Perl Weekly Challenge 213: from here to there!
This post presents my solutions to the Perl Weekly Challenge 213.I keep doing the Perl Weekly Challenge in order to mantain my coding skills in good shape, as well as in order to learn new things, with particular regard to Raku, a language that I love.
This week, I solved the following tasks:
- PWC 213 - Task 1 - Raku
- PWC 213 - Task 2 - Raku
- PWC 213 - Task 1 in PostgreSQL PL/Perl
- PWC 213 - Task 2 in PostgreSQL PL/Perl
- PWC 213 - Task 1 in PostgreSQL PL/PgSQL
- PWC 213 - Task 2 in PostgreSQL PL/PgSQL
Raku Implementations
PWC 213 - Task 1 - Raku Implementation
The first task was really easy: sort a given input list of numbers keeping evens first.sub MAIN( *@n where { @n.elems == @n.grep( { $_ ~~ Int && $_ > 0 } ).elems } ) {
( @n.grep( * %% 2 ).sort.join( ',' ) ~ ',' ~ @n.grep( * !%% 2 ).sort.join( ',' ) ).say;
}
It’s a single line of code:
grep the evens, sort and concatenate with the same logic for odds.
PWC 213 - Task 2 - Raku Implementation
This is a much more complex task, and I’m not sure my solution is totally correct since I don’t know the logic inside a parition. The task gave you a list of partitions made by nodes, each one identified by a number, and a starting and ending value. You need to find the path from the begin to the end point jumping thru partitions.sub MAIN( Int $source, Int $destination, $r ) {
my @routes;
my @current;
for $r.comb( :skip-empty ) {
next if ! $_;
if $_ ~~ "|" {
@routes.push: [@current];
@current = ();
next;
}
@current.push: $_.Int if ( $_.Int > 0 );
}
@routes.push: [@current] if ( @current );
my ( $source-index, $destination-index );
for 0 ..^ @routes.elems -> $index {
$source-index = $index and next if ( @routes[ $index ].grep( $source ) );
$destination-index = $index and next if ( @routes[ $index ].grep( $destination ) );
}
my @current-path;
my $next-route = $source-index;
my $loop = True;
while ( $loop ) {
for @routes[ $next-route ].Array -> $node {
@current-path.push: $node if ( ! @current-path.grep( * ~~ $node ) );
$loop = False;
for $next-route ^..^ @routes.elems -> $jump-to {
if ( @routes[ $jump-to ].grep( { $_ ~~ $node } ) ) {
$next-route = $jump-to;
$loop = True;
last;
}
}
last if $loop;
}
}
@current-path.join( ' -> ' ).say;
}
I assume the list of partitions is passed as a string, where each partition is separated by
|, so the first looping is to get the @routes available. Then I search for the $source-index that identifies the partition that contains the source point.
From there, I loop over the next partitions, adding every node to @current-path. For every node, I scan again the @routes to see if there’s another partition that contains the current $node, and that means is the next partition to jump to.
If I find another partition to jump, I loop again and switch to such partition, until there’s no more partitions to visit. I then print the ending result.
