Restored original behavior of DOMElementValue function combined with …

…XPath (#1220)
flow-php · Sep 12, 2024 · 9c0f8a5 · 9c0f8a5
1 parent e56255c
commit 9c0f8a5
Show file tree

Hide file tree

Showing 6 changed files with 212 additions and 16 deletions.
diff --git a/examples/topics/data_source/xml/output.txt b/examples/topics/data_source/xml/output.txt
@@ -1,15 +1,15 @@
-+----+------+--------+-------+
-| id | name | active | email |
-+----+------+--------+-------+
-|  1 |      |        |       |
-|  2 |      |        |       |
-|  3 |      |        |       |
-|  4 |      |        |       |
-|  5 |      |        |       |
-|  6 |      |        |       |
-|  7 |      |        |       |
-|  8 |      |        |       |
-|  9 |      |        |       |
-| 10 |      |        |       |
-+----+------+--------+-------+
++----+---------+--------+---------------------+
+| id |    name | active |               email |
++----+---------+--------+---------------------+
+|  1 |   Alice |   true |   alice@example.com |
+|  2 |     Bob |  false |     bob@example.com |
+|  3 | Charlie |   true | charlie@example.com |
+|  4 |   David |  false |   david@example.com |
+|  5 |    Emma |   true |    emma@example.com |
+|  6 |   Frank |  false |   frank@example.com |
+|  7 |   Grace |   true |   grace@example.com |
+|  8 |   Harry |  false |   harry@example.com |
+|  9 |    Isla |   true |    isla@example.com |
+| 10 |   James |  false |   james@example.com |
++----+---------+--------+---------------------+
 10 rows
diff --git a/src/core/etl/src/Flow/ETL/Function/DOMElementAttributeValue.php b/src/core/etl/src/Flow/ETL/Function/DOMElementAttributeValue.php
@@ -16,7 +16,20 @@ public function __construct(
 
     public function eval(Row $row) : ?string
     {
-        $domElement = (new Parameter($this->domElement))->asInstanceOf($row, \DOMElement::class);
+        $domElement = Parameter::oneOf(
+            (new Parameter($this->domElement))->asInstanceOf($row, \DOMElement::class),
+            (new Parameter($this->domElement))->asInstanceOf($row, \DOMDocument::class),
+            (new Parameter($this->domElement))->asListOfObjects($row, \DOMElement::class),
+        );
+
+        if ($domElement instanceof \DOMDocument) {
+            $domElement = $domElement->documentElement;
+        }
+
+        if (\is_array($domElement) && \count($domElement)) {
+            $domElement = \reset($domElement);
+        }
+
         $attributeName = (new Parameter($this->attribute))->asString($row);
 
         if ($domElement === null || $attributeName === null) {

diff --git a/src/core/etl/src/Flow/ETL/Function/DOMElementValue.php b/src/core/etl/src/Flow/ETL/Function/DOMElementValue.php
@@ -14,7 +14,19 @@ public function __construct(private readonly ScalarFunction|\DOMElement $domElem
 
     public function eval(Row $row) : mixed
     {
-        $domElement = (new Parameter($this->domElement))->asInstanceOf($row, \DOMElement::class);
+        $domElement = Parameter::oneOf(
+            (new Parameter($this->domElement))->asInstanceOf($row, \DOMElement::class),
+            (new Parameter($this->domElement))->asInstanceOf($row, \DOMDocument::class),
+            (new Parameter($this->domElement))->asListOfObjects($row, \DOMElement::class),
+        );
+
+        if (\is_array($domElement) && \count($domElement)) {
+            $domElement = \reset($domElement);
+        }
+
+        if ($domElement instanceof \DOMDocument) {
+            $domElement = $domElement->documentElement;
+        }
 
         if (!$domElement instanceof \DOMElement) {
             return null;

diff --git a/src/core/etl/src/Flow/ETL/Function/Parameter.php b/src/core/etl/src/Flow/ETL/Function/Parameter.php
@@ -97,6 +97,23 @@ public function asInt(Row $row) : ?int
         return \is_int($result) ? $result : null;
     }
 
