next »

[m0nkeymafia]

any of u regex nutters got a few mins to help a monkey out?

ive got a script to grab all the classes and id's used on a webpage
i then want to pull off any matching css code for each class out of the css files, but i cant figure out a regex to do it

i.e. if i have found a class called "topNav"
i want to search through the css files [already in a buffer] and find the corresponding definitions

i.e.

.topNav {  padding-top: 5px; }

would return just "padding-top: 5px;" obviously with any other lines

this way if i rob someones web layout / design and modify it, i can run it through this and discard any usesless classes
can anyone gimmie a hand on what to do? im stuck!!

cheers

[perkiset]

First - are you doing this in PHP?

(for the rest of this I'll assume so)

Second - you'll want to load all referenced CSS files (and their referenced files) just in case the definitions are not in the immediate page. Best way to find it then would be to create one big mess of all the linked files and do the Regex on <that string>.

Then, assuming you got a class reference, like 'topNav' and have it in the variable $className, you could do something like this:

preg_match('/([#\.]$className[\s]*{[^}]+/', $inputBuff, $matches);

... and then the $matches array would have anything that it found. Important to look for things that assign class by id too, rather than just class name - by this I mean:

<style>
#topNav { font-family: Arial; }
</style>
<div id="topNav">Here is the html</div>

Good luck!
/p

[m0nkeymafia]

cheers perk thats very similar to what i had but wasnt working, the bit at the end is different though.
will let ya know how it goes

and yeah its php
cheers matey

[perkiset]

Hey MM - I forgot something really important:

You really need to add an "m" as a modifier to the regex ie.,

preg_match('/([#\.]$className[\s]*{[^}]+/m', $inputBuff, $matches);

... this dicked with me for a long time. PHP won't let the pattern span multiple lines without it... and if someone defines a class like this:

.topNav {
   font-family: courier;
   font-size: 12px;
}

.. then the first regex I posted won't work.

Hope that helps,
/p

[m0nkeymafia]

excellent perk thanks, just about to try it

[m0nkeymafia]

sorry to be lame but you omitted the closing )
where should it go?

[Bompa]

sorry to be lame but you omitted the closing )
where should it go?

My guess is at the end, but before the semicolon.


Bompa

[m0nkeymafia]

ahh amazin that worked - i think lol
cheers for the help guys
hopefully have this finished and tweaked by tonight \o/

[perkiset]

Right on... BTW ... what's with the missing paren? There's no open parens in my post 

[m0nkeymafia]

off the end apparently

ok so i now have the regex to pull classes and id's out of a webpage
'/(class|id) *= *\"([a-zA-z0-9 ]*)\"/i'

i then explode the results to pull out all classes even if two+ are specified

then i use perks code to grab the classes out of the style sheet
'/([#\.]'.$val.'.*[\s]*{[^}]+)/m'

which all works great

only thing left to do is grab non-class selectors out of the style sheet, i.e. body {} and so forth
I originally tried something like this:

'/([^#\.]'.$val.'.*[\s]*{[^}]+)/m'

but didnt particularily work very well, as it seemed to pull all the classes off :/
anyone have any further ideas?

cheers

[perkiset]

On quick note, I'm assuming that the same code you used for style definitions could be used, except simpole negate the period or pound sign in front and replace the $val with letters/numbers...

/([^#\.][A-Za-z0-9_]+[\s]*{[^}]+)/m

... I think that'd do it, although you might possible get some stuff on a page that you didn't want... that one might take a bit more testing...

[m0nkeymafia]

Yeah perk i tried that [although i now realise i copied teh wrong syntax in my previous post]

for some reason it pulls up class definitions even though they have a preceeding period! very odd?
p.s. i actually get the sytnax u provided previously, when u first posted it i was like WTF !! but now it make sense, i really hope my regex skills keep improving lol [slowly ]

so any ideas why it may not work? even though we say we want text that is NOT directly preceeded by a # or . ??

[Bompa]

The dot is a literal within square brackets.   

(Dont escape it).




Bompa

[thedarkness]

Nice pickup Bomps.

Cheers,
td

[m0nkeymafia]

nice one bompa.
it still doesnt work though?

/([^.#][a-zA-Z0-9_].*[\s]*{[^}])/m

so if we expand it into constituent parts we have

Match a string that follows these rules:
- Starts NOT with a . or #
- Immediately followed by any number of letters or numbers or underscores
- Then match any amount of whitespace
- Then it needs to find an open parenthesis
- It then matches until it hits a closing parenthesis, at which point it stops
-
* Works over multiple lines

I cannot see how this matches ALL classes within my stylesheet?
The critical part, the NOT . or # reads fine to me? What am i missing?

Cheers