rferraz-churn 0.0.16

data/test/data/churn_calculator.rb

@@ -0,0 +1,217 @@
+require 'chronic'
+require 'sexp_processor'
+require 'ruby_parser'
+require 'json'
+require 'fileutils'
+require 'lib/churn/source_control'
+require 'lib/churn/git_analyzer'
+require 'lib/churn/svn_analyzer'
+require 'lib/churn/location_mapping'
+require 'lib/churn/churn_history'
+
+module Churn
+
+  class ChurnCalculator
+
+    def initialize(options={})
+      start_date = options.fetch(:start_date) { '3 months ago' }
+      @minimum_churn_count = options.fetch(:minimum_churn_count) { 5 }
+      puts start_date
+      if self.class.git?
+        @source_control = GitAnalyzer.new(start_date)
+      elsif File.exist?(".svn")
+        @source_control = SvnAnalyzer.new(start_date)
+      else
+        raise "Churning requires a subversion or git repo"
+      end
+      @revision_changes = {}
+      @method_changes   = {}
+      @class_changes    = {}
+    end
+
+    def report
+      self.emit 
+      self.analyze
+      self.to_h
+    end
+    
+    def emit
+      @changes   = parse_log_for_changes.reject {|file, change_count| change_count < @minimum_churn_count}
+      @revisions = parse_log_for_revision_changes  
+    end 
+
+    def analyze
+      @changes = @changes.to_a.sort {|x,y| y[1] <=> x[1]}
+      @changes = @changes.map {|file_path, times_changed| {:file_path => file_path, :times_changed => times_changed }}
+
+      calculate_revision_changes
+
+      @method_changes.to_a.sort {|x,y| y[1] <=> x[1]}
+      @method_changes          = @method_changes.map {|method, times_changed| {'method' => method, 'times_changed' => times_changed }}
+      @class_changes.to_a.sort {|x,y| y[1] <=> x[1]}
+      @class_changes          = @class_changes.map {|klass, times_changed| {'klass' => klass, 'times_changed' => times_changed }}
+    end
+
+    def to_h
+      hash                        = {:churn => {:changes => @changes}}
+      hash[:churn][:method_churn] = @method_changes
+      hash[:churn][:class_churn]  = @class_changes
+      #detail the most recent changes made this revision
+      if @revision_changes[@revisions.first]
+        changes = @revision_changes[@revisions.first]
+        hash[:churn][:changed_files]   = changes[:files]
+        hash[:churn][:changed_classes] = changes[:classes]
+        hash[:churn][:changed_methods] = changes[:methods]
+      end
+      #TODO crappy place to do this but save hash to revision file but while entirely under metric_fu only choice
+      revision = @revisions.first
+      ChurnHistory.store_revision_history(revision, hash)
+      hash
+    end
+
+    private
+
+    def self.git?
+      system("git branch")
+    end
+
+    def calculate_revision_changes
+      @revisions.each do |revision|
+        if revision == @revisions.first
+          #can't iterate through all the changes and tally them up
+          #it only has the current files not the files at the time of the revision
+          #parsing requires the files
+          changed_files, changed_classes, changed_methods = calculate_revision_data(revision)
+        else
+          changed_files, changed_classes, changed_methods = ChurnHistory.load_revision_data(revision)
+        end
+        calculate_changes!(changed_methods, @method_changes) if changed_methods
+        calculate_changes!(changed_classes, @class_changes) if changed_classes
+        
+        @revision_changes[revision] = { :files => changed_files, :classes => changed_classes, :methods => changed_methods }
+      end
+    end
+
+    def calculate_revision_data(revision)
+      changed_files   = parse_logs_for_updated_files(revision, @revisions)
+      
+      changed_classes = []
+      changed_methods = []
+      changed_files.each do |file|
+        classes, methods = get_changes(file)
+        changed_classes += classes
+        changed_methods += methods
+      end
+      changed_files   = changed_files.map { |file, lines| file }
