I should probably say: don't do this at home, kids.
I'm not really sure what sroux is after. I have assumed the 'intention' from the solutions that others baked.
For fun, I open up a postgres SQL interface to the data in this file. I'll admit this postgres spelunking is not for the faint-hearted or for DB beginners. The advantage is getting access to a textfile via the power of SQL.
# bash
file=/tmp/flatfile.dat t=flattable sep='|'
echo "head1|head2|head3
val1|val2|val3
val1|val4|val5
val6|val4|val5
val2|val7|val5
val3|val7|val3" > $file
# postgres extension 'file_fdw' must be installed (but haven't we all)
+:
psql -qXa -c "create server if not exists fs foreign data wrapper file
+_fdw;"
psql -qXa << SQL_PERL
drop foreign table if exists $t;
create foreign table $t ($( < $file perl -ne 'chomp; my $t = "'$t'";
print join("\n , ", map{$_ .= " text"} split(/['${sep}']/)); exit;'
))
server fs options(delimiter'$sep',format'csv',header'TRUE',filename'$f
+ile');
SQL_PERL
running the above, on-screen the SQL that was run, echoed from psql:
create server if not exists fs foreign data wrapper file_fdw;
drop foreign table if exists flattable;
create foreign table flattable (head1 text
, head2 text
, head3 text
)
server fs options(delimiter'|',format'csv',header'TRUE',filename'/tmp/
+flatfile.dat');
Now the file is accessible as a foreign table named 'flattable'.
Once a postgres table, you can run SQL against it, for instance this SELECT:
(select 'head1' "heads", array_agg(distinct head1 order by head1) "val
+ues" from flattable) union all
(select 'head2' "heads", array_agg(distinct head2 order by head2) "val
+ues" from flattable) union all
(select 'head3' "heads", array_agg(distinct head3 order by head3) "val
+ues" from flattable)
-- output:
--
-- heads | values
-- -------+-----------------------
-- head1 | {val1,val2,val3,val6}
-- head2 | {val2,val4,val7}
-- head3 | {val3,val5}
-- (3 rows)
In case of many columns, say hundreds (it happens!), generate the needed SQL. (see psql's \gexec -- left as an exercise for the reader)