git-commits-analyzer 0.1.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA1:
+  metadata.gz: e1c430c515c28a8a1a8f6dea0098f0590101095f
+  data.tar.gz: 01df65d27d647d9dff0207927d9d550a93111b0f
+SHA512:
+  metadata.gz: 21b04677942df99acdef4d4d44cfc907fdeee7907a121c695e616c093d98e99af87ab242aa036009be2644fb629b05d61ebaaaf3c024c4f6fc8cdd5925ffa614
+  data.tar.gz: a647cccacdec435bd1d73f444b76157f363f671e6da468917fb7e637c8dc090e3c9c9fa7c7e4c5612afbc48e0d985119c5b5f609e2269a786a2f564446be07bd

data/bin/analyze_commits

@@ -0,0 +1,51 @@
+#!/usr/bin/env ruby
+
+# Gems.
+require 'logger'
+require 'pp'
+
+# Internal classes.
+require 'git-commits-analyzer'
+require 'git-commits-analyzer/utils'
+
+
+##### CONFIGURATION #####
+
+# Configure logger.
+logger = Logger.new(STDOUT)
+logger.level = Logger::WARN
+
+
+##### MAIN #####
+
+# Parse command line options.
+options = Utils.parse_command_line_options()
+
+# Find git repos to inspect.
+repos = Utils.get_git_repos(path: options[:path])
+puts "Found " + repos.length.to_s + " repos to inspect."
+puts ""
+
+# Inspect git repos.
+puts "===== Inspecting repos ====="
+puts ""
+git_commits_analyzer = GitCommitsAnalyzer.new(logger: logger, author: options[:authors])
+repos.sort.each do |repo|
+  puts "Inspecting repo " + repo
+  git_commits_analyzer.parse_repo(repo: repo)
+  #break
+end
+puts ""
+
+# Display sanity check.
+puts "Found #{git_commits_analyzer.total_commits} commits for author(s) " + options[:authors].join(', ')
+puts ""
+exit if git_commits_analyzer.monthly_commits.keys.length == 0
+
+# Save data.
+puts "===== Save data ====="
+puts ""
+output_file = options[:output];
+File.open(output_file, 'w') { |file| file.write(git_commits_analyzer.to_json) }
+puts "Re-generated #{output_file}."
+puts ""

data/lib/git-commits-analyzer/utils.rb

@@ -0,0 +1,54 @@
+require 'optparse'
+
+class Utils
+  def self.parse_command_line_options()
+    options = {}
+    OptionParser.new do |opts|
+      opts.banner = "Usage: inspect_contributions.rb [options]"
+      options[:authors] = []
+
+      # Parse path.
+      opts.on("-p", "--path PATH", "Specify a path to search for git repositories under") do |path|
+        options[:path] = path
+      end
+
+      # Parse authors.
+      opts.on("-a", "--author EMAIL", "Include this author in statistics") do |email|
+        options[:authors] << email
+      end
+
+      # Parse output directory.
+      opts.on("-p", "--output PATH", "Specify a path to output files with collected data") do |output|
+        options[:output] = output
+      end
+
+      # Show usage
+      opts.on_tail("-h", "--help", "Show this message") do
+        puts opts
+        exit
+      end
+    end.parse!
+
+    # Check mandatory options.
+    raise OptionParser::MissingArgument, '--author' if options[:authors].length == 0
+    raise OptionParser::MissingArgument, '--output' if options[:output].nil?
+    raise OptionParser::MissingArgument, '--path' if options[:path].nil?
+
+    return options
+  end
+
+  def self.get_git_repos(path:)
+    repos = []
+    Dir.glob(File.join(path, '*')) do |dir|
+      # Skip files.
+      next if !File.directory?(dir)
+
+      # Skip directories without .git subdirectory (shortcut to identify repos).
+      next if !File.directory?(File.join(dir, '.git'))
+
+      repos << dir
+    end
+
+    return repos
+  end
+end

