Merge pull request #4585 from nonamethanks/fix_tumblr

Tumblr: support highest res for new image urls

app/logical/sources/strategies/tumblr.rb

@@ -7,6 +7,8 @@
 # https://66.media.tumblr.com/5a2c3fe25c977e2281392752ab971c90/3dbfaec9b9e0c2e3-92/s500x750/4f92bbaaf95c0b4e7970e62b1d2e1415859dd659.png
 #
 # https://superboin.tumblr.com/post/141169066579/photoset_iframe/superboin/tumblr_o45miiAOts1u6rxu8/500/false
+#
+# https://make-do5.tumblr.com/post/619663949657423872 (extremely high res, extractable)
 
 module Sources::Strategies
   class Tumblr < Base
@@ -26,6 +28,11 @@ module Sources::Strategies
     VIDEO = %r{\Ahttps?://(?:vtt|ve|va\.media)\.tumblr\.com/}i
     POST = %r{\Ahttps?://(?<blog_name>[^.]+)\.tumblr\.com/(?:post|image)/(?<post_id>\d+)}i
 
+    NEW_HEADERS = {
+      "user-agent": Danbooru.config.canonical_app_name,
+      "accept": "text/html"
+    }
+
     def self.enabled?
       Danbooru.config.tumblr_consumer_key.present?
     end
@@ -161,14 +168,22 @@ module Sources::Strategies
     # http://media.tumblr.com/tumblr_m24kbxqKAX1rszquso1_1280.jpg
     # => https://media.tumblr.com/tumblr_m24kbxqKAX1rszquso1_1280.jpg
     def find_largest(url, sizes: SIZES)
-      return url unless url =~ OLD_IMAGE
+      if url =~ OLD_IMAGE
+        candidates = sizes.map do |size|
+          "https://media.tumblr.com/#{$~[:dir]}#{$~[:filename]}_#{size}.#{$~[:ext]}"
+        end
 
-      candidates = sizes.map do |size|
+        candidates.find do |candidate|
-        "https://media.tumblr.com/#{$~[:dir]}#{$~[:filename]}_#{size}.#{$~[:ext]}"
+          http_exists?(candidate)
-      end
+        end
+      elsif url =~ %r{/s\d+x\d+/(\w+\.\w+)$}i
+        max_size = Integer.sqrt(Danbooru.config.max_image_resolution)
+        url = url.gsub(%r{/s\d+x\d+/\w+\.\w+$}i, "/s#{max_size}x#{max_size}/#{$1}")
 
-      candidates.find do |candidate|
+        resp = Danbooru::Http.cache(1.minute).get(url, headers: NEW_HEADERS).parse
-        http_exists?(candidate)
+        resp.at("img[src*='/s#{max_size}x#{max_size}/']")["src"]
+      else
+        url
       end
     end
 

test/unit/sources/tumblr_test.rb

@@ -220,14 +220,13 @@ module Sources
 
     context "A Tumblr post with new image URLs" do
       should "return the correct image url" do
-        page_url = "https://emlan.tumblr.com/post/189469423572/kuro-attempts-to-buy-a-racy-book-at-comiket-but"
+        image_url = "https://64.media.tumblr.com/3dfdab77d913ad1ea59f22407d6ac6f3/b1764aa0f9c378d0-23/s1280x1920/46f4af7ec94456f8fef380ee6311eb81178ce7e9.jpg"
-        image1_url = "https://64.media.tumblr.com/168dabd09d5ad69eb5fedcf94c45c31a/3dbfaec9b9e0c2e3-72/s640x960/bf33a1324f3f36d2dc64f011bfeab4867da62bc8.png"
+        page_url = "https://make-do5.tumblr.com/post/619663949657423872"
-        image2_url = "https://64.media.tumblr.com/5a2c3fe25c977e2281392752ab971c90/3dbfaec9b9e0c2e3-92/s540x810/cd270c29db06b5e7fdcee63114fe3eb2c9c0d590.png"
+        strategy = Sources::Strategies.find(image_url, page_url)
-        strategy = Sources::Strategies.find(image2_url, page_url)
 
-        assert_equal([image1_url, image2_url], strategy.image_urls)
+        assert_match(%r{/3dfdab77d913ad1ea59f22407d6ac6f3/b1764aa0f9c378d0-23/s\d+x\d+/}i, image_url)
-        assert_equal(image2_url, strategy.image_url)
+        assert_equal(page_url, strategy.canonical_url)
-        assert_equal("https://emlan.tumblr.com/post/189469423572", strategy.canonical_url)
+        assert_downloaded(7_428_704, strategy.image_url)
       end
     end