Add support for initial names like JJ Abrams and JD Salinger

Fixes tamtamchik#16
jdhartley · Feb 2, 2022 · 8370390 · 8370390
1 parent 461904b
commit 8370390
Show file tree

Hide file tree

Showing 3 changed files with 53 additions and 2 deletions.
diff --git a/README.md b/README.md
@@ -44,6 +44,7 @@ Formatter::nameCase("VON STREIT");                       // => von Streit
 Formatter::nameCase("AP LLWYD DAFYDD");                  // => ap Llwyd Dafydd
 Formatter::nameCase("HENRY VIII");                       // => Henry VIII
 Formatter::nameCase("VAN DYKE");                         // => van Dyke
+Formatter::nameCase("JJ ABRAMS");                        // => JJ Abrams
 Formatter::nameCase("PRINCE PHILIP, DUKE OF EDINBURGH"); // => Prince Philip, Duke of Edinburgh
 
 // Or as an instance

diff --git a/src/Formatter.php b/src/Formatter.php
@@ -98,6 +98,19 @@ class Formatter
     // Excluded post-nominals
     private static $postNominalsExcluded = [];
 
+    // Most two-letter words with no vowels should be kept in all caps as initials
+    private const INITIAL_NAME_REGEX = '\b(Aj|[bcdfghjklmnpqrstvwxyzBCDFGHJKLMNPQRSTVWXYZ]{2})\s';
+    private const INITIAL_NAME_EXCEPTIONS = [
+        'Mr',
+        'Dr',
+        'St',
+        'Jr',
+        'Sr',
+        // FIXME: These collide with POST_NOMINALS
+        // 'Ms',
+        // 'Lt',
+    ];
+
     // Lowercase words
     private const LOWER_CASE_WORDS = ['The', 'Of', 'And'];
 
@@ -169,6 +182,8 @@ public static function nameCase($name = '', array $options = []): string
             $name = mb_ereg_replace($pattern, $replacement, $name);
         }
 
+        $name = self::correctInitialNames($name);
+
         $name = self::correctLowerCaseWords($name);
 
         return self::processOptions($name);
@@ -329,6 +344,26 @@ private static function fixConjunction(string $name): string
         return $name;
     }
 
+    /**
+     * Correct capitalization of initial names like JJ and TJ.
+     *
+     * @param string $name
+     *
+     * @return string
+     */
+    private static function correctInitialNames(string $name): string
+    {
+        return mb_ereg_replace_callback(self::INITIAL_NAME_REGEX, function ($matches) {
+            $match = $matches[0];
+
+            if (in_array($matches[1], self::INITIAL_NAME_EXCEPTIONS)) {
+                return $match;
+            }
+
+            return mb_strtoupper($match);
+        }, $name);
+    }
+
     /**
      * Correct lower-case words of titles.
      *

diff --git a/tests/NameCaseTest.php b/tests/NameCaseTest.php
@@ -36,6 +36,20 @@ final class NameCaseTest extends TestCase
         "Charles II", "Fred XLIX",
     ];
 
+    private $initialNames = [
+        // Two letter initials names should remain capital with and without periods
+        "JJ Abrams", "JD Salinger", "AJ Michalka",
+        "J. F. Kennedy", "J.F. Kennedy",
+        // Except for some specifics
+        "Mr Smith",
+        "Dr Martin Luther King Jr",
+        "St Patrick",
+        "Martin Luther King Sr",
+        // FIXME: These collide with POST_NOMINALS
+        // "Ms Smith",
+        // "Lt Worf",
+    ];
+
     /** Test numbers. */
     public function testNumbersAreUnaffected(): void
     {
@@ -107,8 +121,9 @@ public function testRomanNumerals(): void
     /** Test initials */
     public function testInitials(): void
     {
-        $this->assertEquals('J. F. Kennedy', Formatter::nameCase(mb_strtolower('J. F. Kennedy')));
-        $this->assertEquals('J.F. Kennedy', Formatter::nameCase(mb_strtolower('J.F. Kennedy')));
+        foreach ($this->initialNames as $name) {
+            $this->assertEquals($name, Formatter::nameCase(mb_strtolower($name)));
+        }
     }
 
     private $programmers = [