From 7e465aedaef287b9b9243d87cddc0b7155304cfd Mon Sep 17 00:00:00 2001
From: evazion <noizave@gmail.com>
Date: Sun, 4 Aug 2019 20:23:10 -0500
Subject: [PATCH] Fix #4110: New Twitter image urls are broken in bookmarklet.

---
 app/logical/sources/strategies/twitter.rb | 13 ++++++++++---
 test/unit/sources/twitter_test.rb         | 14 +++++++++++++-
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/app/logical/sources/strategies/twitter.rb b/app/logical/sources/strategies/twitter.rb
index 63dfd489a..279eaba70 100644
--- a/app/logical/sources/strategies/twitter.rb
+++ b/app/logical/sources/strategies/twitter.rb
@@ -4,6 +4,13 @@ module Sources::Strategies
     ASSET = %r!\A(https?://(?:video|pbs)\.twimg\.com/media/)!i
     PROFILE = %r!\Ahttps?://(?:mobile\.)?twitter.com/(?<username>[a-z0-9_]+)!i
 
+    # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb.jpg
+    # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb?format=jpg&name=900x900
+    BASE_IMAGE_URL = %r!\Ahttps?://pbs\.twimg\.com/media!i
+    FILENAME1 = %r!(?<file_name>\w+)\.(?<file_ext>\w+)!i
+    FILENAME2 = %r!(?<file_name>\w+)\?.*format=(?<file_ext>\w+)!i
+    IMAGE_URL = %r!#{BASE_IMAGE_URL}/#{Regexp.union(FILENAME1, FILENAME2)}!i
+
     # Twitter provides a list but it's inaccurate; some names ('intent') aren't
     # included and other names in the list aren't actually reserved.
     # https://developer.twitter.com/en/docs/developer-utilities/configuration/api-reference/get-help-configuration
@@ -40,8 +47,8 @@ module Sources::Strategies
     end
 
     def image_urls
-      if url =~ /(#{ASSET}[^:]+)/
-        [$1 + ":orig" ]
+      if url =~ IMAGE_URL
+        ["https://pbs.twimg.com/media/#{$~[:file_name]}.#{$~[:file_ext]}:orig"]
       elsif api_response.present?
         service.image_urls(api_response)
       else
@@ -57,7 +64,7 @@ module Sources::Strategies
     end
 
     def page_url
-      return "" if status_id.blank? || artist_name.blank?
+      return nil if status_id.blank? || artist_name.blank?
       "https://twitter.com/#{artist_name}/status/#{status_id}"
     end
 
diff --git a/test/unit/sources/twitter_test.rb b/test/unit/sources/twitter_test.rb
index 633d1dfc1..86e789e0e 100644
--- a/test/unit/sources/twitter_test.rb
+++ b/test/unit/sources/twitter_test.rb
@@ -24,7 +24,7 @@ module Sources
         assert_equal(urls, @site.image_urls)
       end
     end
-    
+
     context "A video" do
       setup do
         @site = Sources::Strategies.find("https://twitter.com/CincinnatiZoo/status/859073537713328129")
@@ -179,6 +179,18 @@ module Sources
       end
     end
 
+    context "The source site for a direct image url (pbs.twimg.com/media/*?format=jpg&name=*) without a referer url" do
+      setup do
+        @site = Sources::Strategies.find("https://pbs.twimg.com/media/EBGp2YdUYAA19Uj?format=jpg&name=small")
+      end
+
+      should "work" do
+        assert_equal("https://pbs.twimg.com/media/EBGp2YdUYAA19Uj.jpg:orig", @site.image_url)
+        assert_equal(["https://pbs.twimg.com/media/EBGp2YdUYAA19Uj.jpg:orig"], @site.image_urls)
+        assert_equal("https://pbs.twimg.com/media/EBGp2YdUYAA19Uj.jpg:orig", @site.canonical_url)
+      end
+    end
+
     context "The source site for a https://twitter.com/i/web/status/:id url" do
       setup do
         @site = Sources::Strategies.find("https://twitter.com/i/web/status/943446161586733056")