churn 0.0.9 → 0.0.10

data/README.rdoc

@@ -4,6 +4,13 @@ A Project to give the churn file, class, and method for a project for a given ch
 Over time the tool adds up the history of chruns to give the number of times a file, class, or method is changing during the life of a project.
 Churn for files is immediate, but classes and methods requires buildings up a history using churn between revisions. The history is stored in ./tmp
 
+Currently has Full Git and Mercurial (hg) support, and partial SVN support (supports only file level churn currnetly)
+
+Authors:
+* danmayer
+* ajwalters 
+* cldwalker
+
 == Example Output
   **********************************************************************
   * Revision Changes 
@@ -71,11 +78,15 @@ Churn for files is immediate, but classes and methods requires buildings up a hi
   | lib/churn/churn_calculator.rb | ChurnCalculator | ChurnCalculator#to_s                    | 1             |
   +-------------------------------+-----------------+-----------------------------------------+---------------+
 
+
 TODO:
-* SVN only supports file
+* SVN only supports file, add full SVN support 
+* support bazaar, cvs, and darcs 
 * make storage directory configurable instead of using tmp
 * allow passing in directories to churn, directories to ignore
-* todo add a filter that allows for other files besides. *.rb
+* add a filter that allows for other files besides. *.rb
+* ignore files pattern, so you can ignore things like vendor/, lib/, or docs/
+* finish adding better documenation using YARD
 
 Executable Usage:
 * 'gem install churn'
@@ -100,4 +111,4 @@ Rake Usage:
 
 == Copyright
 
-Copyright (c) 2009 Dan Mayer. See LICENSE for details.
+Copyright (c) 2010 Dan Mayer. See LICENSE for details.

data/VERSION

@@ -1 +1 @@
-0.0.9
+0.0.10

data/lib/churn/churn_calculator.rb

@@ -9,6 +9,7 @@ $LOAD_PATH.unshift(File.dirname(__FILE__))
 require 'source_control'
 require 'git_analyzer'
 require 'svn_analyzer'
+require 'hg_analyzer'
 require 'location_mapping'
 require 'churn_history'
 
@@ -16,6 +17,7 @@ module Churn
 
   class ChurnCalculator
 
+    # intialized the churn calculator object
     def initialize(options={})
       start_date = options.fetch(:start_date) { '3 months ago' }
       @minimum_churn_count = options.fetch(:minimum_churn_count) { 5 }
@@ -26,17 +28,23 @@ module Churn
       @method_changes   = {}
     end
 
+    # prepares the data for the given project to be reported.
+    # reads git/svn logs analyzes the output, generates a report and either formats as a nice string or returns hash.
+    # @param [Bolean] format to return the data, true for string or false for hash
+    # @return [Object] returns either a pretty string or a hash representing the chrun of the project
     def report(print = true)
       self.emit 
       self.analyze
       print ? self.to_s : self.to_h
     end
     
+    # Emits various data from source control to be analyses later... Currently this is broken up like this as a throwback to metric_fu
     def emit
       @changes   = parse_log_for_changes.reject {|file, change_count| change_count < @minimum_churn_count}
       @revisions = parse_log_for_revision_changes  
     end 
 
+    # Analyze the source control data, filter, sort, and find more information on the editted files
     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 }}
@@ -49,6 +57,7 @@ module Churn
       @class_changes  = @class_changes.map {|klass, times_changed| {'klass' => klass, 'times_changed' => times_changed }}
     end
 
+    # collect all the data into a single hash data structure.
     def to_h
       hash                        = {:churn => {:changes => @changes}}
       hash[:churn][:class_churn]  = @class_changes
@@ -65,6 +74,7 @@ module Churn
       hash
     end
 
+    # Pretty print the data as a string for the user
     def to_s
       hash   = to_h
       result = seperator 
@@ -107,9 +117,15 @@ module Churn
       system("git branch")
     end
 
