Improved code climate & coverage

jkphl · Feb 14, 2017 · 0d820b8 · 0d820b8
1 parent 2d3faee
commit 0d820b8
Show file tree

Hide file tree

Showing 10 changed files with 219 additions and 76 deletions.
diff --git a/src/Rdfalite/Domain/Property/PropertyInterface.php b/src/Rdfalite/Domain/Property/PropertyInterface.php
@@ -36,6 +36,7 @@
 
 namespace Jkphl\Rdfalite\Domain\Property;
 
+use Jkphl\Rdfalite\Domain\Thing\ThingInterface;
 use Jkphl\Rdfalite\Domain\Vocabulary\VocabularyInterface;
 
 /**
@@ -61,9 +62,9 @@ public function getName();
     public function getVocabulary();
 
     /**
-     * Property value
+     * Return the property value
      *
-     * @return string Property value
+     * @return string|ThingInterface Property value
      */
     public function getValue();
 }
diff --git a/src/Rdfalite/Infrastructure/Factories/HtmlDocumentFactory.php b/src/Rdfalite/Infrastructure/Factories/HtmlDocumentFactory.php
@@ -55,7 +55,9 @@ class HtmlDocumentFactory implements DocumentFactoryInterface
     public function createDocumentFromString($string)
     {
         $dom = new \DOMDocument();
+        libxml_use_internal_errors(true);
         $dom->loadHTML($string);
+        libxml_clear_errors();
         return $dom;
     }
 }
diff --git a/src/Rdfalite/Infrastructure/Parser/RdfaliteElementProcessor.php b/src/Rdfalite/Infrastructure/Parser/RdfaliteElementProcessor.php
@@ -53,6 +53,28 @@
  */
 class RdfaliteElementProcessor implements ElementProcessorInterface
 {
+    /**
+     * Tag name / attribute map
+     *
+     * @var array
+     */
+    protected static $tagNameAttributes = [
+        'META' => 'content',
+        'AUDIO' => 'src',
+        'EMBED' => 'src',
+        'IFRAME' => 'src',
+        'IMG' => 'src',
+        'SOURCE' => 'src',
+        'TRACK' => 'src',
+        'VIDEO' => 'src',
+        'A' => 'href',
+        'AREA' => 'href',
+        'LINK' => 'href',
+        'OBJECT' => 'data',
+        'DATA' => 'value',
+        'TIME' => 'datetime'
+    ];
+
     /**
      * Process a DOM element
      *
@@ -131,9 +153,18 @@ protected function processProperty(\DOMElement $element, Context $context)
                 // Try to get a resource ID
                 $resourceId = trim($element->getAttribute('resource')) ?: null;
 
+                // Get the property value
+                $propertyValue = $this->getPropertyValue($element, $context);
+                $property = new Property($name, $vocabulary, $propertyValue, $resourceId);
+
                 // Add the property to the current parent thing
-                $property = new Property($name, $vocabulary, $this->getPropertyValue($element, $context), $resourceId);
                 $context->getParentThing()->addProperty($property);
+
+                // If the property value is a thing
+                if ($propertyValue instanceof ThingInterface) {
+                    // Set the thing as parent thing for nested iterations
+                    $context = $context->setParentThing($propertyValue);
+                }
             }
         }
 
@@ -179,18 +210,21 @@ protected function getThing($typeof, $resourceId, Context $context)
         $prefix = array_pop($typeof);
 
         // Determine the vocabulary to use
-        $vocabulary = empty($prefix) ? $context->getDefaultVocabulary() : $context->getVocabulary($prefix);
-        if ($vocabulary instanceof VocabularyInterface) {
-            // Return a new thing
-            return new Thing($type, $vocabulary, $resourceId);
-        }
+        try {
+            $vocabulary = empty($prefix) ? $context->getDefaultVocabulary() : $context->getVocabulary($prefix);
+            if ($vocabulary instanceof VocabularyInterface) {
+                // Return a new thing
+                return new Thing($type, $vocabulary, $resourceId);
+            }
 
