CSS parsing

So, I ran into this guy that was very impressed with himself that he wrote a php script that reads a css file, parses it into a data structure, then prints it out, and that he wrote it in only 20 minutes. I was sitting at o'hare airport waiting for a plane so I decided to impress myself back by writing one in perl shorter, faster and parsing into a fuller data structure. Here it is:

perl -lne'chomp;@a=split/:/;push@{$h{$1}},[@a]if(/(.+){/../}/)&&
length($a[1]);END{for(keys%h){$h{$_}=~s/$_.*{|\t//g;print"$_ {";
for(@{$h{$_}}){print"    $_->[0] : $_->[1]";}print"}\n"}}' test.css
[download]

Comment on CSS parsing Download Code

Replies are listed 'Best First'.
Re: CSS parsing by ambrus (Abbot) on Jun 15, 2005 at 15:11 UTC
I tried this on Common Theme CSS, but it doesn't completely work. It seems that it can't parse single-line blocks such as this: `span.title { float: left }` [download]	[reply] [d/l]
Re^2: CSS parsing by ww (Archbishop) on Jun 15, 2005 at 17:12 UTC
WAG: Could that be for lack of the (NON_required) semicolon in the block? Aemi is not spec'ed for a single element block (nor for the last of multiple items in a block, IIRC) but (mere anecdotal assertion) I've found it wise to include to keep some browsers happy.	[reply]
Re^3: CSS parsing by ambrus (Abbot) on Jun 15, 2005 at 19:15 UTC
No, I doubt that. It has nothing to do with the semicolon, I think the semicolon would even make matters worse. As far as I understand the code, the parsing works line by line. Each line is expected to be either a condition and an opening brace, or two strings separated by a colon, or a closing brace. If a line is more complicated than this, this simple code can't break it apart.	[reply]