Fix remote files not using Content-Type header, streaming (#14184)

3 files changed, 107 insertions, 20 deletions

diff --git a/lib/paperclip/image_extractor.rb b/lib/paperclip/image_extractor.rb
index 114852e8b..f5a54d1a5 100644
--- a/lib/paperclip/image_extractor.rb
+++ b/lib/paperclip/image_extractor.rb
@@ -4,28 +4,10 @@ require 'mime/types/columnar'
 
 module Paperclip
   class ImageExtractor < Paperclip::Processor
-    IMAGE_EXTRACTION_OPTIONS = {
-      convert_options: {
-        output: {
-          'loglevel' => 'fatal',
-          vf: 'scale=\'min(400\, iw):min(400\, ih)\':force_original_aspect_ratio=decrease',
-        }.freeze,
-      }.freeze,
-      format: 'png',
-      time: -1,
-      file_geometry_parser: FastGeometryParser,
-    }.freeze
-
     def make
       return @file unless options[:style] == :original
 
-      image = begin
-        begin
-          Paperclip::Transcoder.make(file, IMAGE_EXTRACTION_OPTIONS.dup, attachment)
-        rescue Paperclip::Error, ::Av::CommandError
-          nil
-        end
-      end
+      image = extract_image_from_file!
 
       unless image.nil?
         begin
@@ -36,7 +18,7 @@ module Paperclip
           # to make sure it's cleaned up
 
           begin
-            FileUtils.rm(image)
+            image.close(true)
           rescue Errno::ENOENT
             nil
           end
@@ -45,5 +27,28 @@ module Paperclip
 
       @file
     end
+
+    private
+
+    def extract_image_from_file!
+      ::Av.logger = Paperclip.logger
+
+      cli = ::Av.cli
+      dst = Tempfile.new([File.basename(@file.path, '.*'), '.png'])
+      dst.binmode
+
+      cli.add_source(@file.path)
+      cli.add_destination(dst.path)
+      cli.add_output_param loglevel: 'fatal'
+
+      begin
+        cli.run
+      rescue Cocaine::ExitStatusError
+        dst.close(true)
+        return nil
+      end
+
+      dst
+    end
   end
 end
diff --git a/lib/paperclip/media_type_spoof_detector_extensions.rb b/lib/paperclip/media_type_spoof_detector_extensions.rb
new file mode 100644
index 000000000..9c0557356
--- /dev/null
+++ b/lib/paperclip/media_type_spoof_detector_extensions.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Paperclip
+  module MediaTypeSpoofDetectorExtensions
+    def calculated_content_type
+      @calculated_content_type ||= type_from_mime_magic || type_from_file_command
+    end
+
+    def type_from_mime_magic
+      @type_from_mime_magic ||= begin
+        begin
+          File.open(@file.path) do |file|
+            MimeMagic.by_magic(file)&.type
+          end
+        rescue Errno::ENOENT
+          ''
+        end
+      end
+    end
+
+    def type_from_file_command
+      @type_from_file_command ||= FileCommandContentTypeDetector.new(@file.path).detect
+    end
+  end
+end
+
+Paperclip::MediaTypeSpoofDetector.prepend(Paperclip::MediaTypeSpoofDetectorExtensions)
diff --git a/lib/paperclip/response_with_limit_adapter.rb b/lib/paperclip/response_with_limit_adapter.rb
new file mode 100644
index 000000000..7d897b8d6
--- /dev/null
+++ b/lib/paperclip/response_with_limit_adapter.rb
@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+module Paperclip
+  class ResponseWithLimitAdapter < AbstractAdapter
+    def self.register
+      Paperclip.io_adapters.register self do |target|
+        target.is_a?(ResponseWithLimit)
+      end
+    end
+
+    def initialize(target, options = {})
+      super
+      cache_current_values
+    end
+
+    private
+
+    def cache_current_values
+      @original_filename = filename_from_content_disposition || filename_from_path || 'data'
+      @size = @target.response.content_length
+      @tempfile = copy_to_tempfile(@target)
+      @content_type = @target.response.mime_type || ContentTypeDetector.new(@tempfile.path).detect
+    end
+
+    def copy_to_tempfile(source)
+      bytes_read = 0
+
+      source.response.body.each do |chunk|
+        bytes_read += chunk.bytesize
+
+        destination.write(chunk)
+        chunk.clear
+
+        raise Mastodon::LengthValidationError if bytes_read > source.limit
+      end
+
+      destination.rewind
+      destination
+    rescue Mastodon::LengthValidationError
+      destination.close(true)
+      raise
+    ensure
+      source.response.connection.close
+    end
+
+    def filename_from_content_disposition
+      disposition = @target.response.headers['content-disposition']
+      disposition&.match(/filename="([^"]*)"/)&.captures&.first
+    end
+
+    def filename_from_path
+      @target.response.uri.path.split('/').last
+    end
+  end
+end