Merge pull request #23 from saloonphp/feature/v1-remove-namespaces

Feature | Remove Namespaces From Reader
saloonphp · Dec 5, 2023 · c26b2f4 · c26b2f4
2 parents fa9f5b0 + 1803f5f
commit c26b2f4
Show file tree

Hide file tree

Showing 3 changed files with 149 additions and 4 deletions.
diff --git a/src/XmlReader.php b/src/XmlReader.php
@@ -44,6 +44,11 @@ class XmlReader
      */
     protected array $xpathNamespaceMap = [];
 
+    /**
+     * Should XML Wrangler keep namespaces?
+     */
+    protected bool $keepNamespaces = true;
+
     /**
      * Constructor
      *
@@ -217,8 +222,16 @@ public function element(string $name, array $withAttributes = []): LazyQuery
             $matchers = [];
 
             foreach ($searchTerms as $index => $searchTerm) {
+                // When the search term is not numeric we must perform an element name
+                // search which will limit the results by a given element name. If
+                // namespaces were removed we must use element_local_name as this
+                // matches on elements no matter the prefix.
+
                 if (! is_numeric($searchTerm)) {
-                    $matchers[$index] = Matcher\element_name($searchTerm);
+                    $matchers[$index] = $this->keepNamespaces === true
+                        ? Matcher\element_name($searchTerm)
+                        : Matcher\element_local_name($searchTerm);
+
                     continue;
                 }
 
@@ -302,13 +315,20 @@ public function xpathElement(string $query): Query
             $xml = $this->reader->provide(Matcher\document_element())->current();
 
             $xpathConfigurators = [];
+            $keepNamespaces = $this->keepNamespaces;
             $namespaceMap = $this->xpathNamespaceMap;
 
+            if ($keepNamespaces === false && ! empty($namespaceMap)) {
+                throw new XmlReaderException('XPath namespace map cannot be used when namespaces are removed.');
+            }
+
             // When the namespace map is empty we will remove the root namespaces
             // because if they are not mapped then you cannot search on them.
 
             if (empty($namespaceMap)) {
-                $xml = Document::fromXmlString($xml, traverse(RemoveNamespaces::unprefixed()))->toXmlString();
+                $namespaceFilter = $keepNamespaces ? RemoveNamespaces::unprefixed() : RemoveNamespaces::all();
+
+                $xml = Document::fromXmlString($xml, traverse($namespaceFilter))->toXmlString();
             } else {
                 $xpathConfigurators[] = namespaces($namespaceMap);
             }
@@ -344,8 +364,8 @@ public function xpathElement(string $query): Query
     /**
      * Convert the XML into an array
      *
-     * @throws \Throwable
      * @return array<string, mixed>
+     * @throws \Throwable
      */
     public function values(): array
     {
@@ -413,7 +433,13 @@ protected function convertElementArrayIntoValues(array|Generator $elements): arr
      */
     protected function parseXml(string $xml): Element|array
     {
-        $decoded = xml_decode($xml);
+        $xmlConfigurators = [];
+
+        if ($this->keepNamespaces === false) {
+            $xmlConfigurators[] = traverse(RemoveNamespaces::all());
+        }
+
+        $decoded = xml_decode($xml, ...$xmlConfigurators);
 
         $firstKey = array_key_first($decoded);
 
@@ -474,6 +500,18 @@ public function setXpathNamespaceMap(array $xpathNamespaceMap): XmlReader
         return $this;
     }
 
+    /**
+     * Remove XML namespaces
+     *
+     * @return $this
+     */
+    public function removeNamespaces(): static
+    {
+        $this->keepNamespaces = false;
+
+        return $this;
+    }
+
     /**
      * Handle destructing the reader
      *

diff --git a/tests/Feature/XmlReaderTest.php b/tests/Feature/XmlReaderTest.php
@@ -553,3 +553,77 @@
 
