Merge pull request #348 from Shopify/load-path-cache-encoding

Set encoding and dedup cached paths so they can be directly used by MRI

lib/bootsnap/load_path_cache/cache.rb

@@ -11,7 +11,7 @@ def initialize(store, path_obj, development_mode: false)
         @development_mode = development_mode
         @store = store
         @mutex = defined?(::Mutex) ? ::Mutex.new : ::Thread::Mutex.new # TODO: Remove once Ruby 2.2 support is dropped.
-        @path_obj = path_obj.map! { |f| File.exist?(f) ? File.realpath(f) : f }
+        @path_obj = path_obj.map! { |f| PathScanner.os_path(File.exist?(f) ? File.realpath(f) : f.dup) }
         @has_relative_paths = nil
         reinitialize
       end
@@ -178,25 +178,42 @@ def now
 
       if DLEXT2
         def search_index(f)
-          try_index("#{f}#{DOT_RB}") || try_index("#{f}#{DLEXT}") || try_index("#{f}#{DLEXT2}") || try_index(f)
+          try_index(f + DOT_RB) || try_index(f + DLEXT) || try_index(f + DLEXT2) || try_index(f)
         end
 
         def maybe_append_extension(f)
-          try_ext("#{f}#{DOT_RB}") || try_ext("#{f}#{DLEXT}") || try_ext("#{f}#{DLEXT2}") || f
+          try_ext(f + DOT_RB) || try_ext(f + DLEXT) || try_ext(f + DLEXT2) || f
         end
       else
         def search_index(f)
-          try_index("#{f}#{DOT_RB}") || try_index("#{f}#{DLEXT}") || try_index(f)
+          try_index(f + DOT_RB) || try_index(f + DLEXT) || try_index(f)
         end
 
         def maybe_append_extension(f)
-          try_ext("#{f}#{DOT_RB}") || try_ext("#{f}#{DLEXT}") || f
+          try_ext(f + DOT_RB) || try_ext(f + DLEXT) || f
         end
       end
 
-      def try_index(f)
-        if (p = @index[f])
-          "#{p}/#{f}"
+      s = rand.to_s.force_encoding(Encoding::US_ASCII).freeze
+      if s.respond_to?(:-@)
+        if (-s).equal?(s) && (-s.dup).equal?(s)
+          def try_index(f)
+            if (p = @index[f])
+              -(File.join(p, f).freeze)
+            end
+          end
+        else
+          def try_index(f)
+            if (p = @index[f])
+              -File.join(p, f).untaint
+            end
+          end
+        end
+      else
+        def try_index(f)
+          if (p = @index[f])
+            File.join(p, f)
+          end
         end
       end
 

lib/bootsnap/load_path_cache/path_scanner.rb

@@ -33,10 +33,10 @@ def call(path)
           requirables = []
           walk(path, nil) do |relative_path, absolute_path, is_directory|
             if is_directory
-              dirs << relative_path
+              dirs << os_path(relative_path)
               !contains_bundle_path || !absolute_path.start_with?(BUNDLE_PATH)
             elsif relative_path.end_with?(*REQUIRABLE_EXTENSIONS)
-              requirables << relative_path
+              requirables << os_path(relative_path)
             end
           end
           [requirables, dirs]
@@ -45,7 +45,7 @@ def call(path)
         def walk(absolute_dir_path, relative_dir_path, &block)
           Dir.foreach(absolute_dir_path) do |name|
             next if name.start_with?('.')
-            relative_path = relative_dir_path ? "#{relative_dir_path}/#{name}" : name.freeze
+            relative_path = relative_dir_path ? File.join(relative_dir_path, name) : name
 
             absolute_path = "#{absolute_dir_path}/#{name}"
             if File.directory?(absolute_path)
@@ -57,6 +57,17 @@ def walk(absolute_dir_path, relative_dir_path, &block)
             end
           end
         end
+
+        if RUBY_VERSION >= '3.1'
+          def os_path(path)
+            path.freeze
+          end
+        else
+          def os_path(path)
+            path.force_encoding(Encoding::US_ASCII) if path.ascii_only?
+            path.freeze
+          end
+        end
       end
     end
   end

lib/bootsnap/load_path_cache/store.rb

@@ -86,7 +86,7 @@ def dump_data
         # `encoding:` looks redundant wrt `binwrite`, but necessary on windows
         # because binary is part of mode.
         File.open(tmp, mode: exclusive_write, encoding: Encoding::BINARY) do |io|
-          MessagePack.dump(@data, io)
+          MessagePack.dump(@data, io, freeze: true)
         end
         FileUtils.mv(tmp, @store_path)
       rescue Errno::EEXIST

test/load_path_cache/cache_test.rb

@@ -16,6 +16,7 @@ def setup
         FileUtils.touch("#{@dir1}/dl#{DLEXT}")
         FileUtils.touch("#{@dir1}/both.rb")
         FileUtils.touch("#{@dir1}/both#{DLEXT}")
+        FileUtils.touch("#{@dir1}/béé.rb")
       end
 
       def teardown
@@ -137,6 +138,36 @@ def test_path_obj_equal?
 
         assert_equal("#{@dir1}/a.rb", cache.find('a'))
       end
+
+      if RUBY_VERSION >= '2.5' && !(RbConfig::CONFIG['host_os'] =~ /mswin|mingw|cygwin/)
+        # https://github.com/ruby/ruby/pull/4061
+        # https://bugs.ruby-lang.org/issues/17517
+        OS_ASCII_PATH_ENCODING = RUBY_VERSION >= '3.1' ? Encoding::UTF_8 : Encoding::US_ASCII
+
+        def test_path_encoding
+          po = [@dir1]
+          cache = Cache.new(NullCache, po)
+
+          path = cache.find('a')
+
+          assert_equal("#{@dir1}/a.rb", path)
+          require path
+          internal_path = $LOADED_FEATURES.last
+          assert_equal(OS_ASCII_PATH_ENCODING, internal_path.encoding)
+          assert_equal(OS_ASCII_PATH_ENCODING, path.encoding)
+          File.write(path, '')
+          assert_same path, internal_path
+
+          utf8_path = cache.find('béé')
+          require utf8_path
+          internal_utf8_path = $LOADED_FEATURES.last
+          assert_equal("#{@dir1}/béé.rb", utf8_path)
+          assert_equal(Encoding::UTF_8, internal_utf8_path.encoding)
+          assert_equal(Encoding::UTF_8, utf8_path.encoding)
+          File.write(utf8_path, '')
+          assert_same utf8_path, internal_utf8_path
+        end
+      end
     end
   end
 end