+      [changed_files, changed_classes, changed_methods]
+    end
+
+    def calculate_changes!(changed, total_changes)
+      if changed
+        changed.each do |change|
+          total_changes.include?(change) ? total_changes[change] = total_changes[change]+1 : total_changes[change] = 1
+        end
+      end
+      total_changes
+    end
+
+    def get_changes(change)
+      begin
+        file = change.first
+        breakdown = LocationMapping.new
+        breakdown.get_info(file)
+        changes = change.last
+        classes = changes_for_type(changes, breakdown, :classes)
+        methods = changes_for_type(changes, breakdown, :methods)
+        #todo move to method
+        classes = classes.map{ |klass| {'file' => file, 'klass' => klass} }
+        methods = methods.map{ |method| {'file' => file, 'klass' => get_klass_for(method), 'method' => method} }
+        [classes, methods]
+      rescue => error
+        [[],[]]
+      end
+    end
+
+    def get_klass_for(method)
+      method.gsub(/(#|\.).*/,'')
+    end
+
+    def changes_for_type(changes, breakdown, type)
+      item_collection = if type == :classes
+                          breakdown.klasses_collection
+                        elsif type == :methods
+                          breakdown.methods_collection
+                        end
+      changed_items  = []
+      item_collection.each_pair do |item, item_lines|
+        item_lines = item_lines[0].to_a
+        changes.each do |change_range|
+          item_lines.each do |line|
+            changed_items << item if change_range.include?(line) && !changed_items.include?(item)
+          end
+        end
+      end
+      changed_items
+    end
+    
+    def parse_log_for_changes
+      changes = {}
+      
+      logs = @source_control.get_logs
+      logs.each do |line|
+        changes[line] ? changes[line] += 1 : changes[line] = 1
+      end
+      changes
+    end
+
+    def parse_log_for_revision_changes
+      @source_control.get_revisions
+    end
+    
+    def parse_logs_for_updated_files(revision, revisions)
+      updated     = {}
+      recent_file = nil
+
+      #SVN doesn't support this
+      return updated unless @source_control.respond_to?(:get_updated_files_from_log)
+      logs = @source_control.get_updated_files_from_log(revision, revisions)
+      logs.each do |line|
+        if line.match(/^---/) || line.match(/^\+\+\+/)
+          line = line.gsub(/^--- /,'').gsub(/^\+\+\+ /,'').gsub(/^a\//,'').gsub(/^b\//,'')
+          unless updated.include?(line)
+            updated[line] = [] 
+          end
+          recent_file = line
+        elsif line.match(/^@@/)
+          #TODO cleanup / refactor
+          #puts "#{recent_file}: #{line}"
+          removed        = line.match(/-[0-9]+/)
+          removed_length = line.match(/-[0-9]+,[0-9]+/)
+          removed        = removed.to_s.gsub(/-/,'')
+          removed_length = removed_length.to_s.gsub(/.*,/,'')
+          added          = line.match(/\+[0-9]+/)
+          added_length   = line.match(/\+[0-9]+,[0-9]+/)
+          added          = added.to_s.gsub(/\+/,'')
+          added_length   = added_length.to_s.gsub(/.*,/,'')
+          removed_range  = if removed_length && removed_length!=''
+                             (removed.to_i..(removed.to_i+removed_length.to_i))
+                           else
+                             (removed.to_i..removed.to_i)
+                           end
+          added_range    = if added_length && added_length!=''
+                             (added.to_i..(added.to_i+added_length.to_i))
+                           else
+                             (added.to_i..added.to_i)
+                           end
+          updated[recent_file] << removed_range
+          updated[recent_file] << added_range
+        else
+          raise "git diff lines that don't match the two patterns aren't expected"
+        end
+      end
+      updated
+    end
+
+  end
+
+end

