git_ownership_insights 0.1.4 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f2cc151f3f515c7ec241175034e18e611a1eedc4df8c322c383e15390f483c96
-  data.tar.gz: 59774d973a6d0aa982f0a8bf426939c0c59caeaa9305c624f19e6cd42b7e92c8
+  metadata.gz: 4bca095a91102d5a60161e08921ba7af125dfc7a5ee528bd527ab57720814d62
+  data.tar.gz: 39e9afe47d9af3c4e1b49a084001c52bef8b55f8f8ee9057bd264c000e466fba
 SHA512:
-  metadata.gz: f146f73f54759ea4e568961158c5b8f938e0f471d0fbc7cfdc84faf6b7597a3352d073e60b8417c672f1dc161c09b0d3dc0282e9e1a459daf7104a6b4bf50c94
-  data.tar.gz: 80779de18c3db314f0ca78bc7b1c07fb8a2290b9720fa2bc2cac7d85ab6965c19f06496c4c1e19ae08bd673d80d9731bcae600d329ef77775924ad542e0b4a98
+  metadata.gz: 207a6053d21c89408c33ca31068ae8d2fb2a88b7bf4bd63d5cdbdafbcbffc6e3df05bee91142adb88030cb2befe7cbfe48258a9a70fd4a952cfa0f3ef2601aed
+  data.tar.gz: 10f59df955f3a5321cf046761303befc0260490069a81cbe40d505ffa82a19dd03d2b03896f7cdd882a39b3a29cc686c6cf911ff1c196abaceb7607ddacaab59

data/Gemfile.lock

@@ -2,6 +2,7 @@ PATH
   remote: .
   specs:
     git_ownership_insights (0.1.4)
+      awesome_print
       date
       pry
 
@@ -9,6 +10,7 @@ GEM
   remote: https://rubygems.org/
   specs:
     ast (2.4.2)
+    awesome_print (1.9.2)
     coderay (1.1.3)
     date (3.3.4)
     diff-lcs (1.5.0)

data/bin/git_ownership_insights

@@ -13,15 +13,31 @@ OptionParser.new do |opts|
     options[:debug] = true
   end
 
-  opts.on('--exclusions STRING', 'Comma-delimited list of exclusions [example: WEB,RAILS,MOBILE]') do |exclusions|
+  opts.on('--ci', 'Do not print the info messages for better CI text parsing [default: false]') do
+    options[:ci] = true
+  end
+
+  opts.on('--codeowners', 'Print CODEOWNERS info [default: false]') do
+    options[:codeowners] = true
+  end
+
+  opts.on('--hotspot-files', 'Print the found hotspot files (big files touched by many) [default: false]') do
+    options[:hotspot_files] = true
+  end
+
+  opts.on('--excluded-contributors STRING', 'Comma-delimited list of excluded contributors [example: WEB,RAILS,MOBILE]') do |exclusions|
     options[:exclusions] = exclusions
   end
 
+  opts.on('--excluded-files STRING', 'Comma-delimited list of excluded files [example: ViewController,AppDelegate.swift]') do |excluded_files|
+    options[:excluded_files] = excluded_files
+  end
+
   opts.on('--steps STRING', 'Number of steps the script will go into the past [default: 1]') do |steps|
     options[:steps] = steps
   end
 
-  opts.on('--duration_in_days STRING',
+  opts.on('--duration-in-days STRING',
           'Number of days to aggregate the changes for [default: 30]') do |duration_in_days|
     options[:duration_in_days] = duration_in_days
   end
@@ -30,28 +46,40 @@ OptionParser.new do |opts|
     options[:path] = path
   end
 
-  opts.on('--team_regex STRING', 'Regex that will identify the team name [default: "[A-Za-z]+"]') do |team_regex|
+  opts.on('--team-regex STRING', 'Regex that will identify the team name [default: "[A-Za-z]+"]') do |team_regex|
     options[:team_regex] = team_regex
   end
 
-  opts.on('--top_contributing_team STRING', 'Limit of top contributed to the directory teams in codeownership data [default: 5]') do |top_contributing_team|
+  opts.on('--top-contributing-team STRING', 'Limit of top contributed to the directory teams in codeownership data [default: 5]') do |top_contributing_team|
     options[:top_contributing_team] = top_contributing_team
   end
 
