git-commits-analyzer 0.1.0 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e1c430c515c28a8a1a8f6dea0098f0590101095f
-  data.tar.gz: 01df65d27d647d9dff0207927d9d550a93111b0f
+  metadata.gz: 3235fa842b69838ddc164316ad8394c08f71b6a5
+  data.tar.gz: 6bb686c2baf86c75abc8d31f752c4968fd1931ed
 SHA512:
-  metadata.gz: 21b04677942df99acdef4d4d44cfc907fdeee7907a121c695e616c093d98e99af87ab242aa036009be2644fb629b05d61ebaaaf3c024c4f6fc8cdd5925ffa614
-  data.tar.gz: a647cccacdec435bd1d73f444b76157f363f671e6da468917fb7e637c8dc090e3c9c9fa7c7e4c5612afbc48e0d985119c5b5f609e2269a786a2f564446be07bd
+  metadata.gz: d7f5f16a6b4b26bcae93456c26d7df31de2fb3b6a35c74b5b268184f529c5303a81b6262a60a6da8d3075a9d901d8c13ef930285bb705978ee836de3ffb296a5
+  data.tar.gz: 82f66448cfe0081c0881cb302dee08c080938e7778a6762fb8119e39143cec4145161412de4c511a5489ca6e1b3f1b1baf64c53aa44eddf368b71fc5c0ab7939

data/bin/analyze_commits

@@ -23,29 +23,36 @@ 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 ""
+printf("Found %s repos to inspect.\n", repos.length.to_s)
+puts ''
 
 # Inspect git repos.
-puts "===== Inspecting repos ====="
-puts ""
-git_commits_analyzer = GitCommitsAnalyzer.new(logger: logger, author: options[:authors])
+puts '===== Inspecting repos ====='
+puts ''
+git_commits_analyzer = GitCommitsAnalyzer.new(
+  logger: logger,
+  author: options[:authors]
+)
 repos.sort.each do |repo|
-  puts "Inspecting repo " + repo
+  printf("Inspecting repo %s.\n", repo)
   git_commits_analyzer.parse_repo(repo: repo)
-  #break
+  # break
 end
-puts ""
+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
+printf(
+  "Found %s commits for author(s) %s.\n",
+  git_commits_analyzer.commits_total,
+  options[:authors].join(', ')
+)
+puts ''
+exit if git_commits_analyzer.commits_by_month.keys.empty?
 
 # Save data.
-puts "===== Save data ====="
-puts ""
-output_file = options[:output];
+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 ""
+puts ''

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

@@ -1,36 +1,44 @@
 require 'optparse'
 
+# Public: supporting functions for the command-line utility.
+#
+# Examples:
+#
+#   require 'git-commits-analyzer/utils'
+#   options = Utils.parse_command_line_options()
+#   repos = Utils.get_git_repos(path)
+#
 class Utils
   def self.parse_command_line_options()
     options = {}
     OptionParser.new do |opts|
-      opts.banner = "Usage: inspect_contributions.rb [options]"
+      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|
+      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|
+      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|
+      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
+      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, '--author' if options[:authors].empty?
     raise OptionParser::MissingArgument, '--output' if options[:output].nil?
     raise OptionParser::MissingArgument, '--path' if options[:path].nil?
 

data/lib/git-commits-analyzer.rb

@@ -11,15 +11,24 @@ require 'json'
 #
 class GitCommitsAnalyzer
   # Public: Returns a hash of commit numbers broken down by month.
-  attr_reader :monthly_commits
+  attr_reader :commits_by_month
 
   # Public: Returns the total number of commits belonging to the author
   # specified.
-  attr_reader :total_commits
+  attr_reader :commits_total
 
   # Public: Returns the number of lines added/removed broken down by language.
   attr_reader :lines_by_language
 
+  # Public: Returns the tally of commits broken down by hour of the day.
+  attr_reader :commit_hours
+
+  # Public: Returns the tally of commits broken down by day.
+  attr_reader :commit_days
+
+  # Public: Returns the tally of commits broken down by weekday and hour.
+  attr_reader :commit_weekdays_hours
+
   # Public: Initialize new GitParser object.
   #
   # logger - A logger object to display git errors/warnings.
@@ -28,10 +37,20 @@ class GitCommitsAnalyzer
   def initialize(logger:, author:)
     @logger = logger
     @author = author
-    @monthly_commits = {}
-    @monthly_commits.default = 0
-    @total_commits = 0
+    @commits_by_month = {}
+    @commits_by_month.default = 0
+    @commits_total = 0
     @lines_by_language = {}
+    @commit_hours = 0.upto(23).map{ |x| [x, 0] }.to_h
+    @commit_days = {}
+    @commit_days.default = 0
+    @commit_weekdays_hours = {}
+    ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun'].each do |weekday|
+      @commit_weekdays_hours[weekday] = {}
+      0.upto(23).each do |hour|
+        @commit_weekdays_hours[weekday][hour] = 0
+      end
+    end
   end
 