+    public function asListOfObjects(Row $row, string $class) : ?array
+    {
+        $result = $this->function->eval($row);
+
+        if (!\is_array($result)) {
+            return null;
+        }
+
+        foreach ($result as $item) {
+            if (!\is_object($item) || !\is_a($item, $class)) {
+                return null;
+            }
+        }
+
+        return $result;
+    }
+
     /**
      * @psalm-suppress InvalidReturnType
      * @psalm-suppress InvalidReturnStatement

diff --git a/src/core/etl/tests/Flow/ETL/Tests/Integration/Function/DOMElementAttributeValueTest.php b/src/core/etl/tests/Flow/ETL/Tests/Integration/Function/DOMElementAttributeValueTest.php
@@ -0,0 +1,77 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Tests\Integration\Function;
+
+use function Flow\ETL\DSL\{df, from_rows, ref, row, rows, xml_element_entry, xml_entry};
+use Flow\ETL\Adapter\Elasticsearch\Tests\Integration\TestCase;
+
+final class DOMElementAttributeValueTest extends TestCase
+{
+    public function test_dom_element_attribute_value() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_element_entry('node', '<name id="1">User Name 01</name>')
+                    )
+                )
+            ))
+            ->withEntry('user_id', ref('node')->domElementAttributeValue('id'))
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_id' => '1'],
+            ],
+            $rows->toArray()
+        );
+    }
+
+    public function test_dom_element_attribute_value_from_dom_document() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_entry('node', '<name id="1">User Name 01</name>')
+                    )
+                )
+            ))
+            ->withEntry('user_id', ref('node')->domElementAttributeValue('id'))
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_id' => '1'],
+            ],
+            $rows->toArray()
+        );
+    }
+
+    public function test_dom_element_attribute_value_on_xpath_result() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_entry('node', '<user><name id="1">User Name 01</name></user>')
+                    )
+                )
+            ))
+            ->withEntry('user_id', ref('node')->xpath('name')->domElementAttributeValue('id'))
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_id' => '1'],
+            ],
+            $rows->toArray()
+        );
+    }
+}
diff --git a/src/core/etl/tests/Flow/ETL/Tests/Integration/Function/DOMElementValueTest.php b/src/core/etl/tests/Flow/ETL/Tests/Integration/Function/DOMElementValueTest.php
@@ -0,0 +1,77 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Tests\Integration\Function;
+
+use function Flow\ETL\DSL\{df, from_rows, ref, row, rows, xml_element_entry, xml_entry};
+use Flow\ETL\Adapter\Elasticsearch\Tests\Integration\TestCase;
+
+final class DOMElementValueTest extends TestCase
+{
+    public function test_dom_element_value() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_element_entry('node', '<name>User Name 01</name>')
+                    )
+                )
+            ))
+            ->withEntry('user_name', ref('node')->domElementValue())
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_name' => 'User Name 01'],
+            ],
+            $rows->toArray()
+        );
+    }
+
+    public function test_dom_element_value_from_dom_document() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_entry('node', '<name>User Name 01</name>')
+                    )
+                )
+            ))
+            ->withEntry('user_name', ref('node')->domElementValue())
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_name' => 'User Name 01'],
+            ],
+            $rows->toArray()
+        );
+    }
+
+    public function test_dom_element_value_on_xpath_result() : void
+    {
+        $rows = df()
+            ->read(from_rows(
+                rows(
+                    row(
+                        xml_entry('node', '<user><name>User Name 01</name></user>')
+                    )
+                )
+            ))
+            ->withEntry('user_name', ref('node')->xpath('name')->domElementValue())
+            ->drop('node')
+            ->fetch();
+
+        self::assertSame(
+            [
+                ['user_name' => 'User Name 01'],
+            ],
+            $rows->toArray()
+        );
+    }
+}