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

2 files changed, 16 insertions, 41 deletions

diff --git a/app/models/concerns/attachmentable.rb b/app/models/concerns/attachmentable.rb
index 18b872c1e..c5febb828 100644
--- a/app/models/concerns/attachmentable.rb
+++ b/app/models/concerns/attachmentable.rb
@@ -8,6 +8,17 @@ module Attachmentable
   MAX_MATRIX_LIMIT = 16_777_216 # 4096x4096px or approx. 16MB
   GIF_MATRIX_LIMIT = 921_600    # 1280x720px
 
+  # For some file extensions, there exist different content
+  # type variants, and browsers often send the wrong one,
+  # for example, sending an audio .ogg file as video/ogg,
+  # likewise, MimeMagic also misreports them as such. For
+  # those files, it is necessary to use the output of the
+  # `file` utility instead
+  INCORRECT_CONTENT_TYPES = %w(
+    video/ogg
+    video/webm
+  ).freeze
+
   included do
     before_post_process :obfuscate_file_name
     before_post_process :set_file_extensions
@@ -21,7 +32,7 @@ module Attachmentable
     self.class.attachment_definitions.each_key do |attachment_name|
       attachment = send(attachment_name)
 
-      next if attachment.blank? || attachment.queued_for_write[:original].blank?
+      next if attachment.blank? || attachment.queued_for_write[:original].blank? || !INCORRECT_CONTENT_TYPES.include?(attachment.instance_read(:content_type))
 
       attachment.instance_write :content_type, calculated_content_type(attachment)
     end
@@ -63,9 +74,7 @@ module Attachmentable
   end
 
   def calculated_content_type(attachment)
-    content_type = Paperclip.run('file', '-b --mime :file', file: attachment.queued_for_write[:original].path).split(/[:;\s]+/).first.chomp
-    content_type = 'video/mp4' if content_type == 'video/x-m4v'
-    content_type
+    Paperclip.run('file', '-b --mime :file', file: attachment.queued_for_write[:original].path).split(/[:;\s]+/).first.chomp
   rescue Terrapin::CommandLineError
     ''
   end
diff --git a/app/models/concerns/remotable.rb b/app/models/concerns/remotable.rb
index 53ebc0835..c6d0c7f1f 100644
--- a/app/models/concerns/remotable.rb
+++ b/app/models/concerns/remotable.rb
@@ -24,28 +24,16 @@ module Remotable
           Request.new(:get, url).perform do |response|
             raise Mastodon::UnexpectedResponseError, response unless (200...300).cover?(response.code)
 
-            content_type = parse_content_type(response.headers.get('content-type').last)
-            extname      = detect_extname_from_content_type(content_type)
-
-            if extname.nil?
-              disposition = response.headers.get('content-disposition').last
-              matches     = disposition&.match(/filename="([^"]*)"/)
-              filename    = matches.nil? ? parsed_url.path.split('/').last : matches[1]
-              extname     = filename.nil? ? '' : File.extname(filename)
-            end
-
-            basename = SecureRandom.hex(8)
-
-            public_send("#{attachment_name}_file_name=", basename + extname)
-            public_send("#{attachment_name}=", StringIO.new(response.body_with_limit(limit)))
+            public_send("#{attachment_name}=", ResponseWithLimit.new(response, limit))
           end
         rescue Mastodon::UnexpectedResponseError, HTTP::TimeoutError, HTTP::ConnectionError, OpenSSL::SSL::SSLError => e
           Rails.logger.debug "Error fetching remote #{attachment_name}: #{e}"
           raise e unless suppress_errors
         rescue Paperclip::Errors::NotIdentifiedByImageMagickError, Addressable::URI::InvalidURIError, Mastodon::HostValidationError, Mastodon::LengthValidationError, Paperclip::Error, Mastodon::DimensionsValidationError => e
           Rails.logger.debug "Error fetching remote #{attachment_name}: #{e}"
-          nil
         end
+
+        nil
       end
 
       define_method("#{attribute_name}=") do |url|
@@ -59,26 +47,4 @@ module Remotable
       alias_method("reset_#{attachment_name}!", "download_#{attachment_name}!")
     end
   end
-
-  private
-
-  def detect_extname_from_content_type(content_type)
-    return if content_type.nil?
-
-    type = MIME::Types[content_type].first
-
-    return if type.nil?
-
-    extname = type.extensions.first
-
-    return if extname.nil?
-
-    ".#{extname}"
-  end
-
-  def parse_content_type(content_type)
-    return if content_type.nil?
-
-    content_type.split(/\s*;\s*/).first
-  end
 end