This is not totally correct, since edge cases, like two partitions having the same next node can screw up the whole thing, as well as a single partition containing both the source and destination nodes…
PL/Perl Implementations
PWC 213 - Task 1 - PL/Perl Implementation
Same approach as the Raku solution. Could have been a single line, but for sake of readability I made it count for three:CREATE OR REPLACE FUNCTION
pwc213.task1_plperl( int[] )
RETURNS int[]
AS $CODE$
my ( $array ) = @_;
my @sorted;
@sorted = ( sort( grep( { $_ % 2 == 0 } $array->@* ) ),
sort( grep( { $_ % 2 != 0 } $array->@* ) ) );
return [ @sorted ];
$CODE$
LANGUAGE plperl;
PWC 213 - Task 2 - PL/Perl Implementation
Pretty much the same stuff as in the Raku solution, but now I consider the function to complete succesfully only if the last node is the end.CREATE OR REPLACE FUNCTION
pwc213.task2_plperl( int, int, int[][] )
RETURNS SETOF int
AS $CODE$
my ( $source, $destination, $routes ) = @_;
my @path;
push @path, $source;
my ( $loop ) = 1;
my ( $current_route_index ) = 0;
while ( $loop ) {
my ( $current_route ) = $routes->@[ $current_route_index ];
# skip this route if there is not a match
$current_route_index++ and next if ( ! grep( { $_ == $path[ -1 ] } $current_route->@* ) );
for my $node ( $current_route->@* ) {
push @path, $node if ( ! grep( { $node == $_ } @path ) );
$loop = 0;
# search for the next route
for my $next_route_index ( ( $current_route_index + 1 ) .. scalar( $routes->@* ) ) {
next if ( ! grep( { $node == $_ } $routes->@[ $next_route_index ]->@* ) );
$current_route_index = $next_route_index;
$loop = 1;
last;
}
last if $loop;
}
last if $current_route_index > scalar( $routes->@* );
}
return undef if $path[ -1 ] != $destination;
return [ @path ];
$CODE$
LANGUAGE plperl;
PostgreSQL Implementations
PWC 213 - Task 1 - PL/PgSQL Implementation
This can be solved with a single, nested, query.CREATE OR REPLACE FUNCTION
pwc213.task1_plpgsql( a int[] )
RETURNS int[]
AS $CODE$
WITH evens AS (
SELECT array_agg( v ) as x
FROM ( SELECT v FROM unnest( a ) v
WHERE v % 2 = 0
ORDER BY 1
) as v
), odds AS (
SELECT array_agg( v ) as x
FROM ( SELECT v FROM unnest( a ) v
WHERE v % 2 <> 0
ORDER BY 1 ) as v
)
SELECT array_cat( e.x, o.x )
FROM evens e, odds o;
$CODE$
LANGUAGE sql;
The CTE selects the odds and evens numbers out of the array, sorting them, and aggregating into a single array by means of
array_agg. The outer query concatenates the two arrays.
PWC 213 - Task 2 - PL/PgSQL Implementation
Cheating here: use the PL/Perl function!CREATE OR REPLACE FUNCTION
pwc213.task2_plpgsql( s int, d int, routes int[] )
RETURNS SETOF int
AS $CODE$
SELECT pwc213.task2_plperl( s, d, routes );
$CODE$
LANGUAGE sql;
The fact is that you cannot easily manage (pass) multidimensional arrays to PL/PgSQL functions. However, I then spent some time in implementing the function, that appears as:
CREATE OR REPLACE FUNCTION
pwc213.task2_plpgsql( s int, d int, routes int[] )
RETURNS SETOF int
AS $CODE$
DECLARE
slice_size int := 3;
current_route_index int;
current_route int[];
next_route_index int;
next_node int;
need_loop boolean;
node int;
path int[];
BEGIN
need_loop := true;
current_route_index := 1;
<<rescan>>
WHILE need_loop LOOP
FOREACH node IN ARRAY routes[ current_route_index : current_route_index ] LOOP
RETURN NEXT node;
IF node = d THEN
EXIT;
END IF;
need_loop := false;
FOR next_route_index IN current_route_index + 1 .. array_length( routes, 1 ) LOOP
FOREACH next_node IN ARRAY routes[ next_route_index : next_route_index ] LOOP
IF next_node = node THEN
current_route_index := next_route_index;
need_loop := true;
CONTINUE rescan;
END IF;
END LOOP;
END LOOP;
END LOOP;
END LOOP;
IF node <> d THEN
RAISE EXCEPTION 'Cannot find the path!';
END IF;
return;
END
$CODE$
LANGUAGE plpgsql;
The trick here is to slice the routes array into its own subarrays one at a time, and then check for joining of the partitions. Since this function returns a
SETOF, it cannot fail easily. because it is already returning results, so I decided to throw an exception thru RAISE EXCEPTION.