cat data.txt | prog1.pl | prog2.pl option | prog3.pl > out.txt
[download]

# input file
good1
good2
bad1
good3

# expected output
good1
good2
good3
[download]

output data (which can be fake, usually) in a nicely plain text, easy to read and easy to edit format. That rules out this, for example:

my %testdata = (
                 'good1' => 'good1',
                 'good2' => 'good2',
                 'bad1' => '',
                 'good3' => 'good3',
               );
[download]

I seems possible all the programs could be tested by the same test script, using different data for each test script. On the other hand, I don't know if that's the best solution, because it seems nice for each test to be self contained.

Another choice is to use a __DATA__ section. That's OK, but it comes at the end of the file, decreasing ease of reading and editing a tad. And it would require another label be embedded and parsed... not that that's hard, but the simpler, the better.

Still another choice is HERE docs. I'm leaning toward that, but wondering if there are better suggestions:

my ($args = <<'ARGS')=~s/^\s+//gm;
arg1
ARGS

my ($input = <<'INPUT')=~s/^\s+//gm;
good1
good2
bad1
good3
INPUT

my ($expected_output = <<'OUTPUT')=~s/^\s+//gm;
good1
good2
good3
OUTPUT
[download]

Separate from the setup question is how to compare the actual and expected results. Or at least I thought at first that these were separate. But there may be some interaction between the two questions. When the data is stored in a hash, each element is being treated individually, with a placeholder for an empty result in some cases. With the HERE doc, when the result is empty, it just means a shorter list. The latter feels like a better approach, and for my application there doesn't need to be a one-to-one correspondence between each input and output item. My main concern is finding a nice way to set up the data. Thoughts or suggested idioms appreciated!