Add the possibility to read blob contents in chunks via IO#popen

Author: siebertm

lib/git/lib.rb

@@ -166,8 +166,8 @@ def process_commit_data(data, sha = nil)
       end
     end
 
-    def object_contents(sha)
-      command('cat-file', ['-p', sha])
+    def object_contents(sha, &block)
+      command('cat-file', ['-p', sha], &block)
     end
 
     def ls_tree(sha)
@@ -596,20 +596,20 @@ def command_lines(cmd, opts = [], chdir = true)
       command(cmd, opts, chdir).split("\n")
     end
 
-    def command(cmd, opts = [], chdir = true)
+    def command(cmd, opts = [], chdir = true, &block)
       ENV['GIT_DIR'] = @git_dir if (@git_dir != ENV['GIT_DIR'])
       ENV['GIT_INDEX_FILE'] = @git_index_file if (@git_index_file != ENV['GIT_INDEX_FILE'])
       ENV['GIT_WORK_TREE'] = @git_work_dir if (@git_work_dir != ENV['GIT_WORK_TREE'])
       path = @git_work_dir || @git_dir || @path
 
       opts = opts.to_a.join(' ')
-      git_cmd = "git #{cmd} #{opts}"
+      git_cmd = "git #{cmd} #{opts} 2>&1"
 
       out = nil
       if chdir && (Dir.getwd != path)
-        Dir.chdir(path) { out = `#{git_cmd} 2>&1`.chomp } 
+        Dir.chdir(path) { out = run_command(git_cmd, &block) } 
       else
-        out = `#{git_cmd} 2>&1`.chomp
+        out = run_command(git_cmd, &block)
       end
 
       if @logger
@@ -626,5 +626,13 @@ def command(cmd, opts = [], chdir = true)
       out
     end
 
+    def run_command(git_cmd, &block)
+      if block_given?
+        IO.popen(git_cmd, &block)
+      else
+        `#{git_cmd}`.chomp
+      end
+    end
+    
   end
 end

lib/git/object.rb

@@ -28,9 +28,17 @@ def size
         @size || @size = @base.lib.object_size(@objectish)
       end
 
-      # caches the contents of this call in memory
-      def contents
-        @contents || @contents = @base.lib.object_contents(@objectish)
+      # get the object's contents
+      # if no block is given, the contents are cached in memory and returned as a string
+      # if a block is given, it yields an IO object (via IO::popen) which could be used to
+      # read a large file in chunks. use this for large files so that they are not held
+      # in memory
+      def contents(&block)
+        if block_given?
+          @base.lib.object_contents(@objectish, &block)
+        else
+          @contents || @contents = @base.lib.object_contents(@objectish)
+        end
       end
 
       def contents_array

tests/units/test_lib.rb

@@ -85,6 +85,33 @@ def test_object_contents
     assert_equal(blob, @lib.object_contents('v2.5:example.txt')) #blob
 
   end
+  
+  def test_object_contents_with_block
+    commit =  "tree 94c827875e2cadb8bc8d4cdd900f19aa9e8634c7\n"
+    commit += "parent 546bec6f8872efa41d5d97a369f669165ecda0de\n"
+    commit += "author scott Chacon <schacon@agadorsparticus.corp.reactrix.com> 1194561188 -0800\n"
+    commit += "committer scott Chacon <schacon@agadorsparticus.corp.reactrix.com> 1194561188 -0800\n"
+    commit += "\ntest"
+    
+    @lib.object_contents('1cc8667014381') do |f|
+      assert_equal(commit, f.read.chomp)
+    end
+    
+     # commit
+    
+    tree =  "040000 tree 6b790ddc5eab30f18cabdd0513e8f8dac0d2d3ed\tex_dir\n"
+    tree += "100644 blob 3aac4b445017a8fc07502670ec2dbf744213dd48\texample.txt"
+
+    @lib.object_contents('1cc8667014381^{tree}') do |f|
+      assert_equal(tree, f.read.chomp) #tree
+    end
+    
+    blob = "1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n2"
+
+    @lib.object_contents('v2.5:example.txt') do |f|
+      assert_equal(blob, f.read.chomp) #blob
+    end
+  end
 
   # returns Git::Branch object array
   def test_branches_all

tests/units/test_object.rb

@@ -102,6 +102,15 @@ def test_blob_contents
     o = @git.gblob('v2.6:example.txt')
     assert_equal('replace with new text', o.contents)
     assert_equal('replace with new text', o.contents)  # this should be cached
+    
+    # make sure the block is called
+    block_called = false
+    o.contents do |f|
+      block_called = true
+      assert_equal('replace with new text', f.read.chomp)
+    end
+    
+    assert(block_called)
   end
 
   def test_revparse