From f1abc38c1744c77fc3f26df5ce47d8d9455be764 Mon Sep 17 00:00:00 2001
From: "Daniel St. Jules" <danielst.jules@gmail.com>
Date: Tue, 6 Aug 2013 22:39:47 -0400
Subject: [PATCH] Fix for collapseWhitespace() not being able to handle
 multibyte space characters

---
 README.md                           | 11 ++++++-----
 src/Stringy/StaticStringy.php       |  7 ++++---
 src/Stringy/Stringy.php             | 13 +++++++++----
 tests/Stringy/CommonTest.php        |  2 ++
 tests/Stringy/StaticStringyTest.php |  4 ++--
 tests/Stringy/StringyTest.php       |  4 ++--
 6 files changed, 25 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index 51535dc..113ede3 100644
--- a/README.md
+++ b/README.md
@@ -97,7 +97,7 @@ of the former is the following:
 
 ```php
 use Stringy\Stringy as S;
-echo S::create("Fòô     Bàř", 'UTF-8')->collapseWhitespace()->swapCase();  // 'fÒÔ bÀŘ'
+echo S::create('Fòô     Bàř', 'UTF-8')->collapseWhitespace()->swapCase();  // 'fÒÔ bÀŘ'
 ```
 
 `Stringy\Stringy` contains a __toString() method, which returns the current
@@ -108,8 +108,8 @@ Using the static wrapper, an alternative is the following:
 
 ```php
 use Stringy\StaticStringy as S;
-$string = S::collapseWhitespace("Fòô     Bàř", 'UTF-8');
-echo S::swapCase($string, 'UTF-8');  // 'fÒÔ bÀŘ''
+$string = S::collapseWhitespace('Fòô     Bàř', 'UTF-8');
+echo S::swapCase($string, 'UTF-8');  // 'fÒÔ bÀŘ'
 ```
 
 ## Methods
@@ -151,10 +151,11 @@ S::camelize('Camel-Case');  // 'camelCase'
 
 $stringy->collapseWhitespace()
 
-S::collapseWhitespace(string $str)
+S::collapseWhitespace(string $str [, string $encoding ])
 
 Trims the string and replaces consecutive whitespace characters with a
-single space. This includes tabs and newline characters.
+single space. This includes tabs and newline characters, as well as
+multibyte whitespace such as the thin space and ideographic space.
 
 ```php
 S::create('   Ο     συγγραφέας  ')->collapseWhitespace();
diff --git a/src/Stringy/StaticStringy.php b/src/Stringy/StaticStringy.php
index f116a21..bb90c77 100644
--- a/src/Stringy/StaticStringy.php
+++ b/src/Stringy/StaticStringy.php
@@ -142,14 +142,15 @@ class StaticStringy
 
     /**
      * Trims the string and replaces consecutive whitespace characters with a
-     * single space. This includes tabs and newline characters.
+     * single space. This includes tabs and newline characters, as well as
+     * multibyte whitespace such as the thin space and ideographic space.
      *
      * @param   string  $str  The string to cleanup whitespace
      * @return  string  The trimmed string with condensed whitespace
      */
-    public static function collapseWhitespace($str)
+    public static function collapseWhitespace($str, $encoding = null)
     {
-        return $result = Stringy::create($str)->collapseWhitespace()->str;
+        return $result = Stringy::create($str, $encoding)->collapseWhitespace()->str;
     }
 
     /**
diff --git a/src/Stringy/Stringy.php b/src/Stringy/Stringy.php
index 7a0a75e..578c305 100644
--- a/src/Stringy/Stringy.php
+++ b/src/Stringy/Stringy.php
@@ -249,16 +249,21 @@ class Stringy
 
     /**
      * Trims the string and replaces consecutive whitespace characters with a
-     * single space. This includes tabs and newline characters.
+     * single space. This includes tabs and newline characters, as well as
+     * multibyte whitespace such as the thin space and ideographic space.
      *
      * @return  Stringy  Object with a trimmed $str and condensed whitespace
      */
     public function collapseWhitespace()
     {
-        $stringy = self::create($this->str, $this->encoding);
-        $stringy->str = preg_replace('/\s+/u', ' ', $stringy->trim());
+        $regexEncoding = mb_regex_encoding();
+        mb_regex_encoding($this->encoding);
 
-        return $stringy;
+        $stringy = self::create($this->str, $this->encoding);
+        $stringy->str = mb_ereg_replace('[[:space:]]+', ' ', $stringy);
+        mb_regex_encoding($regexEncoding);
+
+        return $stringy->trim();
     }
 
     /**
diff --git a/tests/Stringy/CommonTest.php b/tests/Stringy/CommonTest.php
index f2a6a03..07afcbd 100644
--- a/tests/Stringy/CommonTest.php
+++ b/tests/Stringy/CommonTest.php
@@ -173,6 +173,8 @@ abstract class CommonTest extends PHPUnit_Framework_TestCase
             array('test string', 'test string'),
             array('Ο συγγραφέας', '   Ο     συγγραφέας  '),
             array('123', ' 123 '),
+            array('1 2 3', '　　1　　2　　3　　', 'UTF-8'), // ideographic spaces
+            array('', '   ', 'UTF-8'), // thin space and space
             array('', ' '),
             array('', ''),
         );
diff --git a/tests/Stringy/StaticStringyTest.php b/tests/Stringy/StaticStringyTest.php
index d2be4f2..a329497 100644
--- a/tests/Stringy/StaticStringyTest.php
+++ b/tests/Stringy/StaticStringyTest.php
@@ -111,9 +111,9 @@ class StaticStringyTestCase extends CommonTest
     /**
      * @dataProvider stringsForCollapseWhitespace
      */
-    public function testCollapseWhitespace($expected, $str)
+    public function testCollapseWhitespace($expected, $str, $encoding = null)
     {
-        $result = S::collapseWhitespace($str);
+        $result = S::collapseWhitespace($str, $encoding);
         $this->assertInternalType('string', $result);
         $this->assertEquals($expected, $result);
     }
diff --git a/tests/Stringy/StringyTest.php b/tests/Stringy/StringyTest.php
index 45bfdc8..2df7b61 100644
--- a/tests/Stringy/StringyTest.php
+++ b/tests/Stringy/StringyTest.php
@@ -145,9 +145,9 @@ class StringyTestCase extends CommonTest
     /**
      * @dataProvider stringsForCollapseWhitespace
      */
-    public function testCollapseWhitespace($expected, $str)
+    public function testCollapseWhitespace($expected, $str, $encoding = null)
     {
-        $stringy = S::create($str);
+        $stringy = S::create($str, $encoding);
         $result = $stringy->collapseWhitespace();
         $this->assertInstanceOf('Stringy\Stringy', $result);
         $this->assertEquals($expected, $result);