Implement TestPage and upgrade the test suite

2 files changed, 132 insertions, 42 deletions

diff --git a/test/ReadabilityTest.php b/test/ReadabilityTest.php
index 466680e..df2a272 100644
--- a/test/ReadabilityTest.php
+++ b/test/ReadabilityTest.php
@@ -6,12 +6,21 @@ use andreskrey\Readability\Configuration;
 use andreskrey\Readability\ParseException;
 use andreskrey\Readability\Readability;
 
+/**
+ * Class ReadabilityTest
+ * @package andreskrey\Readability\Test
+ */
 class ReadabilityTest extends \PHPUnit\Framework\TestCase
 {
     /**
+     * Test that Readability parses the HTML correctly and matches the expected result
+     *
      * @dataProvider getSamplePages
+     * @param TestPage $testPage
+     *
+     * @throws ParseException
      */
-    public function testReadabilityParsesHTML($html, $expectedResult, $expectedMetadata, $config, $expectedImages)
+    public function testReadabilityParsesHTML(TestPage $testPage)
     {
         $options = ['OriginalURL' => 'http://fakehost/test/test.html',
             'FixRelativeURLs' => true,
@@ -19,28 +28,47 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
             'ArticleByLine' => true
         ];
 
-        if ($config === null || $expectedMetadata === null) {
-            $this->markTestSkipped('Wrong test configuration');
-        }
+        $configuration = new Configuration(array_merge($testPage->getConfiguration(), $options));
 
-        if ($config) {
-            $options = array_merge($config, $options);
-        }
+        $readability = new Readability($configuration);
+        $readability->parse($testPage->getSourceHTML());
 
-        $configuration = new Configuration($options);
+        $this->assertSame($testPage->getExpectedHTML(), $readability->getContent(), 'Parsed text does not match the expected one.');
+    }
 
-        $readability = new Readability($configuration);
-        $readability->parse($html);
 
-        $this->assertSame($expectedResult, $readability->getContent());
+    /**
+     * Test that Readability parses the HTML correctly and matches the expected result
+     *
+     * @dataProvider getSamplePages
+     * @param TestPage $testPage
+     *
+     * @throws ParseException
+     */
+    public function testReadabilityParsesMetadata(TestPage $testPage)
+    {
+        $options = ['OriginalURL' => 'http://fakehost/test/test.html',
+            'FixRelativeURLs' => true,
+            'SubstituteEntities' => true,
+            'ArticleByLine' => true
+        ];
+
+        $configuration = new Configuration(array_merge($testPage->getConfiguration(), $options));
 
-        foreach ($expectedMetadata as $key => $metadata) {
-            $function = 'get' . $key;
-            $this->assertEquals($metadata, $readability->$function(), sprintf('Failed asserting %s metadata', $key));
-        }
+        $readability = new Readability($configuration);
+        $readability->parse($testPage->getSourceHTML());
+
+        $this->assertSame($testPage->getExpectedMetadata()->Author, $readability->getAuthor(), 'Parsed Author does not match expected value.');
+        $this->assertSame($testPage->getExpectedMetadata()->Direction, $readability->getDirection(), 'Parsed Direction does not match expected value.');
+        $this->assertSame($testPage->getExpectedMetadata()->Excerpt, $readability->getExcerpt(), 'Parsed Excerpt does not match expected value.');
+        $this->assertSame($testPage->getExpectedMetadata()->Image, $readability->getImage(), 'Parsed Image does not match expected value.');
+        $this->assertSame($testPage->getExpectedMetadata()->Title, $readability->getTitle(), 'Parsed Title does not match expected value.');
     }
 
     /**
+     * Test that Readability returns all the expected images from the test page
+     *
+     * @param TestPage $testPage
      * @dataProvider getSamplePages
      *
      * @throws ParseException
@@ -52,49 +80,42 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
             'substituteEntities' => true,
         ];
 
-        if ($config) {
-            $options = array_merge($options, $config);
-        }
-
-        $configuration = new Configuration($options);
+        $configuration = new Configuration(array_merge($testPage->getConfiguration(), $options));
 
         $readability = new Readability($configuration);
-        $readability->parse($html);
+        $readability->parse($testPage->getSourceHTML());
 
-        $this->assertSame(json_decode($expectedImages, true), $readability->getImages());
+        $this->assertSame($testPage->getExpectedImages(), $readability->getImages());
     }
 
