require 'chronic' require 'sexp_processor' require 'ruby_parser' require 'json' require 'hirb' require 'fileutils' $LOAD_PATH.unshift(File.dirname(__FILE__)) require 'source_control' require 'git_analyzer' require 'svn_analyzer' require 'hg_analyzer' require 'bzr_analyzer' require 'location_mapping' require 'churn_history' module Churn # The work horse of the the churn library. This class takes user input, determins the SCM the user is using. It then determines changes # made during this revision. Finally it reads all the changes from previous revisions and displays human readable output to the command # line. It can also ouput a yaml format readable by other tools such as metric_fu and Caliper. 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 }.to_i @ignore_files = (options.fetch(:ignore_files){ "" }).to_s.split(',').map(&:strip) @ignore_files << '/dev/null' @source_control = set_source_control(start_date) @changes = {} @revision_changes = {} @class_changes = {} @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 || @ignore_files.include?(file) } @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 = sort_changes(@changes) @changes = @changes.map {|file_path, times_changed| {:file_path => file_path, :times_changed => times_changed }} calculate_revision_changes @method_changes = sort_changes(@method_changes) @method_changes = @method_changes.map {|method, times_changed| {'method' => method, 'times_changed' => times_changed }} @class_changes = sort_changes(@class_changes) @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 hash[:churn][:method_churn] = @method_changes #detail the most recent changes made this revision first_revision = @revisions.first first_revision_changes = @revision_changes[first_revision] if first_revision_changes changes = first_revision_changes 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 ChurnHistory.store_revision_history(first_revision, hash) hash end # Pretty print the data as a string for the user def to_s hash = to_h[:churn] result = seperator result +="* Revision Changes \n" result += seperator result += "Files: \n" result += display_array(hash[:changed_files], :fields=>[:to_str], :headers=>{:to_str=>'file'}) result += "\nClasses: \n" result += display_array(hash[:changed_classes]) result += "\nMethods: \n" result += display_array(hash[:changed_methods]) + "\n" result += seperator result +="* Project Churn \n" result += seperator result += "Files: \n" result += display_array(hash[:changes]) result += "\nClasses: \n" class_churn = collect_items(hash[:class_churn], 'klass') result += display_array(class_churn) result += "\nMethods: \n" method_churn = collect_items(hash[:method_churn], 'method') result += display_array(method_churn) end private def collect_items(collection, match) collection.map {|item| (item.delete(match) || {}).merge(item) } end def sort_changes(changes) changes.to_a.sort! {|first,second| second[1] <=> first[1]} end def filters /.*\.rb/ end def display_array(array, options={}) array ? Hirb::Helpers::AutoTable.render(array, options.merge(:description=>false)) + "\n" : '' end def seperator "*"*70+"\n" end def self.git? system("git branch") end def self.hg? system("hg branch") end def self.bzr? system("bzr nick") end def set_source_control(start_date) if self.class.git? GitAnalyzer.new(start_date) elsif self.class.hg? HgAnalyzer.new(start_date) elsif self.class.bzr? BzrAnalyzer.new(start_date) elsif File.exist?(".svn") SvnAnalyzer.new(start_date) else raise "Churning requires a bazaar, git, mercurial, or subversion repo" end 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_changes| if file_changes.first.match(filters) classes, methods = get_changes(file_changes) changed_classes += classes changed_methods += methods end end changed_files = changed_files.map { |file, lines| file } [changed_files, changed_classes, changed_methods] end def calculate_changes!(changed_objs, total_changes) if changed_objs changed_objs.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) file = change.first breakdown = LocationMapping.new breakdown.get_info(file) changes = change.last classes = changes_for_type(changes, breakdown.klasses_collection) methods = changes_for_type(changes, breakdown.methods_collection) 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 def get_klass_for(method) method.gsub(/(#|\.).*/,'') end def changes_for_type(changes, item_collection) 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 return [] unless @source_control.respond_to?(:get_revisions) @source_control.get_revisions end def parse_logs_for_updated_files(revision, revisions) #TODO SVN doesn't support this return {} unless @source_control.respond_to?(:get_updated_files_change_info) files = @source_control.get_updated_files_change_info(revision, revisions) files.select{ |file, value| !@ignore_files.include?(file) } end end end