git_statistics 0.5.1 → 0.6.0

data/bin/git-statistics

@@ -1,5 +1,6 @@
 #!/usr/bin/env ruby
 
+$:.unshift File.expand_path("../../lib", __FILE__)
 require 'git_statistics'
 
-GitStatistics::GitStatistics.new(ARGV).execute
+GitStatistics::GitStatistics.new.execute

data/bin/git_statistics

@@ -1,5 +1,6 @@
 #!/usr/bin/env ruby
 
+$:.unshift File.expand_path("../../lib", __FILE__)
 require 'git_statistics'
 
-GitStatistics::GitStatistics.new(ARGV).execute
+GitStatistics::GitStatistics.new.execute

data/lib/git_statistics.rb

@@ -1,50 +1,98 @@
+require 'ostruct'
+require 'optparse'
 require 'git_statistics/initialize'
 
 module GitStatistics
   class GitStatistics
-    def initialize(args = nil)
-      @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 :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
+    attr_reader :options
+    def initialize
+      @options = OpenStruct.new(
+        email: false,
+        merges: false,
+        pretty: false,
+        update: false,
+        sort: "commits",
+        top: 0,
+        branch: false,
+        verbose: false,
+        debug: false,
+        limit: 100
+      )
+      parse_options
     end
 
     def execute
+      if options.debug
+        Log.level = Logger::DEBUG
+        Log.use_debug
+      elsif options.verbose
+        Log.level = Logger::INFO
+      end
+
       # Collect data (incremental or fresh) based on presence of old data
-      if @opts[:update]
+      if options.update
         # Ensure commit directory is present
-        collector = Collector.new(@opts[:verbose], @opts[:limit], false, @opts[:pretty])
-        commits_directory = collector.repo_path + ".git_statistics" + File::Separator
+        collector = Collector.new(options.limit, false, options.pretty)
+        commits_directory = File.join(collector.repo_path, ".git_statistics")
         FileUtils.mkdir_p(commits_directory)
         file_count = Utilities.number_of_matching_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}\"")
+          # Only use --since if there is data present
+          collector.collect(options.branch, "--since=\"#{time}\"")
           collected = true
         end
       end
 
       # If no data was collected as there was no present data then start fresh
       unless collected
-        collector = Collector.new(@opts[:verbose], @opts[:limit], true, @opts[:pretty])
-        collector.collect(@opts[:branch])
+        collector = Collector.new(options.limit, true, options.pretty)
+        collector.collect(options.branch)
       end
 
       # Calculate statistics
-      collector.commits.calculate_statistics(@opts[:email], @opts[:merges])
+      collector.commits.calculate_statistics(options.email, options.merges)
 
       # Print results
       results = Formatters::Console.new(collector.commits)
-      puts results.print_summary(@opts[:sort], @opts[:email], @opts[:top])
+      puts results.print_summary(options.sort, options.email, options.top)
+    end
+
+    def parse_options
+      OptionParser.new do |opt|
+        opt.version = VERSION
+        opt.on "-e", "--email", "Use author's email instead of name" do
+          options.email = true
+        end
+        opt.on "-m", "--merges", "Factor in merges when calculating statistics" do
+          options.merges = true
+        end
+        opt.on "-p", "--pretty", "Save the commits in git_repo/.git_statistics in pretty print (larger file size)" do
+          options.pretty = true
+        end
+        opt.on "-u", "--update", "Update saved commits with new data" do
+          options.update = true
+        end
+        opt.on "-s", "--sort TYPE", "Sort authors by {commits, additions, deletions, create, delete, rename, copy, merges}" do |type|
+          options.sort = type
+        end
+        opt.on "-t", "--top N", Float,"Show the top N authors in results" do |value|
+          options.top = value
+        end
+        opt.on "-b", "--branch", "Use current branch for statistics (otherwise all branches)" do
+          options.branch = true
+        end
+        opt.on "-v", "--verbose", "Verbose output (shows INFO level log statements)" do
+          options.verbose = true
+        end
+        opt.on "-d", "--debug", "Debug output (shows DEBUG level log statements)" do
+          options.debug = true
+        end
+        opt.on "-l", "--limit MAX_COMMITS", Float, "The maximum limit of commits to hold in memory at a time" do |number|
+          options.limit = number
+        end
+      end.parse!
     end
   end
 

