Regex to match everything except a pattern

Question

Regex noob here struggling with this, which I know it will be easy for some of you regex gods out there!

Given the following:

title:      Some title
date:       2022-08-15
tags:       <value to extract>
identifier: 1234567
---------------------------

Some text
some more text

I would like a regex to match everything except the value of tags (ie the "<value to extract>" text).

For context, this is supposed to run on emacs (in case it matters).

EDIT: Just to clarify as per @phils question, all I care about extracting the tags value. However, this is via a package setting that asks for a regex string and I don't have much control over how it gets use. It seems to expect a regex to strip what I don't need from the string rather than matching what I do want, which is slightly annoying.. Also, the since it seems to match everything with \\(.\\) , I'm guessing it's using the global flag?

Please let me know if any of this isn't clear.

Answer 1

Emacs regular expressions can't trivially express "not foo" for arbitrary values of foo. (The likes of PCRE have non-regular extensions for zero-width negative look-ahead/behind assertions, but in Emacs that sort of functionality is generally done with the support of lisp code ¹ .)

You can still do it purely with regexp matching, but it's simply very cumbersome. An Emacs regexp which matches any line which does not begin with tags: is:

^\(?:$\|[^t]\|t[^a]\|ta[^g]\|tag[^s]\|tags[^:]\).*

---

¹ In lisp code you would instead simply check each line to see whether it does start with tags: and, if so, skip it (which is why Emacs generally gets away without the feature you're looking for, but of course that doesn't help you here).

Answer 2

After playing around with it for a bit and taken inspiration from @phils' answer, I've come up with the following:

"^\\(?:\\(#\\+\\)?\\(?:filetags:\s+\\|tags:\s+\\|title:.*\\|identifier:.*\\|date:.*\\)\\|.*\\)"

I've also added an extra \\(#\\+\\)? to account for org meta keys which would usually have the format #+key: value .