-  opts.on('--top_touched_files STRING', 'Limit of top touched files by individual contributors in codeownership data [default: 5]') do |top_touched_files|
+  opts.on('--top-touched-files STRING', 'Limit of top touched files by individual contributors in codeownership data [default: 5]') do |top_touched_files|
     options[:top_touched_files] = top_touched_files
   end
 
-  opts.on('--codeowners_path STRING', 'Path to CODEOWNERS file [default: .github/CODEOWNERS]') do |codeowners_path|
+  opts.on('--codeowners-path STRING', 'Path to CODEOWNERS file [default: .github/CODEOWNERS]') do |codeowners_path|
     options[:codeowners_path] = codeowners_path
   end
 
+  opts.on('--big-file-size STRING', 'The amount of lines in the file to be considered big [default: 250]') do |big_file_size|
+    options[:big_file_size] = big_file_size
+  end
+
+  opts.on('--default-branch STRING', 'The default branch to pull and run metrics for [default: master]') do |default_branch|
+    options[:default_branch] = default_branch
+  end
+
+  opts.on('--code-extensions STRING', 'The file extensions that consider to be code [default: ".kt, .swift"]') do |code_extension|
+    options[:code_extension] = code_extension
+  end
+
   opts.on('-h', '--help', 'Display this help message') do
     puts opts
     puts <<~EXAMPLES
 
       Examples:
-        git_ownership_insights --path src/test --exclusions WEB,RAILS --steps 2 --duration_in_days 90 --debug
+        git_ownership_insights --path src/test --exclusions WEB,RAILS --steps 2 --duration-in-days 90 --hotspot-files --debug
     EXAMPLES
     exit
   end
@@ -63,6 +91,18 @@ TEAM_REGEX = options[:team_regex] || '[A-Za-z]+'
 TOP_TOUCHED_FILES = options[:top_touched_files] || 5
 TOP_CONTRIBUTED_TEAMS = options[:top_contributing_team] || 5
 CODEOWNERS_PATH =  options[:codeowners_path] || ".github/CODEOWNERS"
+BIG_FILE_SIZE = options[:big_file_size] || 250
+CI = options[:ci] || false
+DEFAULT_BRANCH = options[:default_branch] || 'master'
+CODEOWNERS = options[:codeowners] || false
+HOTSPOT = options[:hotspot_files] || false
+CODE_EXTENSIONS = options[:code_extension] ? options[:code_extension].split : ['.swift', '.kt']
+EXCLUDED_FILES = options[:excluded_files]
+
+def true?(obj)
+  obj.to_s.downcase == "true"
+end
+
 def read_codeowners_file
   raise "CODEOWNERS file does not exist under #{CODEOWNERS_PATH}" unless File.exist?(CODEOWNERS_PATH)
 
@@ -79,17 +119,50 @@ def read_codeowners_file
 end
 
 def find_owners(file_path, codeowners)
-  matching_patterns = codeowners.keys.select { |pattern| file_path.include?(pattern.sub(%r{^/+}, '').chomp('/')) }
+  matching_patterns = codeowners.keys.select do |pattern|
+    pattern_regex = Regexp.new("^#{Regexp.escape(pattern.sub(%r{^/+}, '').chomp('/')).gsub('\*', '.*').gsub('**', '.*?')}")
+    file_path =~ pattern_regex
+  end
+
   return ['unknown'] if matching_patterns.empty?
 
   # Sort patterns by length in descending order
   sorted_patterns = matching_patterns.sort_by(&:length).reverse
 
   # Find the most specific matching pattern
-  best_match = sorted_patterns.find { |pattern| file_path.include?(pattern.sub(%r{^/+}, '').chomp('/')) }
+  best_match = sorted_patterns.find do |pattern|
+    pattern_regex = Regexp.new("^#{Regexp.escape(pattern.sub(%r{^/+}, '').chomp('/')).gsub('\*', '.*').gsub('**', '.*?')}")
+    file_path =~ pattern_regex
+  end
+
   codeowners[best_match].split(' ')
 end
 