   # Public: Determine the type of a file at the given revision of a repo.
@@ -49,7 +68,9 @@ class GitCommitsAnalyzer
     case filename
     when /\.(pl|pm|t|cgi|pod|run)$/i
       return 'Perl'
-    when /\.rb$/
+    when /\.(?:rb|gemspec)$/
+      return 'Ruby'
+    when /(?:\/|^)Rakefile$/
       return 'Ruby'
     when /\.md$/
       return 'Markdown'
@@ -75,9 +96,15 @@ class GitCommitsAnalyzer
       return 'bash'
     when /(bash|bash_\w+)$/
       return 'bash'
-    when /\.?(SKIP|gitignore|txt|csv|vim|gitmodules|gitattributes|jshintrc|gperf|vimrc|psqlrc|inputrc|screenrc)$/
+    when /\.?(SKIP|gitignore|txt|csv|vim|gitmodules|gitattributes|jshintrc|gperf|vimrc|psqlrc|inputrc|screenrc|curlrc|wgetrc|selected_editor|dmrc|netrc)$/
       return 'Text'
-    when /^(README|MANIFEST|Changes|Gemfile|Gemfile.lock)$/
+    when /(?:\/|^)(?:LICENSE|LICENSE-\w+)$/
+      return nil
+    when /\.(?:0|1|VimballRecord)$/
+      return nil
+    when /^vim\/doc\/tags$/
+      return nil
+    when /(?:\/|^)(?:README|MANIFEST|Changes|Gemfile|Gemfile.lock|CHANGELOG)$/
       return 'Text'
     end
 
@@ -93,6 +120,10 @@ class GitCommitsAnalyzer
     case first_line
     when /perl$/
       return 'Perl'
+    when /ruby$/
+      return 'Ruby'
+    when /^\#!\/usr\/bin\/bash$/
+      return 'Ruby'
     end
 
     # Fall back on the extension in last resort.
@@ -110,23 +141,42 @@ class GitCommitsAnalyzer
   # variables collecting commit metadata.
   #
   def parse_repo(repo:)
-    git_repo = Git.open(repo, :log => @logger)
+    git_repo = Git.open(repo, log: @logger)
 
+    #return if repo != '/git_backups/github-guillaumeaubert/dot-files'
     # 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 commit date and update the corresponding stats.
+      commit_datetime = DateTime.parse(commit.author.date.to_s)
+      commit_hour = commit_datetime.hour
+      @commit_hours[commit_hour] += 1
+      commit_day = commit_datetime.strftime('%Y-%m-%d')
+      @commit_days[commit_day] += 1
+      commit_weekday = commit_datetime.strftime('%a')
+      @commit_weekdays_hours[commit_weekday][commit_hour] += 1
+
       # Parse diff and analyze patches to detect language.
-      diff = commit.diff_parent.to_s
+      diff = git_repo.show(commit.sha)
       diff.encode!('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
 
+      file_properties = git_repo.ls_tree(['-r', commit.sha])
+
       patches = GitDiffParser.parse(diff)
       patches.each do |patch|
+        # Skip submodules.
+        next if file_properties['commit'].has_key?(patch.file);
+
+        # Skip symlinks.
+        next if file_properties['blob'].has_key?(patch.file) &&
+          (file_properties['blob'][patch.file][:mode] == '120000')
+
         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
+        next if language.nil?
         @lines_by_language[language] ||=
         {
           'added'   => 0,
@@ -147,10 +197,10 @@ class GitCommitsAnalyzer
       # 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
+      @commits_by_month[commit.date.strftime("%Y-%m")] += 1
 
       # Add to stats for total commits count.
-      @total_commits += 1
+      @commits_total += 1
     end
   end
 
@@ -160,8 +210,8 @@ class GitCommitsAnalyzer
   #
   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 = @commits_by_month.keys.sort.first.split('-').map { |x| x.to_i }
+    commits_end = @commits_by_month.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]
@@ -178,20 +228,23 @@ class GitCommitsAnalyzer
   # Returns: a JSON string.
   #
   def to_json()
-    formatted_monthly_commits = []
+    formatted_commits_by_month = []
     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 }
+      count = @commits_by_month[data_key].to_s
+      formatted_commits_by_month << { 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,
+        commits_total: @commits_total,
+        commits_by_month: formatted_commits_by_month,
+        commits_by_hour: @commit_hours,
+        commits_by_day: @commit_days,
+        commit_by_weekday_hour: @commit_weekdays_hours,
+        lines_by_language: @lines_by_language
       }
     )
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: git-commits-analyzer
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 1.0.0
 platform: ruby
 authors:
 - Guillaume Aubert
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-03-12 00:00:00.000000000 Z
+date: 2016-03-20 00:00:00.000000000 Z
 dependencies: []
 description: Parse git repos and collect commit statistics/data for a given author.
 email: aubertg@cpan.org