Escape hyphens to ease conversion to Ruby?

[ghost]

I'm getting regex compile warnings on regexes like

mail.[a-zA-Z0-9-.]+.pl\\/o\\/
email.[a-zA-Z0-9-.]+\\/o\\/[a-zA-Z0-9-_.]{64}

Ruby issues a warning when it compiles these:

warning: character class has '-' without escape:...

If I change the original regexes to

mail.[a-zA-Z0-9\\-.]+.pl\\/o\\/
email.[a-zA-Z0-9\\-.]+\\/o\\/[a-zA-Z0-9\\-_.]{64}

to escape the 'naked' dash, the regexes compile just fine. I'm pretty sure this is an artifact of JS that makes this legal but I just don't know for sure...

My question is: Are the second versions legal JS regexes?
My followup question is: Does this change alter the meanings of the original regexes?
My 2nd followup question would be: Which of the two versions are 'more correct™' ?

I'm not trying to start a JS vs Ruby flamewar here...just trying to make the awk script as robust as I can. I'm not a JS guy... So many languages...so little time...

[leggett]

I'll test and let you know.

[leggett]

It seems to be fine. It isn't technically valid as hyphens do not need to be escaped. So I think I could add it.

Another option, first do a replace for a-zA-Z0-9- → a-zA-Z0-9\\- as this will catch all the instances of a hyphen that you would need to escape for the conversion.

[ghost]

The particular statements are:

sub(/-_.]/, "\\\\-_.]")    # substitute the string '-_.]' with the string '\\-_.]'
sub(/-.]/, "\\\\-.]")      # substitute the string '-.]' with the string '\\-.]'        

If you could change your regexes, that would be great...I could get rid of this bit of ugliness.

I've also written a ruby program that uses the hash created by the awk script...like what I asked for here. It's a simple filter but demonstrates the technique. It only uses a few common gems. It should be understandable by anyone using the mail gem for handling mail in ruby.

It detects/removes the tracker <img> tags and adds an X-Trackers-Blocked header to the message. The content of the X-Trackers-Blocked is a JSON-formatted array of the names of the blocked trackers. It also blocks <img> tags with height="1" width="1" and calls those unknown. I'll add all of this as a demonstration after I do some cleanup and add some comments. As always, use at your own risk and no support expressed or implied.

It's based on something I already had (which is much more sophisticated) so I didn't have to re-invent the wheel...

[leggett]

If you have a fix for escaping the hyphens, I'd rather not escape them in JS since they don't need to be escaped there.

Your sub misses a case though:
"Mailcastr.com": "mailcastr.com\\/image\\/[a-zA-Z0-9-_]{64}",

So I think you'd be better off with:
sub(/a-zA-Z0-9-/, "a-zA-Z0-9\\\\-") # substitute the string 'a-zA-Z0-9-' with the string 'a-zA-Z0-9\\-'

I will try to keep the hyphen always after the 0-9 in these patterns.

Nice job removing 1x1 images. I would also remove 0x0 images as those happen sometimes too.

[ghost]

You write:

If you have a fix for escaping the hyphens, I'd rather not escape them in JS since they don't need to be escaped there.

OK.

So I think you'd be better off with:
sub(/a-zA-Z0-9-/, "a-zA-Z0-9\\\\-") # substitute the string 'a-zA-Z0-9-' with the string 'a-zA-Z0-9\\-'

OK... thanks.

Nice job removing 1x1 images. I would also remove 0x0 images as those happen sometimes too.

Thanks for the tip...an easy fix -- change '==' to '<=' in a couple of places...