Allow to rename XML nested structure elements on a loader level (#1203)

* Allow to rename XML nested structure elements on a loader level * Allow to pass xml attributes to XML Loader
flow-php · Aug 31, 2024 · 45c1ad8 · 45c1ad8
1 parent 7fcbbef
commit 45c1ad8
Show file tree

Hide file tree

Showing 4 changed files with 130 additions and 32 deletions.
diff --git a/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/Loader/XMLLoader.php b/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/Loader/XMLLoader.php
@@ -13,17 +13,34 @@
 
 final class XMLLoader implements Closure, Loader, Loader\FileLoader
 {
+    private string $attributePrefix = '_';
+
+    private string $dateTimeFormat = 'Y-m-d\TH:i:s.uP';
+
+    private string $listElementName = 'element';
+
+    private string $mapElementKeyName = 'key';
+
+    private string $mapElementName = 'element';
+
+    private string $mapElementValueName = 'value';
+
+    private string $rootElementName = 'root';
+
+    private string $rowElementName = 'row';
+
     /**
      * @var array<string, int>
      */
     private array $writes = [];
 
+    /**
+     * @var array<string, string>
+     */
+    private array $xmlAttributes = ['version' => '1.0', 'encoding' => 'UTF-8'];
+
     public function __construct(
         private readonly Path $path,
-        private readonly string $rootElementName,
-        private readonly string $rowElementName,
-        private readonly string $attributePrefix,
-        private readonly string $dateTimeFormat,
         private readonly XMLWriter $xmlWriter
     ) {
     }
@@ -48,16 +65,88 @@ public function load(Rows $rows, FlowContext $context) : void
     {
         $normalizer = new RowsNormalizer(
             new EntryNormalizer(
-                new PHPValueNormalizer($context->config->caster(), $this->attributePrefix, $this->dateTimeFormat),
-                $this->attributePrefix,
-                $this->dateTimeFormat
+                new PHPValueNormalizer(
+                    $context->config->caster(),
+                    $this->attributePrefix,
+                    $this->dateTimeFormat,
+                    $this->listElementName,
+                    $this->mapElementName,
+                    $this->mapElementKeyName,
+                    $this->mapElementValueName
+                ),
             ),
             $this->rowElementName
         );
 
         $this->write($rows, $rows->partitions()->toArray(), $context, $normalizer);
     }
 
+    public function withAttributePrefix(string $attributePrefix) : self
+    {
+        $this->attributePrefix = $attributePrefix;
+
+        return $this;
+    }
+
+    public function withDateTimeFormat(string $dateTimeFormat) : self
+    {
+        $this->dateTimeFormat = $dateTimeFormat;
+
+        return $this;
+    }
+
+    public function withListElementName(string $listElementName) : self
+    {
+        $this->listElementName = $listElementName;
+
+        return $this;
+    }
+
+    public function withMapElementKeyName(string $mapElementKeyName) : self
+    {
+        $this->mapElementKeyName = $mapElementKeyName;
+
+        return $this;
+    }
+
+    public function withMapElementName(string $mapElementName) : self
+    {
+        $this->mapElementName = $mapElementName;
+
+        return $this;
+    }
+
+    public function withMapElementValueName(string $mapElementValueName) : self
+    {
+        $this->mapElementValueName = $mapElementValueName;
+
+        return $this;
+    }
+
+    public function withRootElementName(string $rootElementName) : self
+    {
+        $this->rootElementName = $rootElementName;
+
+        return $this;
+    }
+
+    public function withRowElementName(string $rowElementName) : self
+    {
+        $this->rowElementName = $rowElementName;
+
+        return $this;
+    }
+
+    /**
+     * @param array<string, string> $xmlAttributes
+     */
+    public function withXMLAttributes(array $xmlAttributes) : self
+    {
+        $this->xmlAttributes = $xmlAttributes;
+
+        return $this;
+    }
+
     /**
      * @param array<Partition> $partitions
      */
@@ -72,7 +161,9 @@ public function write(Rows $nextRows, array $partitions, FlowContext $context, R
                 $this->writes[$stream->path()->path()] = 0;
             }
 
