git_statistics 0.6.0 → 0.7.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA1:
+  metadata.gz: ac2c8d9e4ff4611dfa6c6b2f9f664fb874a7201f
+  data.tar.gz: 9da9f644f174806900bd8b2a2921906bfc3e3989
+SHA512:
+  metadata.gz: 381f1532eee3178c8a1459ec2f5ca37f1429786c56e4bae54c6139e2e9ed4fea6253fc4960d6b426e360e777c6adad1c48bffac53b1ce4fe9f93b5de27f5e51f
+  data.tar.gz: b7bb4f77e58b6f1bf73cfd09219651c0256a59a8a9b3de25a50c11948063a66c3df88856b14210709a6353f1fd0bf97d5eeaafea95f3d36d0a7587bf4d5fd7cf

data/bin/git-statistics

@@ -3,4 +3,4 @@
 $:.unshift File.expand_path("../../lib", __FILE__)
 require 'git_statistics'
 
-GitStatistics::GitStatistics.new.execute
+GitStatistics::CLI.new(Dir.pwd).execute

data/bin/git_statistics

@@ -3,4 +3,4 @@
 $:.unshift File.expand_path("../../lib", __FILE__)
 require 'git_statistics'
 
-GitStatistics::GitStatistics.new.execute
+GitStatistics::CLI.new(Dir.pwd).execute

data/lib/git_statistics.rb

@@ -1,11 +1,16 @@
-require 'ostruct'
-require 'optparse'
 require 'git_statistics/initialize'
 
 module GitStatistics
-  class GitStatistics
-    attr_reader :options
-    def initialize
+  class CLI
+    attr_reader :repository, :options
+
+    DEFAULT_BRANCH = "master"
+
+    def initialize(dir)
+      repository_location = dir.nil? ? Rugged::Repository.discover(Dir.pwd) : Rugged::Repository.discover(dir)
+      @repository = Rugged::Repository.new(repository_location)
+      @collected = false
+      @collector = nil
       @options = OpenStruct.new(
         email: false,
         merges: false,
@@ -13,7 +18,7 @@ module GitStatistics
         update: false,
         sort: "commits",
         top: 0,
-        branch: false,
+        branch: DEFAULT_BRANCH,
         verbose: false,
         debug: false,
         limit: 100
@@ -22,43 +27,43 @@ module GitStatistics
     end
 
     def execute
-      if options.debug
-        Log.level = Logger::DEBUG
-        Log.use_debug
-      elsif options.verbose
-        Log.level = Logger::INFO
-      end
+      determine_log_level
+      collect_and_only_update
+      fresh_collect! unless @collected
+      calculate!
+      output_results
+    end
 
-      # Collect data (incremental or fresh) based on presence of old data
+    def collect_and_only_update
       if options.update
         # Ensure commit directory is present
-        collector = Collector.new(options.limit, false, options.pretty)
-        commits_directory = File.join(collector.repo_path, ".git_statistics")
+        @collector = Collector.new(repository, options.limit, false, options.pretty)
+        commits_directory = repository.workdir + ".git_statistics/"
         FileUtils.mkdir_p(commits_directory)
         file_count = Utilities.number_of_matching_files(commits_directory, /\d+\.json/) - 1
 
         if file_count >= 0
-          time = Utilities.get_modified_time(commits_directory + "#{file_count}.json")
-          # Only use --since if there is data present
-          collector.collect(options.branch, "--since=\"#{time}\"")
-          collected = true
+          time_since = Utilities.get_modified_time(commits_directory + "#{file_count}.json").to_s
+          @collector.collect({:branch => options.branch, :time_since => time_since})
+          @collected = true
         end
       end
+    end
 
-      # If no data was collected as there was no present data then start fresh
-      unless collected
-        collector = Collector.new(options.limit, true, options.pretty)
-        collector.collect(options.branch)
-      end
-
-      # Calculate statistics
-      collector.commits.calculate_statistics(options.email, options.merges)
+    def calculate!
+      @collector.commits.calculate_statistics(options.email, options.merges)
+    end
 