data/lib/git-commits-analyzer.rb

@@ -0,0 +1,198 @@
+require 'date'
+require 'git'
+require 'git_diff_parser'
+require 'json'
+
+# Public: parse git logs for language and commit metadata.
+#
+# Examples:
+#
+#   git_parser = GitCommitsAnalyzer.new(logger: logger, author: author)
+#
+class GitCommitsAnalyzer
+  # Public: Returns a hash of commit numbers broken down by month.
+  attr_reader :monthly_commits
+
+  # Public: Returns the total number of commits belonging to the author
+  # specified.
+  attr_reader :total_commits
+
+  # Public: Returns the number of lines added/removed broken down by language.
+  attr_reader :lines_by_language
+
+  # Public: Initialize new GitParser object.
+  #
+  # logger - A logger object to display git errors/warnings.
+  # author - The email of the git author for whom we should compile the metadata.
+  #
+  def initialize(logger:, author:)
+    @logger = logger
+    @author = author
+    @monthly_commits = {}
+    @monthly_commits.default = 0
+    @total_commits = 0
+    @lines_by_language = {}
+  end
+
+  # Public: Determine the type of a file at the given revision of a repo.
+  #
+  # filename - The name of the file to analyze.
+  # sha      - The commit ID.
+  # git_repo - A git repo object corresponding to the underlying repo.
+  #
+  # Returns a string corresponding to the language of the file.
+  #
+  def self.determine_language(filename:, sha:, git_repo:)
+    return nil if filename == 'LICENSE'
+
+    # First try to match on known extensions.
+    case filename
+    when /\.(pl|pm|t|cgi|pod|run)$/i
+      return 'Perl'
+    when /\.rb$/
+      return 'Ruby'
+    when /\.md$/
+      return 'Markdown'
+    when /\.json$/
+      return 'JSON'
+    when /\.(yml|yaml)$/
+      return 'YAML'
+    when /\.?(perlcriticrc|githooksrc|ini|editorconfig|gitconfig)$/
+      return 'INI'
+    when /\.css$/
+      return 'CSS'
+    when /\.(tt2|html)$/
+      return 'HTML'
+    when /\.sql$/
+      return 'SQL'
+    when /\.py$/
+      return 'Python'
+    when /\.js$/
+      return 'JavaScript'
+    when /\.c$/
+      return 'C'
+    when /\.sh$/
+      return 'bash'
+    when /(bash|bash_\w+)$/
+      return 'bash'
+    when /\.?(SKIP|gitignore|txt|csv|vim|gitmodules|gitattributes|jshintrc|gperf|vimrc|psqlrc|inputrc|screenrc)$/
+      return 'Text'
+    when /^(README|MANIFEST|Changes|Gemfile|Gemfile.lock)$/
+      return 'Text'
+    end
+
+    # Next, retrieve the file content and infer from that.
+    begin
+      content = git_repo.show(sha, filename)
+    rescue
+      pp "#{$!}"
+    end
+    return nil if content == nil || content == ''
+
+    first_line = content.split(/\n/)[0] || ''
+    case first_line
+    when /perl$/
+      return 'Perl'
+    end
+
+    # Fall back on the extension in last resort.
+    extension = /\.([^\.]+)$/.match(filename)
+    return filename if extension.nil?
+    return nil if extension[0] == 'lock'
+    return extension[0]
+  end
+
+  # Public: Parse the git logs for a repo.
+  #
+  # repo - A git repo object corresponding to the underlying repo.
+  #
+  # This method adds the metadata extracted for this repo to the instance
+  # variables collecting commit metadata.
+  #
+  def parse_repo(repo:)
+    git_repo = Git.open(repo, :log => @logger)
+
+    # Note: override the default of 30 for count(), nil gives the whole git log
+    # history.
+    git_repo.log(count = nil).each do |commit|
+      # Only include the authors specified on the command line.
+      next if !@author.include?(commit.author.email)
+
+      # Parse diff and analyze patches to detect language.
+      diff = commit.diff_parent.to_s
+      diff.encode!('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
+
+      patches = GitDiffParser.parse(diff)
+      patches.each do |patch|
+        body = patch.instance_variable_get :@body
+        language = self.class.determine_language(filename: patch.file, sha: commit.sha, git_repo: git_repo)
+        next if language == nil
+        @lines_by_language[language] ||=
+        {
+          'added'   => 0,
+          'deleted' => 0
+        }
+
+        body.split(/\n/).each do |content|
+          if (/^[+-]/.match(content) && !/^[+-]\s+$/.match(content))
+            if (/^\+/.match(content))
+              @lines_by_language[language]['added'] += 1
+            elsif (/^\-/.match(content))
+              @lines_by_language[language]['deleted'] += 1
+            end
+          end
+        end
+      end
+
+      # Add to stats for monthly commit count.
+      # Note: months are zero-padded to allow easy sorting, even if it's more
+      # work for formatting later on.
+      @monthly_commits[commit.date.strftime("%Y-%m")] += 1
+
+      # Add to stats for total commits count.
+      @total_commits += 1
+    end
+  end
+
+  # Public: Get a range of months from the earliest commit to the latest.
+  #
+  # Returns an array of "YYYY-MM" strings.
+  #
+  def get_month_scale()
+    month_scale = []
+    commits_start = @monthly_commits.keys.sort.first.split('-').map { |x| x.to_i }
+    commits_end = @monthly_commits.keys.sort.last.split('-').map { |x| x.to_i }
+    commits_start[0].upto(commits_end[0]) do |year|
+      1.upto(12) do |month|
+        next if month < commits_start[1] && year == commits_start[0]
+        next if month > commits_end[1] && year == commits_end[0]
+        month_scale << [year, month]
+      end
+    end
+
+    return month_scale
+  end
+
+  # Public: Generate a JSON representation of the parsed data.
+  #
+  # Returns: a JSON string.
+  #
+  def to_json()
+    formatted_monthly_commits = []
+    month_names = Date::ABBR_MONTHNAMES
+    self.get_month_scale.each do |frame|
+      display_key = month_names[frame[1]] + '-' + frame[0].to_s
+      data_key = sprintf('%s-%02d', frame[0], frame[1])
+      count = @monthly_commits[data_key].to_s
+      formatted_monthly_commits << { :month => display_key, :commits => count.to_s }
+    end
+
+    return JSON.pretty_generate(
+      {
+        :monthly_commits   => formatted_monthly_commits,
+        :total_commits     => @total_commits,
+        :lines_by_language => @lines_by_language,
+      }
+    )
+  end
+end

metadata

@@ -0,0 +1,47 @@
+--- !ruby/object:Gem::Specification
+name: git-commits-analyzer
+version: !ruby/object:Gem::Version
+  version: 0.1.0
+platform: ruby
+authors:
+- Guillaume Aubert
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2016-03-12 00:00:00.000000000 Z
+dependencies: []
+description: Parse git repos and collect commit statistics/data for a given author.
+email: aubertg@cpan.org
+executables:
+- analyze_commits
+extensions: []
+extra_rdoc_files: []
+files:
+- bin/analyze_commits
+- lib/git-commits-analyzer.rb
+- lib/git-commits-analyzer/utils.rb
+homepage: http://rubygems.org/gems/
+licenses:
+- GPLv3
+metadata: {}
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 2.4.6
+signing_key: 
+specification_version: 4
+summary: Analyze git commits
+test_files: []