+    /**
+     * Main data provider
+     *
+     * @return \Generator
+     */
     public function getSamplePages()
     {
         $path = pathinfo(__FILE__, PATHINFO_DIRNAME) . DIRECTORY_SEPARATOR . 'test-pages';
         $testPages = scandir($path);
-        if (in_array('.DS_Store', $testPages)) {
-            unset($testPages[array_search('.DS_Store', $testPages)]);
-        }
-
-        $pages = [];
 
         foreach (array_slice($testPages, 2) as $testPage) {
-            $source = file_get_contents($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'source.html');
-            $expectedHTML = file_get_contents($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'expected.html');
-            $expectedImages = file_get_contents($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'expected-images.json');
-
-            $expectedMetadata = json_decode(file_get_contents($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'expected-metadata.json'));
+            $testCasePath = $path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR;
 
-            $config = false;
-            if (file_exists($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'config.json')) {
-                $config = file_get_contents($path . DIRECTORY_SEPARATOR . $testPage . DIRECTORY_SEPARATOR . 'config.json');
-                if ($config) {
-                    $config = json_decode($config, true);
-                }
-            }
+            $source = file_get_contents($testCasePath . 'source.html');
+            $expectedHTML = file_get_contents($testCasePath . 'expected.html');
+            $expectedImages = json_decode(file_get_contents($testCasePath . 'expected-images.json'), true);
+            $expectedMetadata = json_decode(file_get_contents($testCasePath . 'expected-metadata.json'));
+            $configuration = file_exists($testCasePath . 'config.json') ? json_decode(file_get_contents($testCasePath . 'config.json'), true) : [];
 
-            $pages[$testPage] = [$source, $expectedHTML, $expectedMetadata, $config, $expectedImages];
+            yield $testPage => [new TestPage($configuration, $source, $expectedHTML, $expectedImages, $expectedMetadata)];
         }
-
-        return $pages;
     }
 
+    /**
+     * Test that Readability throws an exception with malformed HTML
+     *
+     * @throws ParseException
+     */
     public function testReadabilityThrowsExceptionWithMalformedHTML()
     {
         $parser = new Readability(new Configuration());
@@ -103,6 +124,11 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $parser->parse('<html>');
     }
 
+    /**
+     * Test that Readability throws an exception with incomplete or short HTML
+     *
+     * @throws ParseException
+     */
     public function testReadabilityThrowsExceptionWithUnparseableHTML()
     {
         $parser = new Readability(new Configuration());
@@ -111,6 +137,9 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
         $parser->parse('<html><body><p></p></body></html>');
     }
 
+    /**
+     * Test that the Readability object has no content as soon as it is instantiated
+     */
     public function testReadabilityCallGetContentWithNoContent()
     {
         $parser = new Readability(new Configuration());
diff --git a/test/TestPage.php b/test/TestPage.php
new file mode 100644
index 0000000..7859481
--- /dev/null
+++ b/test/TestPage.php
@@ -0,0 +1,61 @@
+<?php
+
+namespace andreskrey\Readability\Test;
+
+class TestPage
+{
+    private $configuration;
+    private $sourceHTML;
+    private $expectedHTML;
+    private $expectedImages;
+    private $expectedMetadata;
+
+    public function __construct( $configuration, $sourceHTML, $expectedHTML, $expectedImages, $expectedMetadata)
+    {
+        $this->configuration = $configuration;
+        $this->sourceHTML = $sourceHTML;
+        $this->expectedHTML = $expectedHTML;
+        $this->expectedImages = $expectedImages;
+        $this->expectedMetadata = $expectedMetadata;
+    }
+
+    /**
+     * @return array
+     */
+    public function getConfiguration()
+    {
+        return $this->configuration;
+    }
+
+    /**
+     * @return null
+     */
+    public function getSourceHTML()
+    {
+        return $this->sourceHTML;
+    }
+
+    /**
+     * @return null
+     */
+    public function getExpectedHTML()
+    {
+        return $this->expectedHTML;
+    }
+
+    /**
+     * @return mixed
+     */
+    public function getExpectedImages()
+    {
+        return $this->expectedImages;
+    }
+
+    /**
+     * @return \stdClass
+     */
+    public function getExpectedMetadata()
+    {
+        return $this->expectedMetadata;
+    }
+}