-        // If the default vocabulary is empty
-        if (empty($prefix)) {
-            throw new OutOfBoundsException(
-                OutOfBoundsException::EMPTY_DEFAULT_VOCABULARY_STR,
-                OutOfBoundsException::EMPTY_DEFAULT_VOCABULARY
-            );
+            // If the default vocabulary is empty
+            if (empty($prefix)) {
+                throw new OutOfBoundsException(
+                    OutOfBoundsException::EMPTY_DEFAULT_VOCABULARY_STR,
+                    OutOfBoundsException::EMPTY_DEFAULT_VOCABULARY
+                );
+            }
+        } catch (\Jkphl\Rdfalite\Application\Exceptions\OutOfBoundsException $e) {
         }
 
         throw new OutOfBoundsException(
@@ -209,29 +243,17 @@ protected function getPropertyStringValue(\DOMElement $element)
     {
         $tagName = strtoupper($element->tagName);
 
-        // Else: Depend on the tag name
-        switch (true) {
-            case $tagName === 'META':
-                return strval($element->getAttribute('content'));
-            case in_array($tagName, ['AUDIO', 'EMBED', 'IFRAME', 'IMG', 'SOURCE', 'TRACK', 'VIDEO']):
-                return strval($element->getAttribute('src'));
-
-            case in_array($tagName, ['A', 'AREA', 'LINK']):
-                return strval($element->getAttribute('href'));
-            case $tagName === 'OBJECT':
-                return strval($element->getAttribute('data'));
-            case $tagName === 'DATA':
-                return strval($element->getAttribute('value'));
-            case $tagName === 'TIME':
-                $datetime = $element->getAttribute('datetime');
-                if (!empty($datetime)) {
-                    return strval($datetime);
-                }
-            // fall through
-            default:
-//              trigger_error(sprintf('RDFa Lite 1.1 element processor: Unhandled tag name "%s"', $element->tagName), E_USER_WARNING);
-                return $element->textContent;
+        // Map to an attribute (if applicable)
+        if (array_key_exists($tagName, self::$tagNameAttributes)) {
+            $value = strval($element->getAttribute(self::$tagNameAttributes[$tagName]));
+            if (($tagName != 'TIME') || !empty($value)) {
+                return $value;
+            }
         }
+
+        // Return the text content
+//        trigger_error(sprintf('RDFa Lite 1.1 element processor: Unhandled tag name "%s"', $element->tagName), E_USER_WARNING);
+        return $element->textContent;
     }
 
     /**
@@ -256,7 +278,9 @@ public function processElementChildren(\DOMElement $element, Context $context)
      */
     protected function processTypeof(\DOMElement $element, Context $context)
     {
-        if ($element->hasAttribute('typeof')) {
+        if ($element->hasAttribute('typeof') &&
+            (!$element->hasAttribute('property') || empty($element->getAttribute('property')))
+        ) {
             $thing = $this->getThing(
                 $element->getAttribute('typeof'),
                 trim($element->getAttribute('resource')) ?: null,

diff --git a/src/Rdfalite/Tests/Application/DOMIteratorTest.php b/src/Rdfalite/Tests/Application/DOMIteratorTest.php
@@ -55,7 +55,7 @@ class DOMNodeIteratorTest extends ParserIteratorTestBase
     public function testDomNodeIteration()
     {
         $dom = new \DOMDocument();
-        $dom->loadHTML(self::$html);
+        $dom->loadHTML(self::$personRdfa);
         $context = new Context();
 
         $elementProcessor = $this->getMock(ElementProcessorInterface::class);
@@ -87,7 +87,7 @@ public function testDomNodeIteration()
     public function testRdfaLiteProcessor()
     {
         $dom = new \DOMDocument();
-        $dom->loadHTML(self::$html);
+        $dom->loadHTML(self::$personRdfa);
         $context = new Context();
         $domNodeIterator = new DOMIterator($dom->childNodes, $context, new RdfaliteElementProcessor());
         $this->assertInstanceOf(DOMIterator::class, $domNodeIterator);
@@ -106,6 +106,6 @@ public function testRdfaLiteProcessor()
             }
         }
         $this->assertEquals(0, count($elements));
-        $this->validateIteratorResult($context->getChildren());
+        $this->validatePersonResult($context->getChildren());
     }
 }
diff --git a/src/Rdfalite/Tests/Application/ParserIteratorTestBase.php b/src/Rdfalite/Tests/Application/ParserIteratorTestBase.php
@@ -40,6 +40,7 @@
 use Jkphl\Rdfalite\Domain\Property\PropertyInterface;
 use Jkphl\Rdfalite\Domain\Thing\ThingInterface;
 use Jkphl\Rdfalite\Domain\Vocabulary\Vocabulary;
+use Jkphl\Rdfalite\Domain\Vocabulary\VocabularyInterface;
 use Jkphl\Rdfalite\Tests\Domain\VocabularyTest;
 