-            $stream->append("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<" . $this->rootElementName . ">\n");
+            $xmlAttributes = \implode(' ', \array_map(fn (string $key, string $value) => $key . '="' . $value . '"', \array_keys($this->xmlAttributes), \array_values($this->xmlAttributes)));
+
+            $stream->append('<?xml ' . $xmlAttributes . "?>\n<" . $this->rootElementName . ">\n");
         } else {
             $stream = $streams->writeTo($this->path, $partitions);
         }

diff --git a/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/RowsNormalizer/EntryNormalizer.php b/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/RowsNormalizer/EntryNormalizer.php
@@ -16,16 +16,14 @@ final class EntryNormalizer
 {
     public function __construct(
         private readonly PHPValueNormalizer $valueNormalizer,
-        private readonly string $attributePrefix = '_',
-        private readonly string $dateTimeFormat = PHPValueNormalizer::DATE_TIME_FORMAT
     ) {
 
     }
 
     public function normalize(Entry $entry) : XMLNode|XMLAttribute
     {
-        if (\str_starts_with($entry->name(), $this->attributePrefix)) {
-            return new XMLAttribute(\substr($entry->name(), \strlen($this->attributePrefix)), $entry->toString());
+        if (\str_starts_with($entry->name(), $this->valueNormalizer->attributePrefix)) {
+            return new XMLAttribute(\substr($entry->name(), \strlen($this->valueNormalizer->attributePrefix)), $entry->toString());
         }
 
         if ($entry instanceof ListEntry) {
@@ -45,7 +43,7 @@ public function normalize(Entry $entry) : XMLNode|XMLAttribute
             IntegerEntry::class => XMLNode::flatNode($entry->name(), (string) $entry->value()),
             FloatEntry::class => XMLNode::flatNode($entry->name(), (string) $entry->value()),
             BooleanEntry::class => XMLNode::flatNode($entry->name(), $entry->value() ? 'true' : 'false'),
-            DateTimeEntry::class => XMLNode::flatNode($entry->name(), $entry->value()?->format($this->dateTimeFormat)),
+            DateTimeEntry::class => XMLNode::flatNode($entry->name(), $entry->value()?->format($this->valueNormalizer->dateTimeFormat)),
             EnumEntry::class => XMLNode::flatNode($entry->name(), $entry->toString()),
             JsonEntry::class => XMLNode::flatNode($entry->name(), $entry->value()),
             UuidEntry::class => XMLNode::flatNode($entry->name(), $entry->toString()),
@@ -74,7 +72,7 @@ private function listToNode(ListEntry $entry) : XMLNode
         }
 
         foreach ($listValue as $value) {
-            $node = $node->append($this->valueNormalizer->normalize('element', $type->element()->type(), $value));
+            $node = $node->append($this->valueNormalizer->normalize($this->valueNormalizer->listElementName, $type->element()->type(), $value));
         }
 
         return $node;
@@ -125,8 +123,8 @@ private function mapToNode(MapEntry $entry) : XMLNode
         $type = $entry->type();
 
         foreach ($mapValue as $key => $value) {
-            $node = $node->append($this->valueNormalizer->normalize('key', $type->key()->type(), $key));
-            $node = $node->append($this->valueNormalizer->normalize('value', $type->value()->type(), $value));
+            $node = $node->append($this->valueNormalizer->normalize($this->valueNormalizer->mapElementKeyName, $type->key()->type(), $key));
+            $node = $node->append($this->valueNormalizer->normalize($this->valueNormalizer->mapElementValueName, $type->value()->type(), $value));
         }
 
         return $node;

diff --git a/...dapter-xml/src/Flow/ETL/Adapter/XML/RowsNormalizer/EntryNormalizer/PHPValueNormalizer.php b/...dapter-xml/src/Flow/ETL/Adapter/XML/RowsNormalizer/EntryNormalizer/PHPValueNormalizer.php
@@ -14,12 +14,14 @@
 