data/test/data/test_helper.rb

@@ -0,0 +1,14 @@
+require 'rubygems'
+require 'test/unit'
+require 'shoulda'
+require 'construct'
+require 'mocha'
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+require 'churn/churn_calculator'
+Mocha::Configuration.prevent(:stubbing_non_existent_method)
+
+class Test::Unit::TestCase
+  include Construct::Helpers
+end

data/test/test_helper.rb

@@ -0,0 +1,14 @@
+require 'rubygems'
+require 'test/unit'
+require 'shoulda'
+require 'construct'
+require 'mocha'
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+require 'churn/churn_calculator'
+Mocha::Configuration.prevent(:stubbing_non_existent_method)
+
+class Test::Unit::TestCase
+  include Construct::Helpers
+end

data/test/unit/bzr_analyzer_test.rb

@@ -0,0 +1,65 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class BzrAnalyzerTest < Test::Unit::TestCase
+  context "BzrAnalyzer#get_logs" do
+    should "return a list of changed files" do
+      bzr_analyzer = Churn::BzrAnalyzer.new
+      bzr_analyzer.expects(:`).with('bzr log -v --short ').returns(" 1947 Adam Walters    2010-01-16\n   Second commit with 3 files now.\n      M  file1.rb\n      M  file2.rb\n      M  file3.rb\n\n 1946 Adam Walters    2010-01-16\n      First commit\n      A  file1.rb\n")
+      assert_equal ["file1.rb", "file2.rb", "file3.rb", "file1.rb"], bzr_analyzer.get_logs
+    end
+
+    should "scope the changed files to an optional date range" do
+      bzr_analyzer = Churn::BzrAnalyzer.new("1/16/2010")
+      bzr_analyzer.expects(:`).with('bzr log -v --short -r 2010-01-16..').returns(" 1947 Adam Walters 2010-01-16\n      Second commit with 3 files now.\n      M  file1.rb\n      M  file2.rb\n      M  file3.rb\n\n 1946 Adam Walters    2010-01-16\n      First commit\n      A  file1.rb\n")
+      assert_equal ["file1.rb", "file2.rb", "file3.rb", "file1.rb"], bzr_analyzer.get_logs
+    end
+  end
+
+  context "BzrAnalyzer#get_revisions" do
+    should "return a list of changeset ids" do
+      bzr_analyzer = Churn::BzrAnalyzer.new
+      bzr_analyzer.expects(:`).with('bzr log --line ').returns("1947: Adam Walters 2010-01-16 Second commit with 3 files now.\n1946: Adam Walters 2010-01-16 First commit\n")
+      assert_equal ["1947", "1946"], bzr_analyzer.get_revisions
+    end
+
+    should "scope the changesets to an optional date range" do
+      bzr_analyzer = Churn::BzrAnalyzer.new("1/16/2010")
+      bzr_analyzer.expects(:`).with('bzr log --line -r 2010-01-16..').returns("1947: Adam Walters 2010-01-16 Second commit with 3 files now.\n1946: Adam Walters 2010-01-16 First commit\n")
+      assert_equal ["1947", "1946"], bzr_analyzer.get_revisions
+    end
+  end
+
+  context "BzrAnalyzer#get_updated_files_from_log(revision, revisions)" do
+    should "return a list of modified files and the change hunks (chunks)" do
+      bzr_analyzer = Churn::BzrAnalyzer.new
+      bzr_analyzer.expects(:`).with('bzr diff -r 1946..1947').returns("=== modified file 'a/file1.rb'\n--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600\n@@ -1,3 +0,0 @@\n-First\n-Adding sample data\n-Third line\ndiff -r 1947 -r 1946 file2.rb\n=== modified file 'a/file2.rb'\n--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000\n@@ -1,7 +0,0 @@\n-This is the second file.\n-\n-Little more data\n-\n-def cool_method\n-  \"hello\"\n-end\ndiff -r 1947 -r 1946 file3.rb\n--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000\n@@ -1,5 +0,0 @@\n-Third file here.\n-\n-def another_method\n-  \"foo\"\n-end\n")
+      assert_equal ["--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600", "@@ -1,3 +0,0 @@", "--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,7 +0,0 @@", "--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,5 +0,0 @@"], bzr_analyzer.get_updated_files_from_log("1947", ["1947", "1946"])
+    end
+
+    should "return an empty array if it's the final revision" do
+      bzr_analyzer = Churn::BzrAnalyzer.new
+      assert_equal [], bzr_analyzer.get_updated_files_from_log("1946", ["1947", "1946"])
+    end
+  end
+
+  context "BzrAnalyzer#get_updated_files_change_info(revision, revisions)" do
+    setup do
+      @bzr_analyzer = Churn::BzrAnalyzer.new
+    end
+
+    should "return all modified files with their line differences" do
+      @bzr_analyzer.expects(:get_updated_files_from_log).with("1947", ["1947", "1946"]).returns(["--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600", "@@ -1,3 +0,0 @@", "--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,7 +0,0 @@", "--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,5 +0,0 @@"])
+      assert_equal({"/dev/null" => [1..8, 0..0, 1..6, 0..0], "file3.rb" => [], "file1.rb" => [], "file2.rb" => [], "file1.rb" => [1..4, 0..0]}, @bzr_analyzer.get_updated_files_change_info("1947", ["1947", "1946"]))
+    end
+
+    should "raise an error if it encounters a line it cannot parse" do
+      @bzr_analyzer.expects(:get_updated_files_from_log).with("1947", ["1947", "1946"]).returns(["foo"])
+      assert_raise RuntimeError do
+        @bzr_analyzer.stubs(:puts) # supress output from raised error
+        @bzr_analyzer.get_updated_files_change_info("1947", ["1947", "1946"])
+      end
+    end
+  end
+
+end
+