-      # Print results
-      results = Formatters::Console.new(collector.commits)
+    def output_results
+      results = Formatters::Console.new(@collector.commits)
       puts results.print_summary(options.sort, options.email, options.top)
     end
 
+    def fresh_collect!
+      @collector = Collector.new(repository, options.limit, true, options.pretty)
+      @collector.collect({:branch => options.branch})
+    end
+
     def parse_options
       OptionParser.new do |opt|
         opt.version = VERSION
@@ -80,8 +85,8 @@ module GitStatistics
         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
+        opt.on "-b", "--branch BRANCH", "Use the specified branch for statistics (otherwise the master branch is used)" do |branch|
+          options.branch = branch
         end
         opt.on "-v", "--verbose", "Verbose output (shows INFO level log statements)" do
           options.verbose = true
@@ -94,6 +99,17 @@ module GitStatistics
         end
       end.parse!
     end
-  end
 
+    private
+
+      def determine_log_level
+        if options.debug
+          Log.level = Logger::DEBUG
+          Log.use_debug
+        elsif options.verbose
+          Log.level = Logger::INFO
+        end
+      end
+
+  end
 end

data/lib/git_statistics/collector.rb

@@ -1,217 +1,70 @@
 module GitStatistics
   class Collector
 
-    attr_accessor :repo, :repo_path, :commits_path, :commits
+    attr_accessor :repo, :commits_path, :commits
 
-    def initialize(limit, fresh, pretty)
-      @repo = Utilities.get_repository
-      @repo_path = File.expand_path("..", @repo.path)
-      @commits_path = File.join(@repo_path, ".git_statistics")
+    def initialize(repo, limit, fresh, pretty)
+      @repo = repo
+      @commits_path = repo.workdir + ".git_statistics"
       @commits = Commits.new(@commits_path, fresh, limit, pretty)
     end
 
-    def collect(branch, time_since = "", time_until = "")
-      # Collect branches to use for git log
-      branches = branch ? [] : Branches.all
+    def collect(options = {})
+      branch = options[:branch] ? options[:branch] : CLI::DEFAULT_BRANCH
+      branch_head = Rugged::Branch.lookup(repo, branch).tip
 
-      # Create pipe for the git log to acquire commits
-      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\"")
+      walker = Rugged::Walker.new(repo)
+      walker.push(branch_head)
 
-      # Use a buffer approach to queue up lines from the log for each commit
-      buffer = []
-      pipe.each do |line|
-
-        # Extract the buffer (commit) when we match ','x5 in the log format (delimeter)
-        if line.split(',').size == 5
-
-          # Sometimes 'git log' doesn't populate the buffer (i.e., merges), try fallback option if so
-          buffer = fall_back_collect_commit(buffer[0].split(',').first) if buffer.one?
-
-          extract_commit(buffer) unless buffer.empty?
-          buffer = []
-
-          # Save commits to file if size exceeds limit or forced
+      walker.each_with_index do |commit, count|
+        if valid_commit?(commit, options)
+          extract_commit(commit, count + 1)
           @commits.flush_commits
         end
-
-        buffer << line
       end
 
-      # Extract the last commit
-      extract_commit(buffer) unless buffer.empty?
       @commits.flush_commits(true)
     end
 
-    def fall_back_collect_commit(sha)
-      # Create pipe for the git log to acquire commits
-      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 !pipe.empty? && pipe.first.split(',').first == sha
-        pipe.to_a
-      else
-        []
+    def valid_commit?(commit, options)
+      if !options[:time_since].nil?
+        return false unless commit.author[:time] > DateTime.parse(options[:time_since].to_s).to_time
       end
-    end
-
-    def acquire_commit_data(line)
-      # Split up formated line
-      commit_info = line.split(',')
 