data/lib/git_statistics/branches.rb

@@ -0,0 +1,35 @@
+module GitStatistics
+  class Branches
+
+    CURRENT_BRANCH = /\A\*\s/
+    NO_BRANCH = /no branch/i
+
+    def self.all
+      list.collect { |branch| branch.sub(CURRENT_BRANCH, "") }
+    end
+
+    def self.current
+      return '(none)' if detached?
+      list.detect { |branch| branch =~ CURRENT_BRANCH }.sub(CURRENT_BRANCH, "")
+    end
+
+    def self.detached?
+      stripped.grep(NO_BRANCH).any?
+    end
+
+    private
+
+      def self.list
+        stripped.reject { |b| b =~ NO_BRANCH }
+      end
+
+      def self.stripped
+        pipe.map(&:strip)
+      end
+
+      def self.pipe
+        Pipe.new("git --no-pager branch --no-color")
+      end
+
+  end
+end

data/lib/git_statistics/collector.rb

@@ -1,38 +1,29 @@
 module GitStatistics
   class Collector
 
-    attr_accessor :repo, :repo_path, :commits_path, :commits, :verbose
+    attr_accessor :repo, :repo_path, :commits_path, :commits
 
-    def initialize(verbose, limit, fresh, pretty)
-      @verbose = verbose
+    def initialize(limit, fresh, pretty)
       @repo = Utilities.get_repository
-
-      raise "No Git repository found" if @repo.nil?
-
-      @repo_path = File.expand_path("..", @repo.path) + File::Separator
-      @commits_path = @repo_path + ".git_statistics" + File::Separator
+      @repo_path = File.expand_path("..", @repo.path)
+      @commits_path = File.join(@repo_path, ".git_statistics")
       @commits = Commits.new(@commits_path, fresh, limit, pretty)
     end
 
     def collect(branch, time_since = "", time_until = "")
-      # Create pipe for git log to acquire branches
-      pipe = open("|git --no-pager branch --no-color")
-
       # Collect branches to use for git log
-      branches = branch ? ["", ""] : collect_branches(pipe)
+      branches = branch ? [] : Branches.all
 
       # Create pipe for the git log to acquire commits
-      pipe = open("|git --no-pager log #{branches.join(' ')} --date=iso --reverse"\
-                  " --no-color --find-copies-harder --numstat --encoding=utf-8"\
-                  " --summary #{time_since} #{time_until}"\
-                  " --format=\"%H,%an,%ae,%ad,%p\"")
+      pipe = Pipe.new("git --no-pager log #{branches.join(' ')} --date=iso --reverse"\
+                      " --no-color --find-copies-harder --numstat --encoding=utf-8"\
+                      " --summary #{time_since} #{time_until}"\
+                      " --format=\"%H,%an,%ae,%ad,%p\"")
 
       # Use a buffer approach to queue up lines from the log for each commit
       buffer = []
       pipe.each do |line|
 
-        line = line.clean_for_authors
-
         # Extract the buffer (commit) when we match ','x5 in the log format (delimeter)
         if line.split(',').size == 5
 
@@ -44,7 +35,6 @@ module GitStatistics
 
           # Save commits to file if size exceeds limit or forced
           @commits.flush_commits
-          @repo = Utilities.get_repository
         end
 
         buffer << line
@@ -57,33 +47,18 @@ module GitStatistics
 
     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 "\
-                  "--summary --format=\"%H,%an,%ae,%ad,%p\"")
-
-      buffer = pipe.map { |line| line.clean_for_authors }
+      pipe = Pipe.new("git --no-pager show #{sha} --date=iso --reverse"\
+                      " --no-color --find-copies-harder --numstat --encoding=utf-8 "\
+                      "--summary --format=\"%H,%an,%ae,%ad,%p\"")
 
       # Check that the buffer has valid information (i.e., sha was valid)
-      if !buffer.empty? && buffer.first.split(',').first == sha
-        buffer
+      if !pipe.empty? && pipe.first.split(',').first == sha
+        pipe.to_a
       else
-        nil
+        []
       end
     end
 
