git_ownership_insights 0.1.2 → 0.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b6bcceedce35270688443f386c76ec576b007cfd91c53cbda26b4b1166d3bcdb
-  data.tar.gz: c9195552bc37379e3f4cc2277604534d575c2eda7a3f6bf75b78fd0ac906dda0
+  metadata.gz: 0317f67423ade18e56b854a07912a388bece13db418ab72dd8a678ee18c802a5
+  data.tar.gz: f845947afad20fde877e4ed2568934a8bee8299d14d877a352386b5a899928d8
 SHA512:
-  metadata.gz: 0ace817610903ddcf63dafd6d54f365c83129ee24990146259db63f0e6fa959e83e9cdc2b2fb146333ba0f01855cec872656582b8f7a5d964535f652fd64d626
-  data.tar.gz: 929700dcbd1815ac79d0b7b98caba6f41209709c4c29341188c1784fa0083f3d60e21e5263bbf31f3f8f01c60ba59b3aeeef129831942567c8c77ab926977a1a
+  metadata.gz: d3a3d4aedbf40a3928dadd38ab0442382828fe0a3a0712fbf6e15978105ccad2fcbac8102318bd6b70363e779036ff7da7de06e8d55b525858a4b0bf88dbb266
+  data.tar.gz: b67f652e9e3f9676fa6455dc452a04700a48127aef953cb0d99ced1a722e3ad35dbed035fe37435561a349e72cd36e2e5881947f4b5128f2a29a334324144284

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    git_ownership_insights (0.1.1)
+    git_ownership_insights (0.1.3)
       date
       pry
 

data/bin/git_ownership_insights

@@ -3,10 +3,11 @@
 
 require 'date'
 require 'optparse'
+require 'pry'
 
 options = {}
 OptionParser.new do |opts|
-  opts.banner = "Usage: git_ownership_insights [options]"
+  opts.banner = 'Usage: git_ownership_insights [options]'
 
   opts.on('--debug', 'Enable debug mode') do
     options[:debug] = true
@@ -20,7 +21,8 @@ OptionParser.new do |opts|
     options[:steps] = steps
   end
 
-  opts.on('--duration_in_days STRING', 'Number of days to aggregate the changes for [default: 30]') do |duration_in_days|
+  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
 
@@ -32,12 +34,24 @@ OptionParser.new do |opts|
     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|
+    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|
+    options[:top_touched_files] = top_touched_files
+  end
+
+  opts.on('--codeowners_path STRING', 'Path to CODEOWNERS file [default: .github/CODEOWNERS]') do |codeowners_path|
+    options[:codeowners_path] = codeowners_path
+  end
+
   opts.on('-h', '--help', 'Display this help message') do
     puts opts
-    puts <<-EXAMPLES
+    puts <<~EXAMPLES
 
-Examples:
-  git_ownership_insights --path src/test --exclusions WEB,RAILS --steps 2 --duration_in_days 90 --debug
+      Examples:
+        git_ownership_insights --path src/test --exclusions WEB,RAILS --steps 2 --duration_in_days 90 --debug
     EXAMPLES
     exit
   end
@@ -45,7 +59,36 @@ end.parse!
 
 EXCLUSIONS = options[:exclusions]&.split(',')
 REPO_PATH = options[:path] || '.'
-TEAM_REGEX = options[:team_regex] || "[A-Za-z]+"
+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"
+def read_codeowners_file
+  raise "CODEOWNERS file does not exist under #{CODEOWNERS_PATH}" unless File.exist?(CODEOWNERS_PATH)
+
+  codeowners = {}
+  File.readlines(CODEOWNERS_PATH).each do |line|
+    next if line.strip.empty? || line.start_with?('#') # Skip comments and empty lines
+
+    parts = line.split(/\s+/)
+    directory_pattern = parts[0]
+    owner = parts[1..].map { |o| o.start_with?('@') ? o[1..] : o }.join(' ') # Remove leading '@' from team names
+    codeowners[directory_pattern] = owner
+  end
+  codeowners
+end
+
+def find_owners(file_path, codeowners)
+  matching_patterns = codeowners.keys.select { |pattern| file_path.include?(pattern.sub(%r{^/+}, '').chomp('/')) }
+  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('/')) }
+  codeowners[best_match].split(' ')
+end
 
 def contribution_message(directory_path:, duration_in_days:, begin_time:, debug: nil, steps: nil)
   duration_in_days = duration_in_days.to_i
@@ -55,9 +98,10 @@ def contribution_message(directory_path:, duration_in_days:, begin_time:, debug:
   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.each do |file|
+  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_team_map = {}
+  uniq_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
 
@@ -72,7 +116,7 @@ def contribution_message(directory_path:, duration_in_days:, begin_time:, debug:
     all_teams << teams
     teams = teams.uniq
     files_changed_by_many_teams += 1 if teams.count > 1
-
+    file_team_map.merge!("#{filename}" => teams)
     puts "\n#{filename} [#{commit_count}]:#{teams}\n" if debug
   end
 
@@ -80,7 +124,44 @@ 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   Main contributors: #{contributors}\n\n"
+  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"
+
+  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
+
+  owners_data = Hash.new do |hash, key|
+    hash[key] = { directories: Hash.new do |h, k|
+                                 h[k] = { files: [] }
+                               end, total_count: 0 }
+  end
+
+  top_touched_files.each do |file, count|
+    owners = find_owners(file, codeowners)
+    owners.each do |owner|
+      owners_data[owner][:total_count] += count
+
+      dir_path = File.dirname(file)
+      owners_data[owner][:directories][dir_path][:files] << { name: File.basename(file), count: count }
+    end
+  end
+
+  # Sort owners_data by total count in descending order
+  sorted_owners_data = owners_data.sort_by { |_, data| -data[:total_count] }
+
+  # Take the last 5 elements
+  top_owners_data = sorted_owners_data.last(TOP_CONTRIBUTED_TEAMS.to_i)
+
+  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]]}"
+      end
+    end
+  end
 
   steps -= 1
 
@@ -91,7 +172,13 @@ def contribution_message(directory_path:, duration_in_days:, begin_time:, debug:
 end
 
 puts "\nDirectory: #{REPO_PATH}\n"
-puts "Excluded contributors: #{EXCLUSIONS}\n\n" if EXCLUSIONS
+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"
 
 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 = '0.1.2'
+  VERSION = '0.1.3'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: git_ownership_insights
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.3
 platform: ruby
 authors:
 - Serghei Moret
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-01-16 00:00:00.000000000 Z
+date: 2024-01-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: date