+def count_big_files(directory_path, size: BIG_FILE_SIZE)
+  # Get a list of all files in the specified directory
+  files = Dir.glob(File.join(directory_path, '**', '*')).select { |file| File.file?(file) }
+
+  code_files = files.select {|f|
+    extension = File.extname(f)
+    valid_extensions = ['.swift', '.kt']
+    valid_extensions.include?(extension)
+  }
+
+  # Initialize a counter for files that meet the criteria
+  count = 0
+
+  # Iterate through each file and check the line count
+  code_files.each do |file|
+    lines_count = File.foreach(file).reject { |line| line.match(/^\s*(\/\/|\/\*.*\*\/|\s*$)/) }.count
+
+    if lines_count > size
+      count += 1
+    end
+  end
+
+  puts "   Total number of files longer than #{size} lines: #{count}"
+end
+
 def contribution_message(directory_path:, duration_in_days:, begin_time:, debug: nil, steps: nil)
   duration_in_days = duration_in_days.to_i
   all_teams = []
@@ -97,26 +170,40 @@ def contribution_message(directory_path:, duration_in_days:, begin_time:, debug:
   total_changes = 0
   start_date = begin_time.to_time.to_i - duration_in_days * 86_400
   end_date = begin_time.to_time.to_i
-  file_count = `git ls-tree -r --name-only $(git rev-list -1 --before="#{end_date}" HEAD) -- "#{directory_path}" | wc -l`.to_i
-  files_with_changes = `git log --name-only --pretty=format:"" --since="#{start_date}" --until="#{end_date}" "#{directory_path}"`.split.sort
-  uniq_files_with_changes = files_with_changes.uniq
+  file_count = `git ls-tree -r --name-only $(git rev-list -1 --since="#{start_date}" --until="#{end_date}" HEAD) -- "#{directory_path}" | wc -l`.to_i
+  all_files_with_changes = `git log --name-only --pretty=format:"" --since="#{start_date}" --until="#{end_date}" "#{directory_path}"`.split.sort
+  excluded_patterns = EXCLUDED_FILES.split(',')
+
+  code_files_with_changes = all_files_with_changes.select {|f|
+    extension = File.extname(f)
+    valid_extensions = CODE_EXTENSIONS
+    valid_extensions.include?(extension)
+  }.reject do |file|
+    excluded_patterns.any? { |pattern| file.include?(pattern) }
+  end
+
+  uniq_code_files_with_changes = code_files_with_changes.uniq
+
   file_team_map = {}
-  uniq_files_with_changes.each do |file|
+  uniq_code_files_with_changes.each do |file|
     filename = File.basename(file)
     commit_count = `git log --since="#{start_date}" --until="#{end_date}" --follow -- "#{file}" | grep -c '^commit'`.to_i
 
-    next unless commit_count.positive?
-
     # Get the log of the file in the given duration
     git_log = `git log --pretty=format:"%s" --since="#{start_date}" --until="#{end_date}" --follow -- "#{file}"`.split("\n")
     teams = git_log.map do |team|
       team.match(/#{TEAM_REGEX}/)[0].upcase
     end.reject { |e| EXCLUSIONS&.include?(e) }
-    total_changes += teams.count
+
+    total_changes += commit_count
     all_teams << teams
     teams = teams.uniq
-    files_changed_by_many_teams += 1 if teams.count > 1
-    file_team_map.merge!("#{filename}" => teams)
+
+    if teams.count > 1
+      files_changed_by_many_teams += 1
+      file_team_map.merge!("#{file}" => [teams, commit_count])
+    end
+
     puts "\n#{filename} [#{commit_count}]:#{teams}\n" if debug
   end
 
@@ -124,61 +211,99 @@ def contribution_message(directory_path:, duration_in_days:, begin_time:, debug:
   sorted_occurrences = occurrences.sort_by { |element, count| [-count, element] }
   contributors = Hash[sorted_occurrences]
 