-    def collect_branches(pipe)
-      # Acquire all available branches from repository
-      branches = []
-      pipe.each do |line|
-
-        # Remove the '*' leading the current branch
-        line = line[1..-1] if line[0] == '*'
-        branches << line.clean_for_authors
-      end
-
-      return branches
-    end
-
     def acquire_commit_data(line)
       # Split up formated line
       commit_info = line.split(',')
@@ -109,11 +84,11 @@ module GitStatistics
       # Acquire general commit information
       commit_data = acquire_commit_data(buffer[0])
 
-      puts "Extracting #{commit_data[:sha]}" if @verbose
+      Log.info "Extracting #{commit_data[:sha]}"
 
       # Abort if the commit sha extracted form the buffer is invalid
       if commit_data[:sha].scan(/[\d|a-f]{40}/)[0].nil?
-        puts "Invalid buffer containing commit information"
+        Log.warn "Invalid buffer containing commit information"
         return
       end
 
@@ -122,7 +97,7 @@ module GitStatistics
 
       # No files were changed in this commit, abort commit
       if files.nil?
-        puts "No files were changed"
+        Log.debug "No files were changed"
         return
       end
 
@@ -134,9 +109,9 @@ module GitStatistics
         if blob.instance_of?(Grit::Blob)
           process_blob(commit_data[:data], blob, file)
         elsif blob.instance_of?(Grit::Submodule)
-          puts "Ignoring submodule #{blob.name}"
+          Log.debug "Ignoring submodule #{blob.name}"
         else
-          puts "Problem processing file #{file[:file]}"
+          Log.warn "Problem processing file #{file[:file]}"
         end
       end
       return commit_data[:data]

data/lib/git_statistics/commit_line_extractor.rb

@@ -17,31 +17,31 @@ module GitStatistics
         split_file = Utilities.split_old_new_file(changes[2], changes[3])
         {:additions => changes[0].to_i,
           :deletions => changes[1].to_i,
-          :file => split_file[:new_file].clean_for_authors,
-          :old_file => split_file[:old_file].clean_for_authors}
+          :file => split_file[:new_file],
+          :old_file => split_file[:old_file]}
       end
       return modified_or_renamed unless modified_or_renamed.empty?
 
       AdditionsOrDeletions.if_matches(line) do |changes|
         {:additions => changes[0].to_i,
           :deletions => changes[1].to_i,
-          :file => changes[2].clean_for_authors}
+          :file => changes[2]}
       end
     end
 
     def created_or_deleted
       CreatedOrDeleted.if_matches(line) do |changes|
-        {:status => changes[0].clean_for_authors,
-          :file => changes[1].clean_for_authors}
+        {:status => changes[0],
+          :file => changes[1]}
       end
     end
 
     def renamed_or_copied
       RenamedOrCopied.if_matches(line) do |changes|
         split_file = Utilities.split_old_new_file(changes[1], changes[2])
-        {:status => changes[0].clean_for_authors,
-          :old_file => split_file[:old_file].clean_for_authors,
-          :new_file => split_file[:new_file].clean_for_authors,
+        {:status => changes[0],
+          :old_file => split_file[:old_file],
+          :new_file => split_file[:new_file],
           :similar => changes[3].to_i}
       end
     end

data/lib/git_statistics/commits.rb

@@ -14,12 +14,11 @@ module GitStatistics
 
     def clean
       # Ensure the path exists
-      FileUtils.mkdir_p(@path)
+      FileUtils.mkdir_p(path)
 
       # Remove all files within path if saving
-      if @fresh
-        Dir.entries(@path).each do |file|
-          next if %w[. ..].include? file
+      if fresh
+        files_in_path.each do |file|
           File.delete(File.join(path, file))
         end
       end
@@ -30,6 +29,10 @@ module GitStatistics
       @totals[:languages] = {}
     end
 
+    def files_in_path
+      Dir.entries(path).reject { |file| %w[. ..].include?(file) }
+    end
+
     def flush_commits(force = false)
       if size >= limit || force
         file_count = Utilities.number_of_matching_files(path, /\d+\.json/)

data/lib/git_statistics/formatters/console.rb

@@ -2,7 +2,7 @@ module GitStatistics
   module Formatters
     class Console
 
-      attr_accessor :commits
+      attr_accessor :commits, :config
 
       def initialize(commits)
         @commits = commits
@@ -17,7 +17,7 @@ module GitStatistics
         raise "Parameter for --sort is not valid" if data.nil?
 
