PERF: stop downloading images from post processor and lean on uploads

Previously we would unconditionally fetch all images via HTTP to grab
original sizing from cooked post processor in 2 different spots.

This was wasteful as we already calculate and cache this info in upload records.

This also simplifies some specs and reduces use of mocks.

lib/cooked_post_processor.rb

@@ -153,14 +153,21 @@ class CookedPostProcessor
     src = img["src"]
     return if src.blank? || is_a_hyperlink?(img) || is_svg?(img)
 
-    original_width, original_height = (get_size(src) || [0, 0]).map(&:to_i)
-    if original_width == 0 || original_height == 0
-      Rails.logger.info "Can't reach '#{src}' to get its dimension."
-      return
-    end
-
     upload = Upload.get_from_url(src)
 
+    original_width, original_height  = nil
+
+    if (upload.present?)
+      original_width = upload.width || 0
+      original_height = upload.height || 0
+    else
+      original_width, original_height = (get_size(src) || [0, 0]).map(&:to_i)
+      if original_width == 0 || original_height == 0
+        Rails.logger.info "Can't reach '#{src}' to get its dimension."
+        return
+      end
+    end
+
     if (upload.present? && upload.animated?) || src.match?(GIF_SOURCES_REGEXP)
       img.add_class("animated")
     end

lib/cooked_processor_mixin.rb

@@ -45,10 +45,7 @@ module CookedProcessorMixin
     oneboxed_images.each do |img|
       next if img["src"].blank?
 
-      src = img["src"].sub(/^https?:/i, "")
       parent = img.parent
-      img_classes = (img["class"] || "").split(" ")
-      link_classes = ((parent&.name == "a" && parent["class"]) || "").split(" ")
 
       if respond_to?(:process_hotlinked_image, true)
         still_an_image = process_hotlinked_image(img)
@@ -183,7 +180,13 @@ module CookedProcessorMixin
 
     return unless is_valid_image_url?(absolute_url)
 
-    @size_cache[url] = FastImage.size(absolute_url)
+    upload = Upload.get_from_url(absolute_url)
+    if upload && upload.width && upload.width > 0
+      @size_cache[url] = [upload.width, upload.height]
+    else
+      @size_cache[url] = FastImage.size(absolute_url)
+    end
+
   rescue Zlib::BufError, URI::Error, OpenSSL::SSL::SSLError
     # FastImage.size raises BufError for some gifs, leave it.
   end

spec/fabricators/upload_fabricator.rb

@@ -21,6 +21,16 @@ Fabricator(:upload) do
   extension "png"
 end
 
+Fabricator(:large_image_upload, from: :upload) do
+  width 2000
+  height 2000
+  after_create do |upload, _transients|
+    file = file_from_fixtures("2000x2000.png")
+    upload.url = Discourse.store.store_upload(file, upload)
+    upload.sha1 = Upload.generate_digest(file)
+  end
+end
+
 Fabricator(:image_upload, from: :upload) do
   transient color: "white"
 

spec/jobs/pull_hotlinked_images_spec.rb

@@ -271,7 +271,6 @@ RSpec.describe Jobs::PullHotlinkedImages do
           post = Fabricate(:post, raw: "<img src='#{url}'>")
           upload.update(access_control_post: Fabricate(:post))
           FileStore::S3Store.any_instance.stubs(:store_upload).returns(upload.url)
-          FastImage.expects(:size).returns([100, 100]).at_least_once
 
           expect { Jobs::PullHotlinkedImages.new.execute(post_id: post.id) }
             .to change { Upload.count }.by(1)
@@ -374,7 +373,7 @@ RSpec.describe Jobs::PullHotlinkedImages do
         post.rebake!
         post.reload
 
-        expect(post.cooked).to match(/<img src=.*\/uploads/)
+        expect(post.cooked).to match(/<img src=.*\/uploads.*\ class="thumbnail/)
         expect(post.upload_references.count).to eq(1)
       end
 
