From 91bebdb038723b0620b07eef806a825658ae8988 Mon Sep 17 00:00:00 2001
From: Adam Cutler <adam.r.cutler@gmail.com>
Date: Sat, 25 Mar 2023 14:14:01 -0400
Subject: [PATCH] fix: work with extremely large results (#291)

* fix: work with extremely large results

* add test

* back to ff?

* test pls

* waits?

* 60M is too large

* stop comparing html. selenium adds attributes
---
 .../e2e/selenium/spec/api_spec.rb             | 20 ++++++++++----
 packages/axe-core-api/lib/axe/api/run.rb      | 26 ++++++++++++++++---
 2 files changed, 37 insertions(+), 9 deletions(-)

diff --git a/packages/axe-core-api/e2e/selenium/spec/api_spec.rb b/packages/axe-core-api/e2e/selenium/spec/api_spec.rb
index 848a7c4c..c2cb7ce7 100644
--- a/packages/axe-core-api/e2e/selenium/spec/api_spec.rb
+++ b/packages/axe-core-api/e2e/selenium/spec/api_spec.rb
@@ -4,9 +4,9 @@
 require "axe/core"
 require "axe/api/run"
 
-options = Selenium::WebDriver::Firefox::Options.new
-options.add_argument('--headless')
-$driver = Selenium::WebDriver.for :firefox, options: options
+options = Selenium::WebDriver::Chrome::Options.new
+# options.add_argument('--headless')
+$driver = Selenium::WebDriver.for :chrome, options: options
 
 Run = Axe::API::Run
 
@@ -20,6 +20,7 @@ def run_axe(run = Run.new)
 $axe_post_43x = Axe::Configuration.instance.jslib
 $crasher_js = File.read File.join($fixture_root, "axe-crasher.js")
 $force_legacy_js = File.read File.join($fixture_root, "axe-force-legacy.js")
+$large_partial_js = File.read File.join($fixture_root, "axe-large-partial.js")
 
 def fixture(filename)
   "http://localhost:8000" + filename
@@ -56,7 +57,7 @@ def recursive_compact(thing)
   elsif thing.is_a?(Hash)
     thing.each_with_object({}) do |(k,v), h|
       v = recursive_compact(v)
-      h[k] = v unless [nil, [], {}].include?(v)
+      h[k] = v unless ([nil, [], {}].include?(v) or k == :html)
     end
   else
     thing
@@ -92,7 +93,7 @@ def recursive_compact(thing)
 end
 
 describe "frame tests" do
-  it "injects into nested iframes" do
+  it "injects into nested iframes", :fo => true do
     $driver.get fixture "/nested-iframes.html"
     res = run_axe
     expect(res.results.violations).not_to be_empty
@@ -268,6 +269,15 @@ def recursive_compact(thing)
       expect { run_axe }.to raise_error /finishRun failed/
     }
   end
+
+  it "works with large results", :nt => true do
+    $driver.get fixture "/index.html"
+    res = with_js($axe_post_43x + $large_partial_js) { run_axe }
+
+
+    expect(res.results.passes.length).to eq 1
+    expect(res.results.passes[0].id).to eq :'duplicate-id'
+  end
 end
 
 
diff --git a/packages/axe-core-api/lib/axe/api/run.rb b/packages/axe-core-api/lib/axe/api/run.rb
index e7bc5451..6949f306 100644
--- a/packages/axe-core-api/lib/axe/api/run.rb
+++ b/packages/axe-core-api/lib/axe/api/run.rb
@@ -36,9 +36,19 @@ def analyze_post_43x(page, lib)
         partial_results = run_partial_recursive(page, @context, lib, true)
         throw partial_results if partial_results.respond_to?("key?") and partial_results.key?("errorMessage")
         results = within_about_blank_context(page) { |page|
+          partial_res_str = partial_results.to_json
+          size_limit = 20_000_000
+          while not partial_res_str.empty? do
+            chunk_size = size_limit
+            chunk_size = partial_res_str.length if chunk_size > partial_res_str.length
+            chunk = partial_res_str[0..chunk_size-1]
+            partial_res_str = partial_res_str[chunk_size..-1]
+            store_chunk page, chunk
+          end
+
           Common::Loader.new(page, lib).load_top_level Axe::Configuration.instance.jslib
           begin
-            axe_finish_run page, partial_results
+            axe_finish_run page
           rescue
             raise StandardError.new "axe.finishRun failed. Please check out https://github.com/dequelabs/axe-core-gems/blob/develop/error-handling.md"
           end
@@ -138,12 +148,20 @@ def run_partial_recursive(page, context, lib, top_level = false)
         return results
       end
 
-      def axe_finish_run(page, partial_results)
+      def store_chunk(page, chunk)
+        script = <<-JS
+          const chunk = arguments[0];
+          window.partialResults ??= '';
+          window.partialResults += chunk;
+        JS
+        page.execute_script_fixed script, chunk
+      end
+      def axe_finish_run(page)
         script = <<-JS
-          const partialResults = arguments[0];
+          const partialResults = JSON.parse(window.partialResults || '[]');
           return axe.finishRun(partialResults);
         JS
-        page.execute_script_fixed script, partial_results
+        page.execute_script_fixed script
       end
 
       def axe_shadow_select(page, frame_selector)