git_statistics 0.2.0 → 0.3.0

data/lib/git_statistics/commits.rb

@@ -12,38 +12,28 @@ module GitStatistics
       @totals[:languages] = {}
     end
 
-    def identify_authors
-      self.each do |key,value|
-        if not @author_list.include?(value[:author])
-          @author_list << value[:author]
-        end
-      end
-    end
-
-    def identify_authors_email
-      self.each do |key,value|
-        if not @author_list.include?(value[:author_email])
-          @author_list << value[:author_email]
-        end
-      end
-    end
-
-    def author_top_n_type(type, n=0)
-      n = 0 if n < 0
-      return nil if @stats == nil || !@stats.first[1].has_key?(type)
-      return @stats.sorted_hash {|a,b| b[1][type.to_sym] <=> a[1][type]}.to_a[0..n-1]
+    def author_top_n_type(type, top_n=0)
+      top_n = 0 if top_n < 0
+      return nil if @stats.size == 0
+      return nil if !@stats.first[1].has_key?(type)
+      return Hash[*@stats.sorted_hash {|a,b| b[1][type.to_sym] <=> a[1][type]}.to_a[0..top_n-1].flatten]
     end
 
     def calculate_statistics(email, merge)
 
+      # Check that there are statistics to calculate
+      if self.size == 0
+        puts "Cannot calculate statistics as there is no data"
+        return
+      end
+
       # Identify authors and author type
       if email
-        identify_authors_email
         type = :author_email
       else
-        identify_authors
         type = :author
       end
+      @author_list = Utilities.unique_data_in_hash(self, type)
 
       # Initialize the stats hash
       @author_list.each do |author|
@@ -53,7 +43,7 @@ module GitStatistics
 
       # Collect the stats from each commit
       self.each do |key,value|
-        if not merge and value[:merge]
+        if !merge && value[:merge]
           next
         else
 
@@ -88,11 +78,15 @@ module GitStatistics
       if data[:languages][file[:language].to_sym] == nil
         data[:languages][file[:language].to_sym] = Hash.new(0)
       end
+
       data[:languages][file[:language].to_sym][:additions] += file[:additions]
       data[:languages][file[:language].to_sym][:deletions] += file[:deletions]
-      if file[:status] != nil || file[:status] == "submodule"
+
+      if file[:status] != nil
         data[:languages][file[:language].to_sym][file[:status].to_sym] += 1
       end
+
+      return data
     end
 
     def add_commit_stats(data, commit)
@@ -105,6 +99,7 @@ module GitStatistics
       data[:delete] += commit[:delete] if commit[:delete] != nil
       data[:rename] += commit[:rename] if commit[:rename] != nil
       data[:copy] += commit[:copy] if commit[:copy] != nil
+      return data
     end
 
     def load(file)

data/lib/git_statistics/initialize.rb

@@ -2,6 +2,7 @@ require 'json'
 require 'trollop'
 require 'grit'
 require 'linguist'
+require 'os'
 require 'pathname'
 
 # Custom Blob for Grit to enable Linguist

data/lib/git_statistics/results.rb

@@ -0,0 +1,96 @@
+module GitStatistics
+  class Results
+
+    attr_accessor :commits
+
+    def initialize(commits)
+      @commits = commits
+    end
+
+    def prepare_result_summary(sort, email, top_n=0)
+      # Default to a 0 if given a negative number to display
+      top_n = 0 if top_n < 0
+
+      # Acquire data based on sorty type and top # to show
+      data = @commits.author_top_n_type(sort.to_sym, top_n)
+      if data == nil
+        raise "Parameter for --sort is not valid"
+      end
+
+      # Create config
+      config = {:data => data,
+                :author_length => Utilities.find_longest_length(data, 17),
+                :language_length => Utilities.find_longest_length(@commits.language_list, 8),
+                :sort => sort,
+                :email => email,
+                :top_n => top_n}
+
+      # Acquire formatting pattern for output
+      pattern = "%-#{config[:author_length]}s | %-#{config[:language_length]}s | %7s | %9s | %9s | %7s | %7s | %7s | %6s | %6s |"
+      config[:pattern] = pattern
+      return config
+    end
+
+    def print_summary(sort, email, top_n=0)
+      # Prepare and determine the config for the result summary based on parameters
+      config = prepare_result_summary(sort, email, top_n)
+
+      # Print query/header information
+      output = print_header(config)
+
+      # Print per author information
+      config[:data].each do |key,value|
+        output += config[:pattern] % [key, "", value[:commits], value[:additions],
+                        value[:deletions], value[:create], value[:delete],
+                        value[:rename], value[:copy], value[:merges]]
+        output += "\n"
+        output += print_language_data(config[:pattern], value)
+      end
+
+      # Reprint query/header for repository information
+      output += "\n"
+      output += print_header(config)
+      data = @commits.totals
+      output += config[:pattern] % ["Repository Totals", "", data[:commits],
+                      data[:additions], data[:deletions], data[:create],
+                      data[:delete], data[:rename], data[:copy], data[:merges]]
+      output += "\n"
+      output += print_language_data(config[:pattern], data)
+      return output
+    end
+
+    def print_language_data(pattern, data)
+      output = ""
+      # Print information of each language for the data
+      data[:languages].each do |key,value|
+        output += pattern % ["", key, "", value[:additions], value[:deletions],
+                        value[:create], value[:delete], value[:rename],
+                        value[:copy], value[:merges]]
+        output += "\n"
+      end
+      return output
+    end
+
+    def print_header(config)
+      total_authors = @commits.author_list.length
+
+      output = ""
+      # Print summary information of displayed results
+      if config[:top_n] > 0 and config[:top_n] < total_authors
+        output += "Top #{config[:top_n]} authors(#{total_authors}) sorted by #{config[:sort]}\n"
+      else
+        output += "All authors(#{total_authors}) sorted by #{config[:sort]}\n"
+      end
+
+      # Print column headers
+      output += "-"*87 + "-"*config[:author_length] + "-"*config[:language_length]
+      output += "\n"
+      output += config[:pattern] % ['Name/Email', 'Language', 'Commits', 'Additions', 'Deletions', 'Creates', 'Deletes', 'Renames', 'Copies', 'Merges']
+      output += "\n"
+      output +=  "-"*87 + "-"*config[:author_length] + "-"*config[:language_length]
+      output += "\n"
+      return output
+    end
+  end
+end
+

