This produces the desired output for the input you showed:

perl -ple '$_=join",",/\d+/g' input.txt
[download]

But seriously, it seems like you're not telling us the whole story. Not only does this sound like homework, but your example input & output or specification don't indicate:

• whether there can be lines without any of the target characters, and if yes, where they can appear,
• whether there can be any other characters other than the ones you showed on each line, and if yes, where they can appear,
• what happens if the things between the target characters aren't numbers,
• why you said "extract all numbers" but your existing regex appears to only want a specific number of digits.
• There's probably a couple more questions that I'm not thinking of at the moment.
• Update: Are there always three numbers on the line following the single line?

Technically, my example above solves the problem you posed, so if you want a regex that really works for you, you'll have to be a lot more specific.