 /**
@@ -51,19 +52,19 @@
 abstract class ParserIteratorTestBase extends \PHPUnit_Framework_TestCase
 {
     /**
-     * Test HTML
+     * Test HTML with person RDFa Lite 1.1
      *
      * @var string
      */
-    protected static $html;
+    protected static $personRdfa;
 
     /**
      * Setup all tests
      */
     public static function setUpBeforeClass()
     {
-        self::$html = file_get_contents(
-            dirname(__DIR__).DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'rdfa-lite-1.1.html'
+        self::$personRdfa = file_get_contents(
+            dirname(__DIR__).DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'person-rdfa-lite.html'
         );
     }
 
@@ -72,7 +73,7 @@ public static function setUpBeforeClass()
      *
      * @param ThingInterface[] $things Parsing Results
      */
-    protected function validateIteratorResult(array $things)
+    protected function validatePersonResult(array $things)
     {
         $schemaOrgVocabulary = new Vocabulary(VocabularyTest::SCHEMA_ORG);
 
@@ -89,29 +90,26 @@ protected function validateIteratorResult(array $things)
         $this->assertTrue(is_array($properties));
         $this->assertEquals(4, count($properties));
 
-        $name = $thing->getProperty('name');
-        $this->assertTrue(is_array($name));
-        $this->assertEquals(1, count($name));
-        $this->assertInstanceOf(PropertyInterface::class, $name[0]);
-        $this->assertEquals(new Property('name', $schemaOrgVocabulary, 'Joschi Kuphal'), $name[0]);
-
-        $telephone = $thing->getProperty('telephone');
-        $this->assertTrue(is_array($telephone));
-        $this->assertEquals(1, count($telephone));
-        $this->assertInstanceOf(PropertyInterface::class, $telephone[0]);
-        $this->assertEquals(new Property('telephone', $schemaOrgVocabulary, '+49 911 9593945'), $telephone[0]);
-
-        $image = $thing->getProperty('image');
-        $this->assertTrue(is_array($image));
-        $this->assertEquals(1, count($image));
-        $this->assertInstanceOf(PropertyInterface::class, $image[0]);
-        $this->assertEquals(new Property('image', $schemaOrgVocabulary, 'https://jkphl.is/avatar.jpg'), $image[0]);
+        $this->validateProperty($thing, 'name', $schemaOrgVocabulary, 'Joschi Kuphal');
+        $this->validateProperty($thing, 'telephone', $schemaOrgVocabulary, '+49 911 9593945');
+        $this->validateProperty($thing, 'image', $schemaOrgVocabulary, 'https://jkphl.is/avatar.jpg');
+        $this->validateProperty($thing, 'preferredAnimal', new Vocabulary('http://open.vocab.org/terms/'), 'Unicorn');
+    }
 
-        $preferredAnimal = $thing->getProperty('preferredAnimal');
-        $this->assertTrue(is_array($preferredAnimal));
-        $this->assertEquals(1, count($preferredAnimal));
-        $this->assertInstanceOf(PropertyInterface::class, $preferredAnimal[0]);
-        $this->assertEquals(new Property('preferredAnimal', new Vocabulary('http://open.vocab.org/terms/'), 'Unicorn'),
-            $preferredAnimal[0]);
+    /**
+     * Validate a single property
+     *
+     * @param ThingInterface $thing Thing
+     * @param string $name Property name
+     * @param VocabularyInterface $vocabulary Property vocabulary
+     * @param string $value Property value
+     */
+    protected function validateProperty(ThingInterface $thing, $name, VocabularyInterface $vocabulary, $value)
+    {
+        $property = $thing->getProperty($name);
+        $this->assertTrue(is_array($property));
+        $this->assertEquals(1, count($property));
+        $this->assertInstanceOf(PropertyInterface::class, $property[0]);
+        $this->assertEquals(new Property($name, $vocabulary, $value), $property[0]);
     }
 }
diff --git a/src/Rdfalite/Tests/Application/ParserTest.php b/src/Rdfalite/Tests/Application/ParserTest.php
@@ -49,16 +49,74 @@
 class ParserTest extends ParserIteratorTestBase
 {
     /**
-     * Test parser context instantiation
+     * Test person parsing
      */
-    public function testParser()
+    public function testPerson()
     {
         $htmlDocumentFactory = new HtmlDocumentFactory();
         $rdfaElementProcessor = new RdfaliteElementProcessor();
         $parser = new Parser($htmlDocumentFactory, $rdfaElementProcessor);
         $this->assertInstanceOf(Parser::class, $parser);
 
