在PHP 7.1中你可以这样做(接受负字符串偏移量的rfc):

<?php
$silly = 'Mary had a little lamb';
echo $silly[-20];
echo $silly{-6};
echo $silly[-3];
echo $silly[-15];
echo $silly[-13];
echo $silly[-1];
echo $silly[-4];
echo $silly{-10};
echo $silly[-4];
echo $silly[-8];
echo $silly{3}; // <-- this will be deprecated in PHP 7.4
die();

我让你猜一下输出。

同时，我在xenonite的性能代码中添加了以下结果:

Substr()耗时7.0334868431091秒 数组访问花费了2.3111131191254秒 直接字符串访问(负字符串偏移量)花费了1.7971360683441秒

substr("testers", -1); // returns "s"

或者，对于多字节字符串:

mb_substr("multibyte string…", -1); // returns "…"

Siemano，只获取所选目录下的php文件:

$dir    = '/home/zetdoa/ftp/domeny/MY_DOMAIN/projekty/project';
$files = scandir($dir, 1);


foreach($files as $file){
  $n = substr($file, -3);
  if($n == 'php'){
    echo $file.'<br />';
  }
}

不同语言(包括C sharp和PHP)中的字符串也被认为是字符数组。

知道在理论上数组操作应该比字符串操作快，

$foo = "bar";


$lastChar = strlen($foo) -1;
echo $foo[$lastChar];

$firstChar = 0;
echo $foo[$firstChar];

然而，标准数组函数像

count();

对字符串无效。

我建议使用Gordon的解决方案，因为它比substr()性能更好:

<?php 

$string = 'abcdef';
$repetitions = 10000000;

echo "\n\n";
echo "----------------------------------\n";
echo $repetitions . " repetitions...\n";
echo "----------------------------------\n";
echo "\n\n";

$start = microtime(true);
for($i=0; $i<$repetitions; $i++)
    $x = substr($string, -1);

echo "substr() took " . (microtime(true) - $start) . "seconds\n";

$start = microtime(true);
for($i=0; $i<$repetitions; $i++)
    $x = $string[strlen($string)-1];

echo "array access took " . (microtime(true) - $start) . "seconds\n";

die();

输出如下所示

 ---------------------------------- 
 10000000 repetitions...
 ----------------------------------

 substr() took 2.0285921096802seconds 
 array access took 1.7474739551544seconds

substr($string, -1)