better concurrency. better scaling. 20000 entries in 1000 folders sti…

…ll performant.

git_store.gemspec

@@ -16,7 +16,11 @@ LICENSE
 README.md
 git_store.gemspec
 lib/git_store.rb
-spec/git_store_spec.rb
+lib/git_store/blob.rb
+lib/git_store/tree.rb
+lib/git_store/handlers.rb
+test/git_store_spec.rb
+test/benchmark.rb
 }
 end
 

lib/git_store.rb

@@ -1,240 +1,157 @@
+require 'rubygems'
 require 'grit'
+require 'yaml'
 
-# This fix ensures sorted yaml maps.
-class Hash
-	def to_yaml( opts = {} )
-		YAML::quick_emit( object_id, opts ) do |out|
-      out.map( taguri, to_yaml_style ) do |map|
-        sort_by { |k, v| k.to_s }.each do |k, v|
-          map.add( k, v )
-        end
-      end
-    end
-	end
-end
+require 'git_store/blob'
+require 'git_store/tree'
+require 'git_store/handlers'
 
 class GitStore
+  include Enumerable
 
-  class DefaultHandler
-    def read(name, data)
-      data
-    end
+  attr_reader :repo, :index, :root, :last_commit
 
-    def write(data)
-      data
-    end
+  def initialize(path = '.')
+    @repo = Grit::Repo.new(path)
+    @root = Tree.new('')
+    load_last_commit
   end
-
-  class YAMLHandler    
-    def read(name, data)
-      YAML.load(data)
-    end
 
-    def write(data)
-      data.to_yaml
-    end    
+  def load_last_commit
+    @last_commit = @repo.commits('master', 1)[0]
   end
-
-  class RubyHandler
-    def read(name, data)
-      Object.module_eval(data)
-    end
+
+  def commit(message="")
+    head = repo.heads.first
+    commit_index(message, head ? head.commit.id : nil)
   end
-
-  class ERBHandler
-    def read(name, data)
-      ERB.new(data)
-    end
+
+  def [](*args)
+    root[*args]
   end
 
-  Handler = {
-    'yml' => YAMLHandler.new,
-    'rhtml' => ERBHandler.new,
-    'rxml' => ERBHandler.new,
-    'rb' => RubyHandler.new
-  }
+  def []=(*args)
+    value = args.pop
+    root[*args] = value
+  end
 
-  Handler.default = DefaultHandler.new
+  def delete(path)
+    root.delete(path)
+  end
 
-  class Blob
+  def load
+    root.load(repo.tree)
+  end
 
-    attr_reader :id
-    attr_accessor :name
+  def each(&block)
+    root.each(&block)
+  end
 
-    def initialize(*args)
-      if args.first.is_a?(Grit::Blob)
-        @blob = args.first
-        @name = @blob.name
-      else
-        @name = args[0]
-        self.data = args[1]
-      end
-    end
+  def changed?
+    commit = repo.commits('master', 1)[0]
+    commit and (last_commit.nil? or last_commit.id != commit.id)
+  end
 
-    def extname
-      File.extname(name)[1..-1]
-    end
+  def refresh!
+    load if changed?
+  end
 
-    def load(data)
-      @data = handler.read(name, data)
-    end
+  def start_transaction(head = 'master')
+    @lock = open("#{repo.path}/refs/heads/#{head}.lock", "w")
+    @lock.flock(File::LOCK_EX)
+  end
 
-    def handler
-      Handler[extname]
-    end
+  def commit_index(message, parents = nil, actor = nil, head = 'master')
+    start_transaction(head) unless @lock
+
+    tree_sha = write_tree(root)
+
+    contents = []
+    contents << ['tree', tree_sha].join(' ')
 
-    def data
-      @data or (@blob and load(@blob.data))
+    if parents
+      parents.each do |p|
+        contents << ['parent', p].join(' ') if p        
+      end
     end
 
-    def data=(data)
-      @data = data
+    if actor
+      name = actor.name
+      email = actor.email
+    else
+      config = Grit::Config.new(self.repo)
+      name = config['user.name']
+      email = config['user.email']
     end
-
-    def to_s
-      if handler.respond_to?(:write)
-        handler.write(data)
-      else
-        @blob.data
-      end
+
+    author_string = "#{name} <#{email}> #{Time.now.to_i} -0700"
+    contents << ['author', author_string].join(' ')
+    contents << ['committer', author_string].join(' ')
+    contents << ''
+    contents << message
+
+    commit_sha = put_raw_object(contents.join("\n"), 'commit')
+
+    open("#{repo.path}/refs/heads/#{head}", "w") do |file|
+      file.write(commit_sha)
     end
 
+    commit_sha
+  ensure
+    @lock.close if @lock
+    @lock = nil
+    File.unlink("#{repo.path}/refs/heads/#{head}.lock") rescue nil
   end
 
-  class Tree
-    include Enumerable
+  def put_raw_object(data, type)
+    repo.git.ruby_git.put_raw_object(data, type)
+  end
 
-    attr_reader :data
-    attr_accessor :name
+  def write_blob(blob)
+    return if not blob.modified?
 
