Merge pull request #62 from cameron1729/br_in_pre

Fix whitespace problem with pre tags containing brs
mtibben · Mar 16, 2016 · f55104b · f55104b
2 parents 2712226 + 1bc583d
commit f55104b
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 7 deletions.
diff --git a/src/Html2Text.php b/src/Html2Text.php
@@ -436,7 +436,8 @@ protected function convertPre(&$text)
     {
         // get the content of PRE element
         while (preg_match('/<pre[^>]*>(.*)<\/pre>/ismU', $text, $matches)) {
-            $this->preContent = $matches[1];
+            // Replace br tags with newlines to prevent the search-and-replace callback from killing whitespace
+            $this->preContent = preg_replace('/(<br\b[^>]*>)/i', "\n", $matches[1]);
 
             // Run our defined tags search-and-replace with callback
             $this->preContent = preg_replace_callback(

diff --git a/test/PreTest.php b/test/PreTest.php
@@ -4,9 +4,11 @@
 
 class PreTest extends \PHPUnit_Framework_TestCase
 {
-    public function testPre()
+    public function preDataProvider()
     {
-        $html =<<<'EOT'
+        return array(
+            'Basic pre' => array(
+                'html' => <<<EOT
 <p>Before</p>
 <pre>
 
@@ -17,9 +19,9 @@ public function testPre()
 
 </pre>
 <p>After</p>
-EOT;
-
-        $expected =<<<'EOT'
+EOT
+                ,
+                'expected' => <<<EOT
 Before
 
 Foo bar baz
@@ -28,8 +30,36 @@ public function testPre()
 
 After
 
-EOT;
+EOT
+                ,
+            ),
+            'br in pre' => array(
+                'html' => <<<EOT
+<pre>
+some<br />  indented<br />  text<br />    on<br />    several<br />  lines<br />
+</pre>
+EOT
+                ,
+                'expected' => <<<EOT
+some
+  indented
+  text
+    on
+    several
+  lines
+
 
+EOT
+                ,
+            ),
+        );
+    }
+
+    /**
+     * @dataProvider preDataProvider
+     */
+    public function testPre($html, $expected)
+    {
         $html2text = new Html2Text($html);
         $this->assertEquals($expected, $html2text->getText());
     }