更新：Casimir的评论解决方案更直接/更干净。

UPDATE: Casimir's commented solution is more direct/clean.

代码：（演示）（模式演示）

echo preg_replace('~\[url(?|]((https?://[^[]+))|(?:="(https?://[^"]+)")](.+?))\[/url]~i', '<a href=\"$1\">$2</a>', $bbcode);

通过将模式中第一个替代品的捕获加倍，您可以确保始终拥有 $ 1 和 $ 2 申请替换字符串。

By doubling the capture of the first alternative in the pattern, you can ensure that you always have a $1 and $2 to apply to the replacement string.

以下是模式略有扩展的变体，它考虑单引号而不引用。

Here is a slightly extended variation of the pattern that considers single quoting and no quoting.

（上一个解决方案的开始）

(Start of previous solution)

使用 preg_match_callback（）您可以确定在开头 [url] 标签内是否有提供的网址 - 在这种情况下，您将想要保留位于开始标记和结束标记之间的文本。

By using preg_match_callback() you can determine if there was a url provided inside of the opening [url] tag -- in which case, you will want to preserve the text that is located between the opening and closing tags.

如果标记之间的文本是网址，则在< a> 标记字符串。

If the text between the tags IS the url, you use it in both places in the <a> tag string.

无效的字符串不会被转换。

Invalid strings will not be converted.

代码：（演示）（ 模式演示）

$bbcodes = [
    '[URL]www.no.http.example.com[/URL]',
    '[url]https://any.com/any[/url]',
    '[url="nourl"]nourl[/url]',
    '[URL="https://any.com/any?any=333"]text text[/URL]',
    '[url="http://www.emptyTEXT.com"][/url]',
    '[url]http://www.any.com/any?any=44#sss[/url]',
    '[url="https://conflictinglink"]http://differenturl[/url]'
];

foreach ($bbcodes as $bbcode) {
    echo preg_replace_callback('~\[url(?:](https?://[^[]+)|(?:="(https?://[^"]+)")](.+?))\[/url]~i',
                          function($m) {
                              if (isset($m[2])) {
                                  return "<a href=\"{$m[2]}\">{$m[3]}</a>";
                              }
                              return "<a href=\"{$m[1]}\">{$m[1]}</a>";
                          },
                          $bbcode);
    echo "\n---\n";
}

输出：

[URL]www.no.http.example.com[/URL]
---
<a href="https://any.com/any">https://any.com/any</a>
---
[url="nourl"]nourl[/url]
---
<a href="https://any.com/any?any=333">text text</a>
---
[url="http://www.emptyTEXT.com"][/url]
---
<a href="http://www.any.com/any?any=44#sss">http://www.any.com/any?any=44#sss</a>
---
<a href="https://conflictinglink">http://differenturl</a>
---

模式细分：

~                    #start of pattern delimiter
\[url                #match literally [url
(?:                  #start non-capturing group #1
  ]                  #match literally ]
  (https?://[^[]+)   #match and store as Capture Group #1 http , an optional s , colon , two forward slashes, then one or more non-opening square brackets (since valid href values cannot have square brackets)
  |                  #or
  (?:                #start non-capturing group #2
    ="               #match literally ="
    (https?://[^"]+) #match and store as Capture Group #2 (same logic as Capture Group #1)
    "                #match literally "
  )                  #end non-capturing group #2
  ]                  #match literally ]
  (.+?)              #match (lazily) and store as Capture Group #3 one or more characters (this is the innerHTML component)
)                    #end non-capturing group #1
\[/url]              #match literally [/url]
~                    #end of pattern delimiter

回调函数评估匹配数组中的元素（ $ m ）并有条件地生成并返回所需的输出。如果有任何匹配，则输出将包含：

The callback function assesses the elements in the matches array ($m) and conditionally generates and returns the desired output. If there are any matches, the output will either contain:

array(
    0 => [the fullstring match]
    1 => [the url of a bbcode tag that does not have a quoted url]
)

或

array(
    0 => [the fullstring match]
    1 => ''  // <-- empty string
    2 => [the quoted url of the bbcode tag]
    3 => [the text between the opening an closing bbcode tags]
)