data/test/unit/churn_calculator_test.rb

@@ -0,0 +1,90 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class ChurnCalculatorTest < Test::Unit::TestCase
+ 
+  should "use minimum churn count" do
+    within_construct do |container|
+      Churn::ChurnCalculator.stubs(:git?).returns(true)
+      churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+ 
+      churn.stubs(:parse_log_for_changes).returns([['file.rb', 4],['less.rb',1]])
+      churn.stubs(:parse_log_for_revision_changes).returns(['revision'])
+      churn.stubs(:analyze)
+      report = churn.report(false)
+      assert_equal 1, report[:churn][:changes].length
+      assert_equal ["file.rb", 4], report[:churn][:changes].first
+    end
+  end
+
+  should "analize sorts changes" do
+    within_construct do |container|
+      Churn::ChurnCalculator.stubs(:git?).returns(true)
+      churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+ 
+      churn.stubs(:parse_log_for_changes).returns([['file.rb', 4],['most.rb', 9],['less.rb',1]])
+      churn.stubs(:parse_log_for_revision_changes).returns(['revision'])
+      report = churn.report(false)
+      assert_equal 2, report[:churn][:changes].length
+      top = {:file_path => "most.rb", :times_changed => 9}
+      assert_equal top, report[:churn][:changes].first
+      bottom = {:file_path => "file.rb", :times_changed => 4}
+      assert_equal bottom, report[:churn][:changes].last
+    end
+  end
+
+  should "have correct changed_files data" do
+    within_construct do |container|
+      Churn::ChurnCalculator.stubs(:git?).returns(true)
+      churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+      
+      churn.stubs(:parse_log_for_changes).returns([['less.rb',1]])
+      churn.stubs(:parse_log_for_revision_changes).returns(['first'])
+      churn.stubs(:parse_logs_for_updated_files).returns({'fake_file.rb'=>[]})
+      report = churn.report(false)
+      assert_equal ["fake_file.rb"], report[:churn][:changed_files]
+    end
+  end
+
+  should "have correct changed classes and methods data" do
+    within_construct do |container|
+      Churn::ChurnCalculator.stubs(:git?).returns(true)
+      churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+      
+      churn.stubs(:parse_log_for_changes).returns([['less.rb',1]])
+      churn.stubs(:parse_log_for_revision_changes).returns(['first'])
+      churn.stubs(:parse_logs_for_updated_files).returns({'fake_file.rb'=>[]})
+      klasses = [{"klass"=>"LocationMapping", "file"=>"lib/churn/location_mapping.rb"}]
+      methods = [{"klass"=>"LocationMapping", "method"=>"LocationMapping#process_class", "file"=>"lib/churn/location_mapping.rb"}]
+      churn.stubs(:get_changes).returns([klasses,methods])
+      report = churn.report(false)
+      assert_equal [{"klass"=>"LocationMapping", "method"=>"LocationMapping#process_class", "file"=>"lib/churn/location_mapping.rb"}], report[:churn][:changed_methods]
+      assert_equal [{"klass"=>"LocationMapping", "file"=>"lib/churn/location_mapping.rb"}], report[:churn][:changed_classes]
+    end
+  end
+
+  should "have correct churn method and classes at 1 change" do
+    within_construct do |container|
+      Churn::ChurnCalculator.stubs(:git?).returns(true)
+      churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+      
+      churn.stubs(:parse_log_for_changes).returns([['less.rb',1]])
+      churn.stubs(:parse_log_for_revision_changes).returns(['first'])
+      churn.stubs(:parse_logs_for_updated_files).returns({'fake_file.rb'=>[]})
+      klasses = [{"klass"=>"LocationMapping", "file"=>"lib/churn/location_mapping.rb"}]
+      methods = [{"klass"=>"LocationMapping", "method"=>"LocationMapping#process_class", "file"=>"lib/churn/location_mapping.rb"}]
+      churn.stubs(:get_changes).returns([klasses,methods])
+      report = churn.report(false)
+      assert_equal [{"method"=>{"klass"=>"LocationMapping", "method"=>"LocationMapping#process_class", "file"=>"lib/churn/location_mapping.rb"}, "times_changed"=>1}], report[:churn][:method_churn]
+      assert_equal [{"klass"=>{"klass"=>"LocationMapping", "file"=>"lib/churn/location_mapping.rb"}, "times_changed"=>1}], report[:churn][:class_churn]
+    end
+  end
+
+  should "initialize a churn calculator for hg repositories" do
+    Churn::ChurnCalculator.stubs(:git?).returns(false)
+    Churn::ChurnCalculator.stubs(:system).with('hg branch').returns(true)
+    churn = Churn::ChurnCalculator.new({:minimum_churn_count => 3})
+    assert churn.instance_variable_get(:@source_control).is_a?(Churn::HgAnalyzer)
+  end
+
+  
+end

