Fix tests (#1)

markuspoerschke · Dec 23, 2020 · c05f790 · c05f790
1 parent 696c053
commit c05f790
Show file tree

Hide file tree

Showing 6 changed files with 83 additions and 39 deletions.
diff --git a/src/Extractor.php b/src/Extractor.php
@@ -25,8 +25,12 @@
 use Extractum\Parser\DateParser;
 use Extractum\Scorer\DocumentScorer;
 use Extractum\StopWords\StopWords;
+use ML\JsonLD;
 use Symfony\Component\DomCrawler\Crawler;
 
+/**
+ * @SuppressWarnings(PHPMD.CouplingBetweenObjects)
+ */
 final class Extractor
 {
     use ExtractJsonLdTrait;
@@ -76,6 +80,11 @@ public function extract(string $html, string $uri): Essence
 
         $jsonLd = $this->extractJsonLd($document);
 
+        return $this->createEssence($document, $jsonLd, $topNode, $language, $scorerLanguage);
+    }
+
+    private function createEssence(Crawler $document, ?JsonLD\Document $jsonLd, Crawler $topNode, string $language, string $scorerLanguage): Essence
+    {
         $date = $this->dateExtractor->extract($document, $jsonLd);
         $essence = (new Essence())
             ->setDate($date)
@@ -85,11 +94,11 @@ public function extract(string $html, string $uri): Essence
             ->setLinks($this->linksExtractor->extract($topNode))
             ->setText($this->textExtractor->extract($topNode))
             ->setTitle($this->titleExtractor->extract($document))
-            ->setFree($this->freeExtractor->extract($document, $jsonLd))
+            ->setFree($this->freeExtractor->extract($jsonLd))
         ;
 
-        if ($date !== null && $language !== null) {
-            $essence->setParsedDate($this->dateParser->parse($essence->getDate(), $essence->getLanguage()));
+        if ($date !== null) {
+            $essence->setParsedDate($this->dateParser->parse($date, $scorerLanguage));
         }
 
         return $essence;

diff --git a/src/Extractor/DateExtractor.php b/src/Extractor/DateExtractor.php
@@ -50,16 +50,11 @@ final class DateExtractor extends AbstractExtractor
 
     public function extract(Crawler $crawler, ?JsonLD\Document $jsonLd): ?string
     {
-        if ($jsonLd !== null && ($graph = $jsonLd->getGraph()) !== null) {
-            foreach ($graph->getNodes() as $node) {
-                /** @var JsonLD\TypedValue|null $property */
-                $property = $node->getProperty('http://schema.org/datePublished');
-                if ($property instanceof JsonLD\TypedValue) {
-                    return $property->getValue();
-                }
-            }
-        }
+        return $this->extractFromJsonLd($jsonLd) ?? $this->extractFromCrawler($crawler);
+    }
 
+    private function extractFromCrawler(Crawler $crawler): ?string
+    {
         $candidates = $crawler->filter(self::CANDIDATE_SELECTOR);
         /** @var DOMNode|DOMElement $node */
         foreach ($candidates as $node) {
@@ -89,4 +84,26 @@ public function extract(Crawler $crawler, ?JsonLD\Document $jsonLd): ?string
 
         return null;
     }
+
+    private function extractFromJsonLd(?JsonLD\Document $jsonLd): ?string
+    {
+        if ($jsonLd === null) {
+            return null;
+        }
+
+        $graph = $jsonLd->getGraph();
+        if ($graph === null) {
+            return null;
+        }
+
+        foreach ($graph->getNodes() as $node) {
+            /** @var JsonLD\TypedValue|null $property */
+            $property = $node->getProperty('http://schema.org/datePublished');
+            if ($property instanceof JsonLD\TypedValue) {
+                return $property->getValue();
+            }
+        }
+
+        return null;
+    }
 }
diff --git a/src/Extractor/FreeExtractor.php b/src/Extractor/FreeExtractor.php
@@ -12,20 +12,26 @@
 namespace Extractum\Extractor;
 
 use ML\JsonLD;
-use Symfony\Component\DomCrawler\Crawler;
 
