Merge pull request #5 from worksome/JIRA-8030_resolve-plaintext-issue

fix: JIRA-8030 Resolve plain-text encoding
worksome · Nov 15, 2022 · 1d464f3 · 1d464f3
2 parents b4dea21 + a1b23ea
commit 1d464f3
Show file tree

Hide file tree

Showing 4 changed files with 6 additions and 1 deletion.
diff --git a/src/Html2Text.php b/src/Html2Text.php
@@ -116,6 +116,8 @@ public static function getDocument(string $html): DOMDocument
             $html = '<body>' . $html . '</body>';
         }
 
+        $html = (string) preg_replace('/&(?![a-z]+?;)/mi', '&amp;', $html);
+
         $load_result = $doc->loadHTML($html);
 
         if (! $load_result) {

diff --git a/tests/Html2TextTest.php b/tests/Html2TextTest.php
@@ -14,12 +14,13 @@
 
     $output = Html2Text::convert($input, $config);
 
-    expect($output)->toBe($expected);
+    expect(trim($output))->toBe(trim($expected));
 })->with([
     'Basic' => 'basic',
     'Anchor tags' => 'anchors',
     'More anchor tags' => 'more-anchors',
     'Break tag' => 'br',
+    'Ampersand (&)' => 'ampersand',
     'Break tags' => 'brs',
     'Tables' => 'table',
     'Non-breaking spaces (NBSP tag)' => 'nbsp',

diff --git a/tests/fixtures/ampersand.html b/tests/fixtures/ampersand.html
@@ -0,0 +1 @@
+& &amp;
diff --git a/tests/fixtures/ampersand.txt b/tests/fixtures/ampersand.txt
@@ -0,0 +1 @@
+& &