[DomCrawler] Added ability to return empty string on Crawler::text() and Crawler::html() instead of an exception

respinoza · respinoza · commit 571d6bc968d6 · 2018-09-24T23:38:50.000+09:00
diff --git a/src/Symfony/Component/DomCrawler/CHANGELOG.md b/src/Symfony/Component/DomCrawler/CHANGELOG.md
@@ -8,6 +8,7 @@ CHANGELOG
   `Image` classes is now optional.
 * The `Crawler::children()` method will have a new `$selector` argument in version 5.0, 
   not defining it is deprecated since version 4.2.
+* Added ability to return empty string on `Crawler::text()` and `Crawler::html()` instead of an exception
 
 3.1.0
 -----
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -55,14 +55,23 @@ class Crawler implements \Countable, \IteratorAggregate
     private $isHtml = true;
 
     /**
-     * @param mixed  $node     A Node to use as the base for the crawling
-     * @param string $uri      The current URI
-     * @param string $baseHref The base href value
+     * Whether the text() and html() methods will return an empty string
+     *
+     * @var bool
      */
-    public function __construct($node = null, string $uri = null, string $baseHref = null)
+    private $emptyValue;
+
+    /**
+     * @param mixed  $node       A Node to use as the base for the crawling
+     * @param string $uri        The current URI
+     * @param string $baseHref   The base href value
+     * @param bool   $emptyValue Return empty string on text and html methods instead of an exception
+     */
+    public function __construct($node = null, string $uri = null, string $baseHref = null, bool $emptyValue = false)
     {
         $this->uri = $uri;
         $this->baseHref = $baseHref ?: $uri;
+        $this->emptyValue = $emptyValue;
 
         $this->add($node);
     }
@@ -571,12 +580,15 @@ public function nodeName()
      * Returns the node value of the first node of the list.
      *
      * @return string The node value
-     *
-     * @throws \InvalidArgumentException When current node is empty
+     * @throws \InvalidArgumentException When current node is empty unless the emptyValue option is set
      */
     public function text()
     {
         if (!$this->nodes) {
+            if ($this->emptyValue) {
+                return '';
+            }
+
             throw new \InvalidArgumentException('The current node list is empty.');
         }
 
@@ -587,12 +599,15 @@ public function text()
      * Returns the first node of the list as HTML.
      *
      * @return string The node html
-     *
-     * @throws \InvalidArgumentException When current node is empty
+     * @throws \InvalidArgumentException When current node is empty unless the emptyValue option is set
      */
     public function html()
     {
         if (!$this->nodes) {
+            if ($this->emptyValue) {
+                return '';
+            }
+
             throw new \InvalidArgumentException('The current node list is empty.');
         }
 
@@ -1152,7 +1167,7 @@ private function findNamespacePrefixes(string $xpath): array
      */
     private function createSubCrawler($nodes)
     {
-        $crawler = new static($nodes, $this->uri, $this->baseHref);
+        $crawler = new static($nodes, $this->uri, $this->baseHref, $this->emptyValue);
         $crawler->isHtml = $this->isHtml;
         $crawler->document = $this->document;
         $crawler->namespaces = $this->namespaces;
diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -392,6 +392,8 @@ public function testText()
         } catch (\InvalidArgumentException $e) {
             $this->assertTrue(true, '->text() throws an \InvalidArgumentException if the node list is empty');
         }
+
+        $this->assertSame('', $this->createTestCrawler(null, true)->filterXPath('//ol')->text());
     }
 
     public function testHtml()
@@ -405,6 +407,8 @@ public function testHtml()
         } catch (\InvalidArgumentException $e) {
             $this->assertTrue(true, '->html() throws an \InvalidArgumentException if the node list is empty');
         }
+
+        $this->assertSame('', $this->createTestCrawler(null, true)->filterXPath('//ol')->html());
     }
 
     public function testExtract()
@@ -1204,7 +1208,7 @@ public function testInheritedClassCallChildrenWithoutArgument()
         $crawlerChild->children();
     }
 
-    public function createTestCrawler($uri = null)
+    public function createTestCrawler($uri = null, $emptyValue = false)
     {
         $dom = new \DOMDocument();
         $dom->loadHTML('
@@ -1253,7 +1257,7 @@ public function createTestCrawler($uri = null)
             </html>
         ');
 
-        return new Crawler($dom, $uri);
+        return new Crawler($dom, $uri, null, $emptyValue);
     }
 
     protected function createTestXmlCrawler($uri = null)

Original file line number	Diff line number	Diff line change
`@@ -392,6 +392,8 @@ public function testText()`
`392`	`392`	`} catch (\InvalidArgumentException $e) {`
`393`	`393`	`$this->assertTrue(true, '->text() throws an \InvalidArgumentException if the node list is empty');`
`394`	`394`	`}`
	`395`	`+`
	`396`	`+ $this->assertSame('', $this->createTestCrawler(null, true)->filterXPath('//ol')->text());`
`395`	`397`	`}`
`396`	`398`
`397`	`399`	`public function testHtml()`
`@@ -405,6 +407,8 @@ public function testHtml()`
`405`	`407`	`} catch (\InvalidArgumentException $e) {`
`406`	`408`	`$this->assertTrue(true, '->html() throws an \InvalidArgumentException if the node list is empty');`
`407`	`409`	`}`
	`410`	`+`
	`411`	`+ $this->assertSame('', $this->createTestCrawler(null, true)->filterXPath('//ol')->html());`
`408`	`412`	`}`
`409`	`413`
`410`	`414`	`public function testExtract()`
`@@ -1204,7 +1208,7 @@ public function testInheritedClassCallChildrenWithoutArgument()`
`1204`	`1208`	`$crawlerChild->children();`
`1205`	`1209`	`}`
`1206`	`1210`
`1207`		`- public function createTestCrawler($uri = null)`
	`1211`	`+ public function createTestCrawler($uri = null, $emptyValue = false)`
`1208`	`1212`	`{`
`1209`	`1213`	`$dom = new \DOMDocument();`
`1210`	`1214`	`$dom->loadHTML('`
`@@ -1253,7 +1257,7 @@ public function createTestCrawler($uri = null)`
`1253`	`1257`	`</html>`
`1254`	`1258`	`');`
`1255`	`1259`
`1256`		`- return new Crawler($dom, $uri);`
	`1260`	`+ return new Crawler($dom, $uri, null, $emptyValue);`
`1257`	`1261`	`}`
`1258`	`1262`
`1259`	`1263`	`protected function createTestXmlCrawler($uri = null)`