 class FreeExtractor extends AbstractExtractor
 {
-    public function extract(Crawler $crawler, ?JsonLD\Document $jsonLd): bool
+    public function extract(?JsonLD\Document $jsonLd): bool
     {
-        if ($jsonLd !== null && ($graph = $jsonLd->getGraph()) !== null) {
-            $nodes = $graph->getNodes();
-            if (isset($nodes[0])) {
-                $property = $nodes[0]->getProperty('http://schema.org/isAccessibleForFree');
-                if ($property instanceof JsonLD\TypedValue) {
-                    if (strtolower($property->getValue()) === 'false') {
-                        return false;
-                    }
+        if ($jsonLd === null) {
+            return true;
+        }
+
+        $graph = $jsonLd->getGraph();
+        if ($graph === null) {
+            return true;
+        }
+
+        $nodes = $graph->getNodes();
+        if (isset($nodes[0])) {
+            $property = $nodes[0]->getProperty('http://schema.org/isAccessibleForFree');
+            if ($property instanceof JsonLD\TypedValue) {
+                if (strtolower($property->getValue()) === 'false') {
+                    return false;
                 }
             }
         }

diff --git a/src/Extractor/ImageExtractor.php b/src/Extractor/ImageExtractor.php
@@ -27,24 +27,36 @@ final class ImageExtractor extends AbstractExtractor
 
     public function extract(Crawler $crawler, ?JsonLD\Document $jsonLd): ?string
     {
-        if ($jsonLd !== null && ($graph = $jsonLd->getGraph()) !== null) {
-            $nodes = $graph->getNodes();
-            if (isset($nodes[0])) {
-                $images = $nodes[0]->getProperty('http://schema.org/image');
-                if ($images instanceof JsonLD\NodeInterface) {
-                    $images = [$images];
-                }
+        return $this->extractFromJsonLd($jsonLd) ?? $this->extractAttribute('content', self::CANDIDATE_SELECTORS, $crawler);
+    }
+
+    private function extractFromJsonLd(?JsonLD\Document $jsonLd): ?string
+    {
+        if ($jsonLd === null) {
+            return null;
+        }
+
+        $graph = $jsonLd->getGraph();
+        if ($graph === null) {
+            return null;
+        }
+
+        $nodes = $graph->getNodes();
+        if (isset($nodes[0])) {
+            $images = $nodes[0]->getProperty('http://schema.org/image');
+            if ($images instanceof JsonLD\NodeInterface) {
+                $images = [$images];
+            }
 
-                /** @var JsonLD\NodeInterface $image */
-                foreach ($images ?? [] as $image) {
-                    $url = $image->getProperty('http://schema.org/url');
-                    if ($url instanceof JsonLD\NodeInterface) {
-                        return $url->getId();
-                    }
+            /** @var JsonLD\NodeInterface $image */
+            foreach ($images ?? [] as $image) {
+                $url = $image->getProperty('http://schema.org/url');
+                if ($url instanceof JsonLD\NodeInterface) {
+                    return $url->getId();
                 }
             }
         }
 
-        return $this->extractAttribute('content', self::CANDIDATE_SELECTORS, $crawler);
+        return null;
     }
 }
diff --git a/src/Parser/DateParser.php b/src/Parser/DateParser.php
@@ -59,8 +59,8 @@ public function parse(string $dateAsString, string $language = 'en'): ?DateTimeI
                 continue;
             }
 
-            $formattedDateAsString = preg_replace($pattern, $options[self::REPLACEMENT], $dateAsString, 1);
-            $date = DateTimeImmutable::createFromFormat($options[self::FORMAT], $formattedDateAsString);
+            $formattedDate = preg_replace($pattern, $options[self::REPLACEMENT], $dateAsString, 1);
+            $date = DateTimeImmutable::createFromFormat($options[self::FORMAT], $formattedDate);
 
             return $date ?: null;
         }

diff --git a/tests/Extractor/FreeExtractorTest.php b/tests/Extractor/FreeExtractorTest.php
@@ -27,7 +27,7 @@ class FreeExtractorTest extends TestCase
     public function testFreeIsExtracted(string $html, $expected): void
     {
         $crawler = new Crawler($html, 'https://www.example.com');
-        $actual = (new FreeExtractor())->extract($crawler, $this->extractJsonLd($crawler));
+        $actual = (new FreeExtractor())->extract($this->extractJsonLd($crawler));
 
         self::assertSame($expected, $actual);
     }