#398

backend/app/services/preview/thumbnail_fetcher.rb

@@ -1,26 +1,47 @@
 module Preview
   class ThumbnailFetcher
     class GenerationFailed < StandardError; end
+    HTML_MAX_BYTES = 1.megabyte
+    NICONICO_XML_MAX_BYTES = 256.kilobytes
 
     def self.fetch(raw_url)
       uri, = UrlSafety.validate(raw_url)
 
       known_url = KnownSiteExtractor.thumbnail_url(uri)
-      return fetch_image(known_url) if known_url
+      image = fetch_image_or_nil(known_url) if known_url
+      return image if image
 
-      page = HttpFetcher.fetch(uri.to_s)
+      niconico_url = niconico_thumbnail_url(uri)
+      image = fetch_image_or_nil(niconico_url) if niconico_url
+      return image if image
+
+      page = HttpFetcher.fetch(uri.to_s, max_bytes: HTML_MAX_BYTES)
       metadata = HtmlMetadataExtractor.extract(page)
       raise GenerationFailed, 'サムネール画像が見つかりませんでした．' if metadata[:image_url].blank?
 
-      fetch_image(metadata[:image_url])
+      fetch_image!(metadata[:image_url])
     end
 
     def self.title(raw_url)
       uri, = UrlSafety.validate(raw_url)
-      HtmlMetadataExtractor.extract(HttpFetcher.fetch(uri.to_s))[:title]
+      HtmlMetadataExtractor.extract(
+          HttpFetcher.fetch(uri.to_s, max_bytes: HTML_MAX_BYTES))[:title]
     end
 
-    def self.fetch_image(url)
+    def self.fetch_image_or_nil(url)
+      return nil if url.blank?
+
+      response = HttpFetcher.fetch(url)
+      return nil unless response.content_type.downcase.start_with?('image/')
+
+      response.body
+    rescue HttpFetcher::FetchTimeout
+      raise
+    rescue HttpFetcher::FetchFailed
+      nil
+    end
+
+    def self.fetch_image!(url)
       response = HttpFetcher.fetch(url)
       unless response.content_type.downcase.start_with?('image/')
         raise GenerationFailed, 'サムネール画像が見つかりませんでした．'
@@ -29,10 +50,37 @@ module Preview
       response.body
     rescue HttpFetcher::FetchTimeout
       raise
+    rescue HttpFetcher::ResponseTooLarge
+      raise
     rescue HttpFetcher::FetchFailed
       raise GenerationFailed, 'サムネール画像を取得できませんでした．'
     end
 
-    private_class_method :fetch_image
+    def self.niconico_thumbnail_url(uri)
+      video_id = KnownSiteExtractor.niconico_video_id(uri)
+      return nil if video_id.blank?
+
+      response = HttpFetcher.fetch("https://ext.nicovideo.jp/api/getthumbinfo/#{ video_id }",
+                                   max_bytes: NICONICO_XML_MAX_BYTES)
+      xml = Nokogiri::XML(response.body)
+      return nil unless xml.at_xpath('/nicovideo_thumb_response/@status')&.value == 'ok'
+
+      xml.at_xpath('//thumbnail_url')&.text&.strip.presence
+    rescue HttpFetcher::FetchFailed, HttpFetcher::FetchTimeout => e
+      Rails.logger.info("preview_niconico_getthumbinfo_fallback #{ { url: uri.to_s,
+                                                                     video_id:,
+                                                                     error: e.class.name,
+                                                                     message: e.message }.to_json }")
+      nil
+    rescue Nokogiri::XML::SyntaxError => e
+      Rails.logger.info("preview_niconico_getthumbinfo_fallback #{ { url: uri.to_s,
+                                                                     video_id:,
+                                                                     error: e.class.name,
+                                                                     message: e.message }.to_json }")
+      nil
+    end
+
+    private_class_method :fetch_image_or_nil, :fetch_image!,
+                         :niconico_thumbnail_url
   end
 end