适用于所有文本的正则表达式公式，包括HTML标记内，而不是'a'标记内

Question

I am struggling with the formula to replace text in a string, while looking in all tags apart from 'a' tags (links).

This is my current formula:

\\b(?<!-)text\\b(?<!<[^<>]*)(?!-|[^<>]*>)

Example, I would like to replace only the instances of 'text' outside of the 'a' tag, so within the 'p' and 'li' tags:

<p>
    Some text here. <a href="#">Some more text</a>
    <ul>
        <li>Other text</li>
    </ul>
</p>

It must also match on whole words including dashes, which the formula currently does successfully. It also (I believe) doesn't replace anything within the tags themselves, ie:

<li class="text">here some text</li>

It would not replace the 'text' in the class name.

Answer 1

This formula should do what you want:

(?:\A|<\/a>)(?:[^<]|<[^a])*?(text)

The results of the match (meaning text in this case) are stored in the corresponding group (1) of the match. Note however, that this does fail for example if there are links inside the link, as mentioned in the comments, or if you have weird strings inside your document. Also note that this only matches up to the first match in an area outside of a link - if you want to replace something, you'll have to run the expression multiple times. If you only have a maximum of 2 occurances of text in such an area, you can combine this query with this one:

(?:\A|<\/a>)(?:[^<]|<[^a])*(text)

which matches the last text .

I'm aware this isn't an optimal solution, so I'm open to improvement suggestions. However, I'm not sure whether Regex is the best solution here (as Yunnosch pointed out), as Regex does only have a limited power (it is, after all, a type-3 grammar).

Let me explain the formula:

• (?:\\A|<\\/a>) match either the start of the input or the end of a link
• (?:[^<]|<[^a])* match everything that is not an indication of a tag, or - if so - if it's at least not the start of a link
• ? match as little as possible (hence the first match)
• (text) match the actual text (and save it in the group)

You can access the groups for example by using $1 or /1 , depending on your environment.