Add refresh_metadata plugin for re-extracting metadata

CHANGELOG.md

@@ -1,5 +1,7 @@
 ## HEAD
 
+* Add `refresh_metadata` plugin for re-extracting metadata from an uploaded file (@janko-m)
+
 * Allow S3 storage to use parallelized multipart upload for files from FileSystem storage as well (@janko-m)
 
 * Improve default multipart copy threshold for S3 storage (@janko-m)

lib/shrine/plugins/refresh_metadata.rb

@@ -0,0 +1,29 @@
+class Shrine
+  module Plugins
+    # The `refresh_metadata` plugin allows you to re-extract metadata from an
+    # uploaded file.
+    #
+    #     plugin :refresh_metadata
+    #
+    # It provides `UploadedFile#refresh_metadata!` method, which calls
+    # `Shrine#extract_metadata` with the uploaded file opened for reading,
+    # and updates the existing metadata hash with the results.
+    #
+    #     uploaded_file.refresh_metadata!
+    #     uploaded_file.metadata # re-extracted metadata
+    #
+    # For remote storages this will make an HTTP request to open the file for
+    # reading, but only the portion of the file needed for extracting each
+    # metadata value will be downloaded.
+    module RefreshMetadata
+      module FileMethods
+        def refresh_metadata!(context = {})
+          refreshed_metadata = open { uploader.extract_metadata(self, context) }
+          metadata.merge!(refreshed_metadata)
+        end
+      end
+    end
+
+    register_plugin(:refresh_metadata, RefreshMetadata)
+  end
+end

lib/shrine/plugins/restore_cached_data.rb

@@ -1,28 +1,25 @@
 class Shrine
   module Plugins
-    # The `restore_cached_data` plugin re-extracts cached file's metadata on
-    # assignment. This happens when an uploaded file is retained on validation
-    # errors, or when assigning direct uploaded files. In both cases you
-    # usually want to re-extract metadata on the server side, mainly to prevent
+    # The `restore_cached_data` plugin re-extracts metadata when assigning
+    # already cached files, i.e. when the attachment has been retained on
+    # validation errors or assigned from a direct upload. In both cases you may
+    # want to re-extract metadata on the server side, mainly to prevent
     # tempering, but also in case of direct uploads to obtain metadata that
     # couldn't be extracted on the client side.
     #
     #     plugin :restore_cached_data
     #
-    # This will give an opened `UploadedFile` for metadata extraction. For
-    # remote storages this will make an HTTP request, and since metadata is
-    # typically found in the beginning of the file, Shrine will download only
-    # the amount of bytes necessary for extracting the metadata.
+    # It uses the `refresh_metadata` plugin to re-extract metadata.
     module RestoreCachedData
+      def self.load_dependencies(uploader, *)
+        uploader.plugin :refresh_metadata
+      end
+
       module AttacherMethods
         private
 
         def assign_cached(cached_file)
-          uploaded_file(cached_file) do |file|
-            real_metadata = file.open { cache.extract_metadata(file, context) }
-            file.metadata.update(real_metadata)
-          end
-
+          uploaded_file(cached_file) { |file| file.refresh_metadata!(context) }
           super(cached_file)
         end
       end

test/plugin/refresh_metadata_test.rb

@@ -0,0 +1,40 @@
+require "test_helper"
+require "shrine/plugins/refresh_metadata"
+
+describe Shrine::Plugins::RefreshMetadata do
+  before do
+    @uploader = uploader { plugin :refresh_metadata }
+  end
+
+  it "re-extracts metadata" do
+    uploaded_file = @uploader.upload(fakeio("content", filename: "file.txt", content_type: "text/plain"))
+    uploaded_file.metadata.delete("size")
+    uploaded_file.refresh_metadata!
+    assert_equal 7,            uploaded_file.metadata["size"]
+    assert_equal "file.txt",   uploaded_file.metadata["filename"]
+    assert_equal "text/plain", uploaded_file.metadata["mime_type"]
+  end
+
+  it "keeps any custom metadata" do
+    uploaded_file = @uploader.upload(fakeio)
+    uploaded_file.metadata["custom"] = "custom"
+    uploaded_file.refresh_metadata!
+    assert_equal "custom", uploaded_file.metadata["custom"]
+  end
+
+  it "forwards a Shrine::UploadedFile" do
+    uploaded_file = @uploader.upload(fakeio)
+    @uploader.class.plugin :add_metadata
+    @uploader.class.add_metadata(:uploaded_file) { |io| io.is_a?(Shrine::UploadedFile) }
+    uploaded_file.refresh_metadata!
+    assert_equal true, uploaded_file.metadata["uploaded_file"]
+  end
+
+  it "accepts additional context and forwards it" do
+    uploaded_file = @uploader.upload(fakeio)
+    @uploader.class.plugin :add_metadata
+    @uploader.class.add_metadata(:context) { |io, context| context.to_s }
+    uploaded_file.refresh_metadata!(foo: "bar")
+    assert_equal '{:foo=>"bar"}', uploaded_file.metadata["context"]
+  end
+end

test/plugin/restore_cached_data_test.rb

@@ -13,16 +13,17 @@
     assert_equal 1024, @attacher.get.metadata["size"]
   end
 
-  it "keeps any custom metadata" do
-    cached_file = @attacher.cache!(fakeio("image"))
-    cached_file.metadata["custom"] = "custom"
-    @attacher.assign(cached_file.to_json)
-    assert_equal "custom", @attacher.get.metadata["custom"]
-  end
-
   it "skips extracting if the file is not cached" do
-    stored_file = @attacher.store!(fakeio("image"))
+    stored_file = @attacher.store!(fakeio)
     @attacher.cache.expects(:extract_metadata).never
     @attacher.assign(stored_file.to_json)
   end
+
+  it "forwards the context" do
+    cached_file = @attacher.cache!(fakeio)
+    @attacher.shrine_class.plugin :add_metadata
+    @attacher.shrine_class.add_metadata(:context) { |io, context| context.keys.to_s }
+    @attacher.assign(cached_file.to_json)
+    assert_equal "[:record, :name]", @attacher.get.metadata["context"]
+  end
 end

www/_data/plugins.yml

@@ -77,6 +77,10 @@ Metadata:
     name: determine_mime_type
     path: DetermineMimeType.html
     description: Allows you to determine the actual MIME type from file contents.
+  -
+    name: refresh_metadata
+    path: RefreshMetadata.html
+    description: Allows re-extracting metadata of an uploaded file.
   -
     name: restore_cached_data
     path: RestoreCachedData.html