Question

Sometimes my APL familiarity gives me algorithmic ideas for problem solving that I re-implement in a language I have – Perl, for example.

So I have processed a text file to create a boolean vector indicating the used fields in a delimited file, and now I want to output the indexes of those used fields, and the names of the used fields. In APL, I would use the compress operator over the vector of field names, and over the iota of the number of fields.

In Perl, I did this:

my @UsedFieldNames = map { $UsedFields[$_] ? $FieldNames[$_] : () } 0 .. $#UsedFields;

and

say join " ", map { $UsedFields[$_] ? $) : () } 0 .. $#UsedFields;

where @UsedFields is an array containing 0 for unused and 1 for used fields.

1. I don’t really like using map with ?:() to simulate compress – is there a better way (my real program does it a third time when simulating a vertical or reduction over the file)?

2. I don’t really like doing the map over the indexes to get the results – is there a better way to compute that? (I guess one optimization would be to compute the used indexes first, then

   @UsedFieldNames = @FieldNames[@UsedIndexes];

Answer 1

Other ways:

my @UsedFieldNames = map { ( $FieldNames[$_] ) x !!$UsedFields[$_] } 0..$#UsedFields;
my @UsedFieldNames = @FieldNames[ grep $UsedFields[$_], 0..$#UsedFields ];