data/test/unit/churn_history_test.rb

@@ -0,0 +1,24 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class ChurnHistoryTest < Test::Unit::TestCase
+ 
+  should "store results" do
+    within_construct do |container|
+      Churn::ChurnHistory.store_revision_history('aaa','data')
+      assert File.exists?('tmp/aaa.json')
+      data = File.read('tmp/aaa.json')
+      assert data.match(/data/)
+    end
+  end
+ 
+  should "restores results" do
+    within_construct do |container|
+      container.file('tmp/aaa.json', '{"churn":{"changes":[{"file_path":".gitignore","times_changed":2},{"file_path":"lib\/churn.rb","times_changed":2},{"file_path":"Rakefile","times_changed":2},{"file_path":"README.rdoc","times_changed":2},{"file_path":"lib\/churn\/source_control.rb","times_changed":1},{"file_path":"lib\/churn\/svn_analyzer.rb","times_changed":1},{"file_path":"lib\/tasks\/churn_tasks.rb","times_changed":1},{"file_path":"LICENSE","times_changed":1},{"file_path":"test\/churn_test.rb","times_changed":1},{"file_path":"lib\/churn\/locationmapping.rb","times_changed":1},{"file_path":"lib\/churn\/git_analyzer.rb","times_changed":1},{"file_path":".document","times_changed":1},{"file_path":"test\/test_helper.rb","times_changed":1},{"file_path":"lib\/churn\/churn_calculator.rb","times_changed":1}],"method_churn":[],"changed_files":[".gitignore","lib\/churn\/source_control.rb","lib\/tasks\/churn_tasks.rb","lib\/churn\/svn_analyzer.rb","Rakefile","README.rdoc","lib\/churn\/locationmapping.rb","lib\/churn\/git_analyzer.rb","\/dev\/null","lib\/churn\/churn_calculator.rb","lib\/churn.rb"],"class_churn":[],"changed_classes":[{"klass":"ChurnTest","file":"test\/churn_test.rb"},{"klass":"ChurnCalculator","file":"lib\/churn\/churn_calculator.rb"}],"changed_methods":[{"klass":"","method":"#report_churn","file":"lib\/tasks\/churn_tasks.rb"}]}}')
+      changed_files, changed_classes, changed_methods = Churn::ChurnHistory.load_revision_data('aaa')
+      assert changed_files.include?("lib/churn/source_control.rb")
+      assert_equal 2, changed_classes.length
+      assert_equal 1, changed_methods.length
+    end
+  end
+  
+end

data/test/unit/git_analyzer_test.rb

@@ -0,0 +1,17 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class GitAnalyzerTest < Test::Unit::TestCase
+  
+  should "parses logs correctly" do
+    git_analyzer = Churn::GitAnalyzer.new
+    revision     = 'first'
+    revisions    = ['first']
+    lines        = ["--- a/lib/churn/churn_calculator.rb", "+++ b/lib/churn/churn_calculator.rb", "@@ -18,0 +19 @@ module Churn"]
+    git_analyzer.stubs(:get_updated_files_from_log).returns(lines)
+    updated = git_analyzer.get_updated_files_change_info(revision, revisions)
+    expected_hash = {"lib/churn/churn_calculator.rb"=>[18..18, 19..19]}
+    assert_equal = updated
+  end
+  
+end
+

