zip: add support multibyte path

lib/chupa-text/decomposers/zip.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2017  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2017-2019  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -45,14 +45,40 @@ def decompose(data)
             end
             entry_uri = data.uri.dup
             base_path = entry_uri.path.gsub(/\.zip\z/i, "")
-            entry_uri.path = "#{base_path}/#{entry.zip_path}"
+            path = convert_path_encoding(entry.zip_path, base_path.encoding)
+            entry_uri.path = "#{base_path}/#{convert_to_uri_path(path)}"
             entry_data = VirtualFileData.new(entry_uri,
                                              entry.file_data,
                                              source_data: data)
             yield(entry_data)
           end
         end
       end
+
+      private
+      def convert_path_encoding(path, encoding)
+        return path if path.ascii_only?
+
+        candidates = [
+          Encoding::UTF_8,
+          Encoding::Windows_31J,
+        ]
+        candidates.each do |candidate|
+          path.force_encoding(candidate)
+          return path.encode(encoding) if path.valid_encoding?
+        end
+        path.encode(encoding,
+                    Encoding::UTF_8,
+                    invalid: :replace,
+                    undef: :replace)
+      end
+
+      def convert_to_uri_path(path)
+        converted_components = path.split("/").collect do |component|
+          CGI.escape(component)
+        end
+        converted_components.join("/")
+      end
     end
   end
 end

test/decomposers/test-zip.rb

@@ -1,4 +1,4 @@
-# Copyright (C) 2017  Kouhei Sutou <kou@clear-code.com>
+# Copyright (C) 2017-2019  Kouhei Sutou <kou@clear-code.com>
 #
 # This library is free software; you can redistribute it and/or
 # modify it under the terms of the GNU Lesser General Public
@@ -63,6 +63,36 @@ def decompose(data_path)
                    decompose(data_path))
     end
 
+    sub_test_case("multibyte") do
+      test("cp932") do
+        data_path = Pathname.new(fixture_path("cp932.zip"))
+        base_path = data_path.sub_ext("")
+        path = CGI.escape("こんにちは.txt")
+        assert_equal([
+                       {
+                         :uri    => file_uri("#{base_path}/cp932/#{path}").to_s,
+                         :body   => "こんにちは\n".encode("cp932").b,
+                         :source => file_uri(data_path).to_s,
+                       },
+                     ],
+                     decompose(data_path))
+      end
+
+      test("UTF-8") do
+        data_path = Pathname.new(fixture_path("utf-8.zip"))
+        base_path = data_path.sub_ext("")
+        path = CGI.escape("こんにちは.txt")
+        assert_equal([
+                       {
+                         :uri    => file_uri("#{base_path}/utf-8/#{path}").to_s,
+                         :body   => "こんにちは\n".b,
+                         :source => file_uri(data_path).to_s,
+                       },
+                     ],
+                     decompose(data_path))
+      end
+    end
+
     sub_test_case("encrypted") do
       test("without password") do
         data_path = Pathname.new(fixture_path("password.zip"))