将多字节字符串截断为 n 个字符

Question

我试图让这个方法在字符串过滤器中工作：

public function truncate($string, $chars = 50, $terminator = ' …');

我期望这个

$in  = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWYXZ1234567890";
$out = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …";

以及这个

$in  = "âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđĒēĔĕĖėĘęĚěĜĝ";
$out = "âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …";

$chars 减去 $terminator 字符串的字符。

此外，过滤器应该在 $chars 限制以下的第一个单词边界处进行剪切，例如，

$in  = "Answer to the Ultimate Question of Life, the Universe, and Everything.";
$out = "Answer to the Ultimate Question of Life, the …";

我非常确定这应该与以下步骤一起使用：

• 从最大字符数中减去终止符中的字符数
• 验证该字符串比计算的限制长或返回它不变
• 找到字符串中低于计算限制的最后一个空格字符以获取单词边界
• 在最后一个空格或计算的限制处剪切字符串如果没有找到最后一个空格
• 将终止符附加到字符串
• 返回字符串

但是，我尝试了各种现在组合了 str* 和 mb_* 函数，但都产生了错误的结果。这不可能那么困难，所以我显然错过了一些东西。有人会分享一个有效的实现吗或者给我指出一个资源，让我最终能够理解如何做到这一点。

谢谢

​是的，我已经检查了 https://stackoverflow.com/search?q=truncate+string+php前 ：）

Answer 1

刚刚发现 PHP 已经有一个多字节截断

• mb_strimwidth— 获取指定宽度的截断字符串，

但它不遵守单词边界。但还是很方便！

Answer 2

试试这个：

function truncate($string, $chars = 50, $terminator = ' …') {
    $cutPos = $chars - mb_strlen($terminator);
    $boundaryPos = mb_strrpos(mb_substr($string, 0, mb_strpos($string, ' ', $cutPos)), ' ');
    return mb_substr($string, 0, $boundaryPos === false ? $cutPos : $boundaryPos) . $terminator;
}

但是您需要确保您的内部编码设置正确。

Answer 3

我通常不喜欢只编写这样的问题的完整答案。但我也刚刚醒来，我想也许你的问题会让我有一个好心情去参加今天剩下的节目。

我没有尝试运行这个，但它应该可以工作，或者至少可以让你完成 90% 的工作。

function truncate( $string, $chars = 50, $terminate = ' ...' )
{
    $chars -= mb_strlen($terminate);
    if ( $chars <= 0 )
        return $terminate;

    $string = mb_substr($string, 0, $chars);
    $space = mb_strrpos($string, ' ');

    if ($space < mb_strlen($string) / 2)
        return $string . $terminate;
    else
        return mb_substr($string, 0, $space) . $terminate;
}

Answer 4

太棒了；

• 足够短的字符串不应附加省略号。
• 换行符也应该是限定断点。
• 正则表达式一旦被分解和解释，就不会太可怕了。

---

我认为关于这个问题和当前的一系列答案有一些重要的事情需要指出。我将演示答案的比较以及基于戈登样本数据的正则表达式答案和一些其他案例，以揭示一些不同的结果。

首先，明确输入值的质量。 Gordon 说该函数需要是多字节安全的并尊重字边界。示例数据没有公开在确定截断位置时对非空格、非单词字符（例如标点符号）的所需处理，因此我们必须假设定位空白字符就足够了——而且明智的是，因为大多数“阅读更多”字符串在截断时不会担心标点符号的问题。

其次，在相当常见的情况下，需要对包含换行符的大量文本应用省略号。

第三，让我们任意同意一些基本的数据标准化，例如：

• 字符串已经修剪掉所有前导/尾随空白字符
• $chars 的值将始终大于 mb_strlen( ) of $terminator

(演示)

功能：

function truncateGumbo($string, $chars = 50, $terminator = ' …') {
    $cutPos = $chars - mb_strlen($terminator);
    $boundaryPos = mb_strrpos(mb_substr($string, 0, mb_strpos($string, ' ', $cutPos)), ' ');
    return mb_substr($string, 0, $boundaryPos === false ? $cutPos : $boundaryPos) . $terminator;
}

function truncateGordon($string, $chars = 50, $terminator = ' …') {
    return mb_strimwidth($string, 0, $chars, $terminator);
}

function truncateSoapBox($string, $chars = 50, $terminate = ' …')
{
    $chars -= mb_strlen($terminate);
    if ( $chars <= 0 )
        return $terminate;

    $string = mb_substr($string, 0, $chars);
    $space = mb_strrpos($string, ' ');

    if ($space < mb_strlen($string) / 2)
        return $string . $terminate;
    else
        return mb_substr($string, 0, $space) . $terminate;
}

function truncateMickmackusa($string, $max = 50, $terminator = ' …') {
    $trunc = $max - mb_strlen($terminator, 'UTF-8');
    return preg_replace("~(?=.{{$max}})(?:\S{{$trunc}}|.{0,$trunc}(?=\s))\K.+~us", $terminator, $string);
}