     expect($mappedXpathTag)->toBe('1');
 });
+
+test('can remove namespaces and prefixes from xml', function () {
+    $reader = XmlReader::fromFile('tests/Fixtures/prefixed-breakfast-menu.xml');
+
+    expect($reader->element('food.0')->first())->toBeNull();
+
+    expect($reader->element('bkfst:food.0')->first())->toEqual(
+        Element::make()
+            ->setAttributes([
+                'soldOut' => 'false',
+                'bestSeller' => 'true',
+                'xmlns:bkfst' => 'http://breakfast.test/example/doesnt-exist',
+            ])
+            ->setContent([
+                'bkfst:name' => Element::make('Belgian Waffles'),
+                'bkfst:price' => Element::make('$5.95'),
+                'bkfst:description' => Element::make('Two of our famous Belgian Waffles with plenty of real maple syrup'),
+                'bkfst:calories' => Element::make('650'),
+            ])
+    );
+
+    // Now we'll remove the namespaces
+
+    $reader->removeNamespaces();
+
+    expect($reader->element('bkfst:food.0')->first())->toBeNull();
+
+    expect($reader->element('food.0')->first())->toEqual(
+        Element::make()
+            ->setAttributes([
+                'soldOut' => 'false',
+                'bestSeller' => 'true',
+            ])
+            ->setContent([
+                'name' => Element::make('Belgian Waffles'),
+                'price' => Element::make('$5.95'),
+                'description' => Element::make('Two of our famous Belgian Waffles with plenty of real maple syrup'),
+                'calories' => Element::make('650'),
+            ])
+    );
+
+    // We'll also make sure XPath works okay
+
+    expect($reader->xpathValue('//food[1]/name')->sole())->toEqual('Belgian Waffles');
+    expect($reader->xpathElement('//food[1]/name')->sole())->toEqual(Element::make('Belgian Waffles'));
+
+    // Test converting all values
+
+    $unprefixedNodes = 0;
+    $values = $reader->values();
+
+    expect($values)->toHaveKey('breakfast_menu');
+    expect($values['breakfast_menu'])->toHaveKey('food');
+    expect($values['breakfast_menu']['food'])->toHaveCount(5);
+
+    array_walk_recursive($values['breakfast_menu']['food'], function (mixed $value, string $key) use (&$unprefixedNodes) {
+        if (str_contains($key, ':') === false) {
+            $unprefixedNodes++;
+        }
+    });
+
+    expect($unprefixedNodes)->toBe(20);
+});
+
+test('cannot use an xpath namespace map without namespaces', function () {
+    $reader = XmlReader::fromFile('tests/Fixtures/prefixed-breakfast-menu.xml');
+
+    $reader->removeNamespaces();
+    $reader->setXpathNamespaceMap([
+        'bkfast' => 'http://breakfast.test/example/doesnt-exist',
+    ]);
+
+    $reader->xpathValue('//food[1]/name')->sole();
+})->throws(XmlReaderException::class, 'XPath namespace map cannot be used when namespaces are removed.');
diff --git a/tests/Fixtures/prefixed-breakfast-menu.xml b/tests/Fixtures/prefixed-breakfast-menu.xml
@@ -0,0 +1,33 @@
+<?xml version="1.0" encoding="utf-8"?>
+<bkfst:breakfast_menu name="Big G's Breakfasts" xmlns:bkfst="http://breakfast.test/example/doesnt-exist">
+  <bkfst:food soldOut="false" bestSeller="true">
+    <bkfst:name>Belgian Waffles</bkfst:name>
+    <bkfst:price>$5.95</bkfst:price>
+    <bkfst:description>Two of our famous Belgian Waffles with plenty of real maple syrup</bkfst:description>
+    <bkfst:calories>650</bkfst:calories>
+  </bkfst:food>
+  <bkfst:food soldOut="false" bestSeller="false">
+    <bkfst:name>Strawberry Belgian Waffles</bkfst:name>
+    <bkfst:price>$7.95</bkfst:price>
+    <bkfst:description>Light Belgian waffles covered with strawberries and whipped cream</bkfst:description>
+    <bkfst:calories>900</bkfst:calories>
+  </bkfst:food>
+  <bkfst:food soldOut="false" bestSeller="true">
+    <bkfst:name>Berry-Berry Belgian Waffles</bkfst:name>
+    <bkfst:price>$8.95</bkfst:price>
+    <bkfst:description>Light Belgian waffles covered with an assortment of fresh berries and whipped cream</bkfst:description>
+    <bkfst:calories>900</bkfst:calories>
+  </bkfst:food>
+  <bkfst:food soldOut="true" bestSeller="false">
+    <bkfst:name>French Toast</bkfst:name>
+    <bkfst:price>$4.50</bkfst:price>
+    <bkfst:description>Thick slices made from our homemade sourdough bread</bkfst:description>
+    <bkfst:calories>600</bkfst:calories>
+  </bkfst:food>
+  <bkfst:food soldOut="false" bestSeller="false">
+    <bkfst:name>Homestyle Breakfast</bkfst:name>
+    <bkfst:price>$6.95</bkfst:price>
+    <bkfst:description>Two eggs, bacon or sausage, toast, and our ever-popular hash browns</bkfst:description>
+    <bkfst:calories>950</bkfst:calories>
+  </bkfst:food>
+</bkfst:breakfast_menu>