       it {data[:author].should == "Kevin Jalbert"}
       it {data[:author_email].should == "kevin.j.jalbert@gmail.com"}
-      it {data[:time].should == "2012-04-12 14:13:56 -0400"}
+      it {data[:time].should match /\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2} [-|+]\d{4}/}
 
       it {data[:merge].should == false}
       it {data[:additions].should == 30}
       it {data[:deletions].should == 2}
-      it {data[:create].should == 1}
+      it {data[:added_files].should == 1}
+      it {data[:deleted_files].should == 0}
 
-      it {data[:files][0][:name].should == "Gemfile"}
+      it {data[:files][0][:filename].should == "Gemfile"}
       it {data[:files][0][:additions].should == 0}
       it {data[:files][0][:deletions].should == 1}
       it {data[:files][0][:status].should.nil?}
-      it {data[:files][0][:binary].should == false}
-      it {data[:files][0][:image].should == false}
-      it {data[:files][0][:vendored].should == false}
-      it {data[:files][0][:generated].should == false}
       it {data[:files][0][:language].should == "Ruby"}
 
-      it {data[:files][1][:name].should == "Gemfile.lock"}
+      it {data[:files][1][:filename].should == "Gemfile.lock"}
       it {data[:files][1][:additions].should == 30}
       it {data[:files][1][:deletions].should == 0}
-      it {data[:files][1][:status].should == "create"}
-      it {data[:files][1][:binary].should == false}
-      it {data[:files][1][:image].should == false}
-      it {data[:files][1][:vendored].should == false}
-      it {data[:files][1][:generated].should == true}
+      it {data[:files][1][:status].should eq(:added)}
       it {data[:files][1][:language].should == "Unknown"}
 
-      it {data[:files][2][:name].should == "lib/git_statistics/initialize.rb"}
+      it {data[:files][2][:filename].should == "lib/git_statistics/initialize.rb"}
       it {data[:files][2][:additions].should == 0}
       it {data[:files][2][:deletions].should == 1}
       it {data[:files][2][:status].should.nil?}
-      it {data[:files][2][:binary].should == false}
-      it {data[:files][2][:image].should == false}
-      it {data[:files][2][:vendored].should == false}
-      it {data[:files][2][:generated].should == false}
       it {data[:files][2][:language].should == "Ruby"}
     end
 
-    context "with buffer that has no file changes" do
-      let(:fixture_file) {"commit_buffer_information.txt"}
-      it {data.should.nil?}
-    end
-
-    context "with invalid buffer" do
-      let(:buffer) {"invalid input"}
-      it {data.should.nil?}
-    end
-  end
-
-  describe "#fall_back_collect_commit" do
-    subject { collector.fall_back_collect_commit(sha) }
-    context "with valid sha" do
-      let(:fixture_file) { "commit_buffer_whole.txt" }
-      let(:sha) { "260bc61e2c42930d91f3503c5849b0a2351275cf" }
-      it { should == buffer }
+    context "with invalid commit" do
+      let(:sha) {"111111aa111a11111a11aa11aaaa11a111111a11"}
+      it { expect {data}.to raise_error(Rugged::OdbError) }
     end
 
     context "with invalid sha" do
-      let(:sha) { "111111aa111a11111a11aa11aaaa11a111111a11" }
-      it { should be_empty }
-    end
-  end
-
-  describe "#get_blob" do
-    let(:sha) { "695b487432e8a1ede765b4e3efda088ab87a77f8" }  # Commit within repository
-    subject { collector.get_blob(sha, file) }
-
-    context "with valid blob" do
-      let(:file) {{:file => "Gemfile.lock"}}
-      it { should be_a Grit::Blob }
-      its(:name) { should == File.basename(file[:file]) }
-    end
-
-    context "with deleted file" do
-      let(:file) {{:file => "spec/collector_spec.rb"}}
-      it { should be_a Grit::Blob }
-      its(:name) { should == File.basename(file[:file]) }
-    end
-
-    context "with invalid blob" do
-      let(:file) {{:file => "dir/nothing.rb"}}
-      it { should be_nil }
-    end
-  end
-
-  describe "#process_blob" do
-    let(:sha) {"695b487432e8a1ede765b4e3efda088ab87a77f8"}  # Commit within repository
-    let(:blob) {collector.get_blob(sha, file)}
-    let(:data) {
-      data = Hash.new(0)
-      data[:files] = []
-      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]}
-
-      it {data_file[:name].should == file[:file]}
-      it {data_file[:additions].should == file[:additions]}
-      it {data_file[:deletions].should == file[:deletions]}
-      it {data_file[:status].should == file[:status]}
-      it {data_file[:binary].should == false}
-      it {data_file[:image].should == false}
-      it {data_file[:vendored].should == false}
-      it {data_file[:generated].should == false}
-      it {data_file[:language].should == "Ruby"}
-    end
-
-    context "with invalid language blob" do
-      let(:file) {{:file => "Gemfile.lock",
-                   :additions => 33,
-                   :deletions => 11,
-                   :status => nil}}
-
-      it {data[:additions].should == file[:additions]}
-      it {data[:deletions].should == file[:deletions]}
-
-      it {data_file[:name].should == file[:file]}
-      it {data_file[:additions].should == file[:additions]}
-      it {data_file[:deletions].should == file[:deletions]}
-      it {data_file[:status].should.nil?}
-      it {data_file[:binary].should == false}
-      it {data_file[:image].should == false}
-      it {data_file[:vendored].should == false}
-      it {data_file[:generated].should == true}
-      it {data_file[:language].should == "Unknown"}
-    end
-
-    context "with valid language blob" do
-      let(:file) {{:file => "README.md",
-                   :additions => 7,
-                   :deletions => 3,
-                   :status => nil}}
-
-      it {data[:additions].should == file[:additions]}
-      it {data[:deletions].should == file[:deletions]}
-
-      it {data_file[:name].should == file[:file]}
-      it {data_file[:additions].should == file[:additions]}
-      it {data_file[:deletions].should == file[:deletions]}
-      it {data_file[:status].should.nil?}
-      it {data_file[:binary].should == false}
-      it {data_file[:image].should == false}
-      it {data_file[:vendored].should == false}
-      it {data_file[:generated].should == false}
-      it {data_file[:language].should == "Markdown"}
+      let(:sha) {"invalid input"}
+      it { expect {data}.to raise_error(Rugged::InvalidError) }
     end
   end