-  puts "Timeframe: #{(begin_time - duration_in_days).strftime('%Y-%m-%d')} to #{begin_time.strftime('%Y-%m-%d')}\n   Files with a single contributor: #{(100 - ((files_changed_by_many_teams.to_f / files_with_changes.count) * 100)).round(2)}%\n   Amount of commits: #{total_changes}\n   Total files changed: #{files_with_changes.count}\n   Total files in the folder: #{file_count}\n   Contributors: #{contributors}\n"
+  churn_count = file_team_map.values.map { |value| value[1] }.sum
+  hotspot_changes_percentage = (churn_count.to_f / total_changes.to_f)*100
 
-  touched_files = files_with_changes.flatten.compact.tally
-  top_touched_files = touched_files.sort_by { |element, count| [-count, element] }.take(TOP_TOUCHED_FILES.to_i)
-  codeowners = read_codeowners_file
+  puts "Timeframe: #{(begin_time - duration_in_days).strftime('%Y-%m-%d')} to #{begin_time.strftime('%Y-%m-%d')}\n   Code files with a single contributor: #{(100 - ((files_changed_by_many_teams.to_f / code_files_with_changes.count.to_f) * 100)).round(2)}%\n   Hotspot code changes: #{churn_count} (#{hotspot_changes_percentage.round(2)}%)\n   Amount of code changes: #{total_changes}\n   Total files changed: #{code_files_with_changes.count}\n   Total files in the folder: #{file_count}\n   Contributors: #{contributors}\n"
 
-  owners_data = Hash.new do |hash, key|
-    hash[key] = { directories: Hash.new do |h, k|
-                                 h[k] = { files: [] }
-                               end, total_count: 0 }
-  end
+  # Filter files based on extension and size
+  filtered_files = file_team_map.select do |file_path|
+    next unless File.exist?(file_path)
 
-  top_touched_files.each do |file, count|
-    owners = find_owners(file, codeowners)
-    owners.each do |owner|
-      owners_data[owner][:total_count] += count
+    # Check if the file size is more than BIG_FILE_SIZE lines (excluding empty and commented lines)
+    File.foreach(file_path).reject { |line| line.match(/^\s*(\/\/|\/\*.*\*\/|\s*$)/) }.count > BIG_FILE_SIZE.to_i
+  end
 
-      dir_path = File.dirname(file)
-      owners_data[owner][:directories][dir_path][:files] << { name: File.basename(file), count: count }
+  filtered_top_touched_files = filtered_files.sort_by { |element, count| [-count.last, element] }
+  count_big_files(directory_path)
+  puts "   Total files longer than #{BIG_FILE_SIZE} lines with multiple contributors: #{filtered_top_touched_files.count}\n"
+  if HOTSPOT
+    filtered_top_touched_files.each do |line|
+      puts "      #{line.first.gsub(directory_path, '')} Contributors: #{line.last.first} Commits: #{line.last.last}"
     end
   end
+  puts "\n\n"
 
-  # Sort owners_data by total count in descending order
-  sorted_owners_data = owners_data.sort_by { |_, data| -data[:total_count] }
+  if CODEOWNERS
+    codeowners = read_codeowners_file
 
-  # Take the last 5 elements
-  top_owners_data = sorted_owners_data.last(TOP_CONTRIBUTED_TEAMS.to_i)
+    owners_data = Hash.new do |hash, key|
+      hash[key] = { directories: Hash.new do |h, k|
+                                   h[k] = { files: [] }
+                                 end, churn_count: 0 }
+    end
+
+    file_team_map.each do |file, count|
+      owners = find_owners(file, codeowners)
+      owners.each do |owner|
+        owners_data[owner][:churn_count] += count.last
 
-  puts '   Codeownership data:'
-  top_owners_data.each do |owner, data|
-    puts "      #{owner.split('/').last}:\n         Total Count: #{data[:total_count]}"
-    data[:directories].each do |dir, dir_data|
-      puts "            Directory: #{dir}\n               Top files:"
-      dir_data[:files].each do |file_data|
-        puts "                  #{File.basename(file_data[:name])} - #{file_data[:count]} #{file_team_map[file_data[:name]].empty? ? "[ Excluded contributor ]" : file_team_map[file_data[:name]]}"
+        dir_path = File.dirname(file)
+        owners_data[owner][:directories][dir_path][:files] << { name: File.basename(file), count: count }
       end
     end
-  end
 
