From 8662f3db2983ca54deb2e804790733ab302ac450 Mon Sep 17 00:00:00 2001
From: Joseph Bielawski <stloyd@gmail.com>
Date: Mon, 16 Oct 2023 14:44:22 +0200
Subject: [PATCH] Replace hashing algorithm from `sha256` to `xxh128`

---
 src/core/etl/src/Flow/ETL/Cache/LocalFilesystemCache.php   | 2 +-
 src/core/etl/src/Flow/ETL/GroupBy.php                      | 4 ++--
 src/core/etl/tests/Flow/ETL/Tests/Integration/FlowTest.php | 5 ++---
 3 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/src/core/etl/src/Flow/ETL/Cache/LocalFilesystemCache.php b/src/core/etl/src/Flow/ETL/Cache/LocalFilesystemCache.php
index e699c4fcc..831a5c967 100644
--- a/src/core/etl/src/Flow/ETL/Cache/LocalFilesystemCache.php
+++ b/src/core/etl/src/Flow/ETL/Cache/LocalFilesystemCache.php
@@ -90,6 +90,6 @@ public function read(string $id) : \Generator
 
     private function cachePath(string $id) : string
     {
-        return \rtrim($this->path, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR . \hash('sha256', $id);
+        return \rtrim($this->path, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR . \hash('xxh128', $id);
     }
 }
diff --git a/src/core/etl/src/Flow/ETL/GroupBy.php b/src/core/etl/src/Flow/ETL/GroupBy.php
index e69536451..4e176e164 100644
--- a/src/core/etl/src/Flow/ETL/GroupBy.php
+++ b/src/core/etl/src/Flow/ETL/GroupBy.php
@@ -120,12 +120,12 @@ private function hash(array $values) : string
         /** @var mixed $value */
         foreach ($values as $value) {
             if ($value === null) {
-                $stringValues[] =\hash('sha256', 'null');
+                $stringValues[] =\hash('xxh128', 'null');
             } elseif (\is_scalar($value)) {
                 $stringValues[] = (string) $value;
             }
         }
 
-        return \hash('sha256', \implode('', $stringValues));
+        return \hash('xxh128', \implode('', $stringValues));
     }
 }
diff --git a/src/core/etl/tests/Flow/ETL/Tests/Integration/FlowTest.php b/src/core/etl/tests/Flow/ETL/Tests/Integration/FlowTest.php
index 32dd2e8f6..ebe3febee 100644
--- a/src/core/etl/tests/Flow/ETL/Tests/Integration/FlowTest.php
+++ b/src/core/etl/tests/Flow/ETL/Tests/Integration/FlowTest.php
@@ -4,7 +4,6 @@
 
 namespace Flow\ETL\Tests\Integration;
 
-use function Flow\ETL\DSL\col;
 use function Flow\ETL\DSL\ref;
 use Flow\ETL\Cache\PSRSimpleCache;
 use Flow\ETL\Config;
@@ -27,7 +26,7 @@ public function test_etl_cache() : void
 
         $cacheContent = \array_values(\array_diff(\scandir($this->cacheDir), ['..', '.']));
 
-        $this->assertContains(\hash('sha256', 'test_etl_cache'), $cacheContent);
+        $this->assertContains(\hash('xxh128', 'test_etl_cache'), $cacheContent);
     }
 
     public function test_etl_psr_cache() : void
@@ -84,7 +83,7 @@ public function test_etl_sort_by_in_memory() : void
                 ->id($id = 'test_etl_sort_by_in_memory')
                 ->cache($cacheSpy = new CacheSpy(Config::default()->cache()))
         )->extract(new AllRowTypesFakeExtractor($rowsets = 20, $rows = 2))
-            ->sortBy(col('id'))
+            ->sortBy(ref('id'))
             ->fetch();
 
         $cache = \array_diff(\scandir($this->cacheDir), ['..', '.']);