Request for Regex Help

[Omniferum]

I've been trying to further refine my regex

 

[^\r]+((=|"|'|http).*\.exe)

 

That's the furthest i've gotten, it works in the sense of 'preference' i.e. it iwll find " or ' or http first and then go all the way to the extension.

 

The problem is though it doesn't leave out the " or ' and if I try adding a negated character class like [^"'] it stuffs the whole thing up.

 

I want something that should work like this, but it doesn't annoyingly enough.

 

[^\r]+(([^"=']|http).*\.exe)

 

Any pointers? If I can get this down i'll have a regex that finds the download url for ALL my apps. Even tried working from the end, i'll admit MOST of my time finding this was not knowing that Ketarin was singleline regex and case insensitive so that sorta made a lot of time go bye bye. Then again I didn't even know that regex came in flavors so balls to me.

[floele]

I added a note regarding these options to the wiki.

[shawn]

I think what you're after is:

[='"]([^'"\s]*\.exe)[\s'"]

[Omniferum]

Hm, I just realized that perhaps I am a silly bunny.

 

My original regex of [^"']+ worked fine, however all the problem URL's I have which have the http problem are after the = after the a href="

 

I think I remember omitting it because "header information may be in downloads" but forgot.

 

So [^"'=]+64\.exe works fine, finally found my bloody answer.

[Omniferum]

Hm, well that fixed a few. Some others don't actually start with = " ' and just http

 

I keep trying to find a way to work back from the file extension.

 

i.e. find \.exe then match the very first http as a preference, if not found then go through the " ' = group or the very last / without a space after/before it.

[shawn]

You're going to be very hard pressed to find one pattern that works in every situation. Generally, you can find one that'll work across most apps on a site, but even then you'll often need to include a filepart inside (like "x86" or "x64"), so there's no silver bullet.

[Omniferum]

Not trying to get it to fit file extensions or specific builds, I just want it to be able to find the extension and pick the first http or " or ' it finds working its way back from \.exe.

 

[^"'=]+[^"']+\.exe is pretty much the BEST i've come across