data/test/unit/hg_analyzer_test.rb

@@ -0,0 +1,66 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class HgAnalyzerTest < Test::Unit::TestCase
+
+  context "HgAnalyzer#get_logs" do
+    should "return a list of changed files" do
+      hg_analyzer = Churn::HgAnalyzer.new
+      hg_analyzer.expects(:`).with('hg log -v').returns("changeset:   1:4760c1d7cd40\ntag:         tip\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:21:28 2010 -0600\nfiles:       file1.rb file2.rb file3.rb\ndescription:\nSecond commit with 3 files now.\nLong commit\n\n\nchangeset:   0:3cb77114f02a\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:19:32 2010 -0600\nfiles:       file1.rb\ndescription:\nFirst commit\n\n\n")
+      assert_equal ["file1.rb", "file2.rb", "file3.rb", "file1.rb"], hg_analyzer.get_logs
+    end
+
+    should "scope the changed files to an optional date range" do
+      hg_analyzer = Churn::HgAnalyzer.new("1/16/2010")
+      hg_analyzer.expects(:`).with('hg log -v -d "> 2010-01-16"').returns("changeset:   1:4760c1d7cd40\ntag:         tip\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:21:28 2010 -0600\nfiles:       file1.rb file2.rb file3.rb\ndescription:\nSecond commit with 3 files now.\nLong commit\n\n\nchangeset:   0:3cb77114f02a\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:19:32 2010 -0600\nfiles:       file1.rb\ndescription:\nFirst commit\n\n\n")
+      assert_equal ["file1.rb", "file2.rb", "file3.rb", "file1.rb"], hg_analyzer.get_logs
+    end
+  end
+
+  context "HgAnalyzer#get_revisions" do
+    should "return a list of changeset ids" do
+      hg_analyzer = Churn::HgAnalyzer.new
+      hg_analyzer.expects(:`).with('hg log').returns("changeset:   1:4760c1d7cd40\ntag:         tip\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:21:28 2010 -0600\nsummary:     Second commit with 3 files now.\n\nchangeset:   0:3cb77114f02a\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:19:32 2010 -0600\nsummary:     First commit\n\n")
+      assert_equal ["4760c1d7cd40", "3cb77114f02a"], hg_analyzer.get_revisions
+    end
+
+    should "scope the changesets to an optional date range" do
+      hg_analyzer = Churn::HgAnalyzer.new("1/16/2010")
+      hg_analyzer.expects(:`).with('hg log -d "> 2010-01-16"').returns("changeset:   1:4760c1d7cd40\ntag:         tip\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:21:28 2010 -0600\nsummary:     Second commit with 3 files now.\n\nchangeset:   0:3cb77114f02a\nuser:        Adam Walters <awalters@obtiva.com>\ndate:        Sat Jan 16 14:19:32 2010 -0600\nsummary:     First commit\n\n")
+      assert_equal ["4760c1d7cd40", "3cb77114f02a"], hg_analyzer.get_revisions
+    end
+  end
+
+  context "HgAnalyzer#get_updated_files_from_log(revision, revisions)" do
+    should "return a list of modified files and the change hunks (chunks)" do
+      hg_analyzer = Churn::HgAnalyzer.new
+      hg_analyzer.expects(:`).with('hg diff -r 4760c1d7cd40:3cb77114f02a -U 0').returns("diff -r 4760c1d7cd40 -r 3cb77114f02a file1.rb\n--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600\n@@ -1,3 +0,0 @@\n-First\n-Adding sample data\n-Third line\ndiff -r 4760c1d7cd40 -r 3cb77114f02a file2.rb\n--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000\n@@ -1,7 +0,0 @@\n-This is the second file.\n-\n-Little more data\n-\n-def cool_method\n-  \"hello\"\n-end\ndiff -r 4760c1d7cd40 -r 3cb77114f02a file3.rb\n--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600\n+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000\n@@ -1,5 +0,0 @@\n-Third file here.\n-\n-def another_method\n-  \"foo\"\n-end\n")
+      assert_equal ["--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600", "@@ -1,3 +0,0 @@", "--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,7 +0,0 @@", "--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,5 +0,0 @@"], hg_analyzer.get_updated_files_from_log("4760c1d7cd40", ["4760c1d7cd40", "3cb77114f02a"])
+    end
+
+    should "return an empty array if it's the final revision" do
+      hg_analyzer = Churn::HgAnalyzer.new
+      assert_equal [], hg_analyzer.get_updated_files_from_log("3cb77114f02a", ["4760c1d7cd40", "3cb77114f02a"])
+    end
+  end
+
+  context "HgAnalyzer#get_updated_files_change_info(revision, revisions)" do
+    setup do
+      @hg_analyzer = Churn::HgAnalyzer.new
+    end
+
+    should "return all modified files with their line differences" do
+      @hg_analyzer.expects(:get_updated_files_from_log).with("4760c1d7cd40", ["4760c1d7cd40", "3cb77114f02a"]).returns(["--- a/file1.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ b/file1.rb\tSat Jan 16 14:19:32 2010 -0600", "@@ -1,3 +0,0 @@", "--- a/file2.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,7 +0,0 @@", "--- a/file3.rb\tSat Jan 16 14:21:28 2010 -0600", "+++ /dev/null\tThu Jan 01 00:00:00 1970 +0000", "@@ -1,5 +0,0 @@"])
+      assert_equal({"/dev/null" => [1..8, 0..0, 1..6, 0..0], "file3.rb" => [], "file1.rb" => [], "file2.rb" => [], "file1.rb" => [1..4, 0..0]}, @hg_analyzer.get_updated_files_change_info("4760c1d7cd40", ["4760c1d7cd40", "3cb77114f02a"]))
+    end
+
+    should "raise an error if it encounters a line it cannot parse" do
+      @hg_analyzer.expects(:get_updated_files_from_log).with("4760c1d7cd40", ["4760c1d7cd40", "3cb77114f02a"]).returns(["foo"])
+      assert_raise RuntimeError do
+        @hg_analyzer.stubs(:puts) # supress output from raised error
+        @hg_analyzer.get_updated_files_change_info("4760c1d7cd40", ["4760c1d7cd40", "3cb77114f02a"])
+      end
+    end
+  end
+
+end
+

