From 606b31184189b8344c5fe120aa0839c5c759b643 Mon Sep 17 00:00:00 2001
From: nonamethanks <hellafrickingepic@gmail.com>
Date: Wed, 21 Jul 2021 14:07:45 +0200
Subject: [PATCH] Fix skeb strategy

---
 app/logical/sources/strategies/skeb.rb |  7 ++++++-
 test/unit/sources/skeb_test.rb         | 28 +++++++++++++++++++++++++-
 2 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/app/logical/sources/strategies/skeb.rb b/app/logical/sources/strategies/skeb.rb
index 41200483d..12a9ef877 100644
--- a/app/logical/sources/strategies/skeb.rb
+++ b/app/logical/sources/strategies/skeb.rb
@@ -49,7 +49,12 @@ module Sources
           # Heavy heuristic to extract the uncropped image among the nighmare that is the skeb minified json
           candidates = page&.css("script")&.map { |script| script.text&.scan(/(https:\\u002F\\u002Fskeb\.imgix\.net.*?)(?:"|,|\s)/) }
           candidates = candidates.to_a.flatten.compact.uniq.reject { |match| match.include? "crop=" }
-          candidates.map { |img| img.gsub("\\u002F", "/") }
+          # sometimes skeb offers a slightly-smaller, non-watermarked version picture
+          unwatermarked = candidates.reject { |match| match.include? "=SAMPLE" }
+          unsampled = unwatermarked.reject { |match| match.include? "q=" }
+
+          final_candidates = [unsampled, unwatermarked, candidates].reject(&:empty?).first&.to_a
+          final_candidates.map { |img| img.gsub("\\u002F", "/") }
         else
           []
         end
diff --git a/test/unit/sources/skeb_test.rb b/test/unit/sources/skeb_test.rb
index f23d49bfc..a076f12e5 100644
--- a/test/unit/sources/skeb_test.rb
+++ b/test/unit/sources/skeb_test.rb
@@ -1,4 +1,4 @@
-require 'test_helper'
+require "test_helper"
 
 module Sources
   class SkebTest < ActiveSupport::TestCase
@@ -65,6 +65,32 @@ module Sources
       end
     end
 
+    context "A post with a smaller unwatermarked version" do
+      should "get the smaller but clean picture" do
+        site = Sources::Strategies.find("https://skeb.jp/@2gi0gi_/works/13")
+        assert_equal(["https://skeb.imgix.net/requests/191942_0?bg=%23fff&fm=jpg&q=45&w=696&s=5783ee951cc55d183713395926389453"], site.image_urls)
+      end
+    end
+
+    context "A post with both the small and large version clean" do
+      should "just get the bigger image" do
+        site = Sources::Strategies.find("https://skeb.jp/@LambOic029/works/149")
+        assert_equal(["https://skeb.imgix.net/uploads/origins/ebe94108-7ca7-4b3d-b80c-b37759ffd695?bg=%23fff&auto=format&w=800&s=25a889a808e6062d03985f7408201a4d"], site.image_urls)
+      end
+    end
+
+    context "A post with two images" do
+      should "get both correctly" do
+        site = Sources::Strategies.find("https://skeb.jp/@LambOic029/works/146")
+        image_urls = %w[
+          https://skeb.imgix.net/uploads/origins/e888bb27-e1a6-48ec-a317-7615252ff818?bg=%23fff&auto=format&w=800&s=7c518083d3fb19c8d5e7376f628f0fb0
+          https://skeb.imgix.net/uploads/origins/3fc062c5-231d-400f-921f-22d77cde54df?bg=%23fff&auto=format&w=800&s=f20697609ca2923f96fc49ca7eba22b6
+        ]
+
+        assert_equal(image_urls, site.image_urls)
+      end
+    end
+
     context "normalizing for source" do
       should "avoid normalizing unnormalizable urls" do
         bad_source = "https://skeb.imgix.net/requests/229088_2?bg=%23fff&auto=format&w=800&s=9cac8b76c0838f2df4f19ebc41c1ae0a"