-      # Initialize commit data
-      data = (@commits[commit_info[0]] ||= Hash.new(0))
-      data[:author] = commit_info[1]
-      data[:author_email] = commit_info[2]
-      data[:time] = commit_info[3]
-      data[:files] = []
-
-      # Flag commit as merge if necessary (determined if two parents)
-      if commit_info[4].nil? || commit_info[4].split(' ').one?
-        data[:merge] = false
-      else
-        data[:merge] = true
+      if !options[:time_until].nil?
+        return false unless commit.author[:time] < DateTime.parse(options[:time_until].to_s).to_time
       end
 
-      return {:sha => commit_info[0], :data => data}
+      return true
     end
 
-    def extract_commit(buffer)
-      # Acquire general commit information
-      commit_data = acquire_commit_data(buffer[0])
-
-      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?
-        Log.warn "Invalid buffer containing commit information"
-        return
-      end
-
-      # Identify all changed files for this commit
-      files = identify_changed_files(buffer[2..-1])
-
-      # No files were changed in this commit, abort commit
-      if files.nil?
-        Log.debug "No files were changed"
-        return
-      end
-
-      # Acquire blob for each changed file and process it
-      files.each do |file|
-        blob = get_blob(commit_data[:sha], file)
-
-        # Only process blobs, or log the submodules and problematic files
-        if blob.instance_of?(Grit::Blob)
-          process_blob(commit_data[:data], blob, file)
-        elsif blob.instance_of?(Grit::Submodule)
-          Log.debug "Ignoring submodule #{blob.name}"
-        else
-          Log.warn "Problem processing file #{file[:file]}"
-        end
-      end
-      return commit_data[:data]
-    end
-
-    def get_blob(sha, file)
-      # Split up file for Grit navigation
-      file = file[:file].split(File::Separator)
-
-      # Acquire blob of the file for this specific commit
-      blob = Utilities.find_blob_in_tree(@repo.tree(sha), file)
-
-      # If we cannot find blob in current commit (deleted file), check previous commit
-      if blob.nil? || blob.instance_of?(Grit::Tree)
-        prev_commit = @repo.commits(sha).first.parents[0]
-        return nil if prev_commit.nil?
+    def acquire_commit_meta(commit_summary)
+      # Initialize commit data
+      data = (@commits[commit_summary.oid] ||= Hash.new(0))
+
+      data[:author] = commit_summary.author[:name]
+      data[:author_email] = commit_summary.author[:email]
+      data[:time] = commit_summary.author[:time].to_s
+      data[:merge] = commit_summary.merge?
+      data[:additions] = commit_summary.additions
+      data[:deletions] = commit_summary.deletions
+      data[:net] = commit_summary.net
+      data[:added_files] = commit_summary.added_files
+      data[:deleted_files] = commit_summary.deleted_files
+      data[:modified_files] = commit_summary.modified_files
+      data[:files] = commit_summary.file_stats.map{ |file| file.to_json }
 
-        prev_tree = @repo.tree(prev_commit.id)
-        blob = Utilities.find_blob_in_tree(prev_tree, file)
-      end
-      return blob
+      return data
     end
 
-    def identify_changed_files(buffer)
-      return buffer if buffer.nil?
-
-      # For each modification extract the details
-      changed_files = []
-      buffer.each do |line|
-        extracted_line = CommitLineExtractor.new(line)
+    def extract_commit(commit, count)
+      Log.info "Extracting(#{count}) #{commit.oid}"
+      commit_summary = CommitSummary.new(@repo, commit)
 
-        # Extract changed file information if it exists
-        changed_file_information = extracted_line.changed
-        if changed_file_information.any?
-          changed_files << changed_file_information
-          next  # This line is processed, skip to next
-        end
+      # Acquire meta information about commit
+      commit_data = acquire_commit_meta(commit_summary)
 