data/test/unit/location_mapping_test.rb

@@ -0,0 +1,35 @@
+require File.expand_path('../test_helper', File.dirname(__FILE__))
+
+class LocationMappingTest < Test::Unit::TestCase
+  
+  #todo unfortunately it looks like ruby parser can't handle construct tmp dirs
+  #<Pathname:/private/var/folders/gl/glhHkYYSGgG5nb6+4OG0yU+++TI/-Tmp-/construct_container-56784-851001101/fake_class.rb>
+  #(rdb:1) p locationmapping.get_info(file.to_s)
+  #RegexpError Exception: invalid regular expression; there's no previous pattern, to which '+' would define cardinality at 2: /^+++/
+
+  should "location_mapping gets correct classes info" do
+    file = 'test/data/churn_calculator.rb'
+    locationmapping = Churn::LocationMapping.new
+    locationmapping.get_info(file.to_s)
+    klass_hash = {"ChurnCalculator"=>[14..215]}
+    assert_equal klass_hash, locationmapping.klasses_collection
+  end
+
+  should "location_mapping gets correct methods info" do
+    file = 'test/data/churn_calculator.rb'
+    locationmapping = Churn::LocationMapping.new
+    locationmapping.get_info(file.to_s)
+    methods_hash = {"ChurnCalculator#report"=>[32..36], "ChurnCalculator#emit"=>[38..41], "ChurnCalculator#changes_for_type"=>[139..155], "ChurnCalculator#get_klass_for"=>[135..137], "ChurnCalculator#calculate_changes!"=>[109..116], "ChurnCalculator#analyze"=>[43..53], "ChurnCalculator#calculate_revision_data"=>[95..107], "ChurnCalculator#calculate_revision_changes"=>[78..93], "ChurnCalculator#parse_logs_for_updated_files"=>[171..213], "ChurnCalculator#to_h"=>[55..70], "ChurnCalculator#parse_log_for_revision_changes"=>[167..169], "ChurnCalculator#get_changes"=>[118..133], "ChurnCalculator#parse_log_for_changes"=>[157..165], "ChurnCalculator#initialize"=>[16..30]}
+    assert_equal methods_hash, locationmapping.methods_collection
+  end
+  
+  should "location_mapping gets correct classes info for test helper files" do
+    file = 'test/data/test_helper.rb'
+    locationmapping = Churn::LocationMapping.new
+    locationmapping.get_info(file.to_s)
+    klass_hash = {"TestCase"=>[12..14]}
+    assert_equal klass_hash, locationmapping.klasses_collection
+  end
+  
+end
+