@@ -418,7 +417,7 @@ RSpec.describe Jobs::PullHotlinkedImages do
         MD
 
         expect(post.cooked).to match(/<p><img src=.*\/uploads/)
-        expect(post.cooked).to match(/<img src=.*\/uploads.*\ class="thumbnail"/)
+        expect(post.cooked).to match(/<img src=.*\/uploads.*\ class="thumbnail/)
         expect(post.cooked).to match(/<span class="broken-image/)
         expect(post.cooked).to match(/<div class="large-image-placeholder">/)
       end
@@ -542,12 +541,10 @@ RSpec.describe Jobs::PullHotlinkedImages do
   end
 
   describe "with a lightboxed image" do
-    fab!(:upload) { Fabricate(:upload) }
+    fab!(:upload) { Fabricate(:large_image_upload) }
     fab!(:user) { Fabricate(:user) }
 
     before do
-      FastImage.expects(:size).returns([1750, 2000]).at_least_once
-      OptimizedImage.stubs(:resize).returns(true)
       Jobs.run_immediately!
     end
 

spec/lib/cooked_post_processor_spec.rb

@@ -5,6 +5,7 @@ require "file_store/s3_store"
 
 RSpec.describe CookedPostProcessor do
   fab!(:upload) { Fabricate(:upload) }
+  fab!(:large_image_upload) { Fabricate(:large_image_upload)  }
   let(:upload_path) { Discourse.store.upload_path }
 
   describe "#post_process" do
@@ -764,7 +765,6 @@ RSpec.describe CookedPostProcessor do
         end
 
         it "won't remove the original image if another post doesn't have an image" do
-          FastImage.stubs(:size)
           topic = post.topic
 
           cpp.post_process
@@ -781,8 +781,7 @@ RSpec.describe CookedPostProcessor do
         end
 
         it "generates thumbnails correctly" do
-          FastImage.expects(:size).returns([1750, 2000])
-
+          # image size in cooked is 1500*2000
           topic = post.topic
           cpp.post_process
           topic.reload
@@ -941,17 +940,13 @@ RSpec.describe CookedPostProcessor do
   describe "#convert_to_link" do
     fab!(:thumbnail) { Fabricate(:optimized_image, upload: upload, width: 512, height: 384) }
 
-    before do
-      CookedPostProcessor.any_instance.stubs(:get_size).with(upload.url).returns([1024, 768])
-    end
-
     it "adds lightbox and optimizes images" do
-      post = Fabricate(:post, raw: "![image|1024x768, 50%](#{upload.short_url})")
-
+      post = Fabricate(:post, raw: "![image|1024x768, 50%](#{large_image_upload.short_url})")
       cpp = CookedPostProcessor.new(post, disable_dominant_color: true)
       cpp.post_process
 
       doc = Nokogiri::HTML5::fragment(cpp.html)
+
       expect(doc.css('.lightbox-wrapper').size).to eq(1)
       expect(doc.css('img').first['srcset']).to_not eq(nil)
     end
@@ -1001,7 +996,7 @@ RSpec.describe CookedPostProcessor do
     it "optimizes images in quotes" do
       post = Fabricate(:post, raw: <<~MD)
         [quote]
-        ![image|1024x768, 50%](#{upload.short_url})
+        ![image|1024x768, 50%](#{large_image_upload.short_url})
         [/quote]
       MD
 
@@ -1016,7 +1011,7 @@ RSpec.describe CookedPostProcessor do
     it "optimizes images in Onebox" do
       Oneboxer.expects(:onebox)
         .with("https://discourse.org", anything)
-        .returns("<aside class='onebox'><img src='#{upload.url}' width='512' height='384'></aside>")
+        .returns("<aside class='onebox'><img src='#{large_image_upload.url}' width='512' height='384'></aside>")
 
       post = Fabricate(:post, raw: "https://discourse.org")