         # Create config
-        config = {:data => data,
+        @config = {:data => data,
                   :author_length => Utilities.max_length_in_list(data.keys, 17),
                   :language_length => Utilities.max_length_in_list(@commits.totals[:languages].keys, 8),
                   :sort => sort,
@@ -25,70 +25,73 @@ module GitStatistics
                   :top_n => top_n}
 
         # Acquire formatting pattern for output
-        pattern = "%-#{config[:author_length]}s | %-#{config[:language_length]}s | %7s | %9s | %9s | %7s | %7s | %7s | %6s | %6s |"
-        config[:pattern] = pattern
-        return config
+        @pattern = "| %-#{config[:author_length]}s | %-#{config[:language_length]}s | %7s | %9s | %9s | %7s | %7s | %7s | %6s | %6s |"
+        config
       end
 
       def print_summary(sort, email, top_n = 0)
         # Prepare and determine the config for the result summary based on parameters
+        commit_totals = @commits.totals
         config = prepare_result_summary(sort, email, top_n)
 
         # Print query/header information
-        output = print_header(config)
+        output = print_header
 
         # Print per author information
-        config[:data].each do |key,value|
-          output += config[:pattern] % [key, "", value[:commits], value[:additions],
-                          value[:deletions], value[:create], value[:delete],
-                          value[:rename], value[:copy], value[:merges]]
-          output += "\n"
-          output += print_language_data(config[:pattern], value)
+        config[:data].each do |name, commit_data|
+          output << print_row(name, commit_data)
+          output << print_language_data(commit_data)
         end
-
-        # Reprint query/header for repository information
-        output += "\n"
-        output += print_header(config)
-        data = @commits.totals
-        output += config[:pattern] % ["Repository Totals", "", data[:commits],
-                        data[:additions], data[:deletions], data[:create],
-                        data[:delete], data[:rename], data[:copy], data[:merges]]
-        output += "\n"
-        output += print_language_data(config[:pattern], data)
-        return output
+        output << separator
+        output << print_row("Repository Totals", commit_totals)
+        output << print_language_data(commit_totals)
+        output.flatten.join("\n")
       end
 
-      def print_language_data(pattern, data)
-        output = ""
+      def print_language_data(data)
+        output = []
         # Print information of each language for the data
-        data[:languages].each do |key,value|
-          output += pattern % ["", key, "", value[:additions], value[:deletions],
-                          value[:create], value[:delete], value[:rename],
-                          value[:copy], value[:merges]]
-          output += "\n"
+        data[:languages].each do |language, commit_data|
+          output << print_row("", commit_data, language)
         end
-        return output
+        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])
+      end
+
+      def print_header
+        output = []
+        output << get_author_info(@commits.stats.size)
+        output << get_header_info
+        output
       end
 
-      def print_header(config)
-        total_authors = @commits.stats.size
+      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
+      end
+
+      def format_for_row(*columns)
+        @pattern % columns
+      end
+
+      def separator
+        "-" * 89 + "-"*config[:author_length] + "-"*config[:language_length]
+      end
 
-        output = ""
-        # Print summary information of displayed results
-        if config[:top_n] > 0 and config[:top_n] < total_authors
-          output += "Top #{config[:top_n]} authors(#{total_authors}) sorted by #{config[:sort]}\n"
-        else
-          output += "All authors(#{total_authors}) sorted by #{config[:sort]}\n"
+      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]}"
         end
 
-        # Print column headers
-        output += "-"*87 + "-"*config[:author_length] + "-"*config[:language_length]
-        output += "\n"
-        output += config[:pattern] % ['Name/Email', 'Language', 'Commits', 'Additions', 'Deletions', 'Creates', 'Deletes', 'Renames', 'Copies', 'Merges']
-        output += "\n"
-        output +=  "-"*87 + "-"*config[:author_length] + "-"*config[:language_length]
-        output += "\n"
-        return output
+        "All authors(#{total_authors}) sorted by #{config[:sort]}"
       end
     end
   end