1 files changed, 128 insertions, 0 deletions
diff --git a/app/helpers/img_proxy_helper.rb b/app/helpers/img_proxy_helper.rb
new file mode 100644
index 000000000..5ea4bcd93
--- /dev/null
+++ b/app/helpers/img_proxy_helper.rb
@@ -0,0 +1,128 @@
+# frozen_string_literal: true
+
+#                  .~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.                  #
+###################              Cthulhu Code!              ###################
+#                  `~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~`                  #
+# - Has a high complexity level and needs tests.                              #
+# - Makes many assumptions the environment it's included into.                #
+# - Incurs a high performance penalty.                                        #
+#                                                                             #
+###############################################################################
+
+module ImgProxyHelper
+  def process_inline_images!
+    raise NameError('@status must be defined by the instance this method is being called from.') unless defined?(@status)
+    return if @status.text&.strip.blank? || @status.content_type == 'text/plain'
+
+    replace_markdown_images_with_html!
+
+    handler = ImgTagHandler.new
+    Ox.sax_parse(handler, StringIO.new(@status.text, 'r'))
+    return if handler.srcs.blank?
+
+    @skip_download_from = { @status.account.domain => DomainBlock.reject_media?(@status.account.domain) }
+    @redownload_attachment_ids = Set[]
+
+    handler.srcs.each do |src|
+      alt                   = handler.alts[src]
+      normalized_src_parts  = begin
+                                Addressable::URI.parse(src&.strip).normalize
+                              rescue Addressable::URI::InvalidURIError
+                                nil
+                              end
+      normalized_src        = normalized_src_parts.to_s
+
+      next replace_text!(src) if normalized_src.blank? || skip_download_from?(normalized_src_parts.host)
+
+      file_name             = normalized_src_parts.path.split('/').last
+      media_attachment      = find_media_attachment(normalized_src, file_name)
+
+      if media_attachment.present?
+        media_attachment.update(description: alt) if alt_more_descriptive?(alt, media_attachment.description)
+      elsif normalized_src_parts.scheme.blank? || !file_name.match?(/\S\.\w{3,}/)
+        next replace_text!(src)
+      else
+        media_attachment = create_media_attachment!(normalized_src, alt)
+      end
+
+      next replace_text!(src) if media_attachment.blank? || media_attachment.destroyed?
+
+      if media_attachment.needs_redownload?
+        replace_text!(src, "#{media_attachment.file.url(:small)}##{media_attachment.id}")
+      else
+        replace_text!(src, media_attachment.file.url(:small))
+      end
+    end
+  end
+
+  private
+
+  def skip_download_from?(domain)
+    return true if @skip_download_from[@status.account.domain]
+    return @skip_download_from[domain] if @skip_download_from[domain]
+
+    @skip_download_from[domain] = DomainBlock.reject_media?(domain)
+  end
+
+  def unsupported_media_type?(mime_type)
+    mime_type.present? && !MediaAttachment.supported_mime_types.include?(mime_type)
+  end
+
+  def html_entities
+    @html_entities ||= HTMLEntities.new
+  end
+
+  def replace_markdown_images_with_html!
+    return unless @status.content_type == 'text/markdown'
+
+    @status.text.gsub!(/!\[(\S+)\]\(\s*(\S+)\s*\)/) do
+      begin
+        alt = html_entities.encode(Regexp.last_match(1).strip)
+        url = Addressable::URI.parse(Regexp.last_match(2)).normalize.to_s
+        "<img title=\"#{alt}\" alt=\"#{alt}\" src=\"#{url}\" />"
+      rescue Addressable::URI::InvalidURIError
+        ''
+      end
+    end
+  end
+
+  def replace_text!(text, replacement = '')
+    @status.text.gsub!(text, replacement)
+  end
+
+  def alt_more_descriptive?(alt, description)
+    return false unless alt.present? && description != alt
+    return true if description.blank? || alt.split(/[\s\n\r]+/).count > description.split(/[\s\n\r]+/).count
+  end
+
+  def find_media_attachment(src, file_name)
+    media_attachment = src.start_with?('http') ? MediaAttachment.find_by(account: @account, remote_url: src, inline: true) : nil
+    return media_attachment if media_attachment.present?
+
+    MediaAttachment.where(account: @status.account, file_file_name: file_name, inline: true)
+                   .find { |m| [m.file.url(:small), m.file.url(:original)].include?(src) || m.status_id == @status.id }
+  end
+
+  def create_media_attachment!(src, alt)
+    media_attachment = MediaAttachment.create!(account: @status.account, remote_url: src, description: alt, focus: nil, inline: true)
+    media_attachment = process_media_attachment!(media_attachment)
+    return if media_attachment.destroyed?
+
+    @status.inlined_attachments.first_or_create!(media_attachment: media_attachment)
+    media_attachment
+  end
+
+  def process_media_attachment!(media_attachment)
+    media_attachment.download_file!
+    media_attachment.download_thumbnail!
+    media_attachment.save!
+    media_attachment.destroy! if unsupported_media_type?(media_attachment.file.content_type)
+    media_attachment
+  rescue Mastodon::UnexpectedResponseError, HTTP::TimeoutError, HTTP::ConnectionError, OpenSSL::SSL::SSLError
+    return if @redownload_attachment_ids.include?(media_attachment.id)
+
+    RedownloadMediaWorker.perform_in(rand(30..60).seconds, media_attachment.id)
+    @redownload_attachment_ids << media_attachment.id
+    media_attachment
+  end
+end