As your analysis gets more complex, you might find gffutils useful for accessing GFF attributes. It has a fairly robust GFF/GTF parser. Here's an example that only keeps >1kb features that are in the LINE superfamily (as you suggest is your goal in a comment on another answer).

(edit: disclaimer, I'm the primary author)

This is a simple perl script to filter by size which I had lying around which could do that, but python would be just as simple to learn for the long run. I recommend www.pythonforbiologists.com

#!/usr/bin/perl
use strict;
use warnings;

my $usage = "## USAGE:  filter_csv.pl infile.txt [min value]\n";

my $infile = $ARGV[0] or die $usage;
my $filter = $ARGV[1] or die $usage;

open INFILE, "<", "$infile" or die "$usage | check $infile\n";

my $n=0;

while (my $line = <INFILE>){
    $n++;
    if ($n == 1){
        ## Don't process on header line (Maybe you don't need this bit)
        print $line;
    }else{
        my @l = split(/\t/, $line);
        if ($l[4] - $l[3] >= $filter){
            print $line;
        }
    }
}

You would run it like so:

./filter_csv.pl my_file.gff 1000

You could add in a filter for the right term by using a similar if clause and adding it in the size loop.

EDIT: For example (but not tested):

while (my $line = <INFILE>){
    $n++;
    if ($n == 1){
        ## Don't process on header line (Maybe you don't need this bit)
        print $line;
    }else{
       my @l = split(/\t/, $line);
       if ($l[4] - $l[3] >= $filter) && ($line ~= /$term/`){
            print $line;
        }
    }
}

Since you asked for Python, this simple script should work.

import sys
for line in sys.stdin:
    line = line.rstrip()
    values = line.split("\t")
    start, end = int(values[3]), int(values[4])
    length = end - start + 1
    if length >= 1000:
        print line

It was not tested on a file containing comments or pragmas, but that should be easy to account for.

In your question, you mention semicolon-separated attributes. If you're trying to further process the data (rather than just filtering out lines < 1 kb), you'll need to be more specific about what you need.