Merge pull request cakephp#17869 from cakephp/feature/5.1-map-reduce

Allow specifying key when emitting intermediate value for MapReduce
CakeDC · Aug 27, 2024 · 910824d · 910824d
2 parents ddd9dc7 + 58e7fb4
commit 910824d
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 2 deletions.
diff --git a/src/Collection/Iterator/MapReduce.php b/src/Collection/Iterator/MapReduce.php
@@ -144,11 +144,18 @@ public function getIterator(): Traversable
      *
      * @param mixed $val The record itself to store in the bucket
      * @param mixed $bucket the name of the bucket where to put the record
+     * @param mixed $key An optional key to assign to the value
      * @return void
      */
-    public function emitIntermediate(mixed $val, mixed $bucket): void
+    public function emitIntermediate(mixed $val, mixed $bucket, mixed $key = null): void
     {
-        $this->_intermediate[$bucket][] = $val;
+        if ($key === null) {
+            $this->_intermediate[$bucket][] = $val;
+
+            return;
+        }
+
+        $this->_intermediate[$bucket][$key] = $val;
     }
 
     /**

diff --git a/tests/TestCase/Collection/Iterator/MapReduceTest.php b/tests/TestCase/Collection/Iterator/MapReduceTest.php
@@ -67,6 +67,32 @@ public function testInvertedIndexCreation(): void
         $this->assertEquals($expected, iterator_to_array($results));
     }
 
+    public function testSpecifyingKeyWhenEmittingIntermediate(): void
+    {
+        $data = [
+            'document_1' => 'one two three',
+            'document_2' => 'one three',
+            'document_3' => 'two four',
+        ];
+        $mapper = function ($row, $document, $mr): void {
+            $words = array_map('strtolower', explode(' ', $row));
+            foreach ($words as $word) {
+                $mr->emitIntermediate($document, $word, $document);
+            }
+        };
+        $reducer = function ($documents, $word, $mr): void {
+            $mr->emit(array_unique($documents), $word);
+        };
+        $results = new MapReduce(new ArrayIterator($data), $mapper, $reducer);
+        $expected = [
+            'one' => ['document_1' => 'document_1', 'document_2' => 'document_2'],
+            'two' => ['document_1' => 'document_1', 'document_3' => 'document_3'],
+            'three' => ['document_1' => 'document_1', 'document_2' => 'document_2'],
+            'four' => ['document_3' => 'document_3'],
+        ];
+        $this->assertEquals($expected, iterator_to_array($results));
+    }
+
     /**
      * Tests that it is possible to use the emit function directly in the mapper
      */