Handle encoding header of Git commits when possible.

lib/grit.rb

@@ -28,6 +28,7 @@
 require 'grit/ruby1.9'
 
 # internal requires
+require 'grit/encode'
 require 'grit/lazy'
 require 'grit/errors'
 require 'grit/git-ruby'

lib/grit/commit.rb

@@ -2,6 +2,7 @@ module Grit
 
   class Commit
     extend Lazy
+    extend Encode
 
     attr_reader :id
     attr_reader :repo
@@ -25,12 +26,17 @@ class Commit
     def self.parse_batch(repo, sha, size, object)
       info, message = object.split("\n\n", 2)
 
+      encoding = nil
+
       lines = info.split("\n")
       tree = lines.shift.split(' ', 2).last
       parents = []
       parents << lines.shift[7..-1] while lines.first[0, 6] == 'parent'
       author,    authored_date  = Grit::Commit.actor(lines.shift)
       committer, committed_date = Grit::Commit.actor(lines.shift)
+      encoding = $1  if lines.detect { |line| line =~ /\Aencoding\s*(.+)\z/ }
+
+      message = message_in_utf8(message, encoding)
 
       Grit::Commit.new(
         repo, sha, parents, tree,
@@ -176,6 +182,8 @@ def self.list_from_string(repo, text)
 
         lines.shift while lines.first && lines.first.empty?
 
+        message_lines = message_lines.map { |line| message_in_utf8(line, encoding) }  if encoding
+
         commits << Commit.new(repo, id, parents, tree, author, authored_date, committer, committed_date, message_lines)
       end
 

lib/grit/encode.rb

@@ -0,0 +1,35 @@
+module Grit
+  module Encode
+    ENCODING_MAP = {
+      'koi8r' => 'KOI8-R',
+    }
+
+    def find_ruby_encoding(encoding)
+      return nil  unless defined?(Encoding)
+      Encoding.find(ENCODING_MAP[encoding] || encoding)
+    rescue ArgumentError
+      warn "Cannot map git encoding to ruby one: #{encoding}"
+      nil
+    end
+
+   def message_in_utf8(message, encoding)
+      message = message.dup
+      message.force_encoding('UTF-8')  if message.respond_to?(:force_encoding)
+      if encoding && encoding !~ /\Autf\-8i\z/i && message && message.respond_to?(:encode!)
+        ruby_encoding = find_ruby_encoding(encoding)
+        if ruby_encoding
+          message.encode!('UTF-8', ruby_encoding, :invalid => :replace, :undef => :replace)
+          unless message.valid_encoding?
+            message.encode!('UTF-8', 'ISO-8859-1', :invalid => :replace, :undef => :replace)
+          end
+        else
+          unless message.valid_encoding?
+            message.encode!('UTF-8', 'ISO-8859-1', :invalid => :replace, :undef => :replace)
+          end
+        end
+      end
+      message
+    end
+  end
+end
+

lib/grit/git-ruby/git_object.rb

@@ -1,4 +1,5 @@
-#
+# --- encoding: utf-8 ---
+
 # converted from the gitrb project
 #
 # authors:
@@ -232,11 +233,13 @@ def actual_raw
   end
 
   class Commit < GitObject
-    attr_accessor :author, :committer, :tree, :parent, :message, :headers
+    extend Grit::Encode
+
+    attr_accessor :author, :committer, :tree, :parent, :message, :encoding, :headers
 
     def self.from_raw(rawobject, repository=nil)
       parent = []
-      tree = author = committer = nil
+      tree = author = committer = encoding = nil
 
       headers, message = rawobject.content.split(/\n\n/, 2)
       all_headers = headers.split(/\n/).map { |header| header.split(/ /, 2) }
@@ -250,6 +253,8 @@ def self.from_raw(rawobject, repository=nil)
           author = UserInfo.new(value)
         when "committer"
           committer = UserInfo.new(value)
+        when "encoding"
+          encoding = value
         else
           warn "unknown header '%s' in commit %s" % \
             [key, rawobject.sha1.unpack("H*")[0]]
@@ -258,17 +263,22 @@ def self.from_raw(rawobject, repository=nil)
       if not tree && author && committer
         raise RuntimeError, "incomplete raw commit object"
       end
-      new(tree, parent, author, committer, message, headers, repository)
+      new(tree, parent, author, committer, message, headers, repository, encoding)
     end
 
-    def initialize(tree, parent, author, committer, message, headers, repository=nil)
+    def initialize(tree, parent, author, committer, message, headers, repository=nil, encoding=nil)
       @tree = tree
       @author = author
       @parent = parent
       @committer = committer
       @message = message
       @headers = headers
       @repository = repository
+      @encoding = encoding
+    end
+
+    def decoded_message
+      message_in_utf8(message, encoding)
     end
 
     def type

test/test_git.rb

@@ -10,7 +10,7 @@ def teardown
   end
 
   def test_method_missing
-    assert_match(/^git version [\w\.]*$/, @git.version)
+    assert_match(/\Agit version [\w\.\(\)\s\-]*\z/, @git.version)
   end
 
   def test_logs_stderr