 final class PHPValueNormalizer
 {
-    public const DATE_TIME_FORMAT = 'Y-m-d\TH:i:s.uP';
-
     public function __construct(
         private readonly Caster $caster,
-        private readonly string $attributePrefix = '_',
-        private readonly string $dateTimeFormat = self::DATE_TIME_FORMAT
+        public readonly string $attributePrefix = '_',
+        public readonly string $dateTimeFormat = 'Y-m-d\TH:i:s.uP',
+        public readonly string $listElementName = 'element',
+        public readonly string $mapElementName = 'element',
+        public readonly string $mapElementKeyName = 'key',
+        public readonly string $mapElementValueName = 'value',
     ) {
 
     }
@@ -42,7 +44,7 @@ public function normalize(string $name, Type $type, mixed $value) : XMLNode|XMLA
             }
 
             foreach ($value as $elementValue) {
-                $listNode = $listNode->append($this->normalize('element', $type->element()->type(), $elementValue));
+                $listNode = $listNode->append($this->normalize($this->listElementName, $type->element()->type(), $elementValue));
             }
 
             return $listNode;
@@ -57,9 +59,9 @@ public function normalize(string $name, Type $type, mixed $value) : XMLNode|XMLA
 
             foreach ($value as $key => $elementValue) {
                 $mapNode = $mapNode->append(
-                    XMLNode::nestedNode('element')
-                        ->append($this->normalize('key', $type->key()->type(), $key))
-                        ->append($this->normalize('value', $type->value()->type(), $elementValue))
+                    XMLNode::nestedNode($this->mapElementName)
+                        ->append($this->normalize($this->mapElementKeyName, $type->key()->type(), $key))
+                        ->append($this->normalize($this->mapElementValueName, $type->value()->type(), $elementValue))
                 );
             }
 

diff --git a/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/functions.php b/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/functions.php
@@ -6,7 +6,6 @@
 
 use function Flow\ETL\DSL\from_all;
 use Flow\ETL\{Adapter\XML\Loader\XMLLoader,
-    Adapter\XML\RowsNormalizer\EntryNormalizer\PHPValueNormalizer,
     Adapter\XML\XMLWriter\DOMDocumentWriter,
     Attribute\DocumentationDSL,
     Attribute\Module,
@@ -42,21 +41,29 @@ function from_xml(
     );
 }
 
+/**
+ * @param Path|string $path
+ * @param string $root_element_name - @deprecated use `withRootElementName()` method instead
+ * @param string $row_element_name - @deprecated use `withRowElementName()` method instead
+ * @param string $attribute_prefix - @deprecated use `withAttributePrefix()` method instead
+ * @param string $date_time_format - @deprecated use `withDateTimeFormat()` method instead
+ * @param DOMDocumentWriter $xml_writer
+ */
 #[DocumentationDSL(module: Module::XML, type: DSLType::LOADER)]
 function to_xml(
     string|Path $path,
     string $root_element_name = 'rows',
     string $row_element_name = 'row',
     string $attribute_prefix = '_',
-    string $date_time_format = PHPValueNormalizer::DATE_TIME_FORMAT,
+    string $date_time_format = 'Y-m-d\TH:i:s.uP',
     XMLWriter $xml_writer = new DOMDocumentWriter()
 ) : XMLLoader {
-    return new XMLLoader(
+    return (new XMLLoader(
         \is_string($path) ? Path::realpath($path) : $path,
-        $root_element_name,
-        $row_element_name,
-        $attribute_prefix,
-        $date_time_format,
         $xml_writer
-    );
+    ))
+        ->withRootElementName($root_element_name)
+        ->withRowElementName($row_element_name)
+        ->withAttributePrefix($attribute_prefix)
+        ->withDateTimeFormat($date_time_format);
 }