+    def self.hg?
+      system("hg branch")
+    end
+
     def set_source_control(start_date)
       if self.class.git?
         GitAnalyzer.new(start_date)
+      elsif self.class.hg?
+        HgAnalyzer.new(start_date)
       elsif File.exist?(".svn")
         SvnAnalyzer.new(start_date)
       else

data/lib/churn/hg_analyzer.rb

@@ -0,0 +1,71 @@
+module Churn
+  class HgAnalyzer < SourceControl
+    def get_logs
+      `hg log -v#{date_range}`.split("\n").reject{|line| line !~ /^files:/}.map{|l| l.split(" ")[1..-1]}.flatten
+    end
+
+    def get_revisions
+      `hg log#{date_range}`.split("\n").reject{|line| line !~ /^changeset:/}.map{|l| l[/:(\S+)$/, 1] }
+    end
+
+    def get_updated_files_from_log(revision, revisions)
+      current_index = revisions.index(revision)
+      previous_index = current_index+1
+      previous_revision = revisions[previous_index] unless revisions.length < previous_index
+      if revision && previous_revision
+        `hg diff -r #{revision}:#{previous_revision} -U 0`.split(/\n/).select{|line| line.match(/^@@/) || line.match(/^---/) || line.match(/^\+\+\+/) }
+      else
+        []
+      end
+    end
+
+    def get_updated_files_change_info(revision, revisions)
+      updated     = {}
+      logs        = get_updated_files_from_log(revision, revisions)
+      recent_file = nil
+      logs.each do |line|
+        if line.match(/^---/) || line.match(/^\+\+\+/)
+          # Remove the --- a/ and +++ b/ if present
+          recent_file = get_recent_file(line)
+          updated[recent_file] = [] unless updated.include?(recent_file)
+        elsif line.match(/^@@/)
+          # Now add the added/removed ranges for the line
+          removed_range = get_changed_range(line, '-')
+          added_range   = get_changed_range(line, '\+')
+          updated[recent_file] << removed_range
+          updated[recent_file] << added_range
+        else
+          puts line.match(/^---/)
+          raise "hg diff lines that don't match the two patterns aren't expected: '#{line}'"
+        end
+      end
+      updated
+    end
+
+  private
+    def date_range
+      if @start_date
+        date = Chronic.parse(@start_date)
+        " -d \"> #{date.strftime('%Y-%m-%d')}\""
+      end
+    end
+
+    def get_recent_file(line)
+      line = line.gsub(/^--- /,'').gsub(/^\+\+\+ /,'').gsub(/^a\//,'').gsub(/^b\//,'').split("\t")[0]
+    end
+
+    def get_changed_range(line, matcher)
+      change_start = line.match(/#{matcher}[0-9]+/)
+      change_end   = line.match(/#{matcher}[0-9]+,[0-9]+/)
+      change_start = change_start.to_s.gsub(/#{matcher}/,'')
+      change_end   = change_end.to_s.gsub(/.*,/,'')
+
+      range  = if change_end && change_end!=''
+                 (change_start.to_i..(change_start.to_i+change_end.to_i))
+               else
+                 (change_start.to_i..change_start.to_i)
+               end
+      range
+    end
+  end
+end

data/test/unit/churn_calculator_test.rb

@@ -78,5 +78,13 @@ class ChurnCalculatorTest < Test::Unit::TestCase
       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/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
+

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: churn
 version: !ruby/object:Gem::Version 
-  version: 0.0.9
+  version: 0.0.10
 platform: ruby
 authors: 
 - Dan Mayer
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-01-19 00:00:00 -05:00
+date: 2010-01-20 00:00:00 -05:00
 default_executable: churn
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -123,6 +123,7 @@ files:
 - lib/churn/churn_calculator.rb
 - lib/churn/churn_history.rb
 - lib/churn/git_analyzer.rb
+- lib/churn/hg_analyzer.rb
 - lib/churn/location_mapping.rb
 - lib/churn/source_control.rb
 - lib/churn/svn_analyzer.rb
@@ -133,6 +134,7 @@ files:
 - test/unit/churn_calculator_test.rb
 - test/unit/churn_history_test.rb
 - test/unit/git_analyzer_test.rb
+- test/unit/hg_analyzer_test.rb
 - test/unit/location_mapping_test.rb
 has_rdoc: true
 homepage: http://github.com/danmayer/churn
@@ -169,4 +171,5 @@ test_files:
 - test/unit/churn_calculator_test.rb
 - test/unit/churn_history_test.rb
 - test/unit/git_analyzer_test.rb
+- test/unit/hg_analyzer_test.rb
 - test/unit/location_mapping_test.rb