+    # Sort owners_data by total count in descending order
+    sorted_owners_data = owners_data.sort_by { |_, data| -data[:churn_count] }
+
+    # Take the last 5 elements
+    top_owners_data = sorted_owners_data.last(TOP_CONTRIBUTED_TEAMS.to_i)
+
+    converted_team_map = file_team_map.transform_keys { |key| File.basename(key) }
+
+    puts '   Codeownership data:'
+    top_owners_data.each do |owner, data|
+      puts "      #{owner.split('/').last}:\n         Total Count: #{data[:churn_count]}"
+      data[:directories].each do |dir, dir_data|
+        puts "            Directory: #{dir}\n               Top files:"
+        dir_data[:files].each do |file_data|
+          next if converted_team_map[File.basename(file_data[:name])].nil?
+
+          contributors = converted_team_map[file_data[:name]]&.first&.empty? ? [ "Excluded contributor" ] : converted_team_map[file_data[:name]].first
+          puts "                  #{File.basename(file_data[:name])} - #{file_data[:count].last} #{contributors}}"
+        end
+      end
+    end
+  end
   steps -= 1
 
   return unless steps.positive?
 
+  system("git checkout `git rev-list -1 --before='#{(begin_time - duration_in_days).strftime("%B %d %Y")}' HEAD`", [ :out, :err ] => File::NULL)
   contribution_message(duration_in_days: duration_in_days, directory_path: directory_path,
                        begin_time: begin_time - duration_in_days, steps: steps, debug: debug)
 end
 
-puts "\nDirectory: #{REPO_PATH}\n"
-puts "Time period that data is aggregated by: #{options[:duration_in_days]} days"
-puts "Steps to jump in the past: #{options[:steps]}"
-puts "Limit of the teams shown in codeownership data: #{TOP_CONTRIBUTED_TEAMS}"
-puts "Limit of the files shown in codeownership data: #{TOP_TOUCHED_FILES}"
-puts "Regex to detect the teams identifiers: #{TEAM_REGEX}"
-puts "Excluded contributors: #{EXCLUSIONS}\n" if EXCLUSIONS
-puts "Debug mode is: #{options[:debug] ? 'on' : 'off'}\n\n"
+unless CI
+  puts "\nDirectory: #{REPO_PATH}\n"
+  puts "Time period that data is aggregated by: #{options[:duration_in_days]} days"
+  puts "Steps to jump in the past: #{options[:steps].to_i}"
+  puts "Runs against: #{DEFAULT_BRANCH}"
+  puts "Code extensions: #{CODE_EXTENSIONS}"
+  puts "Regex to detect the teams identifiers: #{TEAM_REGEX}"
+  puts "Excluded contributors: #{EXCLUSIONS}\n" if EXCLUSIONS
+  puts "Excluded file patterns: #{EXCLUDED_FILES.split(',')}\n" if EXCLUDED_FILES
+  puts "Lines of code limit (big files) for the hotspot calculation: #{BIG_FILE_SIZE}"
+  puts "Hotspot detailed output is: #{options[:hotspot_files] ? 'on' : 'off'}\n"
+  puts "CODEOWNERS output is: #{options[:codeowners] ? 'on' : 'off'}\n"
+  puts "Limit of the teams shown in codeownership data: #{TOP_CONTRIBUTED_TEAMS}"
+  puts "Limit of the files shown in codeownership data: #{TOP_TOUCHED_FILES}"
+  puts "CI mode is: #{options[:ci] ? 'on' : 'off'}\n"
+  puts "Debug mode is: #{options[:debug] ? 'on' : 'off'}\n\n"
+end
+
+system("git checkout #{DEFAULT_BRANCH}", [ :out ] => File::NULL)
+system("git pull", [ :out ] => File::NULL)
 
 contribution_message(duration_in_days: options[:duration_in_days] || 30, directory_path: REPO_PATH,
                      begin_time: DateTime.now, steps: options[:steps].to_i, debug: options[:debug])

data/lib/git_ownership_insights/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module GitOwnershipInsights
-  VERSION = '0.1.4'
+  VERSION = '1.0.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: git_ownership_insights
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 1.0.0
 platform: ruby
 authors:
 - Serghei Moret
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-01-17 00:00:00.000000000 Z
+date: 2024-01-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: date