fix: detect spaced profanity like "f u c k i n g" (#36) (#37)

* fix: detect spaced profanity like "f u c k i n g" (#36)

The isSpanningWordBoundary() method was incorrectly rejecting spaced-out
profanity because it checked if the first or last part was a single
character. This caused intentional obfuscation like "f u c k i n g" to
be missed.

The fix distinguishes between:
- Intentional obfuscation: ALL parts are single chars → allow detection
- Cross-word accidents: only SOME parts are single chars → reject

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* docs: improve docstring for isSpanningWordBoundary method

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

Author: deemonic

src/BlaspService.php

@@ -352,34 +352,47 @@ private function handle(): self
     }
 
     /**
-     * Check if a match inappropriately spans across word boundaries.
-     * 
-     * @param string $matchedText The text that was matched by the regex
-     * @return bool
+     * Determine whether a matched substring inappropriately spans word boundaries (and should be treated as a cross-word match).
+     *
+     * @param string $matchedText The substring captured by the detector, possibly containing internal whitespace or obfuscation.
+     * @return bool `true` if the match spans word boundaries and should be rejected, `false` otherwise.
      */
     private function isSpanningWordBoundary(string $matchedText): bool
     {
         // If the match contains spaces, it might be spanning word boundaries
         if (preg_match('/\s+/', $matchedText)) {
-            // Split by spaces to check the word structure
             $parts = preg_split('/\s+/', $matchedText);
-            
-            // If we have multiple parts and the last part is just a single character,
-            // it's likely the beginning of the next word
+
             if (count($parts) > 1) {
+                // Count how many parts are single characters
+                $singleCharCount = 0;
+                foreach ($parts as $part) {
+                    if (strlen($part) === 1 && preg_match('/[a-z]/i', $part)) {
+                        $singleCharCount++;
+                    }
+                }
+
+                // If ALL parts are single characters, this is intentional obfuscation
+                // (e.g., "f u c k i n g") - allow it
+                if ($singleCharCount === count($parts)) {
+                    return false;
+                }
+
+                // If SOME parts are single characters at edges, this is likely
+                // a cross-word match (e.g., "t êt" from "pourrait être") - reject it
+                $firstPart = $parts[0];
                 $lastPart = end($parts);
+
                 if (strlen($lastPart) === 1 && preg_match('/[a-z]/i', $lastPart)) {
-                    return true;  // Last part is single char - likely from next word
+                    return true;
                 }
-                
-                // Also check if first part is single char (less common but possible)
-                $firstPart = $parts[0];
+
                 if (strlen($firstPart) === 1 && preg_match('/[a-z]/i', $firstPart)) {
-                    return true;  // First part is single char - likely from previous word
+                    return true;
                 }
             }
         }
-        
+
         return false;
     }
 

tests/BlaspCheckTest.php

@@ -290,4 +290,20 @@ public function test_multiple_profanities_with_spaces()
         $this->assertCount(2, $result->uniqueProfanitiesFound);
         $this->assertSame('This is a ******* **** sentence', $result->cleanString);
     }
+
+    public function test_spaced_profanity_with_substitution()
+    {
+        // Issue #36 - README example should be detected
+        $result = $this->blaspService->check('This is f u c k 1 n g awesome!');
+
+        $this->assertTrue($result->hasProfanity);
+        $this->assertStringContainsString('*', $result->cleanString);
+    }
+
+    public function test_spaced_profanity_without_substitution()
+    {
+        $result = $this->blaspService->check('f u c k i n g');
+
+        $this->assertTrue($result->hasProfanity);
+    }
 }