-        # Extract details of create/delete files if it exists
-        created_or_deleted = extracted_line.created_or_deleted
-        if created_or_deleted.any?
-          augmented = false
-          # Augment changed file with create/delete information if possible
-          changed_files.each do |file|
-            if file[:file] == created_or_deleted[:file]
-              file[:status] = created_or_deleted[:status]
-              augmented = true
-              break
-            end
-          end
-          changed_files << created_or_deleted unless augmented
-          next  # This line is processed, skip to next
-        end
-
-        # Extract details of rename/copy files if it exists
-        renamed_or_copied = extracted_line.renamed_or_copied
-        if renamed_or_copied.any?
-          augmented = false
-          # Augment changed file with rename/copy information if possible
-          changed_files.each do |file|
-            if file[:file] == renamed_or_copied[:new_file]
-              file[:status] = renamed_or_copied[:status]
-              file[:old_file] = renamed_or_copied[:old_file]
-              file[:similar] = renamed_or_copied[:similar]
-              augmented = true
-              break
-            end
-          end
-          changed_files << renamed_or_copied unless augmented
-          next  # This line is processed, skip to next
-        end
-      end
-
-      changed_files
-    end
-
-    def process_blob(data, blob, file)
-      # Initialize a hash to hold information regarding the file
-      file_hash = Hash.new(0)
-      file_hash[:name] = file[:file]
-      file_hash[:additions] = file[:additions]
-      file_hash[:deletions] = file[:deletions]
-      file_hash[:status] = file[:status]
-
-      # Add file information to commit itself
-      data[file[:status].to_sym] += 1 if file[:status] != nil
-      data[:additions] += file[:additions]
-      data[:deletions] += file[:deletions]
-
-      # Acquire specifics on blob
-      file_hash[:binary] = blob.binary?
-      file_hash[:image] = blob.image?
-      file_hash[:vendored] = blob.vendored?
-      file_hash[:generated] = blob.generated?
-
-      # Identify the language of the blob if possible
-      file_hash[:language] = blob.language.nil? ? "Unknown" : blob.language.name
-      data[:files] << file_hash
-
-      return data
+      return commit_data
     end
 
   end

data/lib/git_statistics/commit_summary.rb

@@ -0,0 +1,86 @@
+module GitStatistics
+  class CommitSummary < SimpleDelegator
+    def initialize(repo, commit)
+      super(commit)
+      @repo = repo
+      @diff = diff(commit.parents.first)
+      @patches = @diff.patches
+    end
+
+    # A Git commit is a merge if it has more than one parent
+    def merge?
+      parents.size > 1
+    end
+
+    # How many files were removed in this commit
+    def deleted_files
+      file_stats.select { |file| file.status == :deleted }.count
+    end
+
+    # How many files were added in this commit
+    def added_files
+      file_stats.select { |file| file.status == :added }.count
+    end
+
+    # How many files were modified (not added/deleted) in this commit
+    def modified_files
+      file_stats.select { |file| file.status == :modified }.count
+    end
+
+    # How many total additions in this commit?
+    def additions
+      commit_summary(:additions)
+    end
+
+    # How many total deletions in this commit?
+    def deletions
+      commit_summary(:deletions)
+    end
+
+    # What is the net # of lines changes in this commit?
+    def net
+      commit_summary(:net)
+    end
+
+    def file_stats
+      @cached_file_stats ||= diffstats.map { |diff| DiffSummary.new(@repo, diff) }
+    end
+
+    LanguageSummary = Struct.new(:name, :additions, :deletions, :net, :added_files, :deleted_files, :modified_files)
+
+    # Array of LanguageSummary objects (one for each language) for simple calculations
+    def languages
+      grouped_language_files.collect do |language, stats|
+        additions = summarize(stats, :additions)
+        deletions = summarize(stats, :deletions)
+        net       = summarize(stats, :net)
+        LanguageSummary.new(language, additions, deletions, net, added_files, deleted_files, modified_files)
+      end
+    end
+
+    # Group file statistics by language
+    def grouped_language_files
+      file_stats.group_by(&:language)
+    end
+
+    # Files touched in this commit
+    def filenames
+      file_stats.map(&:filename)
+    end
+
+    private
+
+      def summarize(stats, what)
+        stats.map(&what).inject(0, :+)
+      end
+
+      def commit_summary(what)
+        summarize(file_stats, what)
+      end
+
+      def diffstats
+        @patches
+      end
+
+  end
+end