-    def initialize(name = nil)
-      @data = {}
-      @name = name
-    end
-
-    def load(tree)
-      @name = tree.name
-      @data = tree.contents.inject({}) do |hash, file|
-        if file.is_a?(Grit::Tree)
-          hash[file.name] = (@data[file.name] || Tree.new).load(file)
-        else          
-          hash[file.name] = Blob.new(file)
-        end
-        hash
-      end
-      self
-    end
-
-    def inspect
-      "#<GitStore::Tree #{@data.inspect}>"
-    end
-
-    def fetch(name)
-      name = name.to_s
-      entry = @data[name]
+    blob.sha1 = put_raw_object(blob.serialize, 'blob')
+    blob.modified = false
+  end
+
+  def write_tree(tree)
+    return if not tree.modified?
+
+    contents = tree.data.map do |name, entry|
       case entry
-      when Blob then entry.data
-      when Tree then entry
-      end      
-    end
-
-    def store(name, value)
-      name = name.to_s
-      if value.is_a?(Tree)
-        value.name = name
-        @data[name] = value
-      else
-        @data[name] = Blob.new(name, value)
+      when Blob; write_blob(entry)
+      when Tree; write_tree(entry)
       end
+      "%s %s\0%s" % [entry.mode, name, [entry.sha1].pack("H*")]
     end
 
-    def has_key?(name)
-      @data.has_key?(name)
-    end
+    tree.modified = false
+    tree.sha1 = put_raw_object(contents.join, 'tree')
+  end
+
+  class FileStore < GitStore
 
-    def [](*args)
-      args = args.first.to_s.split('/') if args.size == 1
-      args.inject(self) { |tree, key| tree.fetch(key) or return nil }
+    attr_reader :path
+
+    def initialize(path = '.')
+      @path = path
+      @root = Tree.new('')
     end
 
-    def []=(*args)
-      value = args.pop
-      args = args.first.to_s.split('/') if args.size == 1
-      tree = args[0..-2].to_a.inject(self) do |tree, key|
-        tree.has_key?(key) ? tree.fetch(key) : tree.store(key, Tree.new(key))
-      end
-      tree.store(args.last, value)
+    def load
+      root.load_from_disk
     end
 
-    def delete(name)
-      @data.delete(name)
-    end
-
-    def each(&block)
-      @data.values.each do |entry|
-        case entry
-        when Blob then yield entry.data
-        when Tree then entry.each(&block)
-        end
+    def refresh!
+      root.each_blob do |blob|
+
       end
     end
 
-    def each_with_path(path = [], &block)
-      @data.each do |name, entry|
-        child_path = path + [name]
-        case entry
-        when Blob then yield entry, child_path.join('/')
-        when Tree then entry.each_with_path(child_path, &block)
-        end        
-      end
-    end
-
-    def to_hash
-      @data.inject({}) do |hash, (name, entry)|
-        hash[name] = entry.is_a?(Tree) ? entry.to_hash : entry.to_s
-        hash
-      end
+    def commit(message="")
+      root.write_to_disk
     end
 
   end
 
-  attr_reader :repo, :index, :tree
-
-  def initialize(path, &block)
-    @repo = Grit::Repo.new(path)
-    @index = Grit::Index.new(@repo)
-    @tree = Tree.new
-  end
-
-  def commit(message="")
-    index.tree = tree.to_hash
-    head = repo.heads.first
-    index.commit(message, head ? head.commit.id : nil)
-  end
-
-  def [](*args)
-    tree[*args]
-  end
-
-  def []=(*args)
-    value = args.pop
-    tree[*args] = value
-  end
-
-  def delete(path)
-    tree.delete(path)
-  end
-
-  def load
-    tree.load(repo.tree)
-  end
-
-  def each(&block)
-    tree.each(&block)
-  end
-
-  def each_with_path(&block)
-    tree.each_with_path(&block)
-  end
-
 end