Decompiler Middleware

app/models/repository.rb

@@ -2,7 +2,7 @@ class Repository < Rugged::Repository
   REPO_PATH = Rails.root.join 'repos'
   class EmptyCommit < RuntimeError; end
 
-  attr_accessor :app_id
+  attr_accessor :app_id, :path
 
   # Rugged gets its hands on new, not initialize
   def self.new(app_id, options={})
@@ -14,6 +14,7 @@ def self.new(app_id, options={})
 
     super(path.to_s).instance_eval do
       @app_id = app_id
+      @path = path
       self
     end
   end

lib/stack.rb

@@ -7,7 +7,7 @@ def self.process_free_app(app_id)
       use PrepareFS
       use FetchMarketDetails
       use DownloadApk
-      # use DecompileApk
+      use DecompileApk
     end
     @create_app_stack.call :app_id => app_id, :crawl_date => Date.today
   end

lib/stack/base_git.rb

@@ -1,5 +1,6 @@
 class Stack::BaseGit < Stack::Base
   class Git
+    class CommitError < RuntimeError; end
     include ActionView::Helpers::TextHelper
     include ActionView::Helpers::SanitizeHelper
 
@@ -12,7 +13,6 @@ def initialize(env, options={})
       @branch   = options[:branch]
       @role     = options[:role]
       @tag_with = options[:tag_with]
-      @tree_builder = TreeBuilderProxy.new(repo)
     end
 
     def app
@@ -43,6 +43,10 @@ def last_commit_sha
       Rugged::Reference.lookup(repo, branch_ref).try(:target)
     end
 
+    def last_committed_tree
+      repo.lookup(last_commit_sha).tree
+    end
+
     def new_branch?
       !last_commit_sha
     end
@@ -59,37 +63,36 @@ def full_message
       msg += "\n\nVersion Code: #{app.version_code}"
     end
 
-    class TreeBuilderProxy < Rugged::Tree::Builder
-      def initialize(repo)
-        @repo = repo
-        super()
+    class Index < Rugged::Index
+      def self.new(repo)
+        super().instance_eval do
+          @repo = repo
+          self
+        end
       end
 
-      def add_file(name, content, mode=0100644)
+      def add_file(filename, content, mode=0100644)
         oid = @repo.write(content, :blob)
-        self.insert(:type => :blob, :name => name.to_s, :oid => oid, :filemode => mode)
-        @has_file = true
+        self.add(:path => filename.to_s, :oid => oid, :mode => mode)
       end
 
-      def has_files?
-        @has_file
+      def add_dir(dir)
+        prefix_regexp = Regexp.new("^#{dir}/")
+        Dir["#{dir}/**/*"].each do |filename|
+          next unless File.file?(filename)
+          add_file(filename.gsub(prefix_regexp, ''), File.open(filename, 'rb') { |f| f.read })
+        end
       end
 
-      def write
+      def write_tree
         super(@repo)
       end
     end
 
-    def read_file(file_name)
-      file = repo.lookup(last_commit_sha).tree[file_name]
-      return nil unless file
-      repo.lookup(file[:oid]).read_raw.data
-    end
-
     def commit(&block)
-      block.call(tree_builder)
-
-      raise "No files to commit" unless tree_builder.has_files?
+      index = Index.new(repo)
+      block.call(index)
+      raise CommitError.new "No files to commit" unless index.count > 0
 
       commit = {}
 
@@ -107,12 +110,27 @@ def commit(&block)
 
       commit[:parents]    = [last_commit_sha].compact
       commit[:update_ref] = branch_ref
-      commit[:tree]       = tree_builder.write
+      commit[:tree]       = index.write_tree
 
       Rugged::Commit.create(repo, commit).tap do |commit_sha|
         Rugged::Reference.create(repo, tag_ref, commit_sha)
       end
     end
+
+    def read_file(file_name)
+      file = last_committed_tree[file_name]
+      return nil unless file
+      repo.lookup(file[:oid]).read_raw.data
+    end
+
+    def read_files(&block)
+      last_committed_tree.walk(:preorder) do |dir, entry|
+        case entry[:type]
+        when :tree then block.call("#{dir}#{entry[:name]}", nil)
+        when :blob then block.call("#{dir}#{entry[:name]}", repo.lookup(entry[:oid]).read_raw.data)
+        end
+      end
+    end
   end
 
   class << self

lib/stack/decompile_apk.rb

@@ -1,5 +1,32 @@
-class Stack::DecompileApk < Stack::Base
-  def call(env)
-    @stack.call(env)
+require 'fileutils'
+
+class Stack::DecompileApk < Stack::BaseGit
+  class DecompilationError < RuntimeError; end
+
+  use_git :branch => :src
+
+  def persist_to_git(env, git)
+    output = `script/decompile #{env[:scratch]} #{env[:apk_path].basename}`
+    raise DecompilationError.new(output) unless $?.success?
+
+    env[:src_dir] = env[:scratch].join('src')
+
+    git.commit do |index|
+      index.add_dir(env[:src_dir])
+    end
+  end
+
+  def parse_from_git(env, git)
+    env[:src_dir] = env[:scratch].join('src')
+    FileUtils.mkpath(env[:src_dir])
+
+    git.read_files do |filename, content|
+      path = env[:src_dir].join(filename)
+      if content
+        File.open(path, 'wb') { |f| f.write(content) }
+      else
+        FileUtils.mkpath(path)
+      end
+    end
   end
 end

lib/stack/download_apk.rb

@@ -32,8 +32,8 @@ def persist_to_git(env, git)
 
     apk_filename = "#{env[:app].id_version}.apk"
 
-    git.commit do |tree|
-      tree.add_file(apk_filename, response.body)
+    git.commit do |index|
+      index.add_file(apk_filename, response.body)
     end
 
     env[:apk_path] = env[:scratch].join(apk_filename)

lib/stack/fetch_market_details.rb

@@ -9,8 +9,8 @@ def persist_to_git(env, git)
 
     env[:app] = App.from_market(app_details.raw_app)
 
-    git.commit do |tree|
-      tree.add_file('metadata.json', MultiJson.dump(app_details.raw_app, :pretty => true))
+    git.commit do |index|
+      index.add_file('metadata.json', MultiJson.dump(app_details.raw_app, :pretty => true))
     end
   end
 

lib/stack/prepare_fs.rb

@@ -9,6 +9,11 @@ def call(env)
     Dir.mktmpdir "#{env[:app_id]}", Rails.root.join('scratch') do |dir|
       env[:scratch] = Pathname.new(dir)
       @stack.call(env)
+
+      # It would be much more efficient to write a pack directly.
+      # Expect horrible performance when saving sources.
+      output = `cd #{env[:repo].path} 2>&1 && git gc --prune=now -q 2>&1`
+      Rails.logger.info "Cannot garbage collect the repository: #{output}" unless $?.success?
     end
   end
 end

script/decompile

@@ -5,4 +5,5 @@
 APK2GOLD=`pwd`/vendor/apk2gold/apk2gold
 ulimit -m $((1024*1024))
 cd $1
-$APK2GOLD $2 2>&1
+$APK2GOLD $2 2>&1 || exit 1
+mv "${2%.*}" src