Merge pull request #61 from kevinjalbert/move-to-rugged

Move from Grit to Rugged for git processing

Author: kevinjalbert

.ruby-version

@@ -1 +1 @@
-1.9.3-p392
+2.0.0

.travis.yml

@@ -1,8 +1,8 @@
 language: ruby
 
 rvm:
- - 1.9.2
  - 1.9.3
+ - 2.0.0
 
 bundler_args: --without darwin debug
 

Rakefile

@@ -16,7 +16,6 @@ end
 
 desc "Run git_statistics on current/specified directory (for debugging)"
 task :run, :dir do |t, args|
- puts args[:dir]
  Bundler.require(:debug)
  require 'git_statistics'
  GitStatistics::CLI.new(args[:dir]).execute

git_statistics.gemspec

@@ -14,10 +14,10 @@ Gem::Specification.new do |gem|
  gem.files = Dir['lib/**/*']
  gem.test_files = Dir['spec/**/*_spec.rb']
  gem.executables = %w[ git_statistics git-statistics ]
- gem.required_ruby_version = '>= 1.9.1'
+ gem.required_ruby_version = '>= 1.9.3'
 
  gem.add_dependency('json')
- gem.add_dependency('grit')
+ gem.add_dependency('rugged')
  gem.add_dependency('language_sniffer')
 
  gem.add_development_dependency "rspec", "~> 2.12.0"

lib/git_statistics.rb

@@ -4,8 +4,11 @@ module GitStatistics
  class CLI
  attr_reader :repository, :options
 
+ DEFAULT_BRANCH = "master"
+
  def initialize(dir)
- @repository = dir.nil? ? Repo.new(Dir.pwd) : Repo.new(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(
@@ -15,7 +18,7 @@ def initialize(dir)
  update: false,
  sort: "commits",
  top: 0,
- branch: false,
+ branch: DEFAULT_BRANCH,
  verbose: false,
  debug: false,
  limit: 100
@@ -35,14 +38,13 @@ def collect_and_only_update
  if options.update
  # Ensure commit directory is present
  @collector = Collector.new(repository, options.limit, false, options.pretty)
- commits_directory = repository.working_dir + ".git_statistics"
+ 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_since = Utilities.get_modified_time(commits_directory + "#{file_count}.json")
- # Only use --since if there is data present
- @collector.collect(options.branch, {:since => time_since})
+ 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
@@ -59,7 +61,7 @@ def output_results
 
  def fresh_collect!
  @collector = Collector.new(repository, options.limit, true, options.pretty)
- @collector.collect(options.branch)
+ @collector.collect({:branch => options.branch})
  end
 
  def parse_options
@@ -83,8 +85,8 @@ def parse_options
  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

lib/git_statistics/blob_finder.rb

@@ -4,52 +4,67 @@ class Collector
  attr_accessor :repo, :commits_path, :commits
 
  def initialize(repo, limit, fresh, pretty)
- Grit::Git.git_timeout = 0
- Grit::Git.git_max_size = 0
-
  @repo = repo
- @commits_path = repo.working_dir + ".git_statistics"
+ @commits_path = repo.workdir + ".git_statistics"
  @commits = Commits.new(@commits_path, fresh, limit, pretty)
  end
 
- def collect(branch, options = {})
- # Collect branches to use for git log
- branches = branch ? [] : @repo.branches.compact.map(&:name)
- Grit::Commit.find_all(@repo, nil, options).each do |commit|
- extract_commit(commit)
- @commits.flush_commits
+ def collect(options = {})
+ branch = options[:branch] ? options[:branch] : CLI::DEFAULT_BRANCH
+ branch_head = Rugged::Branch.lookup(repo, branch).tip
+
+ walker = Rugged::Walker.new(repo)
+ walker.push(branch_head)
+
+ walker.each_with_index do |commit, count|
+ if valid_commit?(commit, options)
+ extract_commit(commit, count + 1)
+ @commits.flush_commits
+ end
  end
+
  @commits.flush_commits(true)
  end
 
+ 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
+
+ if !options[:time_until].nil?
+ return false unless commit.author[:time] < DateTime.parse(options[:time_until].to_s).to_time
+ end
+
+ return true
+ end
+
  def acquire_commit_meta(commit_summary)
  # Initialize commit data
- data = (@commits[commit_summary.sha] ||= Hash.new(0))
+ 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.authored_date.to_s
+ 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[:new_files] = commit_summary.new_files
- data[:removed_files] = commit_summary.removed_files
- data[:files] = commit_summary.files
+ 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 }
 
  return data
  end
 
- def extract_commit(commit)
- unless commit.nil?
- commit_summary = CommitSummary.new(commit)
- Log.info "Extracting #{commit_summary.sha}"
+ def extract_commit(commit, count)
+ Log.info "Extracting(#{count}) #{commit.oid}"
+ commit_summary = CommitSummary.new(@repo, commit)
 