测试用例：

$tests = [
    [
        'testCase' => "Answer to the Ultimate Question of Life, the Universe, and Everything.",
        // 50th char ---------------------------------------------------^
        'expected' => "Answer to the Ultimate Question of Life, the …",
    ],
    [
        'testCase' => "A single line of text to be followed by another\nline of text",
        // 50th char ----------------------------------------------------^
        'expected' => "A single line of text to be followed by another …",
    ],
    [
        'testCase' => "âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđĒēĔĕĖėĘęĚěĜĝ",
        // 50th char ---------------------------------------------------^
        'expected' => "âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …",
    ],
    [
        'testCase' => "123456789 123456789 123456789 123456789 123456789",
        // 50th char doesn't exist -------------------------------------^
        'expected' => "123456789 123456789 123456789 123456789 123456789",
    ],
    [
        'testCase' => "Hello worldly world",
        // 50th char doesn't exist -------------------------------------^
        'expected' => "Hello worldly world",
    ],
    [
        'testCase' => "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWYXZ1234567890",
        // 50th char ---------------------------------------------------^
        'expected' => "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …",
    ],
];

执行：

foreach ($tests as ['testCase' => $testCase, 'expected' => $expected]) {
    echo "\tSample Input:\t\t$testCase\n";
    echo "\n\ttruncateGumbo:\t\t" , truncateGumbo($testCase);
    echo "\n\ttruncateGordon:\t\t" , truncateGordon($testCase);
    echo "\n\ttruncateSoapBox:\t" , truncateSoapBox($testCase);
    echo "\n\ttruncateMickmackusa:\t" , truncateMickmackusa($testCase);
    echo "\n\tExpected Result:\t{$expected}";
    echo "\n-----------------------------------------------------\n";
}

输出：

    Sample Input:           Answer to the Ultimate Question of Life, the Universe, and Everything.

    truncateGumbo:          Answer to the Ultimate Question of Life, the …
    truncateGordon:         Answer to the Ultimate Question of Life, the Uni …
    truncateSoapBox:        Answer to the Ultimate Question of Life, the …
    truncateMickmackusa:    Answer to the Ultimate Question of Life, the …
    Expected Result:        Answer to the Ultimate Question of Life, the …
-----------------------------------------------------
    Sample Input:           A single line of text to be followed by another
line of text

    truncateGumbo:          A single line of text to be followed by …
    truncateGordon:         A single line of text to be followed by another
 …
    truncateSoapBox:        A single line of text to be followed by …
    truncateMickmackusa:    A single line of text to be followed by another …
    Expected Result:        A single line of text to be followed by another …
-----------------------------------------------------
    Sample Input:           âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđĒēĔĕĖėĘęĚěĜĝ

    truncateGumbo:          âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …
    truncateGordon:         âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …
    truncateSoapBox:        âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …
    truncateMickmackusa:    âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …
    Expected Result:        âãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿĀāĂ㥹ĆćĈĉĊċČčĎďĐđ …
-----------------------------------------------------
    Sample Input:           123456789 123456789 123456789 123456789 123456789

    truncateGumbo:          123456789 123456789 123456789 123456789 12345678 …
    truncateGordon:         123456789 123456789 123456789 123456789 123456789
    truncateSoapBox:        123456789 123456789 123456789 123456789 …
    truncateMickmackusa:    123456789 123456789 123456789 123456789 123456789
    Expected Result:        123456789 123456789 123456789 123456789 123456789
-----------------------------------------------------
    Sample Input:           Hello worldly world

    truncateGumbo:          
Warning: mb_strpos(): Offset not contained in string in /in/ibFH5 on line 4
Hello worldly world …
    truncateGordon:         Hello worldly world
    truncateSoapBox:        Hello worldly …
    truncateMickmackusa:    Hello worldly world
    Expected Result:        Hello worldly world
-----------------------------------------------------
    Sample Input:           abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWYXZ1234567890

    truncateGumbo:          abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …
    truncateGordon:         abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …
    truncateSoapBox:        abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …
    truncateMickmackusa:    abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …
    Expected Result:        abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUV …
-----------------------------------------------------

我的模式解释：

虽然看起来确实相当难看，但大多数乱码模式语法都是将数值作为动态量词插入的问题。

我也可以将其写为：

'~(?:\S{' . $trunc . '}|(?=.{' . $max . '}).{0,' . $trunc . '}(?=\s))\K.+~us'

为了简单起见，我将 $trunc 替换为 48，将 $max 替换为 50代码>.

~                 #opening pattern delimiter
(?=.{50})         #lookahead to ensure that the string has a minimum of 50 characters
(?:               #start of non-capturing group -- to maintain pattern logic only
  \S{48}          #the string starts with at least 48 non-white-space characters
  |               #or
  .{0,48}(?=\s)   #the string starts with upto 48 characters followed by a whitespace
)                 #end of non-capturing group
\K                #restart the fullstring match (aka "forget" the previously matched characters)
.+                #match the remaining characters (these characters will be replaced)
~                 #closing pattern delimiter
us                #pattern modifiers: unicode/multibyte flag & dot matches newlines flag