PCRE2正则优化：将引号/无引号src值捕获至同一分组

环境与需求 #

正则环境：PCRE2（PHP >= 7.3）

现有一段包含多个<img>标签的多行字符串，需实现以下正则捕获：

• 捕获所有包含src属性的完整<img>标签；
• 捕获上述src属性的属性值。

src属性格式规则：

• 可使用双引号、单引号包裹，或无引号包裹；
• 带引号时，属性值内不得包含对应引号；
• 无引号时，属性值内不得包含空白字符或>。

当前问题 #

现有正则已实现基本匹配，但带引号的src值被捕获至$matches[2]，无引号的被捕获至$matches[3]，需要手动复制数据到同一分组，现需优化正则，让两类src值直接捕获到同一个分组中。

原实现代码 #

$code = <<<EOD
1. <img width src = one height>
2. <notAtStart><img src=two height>
3. NotAtAll
4. <img width src="three"><notAtEnd>
5. <notAtStart><img src = 'four' /><notAtEnd>
6. <img src =five><test>
7. <img WithoutSrc>
EOD;

$regex='/(<IMG(?=\s).*\sSRC\s*=\s*
    (?(?=["\'])
        .(.+?) ["\'] 
     |         
        (.+?) [\s>]
    )
    (?(?<!>).*?>)
)/ix';

preg_match_all($regex, $code, $matches);
echo PHP_EOL . "Matches:";
// print all groups:
print_r($matches);

// copy matches captures in $matches[2] to $matches[3]
foreach($matches[2] as $a=>$b) 
    if ($b != "")
        $matches[3][$a] = $b;

// print the whole captured img tags:
print_r($matches[1]);
// print just the captured paths:
print_r($matches[3]);

原输出 #

Matches:Array
(
    [0] => Array
        (
            [0] => <img width src = one height>
            [1] => <img src=two height>
            [2] => <img width src="three">
            [3] => <img src = 'four' />
            [4] => <img src =five>
        )

    [1] => Array
        (
            [0] => <img width src = one height>
            [1] => <img src=two height>
            [2] => <img width src="three">
            [3] => <img src = 'four' />
            [4] => <img src =five>
        )

    [2] => Array
        (
            [0] => 
            [1] => 
            [2] => three
            [3] => four
            [4] => 
        )

    [3] => Array
        (
            [0] => one
            [1] => two
            [2] => 
            [3] => 
            [4] => five
        )

)
Array
(
    [0] => <img width src = one height>
    [1] => <img src=two height>
    [2] => <img width src="three">
    [3] => <img src = 'four' />
    [4] => <img src =five>
)
Array
(
    [0] => one
    [1] => two
    [2] => three
    [3] => four
    [4] => five
)

优化方案 #

使用PCRE2支持的分支重置组((?|...))，这个语法可以让不同分支中的捕获组复用同一个编号，从而把带引号和无引号的src值统一捕获到同一个分组里。

优化后的正则表达式：

$regex='/(<IMG(?=\s).*\sSRC\s*=\s*
    (?|
        ["\'](.+?)["\'] 
     |         
        (.+?)[\s>]
    )
    (?(?<!>).*?>)
)/ix';

优化后代码与输出 #

$code = <<<EOD
1. <img width src = one height>
2. <notAtStart><img src=two height>
3. NotAtAll
4. <img width src="three"><notAtEnd>
5. <notAtStart><img src = 'four' /><notAtEnd>
6. <img src =five><test>
7. <img WithoutSrc>
EOD;

$regex='/(<IMG(?=\s).*\sSRC\s*=\s*
    (?|
        ["\'](.+?)["\'] 
     |         
        (.+?)[\s>]
    )
    (?(?<!>).*?>)
)/ix';

preg_match_all($regex, $code, $matches);
echo PHP_EOL . "Matches:";
print_r($matches);

// 直接获取统一的src值分组，无需手动复制
echo PHP_EOL . "Captured img tags:";
print_r($matches[1]);
echo PHP_EOL . "Captured src values:";
print_r($matches[2]);

输出结果：

Matches:Array
(
    [0] => Array
        (
            [0] => <img width src = one height>
            [1] => <img src=two height>
            [2] => <img width src="three">
            [3] => <img src = 'four' />
            [4] => <img src =five>
        )

    [1] => Array
        (
            [0] => <img width src = one height>
            [1] => <img src=two height>
            [2] => <img width src="three">
            [3] => <img src = 'four' />
            [4] => <img src =five>
        )

    [2] => Array
        (
            [0] => one
            [1] => two
            [2] => three
            [3] => four
            [4] => five
        )

)
Captured img tags:Array
(
    [0] => <img width src = one height>
    [1] => <img src=two height>
    [2] => <img width src="three">
    [3] => <img src = 'four' />
    [4] => <img src =five>
)
Captured src values:Array
(
    [0] => one
    [1] => two
    [2] => three
    [3] => four
    [4] => five
)

说明 #

• 分支重置组(?|...)内部的两个分支，各自的捕获组(.+?)会被视为同一个分组（编号为2），不管匹配的是带引号还是无引号的src值，都会被存入$matches[2]；
• 同时保留了原正则对完整<img>标签的捕获（分组1），满足需求；
• 无需再手动复制分组数据，简化了后续处理逻辑。

内容的提问来源于stack exchange，提问作者jamacoe