Regex list discussion

General Customizing Pi-hole
1

Continuing the discussion from Collection of RegEx for blacklisting:

@Smolka_J I just skimmed over the list but found a few entries, I do not really understand, e.g.

The first * means "any number of initial dots or line beginnings". This is not meaningful at all but makes the regex more costly to evaluate. Similarly for the second *, it means "match zero ts" as in "-aler", one t as in -alert or any number of ts like -alerttttt. I don't think this is what you want.

This is just an example for many similar things in there.

2

Thank you for the input. I agree this list needs some further review to make the most effective. I've just been combining what I could find from multiple blacklist related sources into one list, alphabetized, de-duplicated, and then over the last couple years removed most any entries I found that were creating excess false positive alerts, I'm not the original writer of them nor am I a regex expert. My main goal here was testing one lists syntax vs another from what I came across and getting regex alerts of any kind to pop more often than practically not at all like I was first was finding. Been running this list for over a year now without "needing" much edited or whitelisted and am always open to suggestions for improvements, still a work in progress and wanted to get a bigger list out there compared to most regex lists I found about that are just a dozen or two lines or less

3

Just be careful.
In regex, the asterisk (*) is not a wildcard, but a dot (.) is. There are many other rules to learn.

Apparently you are adding regex entries that will block more than you expect.

Another example from your list:
((^)|(.))adchoice.

The first dot will match any character once (not just a dot). The last dot will also match any character. Note: it could be simplified as (^|.)adchoice.

This will match any domains containing words with the following pattern: _adchoice_ (where _ is a letter, a digit, a dot, an hyphen or _).

I think the intention was to block only the domains and subdomains with adchoice between dots, like: www.adchoice.com.

In this case, your regex should be (^|\.)adchoice\.

4

Here's some advice: it seems like you're using too many pantheist beliefs. For instance, using “[-.]” or “[.-]” in this example : “^banners?[-.], ^banners?[.-]” which is twice the same search load, which in turn produce the same single results. So, choose one and use it instead, I was able to check some of the regex in the list, coming up with the same result, hope this helps.