A little config parsing.

rendler has asked for the wisdom of the Perl Monks concerning the following question:

I've been working on a script in the past couple of days thats news boxes of other site. I've put the script in my stratchpad. So now I'm trying to improve the code and make it less spaghetti like. First of all I'd like to improve the config file parsing. I've been trying to redo it and so far I have

my $sites_conf = "$ENV{'HOME'}/.newsboxes/sites.conf";
my %sites_config = parse_sites_config();
sub parse_sites_config {
    open SITES_CFG, "< $sites_conf" or die "Couldn't open $sites_conf:
+ $!\n";
    my %options;

    while(<SITES_CFG>) {
    chomp;
    s/\/\/\s+.*//g;    # Get rid of comments;
    next unless length;

    if (my($key, $operator, $value) = /^\s*?(\S+)\s*?(\=\>?)\s*?(\S+)\
+s*$/) {
        if ($operator eq "=>") {
        ++$options{'total_sites'};
        $options{$options{'total_sites'}}{'site_name'} = $key;
        $options{$options{'total_sites'}}{'site_url'}  = $value;
        } elsif ($operator eq "=") {
        $options{$options{'total_sites'}}{$key} = $value;
        } else {
        die "Ooops you might want to check $sites_conf for errors for 
+site: $options{$options{'total_sites'}}{'site_name'}\n";
        }
    }
    }
    return %options;
}
[download]

Also the sites entry in the config file usually looks like this

Kuro5hin => http://www.kuro5hin.org
  key       = k5
  story_tag = item
  title_tag = title
  link_tag  = link
  site_xml  = http://www.kuro5hin.org/backend.rdf
  refresh   = 1800
  colour_1  = #302BA2
  colour_2  = #FFFFFF
  colour_3  = #E0E0E0
  colour_4  = #302BA2
[download]

But sometimes such as for "Linux Today" it has the space in the site name and because I'm using \S+ to match that it's not working for those sites that have space(s) in their names. So I was wondering what I should be using instead. Thanks.

Comment on A little config parsing. Select or Download Code

Replies are listed 'Best First'.
Re: A little config parsing. by perrin (Chancellor) on Jan 11, 2002 at 08:59 UTC
Why not just make the config in Perl code and require it? `my %conf = ( Kuro5hin => { url => 'http://www.kuro5hin.org', key => 'k5', story_tag => 'item', } );` [download] Or use Data::Denter? Or one of the millions of config parsing modules on CPAN?	[reply] [d/l]
Re: Re: A little config parsing. by rendler (Pilgrim) on Jan 11, 2002 at 09:10 UTC
That's a good idea, never thought of it ;) As for using a module, I just needed something simple without needing too many lines to work it or use anything external.	[reply]
Re: Re: A little config parsing. by cforde (Monk) on Jan 12, 2002 at 01:44 UTC
I second the suggestion of using a prewritten parser. I like Config::IniFiles which parses INI format files like this: `[Linux Today] url = http://linuxtoday.org story_tag = item title_tag = title link_tag = link refresh = 1800` [download] If you really can't use an external module, this format is very simple to parse yourself. It's also easy to understand, and probably familiar to the person following you... Have fun, Carl Forde	[reply] [d/l]
Re: A little config parsing. by Chmrr (Vicar) on Jan 11, 2002 at 06:37 UTC
I think `split` is your friend in this case. Perhaps something along the lines of `my ($key,$operator,$value) = split /\s(=>?)\s/,$_,2;` will set you in the right direction. The rest is just trimming leading and trailing whitespace. Update: In the interests of TIMTOWDI, I do believe that one can modify your above regex to the following: `/(\S.?)\s(\=\>?)\s(.+?)\s/` However, due to the amount of backtracking involved in the regex, along with the general ugliness of it, I'd personally use split in this case. In my mind, it also better captures the essence of what you're trying to do -- you've got these two things, and they'er seperated by something. You want to split them apart, so.. perl -pe '"I lo`+$^X$\"$]!$/"=~m%(.)%s;$_=$1;y^`+*^e v^#$&V"+@( NO CARRIER'	[reply] [d/l] [select]
Re: Re: A little config parsing. by rendler (Pilgrim) on Jan 11, 2002 at 08:05 UTC
Thanks that worked great, but then I went to test the results with `for my $i (1 .. $sites_config{'total_sites'}) { print "$sites_config{$i}{'site_name'}\n"; }` [download] What I don't get is how `$i` is getting auto incremented.	[reply] [d/l] [select]
Re: Re: Re: A little config parsing. by Chmrr (Vicar) on Jan 11, 2002 at 08:14 UTC
`$i` is the loop variable for your for loop. Perl sets it to every value in the list that the loop is iterating over -- in this case, the list of numbers from 1 to `$sites_config{'total_sites'}`. As an aside, any particular reason you're using a hash and not an array? It seems like you're mixing metaphors a bit to have </code>$site_config{'some_string'}</code> be at the same level as `$site_config{$some_site_number}`. perl -pe '"I lo`+$^X$\"$]!$/"=~m%(.)%s;$_=$1;y^`+*^e v^#$&V"+@( NO CARRIER'	[reply] [d/l] [select]
Re: Re: Re: Re: A little config parsing. by rendler (Pilgrim) on Jan 11, 2002 at 08:39 UTC
Re: Re: Re: Re: Re: A little config parsing. by Chmrr (Vicar) on Jan 11, 2002 at 08:59 UTC
Re: A little config parsing. by blokhead (Monsignor) on Jan 11, 2002 at 08:17 UTC
`(my($key, $operator, $value) = /^\s?(\S+)\s?(\=\>?)\s?(\S+)\s$/)` [download] as long as the site names don't have an equals sign in them, i don't see why this wouldn't work: `(my($key, $operator, $value) = /^\s?([^=]+?)\s?(\=\>?)\s?(\S+)\s$/ +)` [download] I changed the \s+ to `[^=]+?`	[reply] [d/l] [select]