-  # Acquire meta information about commit
-  commit_data = acquire_commit_meta(commit_summary)
+ # Acquire meta information about commit
+ commit_data = acquire_commit_meta(commit_summary)
 
- return commit_data
- end
+ return commit_data
  end
 
  end

lib/git_statistics/collector.rb

@@ -1,7 +1,10 @@
 module GitStatistics
  class CommitSummary < SimpleDelegator
- def initialize(commit)
+ 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
@@ -10,13 +13,18 @@ def merge?
  end
 
  # How many files were removed in this commit
- def removed_files
- cached_show.select { |diff| diff.deleted_file == true }.count
+ def deleted_files
+ file_stats.select { |file| file.status == :deleted }.count
  end
 
  # How many files were added in this commit
- def new_files
- cached_show.select { |diff| diff.new_file == true }.count
+ 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?
@@ -35,18 +43,18 @@ def net
  end
 
  def file_stats
- @cached_file_stats ||= diffstats.map { |diff| DiffSummary.new(diff, current_tree) }
+ @cached_file_stats ||= diffstats.map { |diff| DiffSummary.new(@repo, diff) }
  end
 
- LanguageSummary = Struct.new(:name, :additions, :deletions, :net)
+ 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)
+ LanguageSummary.new(language, additions, deletions, net, added_files, deleted_files, modified_files)
  end
  end
 
@@ -55,72 +63,13 @@ def grouped_language_files
  file_stats.group_by(&:language)
  end
 
- FileSummary = Struct.new(:name, :language, :additions, :deletions, :net, :filestatus)
-
- # Array of FileSummary objects (one for each file) for simple calculations
- def files
- file_stats.collect{ |stats| determine_file_summary(stats) }
- end
-
- def cached_show
- @cached_commit_show ||= show
- end
-
  # Files touched in this commit
- def file_names
- diffstats.map(&:filename)
- end
-
- # Fetch the current Grit::Repo tree from this commit
- def current_tree
- @current_tree ||= repo.tree(sha)
+ def filenames
+ file_stats.map(&:filename)
  end
 
  private
 
- def determine_file_summary(stats)
- filestatus = :modified
- language = stats.language
-
- # Determine if this file could be a new or deleted file
- if (stats.additions > 0 && stats.deletions == 0) || (stats.additions == 0 && stats.deletions > 0)
- # Extract file status from commit's diff object
- cached_show.each do |diff|
- if stats.filename == diff.b_path
- filestatus = :create if diff.new_file
- filestatus = :delete if diff.deleted_file
- break
- end
- end
- end
-
- # Determine language of blob
- if stats.tree?
- # Trees have no language (the tree's blobs are still processed via the remainder diffstats)
- language = "Unknown"
- elsif stats.submodule?
- language = "Submodule"
- elsif stats.blob.nil?
- # If blob is nil (i.e., deleted file) grab the previous version of this blob using the parents of the current commit
- blob = BlobFinder.get_blob(self.parents.first, stats.filename)
- blob = BlobFinder.get_blob(self.parents.last, stats.filename) if blob.nil?
-
- # Determine language of newly found blob
- if blob.kind_of? Grit::Tree
- language = "Unknown"
- elsif blob.kind_of? Grit::Submodule
- language = "Submodule"
- elsif blob.nil? || blob.language.nil?
- language = "Unknown"
- else
- language = blob.language.to_s
- end
- end
-
- # TODO Converts file summary into hash to keep json compatibility (for now)
- Hash[FileSummary.new(stats.filename, language, stats.additions, stats.deletions, stats.net, filestatus).each_pair.to_a]
- end
-
  def summarize(stats, what)
  stats.map(&what).inject(0, :+)
  end
@@ -130,19 +79,7 @@ def commit_summary(what)
  end
 
  def diffstats
- if merge?
- merge_diffstats
- else
- stats.to_diffstat
- end
- end
-
- # Hackery coming...
- DIFFSTAT_REGEX = /([-|\d]+)\s+([-|\d]+)\s+(.+)/i
- def merge_diffstats
- native_diff = repo.git.native(:diff, {numstat: true}, parents.join("..."))
- per_file_info = native_diff.scan(DIFFSTAT_REGEX)
- per_file_info.map { |add, del, file| Grit::DiffStat.new(file, add.to_i, del.to_i) }
+ @patches
  end
 
  end