add option to hide alt text

mtibben · Sep 24, 2024 · 81999a8 · 81999a8
1 parent 3b443cb
commit 81999a8
Show file tree

Hide file tree

Showing 2 changed files with 55 additions and 3 deletions.
diff --git a/src/Html2Text.php b/src/Html2Text.php
@@ -68,7 +68,6 @@ class Html2Text
         '/(<tr\b[^>]*>|<\/tr>)/i',                        // <tr> and </tr>
         '/<td\b[^>]*>(.*?)<\/td>/i',                      // <td> and </td>
         '/<span class="_html2text_ignore">.+?<\/span>/i', // <span class="_html2text_ignore">...</span>
-        '/<(img)\b[^>]*alt=\"([^>"]+)\"[^>]*>/i',         // <img> with alt tag
     );
 
     /**
@@ -99,7 +98,6 @@ class Html2Text
         "\n",                            // <tr> and </tr>
         "\t\t\\1\n",                     // <td> and </td>
         "",                              // <span class="_html2text_ignore">...</span>
-        '[\\2]',                         // <img> with alt tag
     );
 
     /**
@@ -222,6 +220,9 @@ class Html2Text
         'width' => 70,          //  Maximum width of the formatted text, in columns.
                                 //  Set this value to 0 (or less) to ignore word wrapping
                                 //  and not constrain text to a fixed-width column.
+
+        'images'=>'alt',        //  'alt' (show the alt text in brackets)
+                                //  'none' (don't show the alt text)
     );
 
     private function legacyConstruct($html = '', $fromFile = false, array $options = array())
@@ -376,6 +377,7 @@ protected function doConvert()
 
     protected function converter(&$text)
     {
+        $this->setImgAltPreference();
         $this->convertBlockquotes($text);
         $this->convertPre($text);
         $text = preg_replace($this->search, $this->replace, $text);
@@ -406,6 +408,16 @@ protected function converter(&$text)
         }
     }
 
+    /**
+     * show/hide alt text for images
+     */
+
+    protected function setImgAltPreference()
+    {
+        $this->search[] = '/<(img)\b[^>]*alt=\"([^>"]+)\"[^>]*>/i';
+        $this->replace[] = $this->options['images'] == 'alt'?'[\\2]':'';
+    }
+
     /**
      * Helper function called by preg_replace() on link replacement.
      *

diff --git a/test/ImageTest.php b/test/ImageTest.php
@@ -35,14 +35,54 @@ public function imageDataProvider() {
         );
     }
 
+    public function ImageDataWithoutAltProvider() {
+        return array(
+            'Without alt tag' => array(
+                'html' => '<img src="http://example.com/example.jpg">',
+                'expected'  => '',
+            ),
+            'Without alt tag, wrapped in text' => array(
+                'html' => 'xx<img src="http://example.com/example.jpg">xx',
+                'expected'  => 'xxxx',
+            ),
+            'With alt tag' => array(
+                'html' => '<img src="http://example.com/example.jpg" alt="An example image">',
+                'expected'  => '',
+            ),
+            'With alt, and title tags' => array(
+                'html' => '<img src="http://example.com/example.jpg" alt="An example image" title="Should be ignored">',
+                'expected'  => '',
+            ),
+            'With alt tag, wrapped in text' => array(
+                'html' => 'xx<img src="http://example.com/example.jpg" alt="An example image">xx',
+                'expected'  => 'xxxx',
+            ),
+            'With italics' => array(
+                'html' => '<img src="shrek.jpg" alt="the ogrelord" /> Blah <i>blah</i> blah',
+                'expected' => ' Blah _blah_ blah'
+            )
+        );
+    }
+
     /**
      * @dataProvider imageDataProvider
      */
-    public function testImages($html, $expected)
+    public function testImagesAlt($html, $expected)
     {
         $html2text = new Html2Text($html);
         $output = $html2text->getText();
 
         $this->assertEquals($expected, $output);
     }
+
+    /**
+     * @dataProvider ImageDataWithoutAltProvider
+     */
+    public function testImagesNoAlt($html, $expected)
+    {
+        $html2text = new Html2Text($html,array('images' => 'none'));
+        $output = $html2text->getText();
+
+        $this->assertEquals($expected, $output);
+    }
 }