-        $things = $parser->parse(self::$html);
-        $this->validateIteratorResult($things);
+        $things = $parser->parse(self::$personRdfa);
+        $this->validatePersonResult($things);
+    }
+
+    /**
+     * Test article parsing
+     */
+    public function testArticle()
+    {
+        $htmlDocumentFactory = new HtmlDocumentFactory();
+        $rdfaElementProcessor = new RdfaliteElementProcessor();
+        $parser = new Parser($htmlDocumentFactory, $rdfaElementProcessor);
+        $this->assertInstanceOf(Parser::class, $parser);
+
+        // TODO: Validate result
+        $parser->parse(
+            file_get_contents(
+                dirname(__DIR__).DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'article-rdfa-lite.html'
+            )
+        );
+    }
+
+    /**
+     * Test empty default vocabulary parsing
+     *
+     * @expectedException \Jkphl\Rdfalite\Infrastructure\Exceptions\OutOfBoundsException
+     * @expectedExceptionCode 1487030264
+     */
+    public function testEmptyDefaultVocabulary()
+    {
+        $htmlDocumentFactory = new HtmlDocumentFactory();
+        $rdfaElementProcessor = new RdfaliteElementProcessor();
+        $parser = new Parser($htmlDocumentFactory, $rdfaElementProcessor);
+        $this->assertInstanceOf(Parser::class, $parser);
+
+        $parser->parse(
+            file_get_contents(
+                dirname(__DIR__).DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'empty-default-vocab-rdfa-lite.html'
+            )
+        );
+    }
+
+    /**
+     * Test unknown vocabulary prefix parsing
+     *
+     * @expectedException \Jkphl\Rdfalite\Infrastructure\Exceptions\OutOfBoundsException
+     * @expectedExceptionCode 1486928423
+     */
+    public function testUnknownVocabularyPrefix()
+    {
+        $htmlDocumentFactory = new HtmlDocumentFactory();
+        $rdfaElementProcessor = new RdfaliteElementProcessor();
+        $parser = new Parser($htmlDocumentFactory, $rdfaElementProcessor);
+        $this->assertInstanceOf(Parser::class, $parser);
+
+        $parser->parse(
+            file_get_contents(
+                dirname(__DIR__).DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'unknown-vocab-prefix-rdfa-lite.html'
+            )
+        );
     }
 }
diff --git a/src/Rdfalite/Tests/Fixtures/article-rdfa-lite.html b/src/Rdfalite/Tests/Fixtures/article-rdfa-lite.html
@@ -0,0 +1,25 @@
+<!DOCTYPE html>
+<html lang="en">
+    <head>
+        <meta charset="UTF-8">
+        <title>News article</title>
+    </head>
+    <body vocab="http://schema.org/">
+        <article typeof="NewsArticle">
+            <h1 property="name">Nam dictum eros at libero tempor semper</h1>
+            <p>Published on
+                <time property="datePublished" datetime="2014-03-04T09:00:00+00:00">4th March 2014</time>
+            </p>
+            <p property="description">Nullam in convallis metus, quis hendrerit velit. Nam mollis congue orci, a
+                lobortis nibh feugiat id.</p>
+
+            <div property="articleBody">
+                <figure property="associatedMedia" typeof="ImageObject">
+                    <img property="contentURL" src="image.jpg">
+                    <figcaption property="caption">Lorem ipsum dolor.</figcaption>
+                </figure>
+                <p>News article content here.</p>
+            </div>
+        </article>
+    </body>
+</html>