Merge pull request #325 from jmcclell/mbstring-ucword-congruency

Ensure mb formatString is congruent to standard ucwords() functionality

Author: willdurand

src/Geocoder/Result/AbstractResult.php

@@ -60,8 +60,27 @@ public function offsetUnset($offset)
      */
     protected function formatString($str)
     {
-        if (function_exists('mb_convert_case')) {
+        if (extension_loaded('mbstring')) {
+            $originalStr = $str;
             $str = mb_convert_case($str, MB_CASE_TITLE, 'UTF-8');
+
+            // Correct for MB_TITLE_CASE's insistence on uppercasing letters
+            // immediately preceded by numerals, eg: 1st -> 1St
+            $originalEncoding = mb_regex_encoding();
+            mb_regex_encoding('UTF-8');
+
+            // matches an upper case letter character immediately preceded by a numeral
+            mb_ereg_search_init($str, '[0-9]\p{Lu}');
+
+            while($match = mb_ereg_search_pos()) {
+                $charPos = $match[0] + 1;
+                // Only swap it back to lowercase if it was lowercase to begin with
+                if (mb_ereg_match('\p{Ll}', $originalStr[$charPos])) {
+                    $str[$charPos] = mb_strtolower($str[$charPos]);
+                }
+            }
+
+            mb_regex_encoding($originalEncoding);
         } else {
             $str = $this->lowerize($str);
             $str = ucwords($str);
@@ -82,7 +101,7 @@ protected function formatString($str)
      */
     protected function lowerize($str)
     {
-        return function_exists('mb_strtolower') ? mb_strtolower($str, 'UTF-8') : strtolower($str);
+        return extension_loaded('mbstring') ? mb_strtolower($str, 'UTF-8') : strtolower($str);
     }
 
     /**
@@ -94,6 +113,6 @@ protected function lowerize($str)
      */
     protected function upperize($str)
     {
-        return function_exists('mb_strtoupper') ? mb_strtoupper($str, 'UTF-8') : strtoupper($str);
+        return extension_loaded('mbstring') ? mb_strtoupper($str, 'UTF-8') : strtoupper($str);
     }
 }

tests/Geocoder/Tests/Result/GeocodedTest.php

@@ -269,4 +269,19 @@ public function testUpperizeViaReflection()
 
         $this->assertEquals('FOO', $method->invoke($this->geocoded, 'foo'));
     }
+
+    public function testResultFormattingWithLeadingNumeral()
+    {
+        if (version_compare(phpversion(), '5.5.16', '<')) {
+            $this->markTestSkipped("Character property matching for mb_ereg doesn't work for PHP < 5.5");
+        }
+        // MB_TITLE_CASE Will turn this into 1St so let's test to ensure we are correcting that
+        // We do not want to "correct" 5C, however, as it is part of the original string
+        $array = array(
+            'streetName' => '1st ave 1A',
+        );
+
+        $this->geocoded->fromArray($array);
+        $this->assertEquals('1st Ave 1A', $this->geocoded->getStreetName());
+    }
 }