data/lib/git_statistics/utilities.rb

@@ -0,0 +1,108 @@
+module GitStatistics
+  module Utilities
+    def self.get_repository(path=Dir.pwd)
+      # Connect to git repository if it exists
+      directory = Pathname.new(path)
+      repo = nil
+      while !directory.root? do
+        begin
+          repo = Grit::Repo.new(directory)
+          return repo
+        rescue
+          directory = directory.parent
+        end
+      end
+    end
+
+    def self.find_longest_length(list, max=nil)
+      return list if list == nil
+      list.each do |key,value|
+        max = key.length if max == nil || key.length > max
+      end
+      return max
+    end
+
+    def self.unique_data_in_hash(data, type)
+      list = []
+      data.each do |key,value|
+        if not list.include?(value[type])
+          list << value[type]
+        end
+      end
+      return list
+    end
+
+    def self.clean_string(string)
+      return string.strip.force_encoding("iso-8859-1").encode("utf-8")
+    end
+
+    def self.split_old_new_file(old, new)
+      # Split the old and new chunks up (separted by the =>)
+      split_old = old.split('{')
+      split_new = new.split('}')
+
+      # Handle recombine the file splits into their whole paths)
+      if split_old.size == 1 && split_new.size == 1
+        old_file = split_old[0]
+        new_file = split_new[0]
+      elsif split_new.size == 1
+        old_file = split_old[0] + split_old[1]
+        new_file = split_old[0] + split_new[0]
+      elsif split_old.size == 1
+        old_file = split_old[0] + split_new[1]
+        new_file = split_old[0] + split_new[0] + split_new[1]
+      else
+        old_file = split_old[0] + split_old[1] + split_new[1]
+        new_file = split_old[0] + split_new[0] + split_new[1]
+      end
+
+      # Return files, yet remove the '//' if present from combining splits
+      return {:old_file => old_file.gsub('//', '/'),
+              :new_file => new_file.gsub('//', '/')}
+    end
+
+    def self.find_blob_in_tree(tree, file)
+      # Check If cannot find tree in commit or if we found a submodule as the changed file
+      if tree == nil
+        return nil
+      elsif file == nil
+        return nil
+      elsif tree.instance_of?(Grit::Submodule)
+        return tree
+      end
+
+      # If the blob is within the current directory (tree)
+      if file.size == 1
+        blob = tree / file.first
+
+        # Check if blob is nil (could not find changed file in tree)
+        if blob == nil
+
+          # Try looking for submodules as they cannot be found using tree / file notation
+          tree.contents.each do |content|
+            if file.first == content.name
+              return tree
+            end
+          end
+
+          # Exit through recusion with the base case of a nil tree/blob
+          return find_blob_in_tree(blob, file)
+        end
+        return blob
+      else
+        # Explore deeper in the tree to find the blob of the changed file
+        return find_blob_in_tree(tree / file.first, file[1..-1])
+      end
+    end
+
+    def self.get_modified_time(file)
+      if OS.mac?
+        Time.at(`stat -f %m commits.json`.to_i)
+      elsif OS.linux?
+        Time.at(`stat -c %Y commits.json`.to_i)
+      else
+        raise "Update on the Windows operating system is not supported"
+      end
+    end
+  end
+end

data/lib/git_statistics/version.rb

@@ -1,3 +1,3 @@
 module GitStatistics
-  VERSION = "0.2.0"
+  VERSION = "0.3.0"
 end

data/lib/git_statistics.rb

@@ -7,7 +7,7 @@ module GitStatistics
         opt :email, "Use author's email instead of name", :default => false
         opt :merges, "Factor in merges when calculating statistics", :default => false
         opt :save, "Save the commits as commits.json", :default => false
-        opt :pretty, "Pretty print saved commits (larger file size)", :default => false
+        opt :pretty, "Save the commits as commits.json in pretty print (larger file size)", :default => false
         opt :load, "Load commits.json instead of re-collecting data", :default => false
         opt :update, "Update commits.json with new data (same as save and load together)", :default => false
         opt :sort, "Sort authors by {commits, additions, deletions, create, delete, rename, copy, merges}", :default => "commits"
@@ -29,17 +29,20 @@ module GitStatistics
 
       # Collect incremental recent data
       if @opts[:update]
-        collector.collect(@opts[:branch], "--since=\"`date -r commits.json \"+%F %T\"`\"")
+        time = Utilities.get_modified_time("commit.json")
+        collector.collect(@opts[:branch], "--since=\"#{time}\"")
       end
 
       # Save data
-      if @opts[:save] || @opts[:update]
+      if @opts[:save] || @opts[:update] || @opts[:pretty]
         collector.commits.save("commits.json", @opts[:pretty])
       end
 
       collector.commits.calculate_statistics(@opts[:email], @opts[:merges])
 
-      collector.print_summary(@opts[:sort].to_sym, @opts[:email], @opts[:top])
+      # Print results
+      results = Results.new(collector.commits)
+      puts results.print_summary(@opts[:sort], @opts[:email], @opts[:top])
     end
   end
 end