fix: update some preg_replace calls in MarkdownRemover and add tests (#…

…57)
tigitz · Jan 11, 2025 · 88e1d08 · 88e1d08
1 parent 6eca2fb
commit 88e1d08
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 12 deletions.
diff --git a/src/TextProcessor/MarkdownRemover.php b/src/TextProcessor/MarkdownRemover.php
@@ -45,19 +45,15 @@ public function process(TextInterface $text): TextInterface
         // Remove blockquotes
         $output = \PhpSpellcheck\preg_replace('/^\s{0,3}>\s?/', '', $output);
         // Remove reference-style links?
+
         $output = \PhpSpellcheck\preg_replace('/^\s{1,2}\[(.*?)\]: (\S+)( ".*?")?\s*$/', '', $output);
-        /**
-         * Remove atx-style headers.
-         *
-         *@TODO find a way to merge the two regex below
-         * remove ## Heading ##
-         */
-        $output = \PhpSpellcheck\preg_replace('/^#{1,6}\s+(.*)(\s+#{1,6})$/m', '$1', $output);
-        // remove ## Heading
-        $output = \PhpSpellcheck\preg_replace('/^#{1,6}\s+(.*)$/m', '$1', $output);
-        // Remove emphasis (repeat the line to remove double emphasis)
-        $output = \PhpSpellcheck\preg_replace('/([\*_]{1,3})(\S.*?\S{0,1})\1/', '$2', $output);
-        $output = \PhpSpellcheck\preg_replace('/([\*_]{1,3})(\S.*?\S{0,1})\1/', '$2', $output);
+        // Remove ## Heading
+        $output = \PhpSpellcheck\preg_replace('/^#{1,6}\s+(.*?)(?:\s+#{1,6})?$/m', '$1', $output);
+        // Remove all layers of emphasis
+        while (\PhpSpellcheck\preg_match('/([\*_]{1,3})(\S.*?\S{0,1})\1/', $output)) {
+            $output = \PhpSpellcheck\preg_replace('/([\*_]{1,3})(\S.*?\S{0,1})\1/', '$2', $output);
+        }
+
         // Remove list items
         $output = \PhpSpellcheck\preg_replace('/^([^\S\r\n]*)\*\s/m', '$1', $output);
         // Remove code blocks

diff --git a/tests/TextProcessor/MarkdownRemoverTest.php b/tests/TextProcessor/MarkdownRemoverTest.php
@@ -113,6 +113,22 @@ public function testShouldRemoveDoubleEmphasis(): void
         $this->assertSame($expected, (new MarkdownRemover())->process(t($string))->getContent());
     }
 
+    public function testShouldRemoveTripleEmphasis(): void
+    {
+        $string = 'This text is ***really important***.';
+        $expected = 'This text is really important.';
+
+        $this->assertSame($expected, (new MarkdownRemover())->process(t($string))->getContent());
+    }
+
+    public function testShouldRemoveLongerEmphasis(): void
+    {
+        $string = 'This text is ******really important******.';
+        $expected = 'This text is really important.';
+
+        $this->assertSame($expected, (new MarkdownRemover())->process(t($string))->getContent());
+    }
+
     public function testShouldRemoveHorizontalRules(): void
     {
         $string = "Some text on a line\n\n---\n\nA line below";