Perl SQL::Parser table alias substitution: works for SELECT column names but not for WHERE column names - sql

Question

I'm trying to parse some SQL queries stored in a log database -- I don't want to submit them to a SQL database, just to extract the fields used in the SELECT and WHERE clause.
I've been fiddling with several SQL parsers in Java, Python and Perl. The one that seems to work better for my problem are SQL::Parser and SQL::Statement. With those I was able to write the following code:
#!/usr/bin/perl
use strict;
use SQL::Parser;
use SQL::Statement;
use Data::Dumper;
my $sql = "SELECT sl.plate,sp.fehadop FROM sppLines AS sl ".
"JOIN sppParams AS sp ON sl.specobjid = sp.specobjid ".
"WHERE fehadop < -3.5 ";
my $parser = SQL::Parser->new();
my $stmt = SQL::Statement->new($sql,$parser);
printf("COMMAND [%s]\n",$stmt->command);
printf("COLUMNS \n");
my #columns = #{$stmt->column_defs()};
foreach my $column ( #columns)
{
print " ".$column->{value}."\n";
}
printf("TABLES \n");
my #tables = $stmt->tables();
foreach my $table ( #tables)
{
print " ".$table->{name}."\n";
}
printf("WHERE COLUMNS\n");
my $where_hash = $stmt->where_hash();
print Dumper($where_hash);
Sorry if it is too long, it is the smallest, self-contained example I could devise.
The output of this code is:
COMMAND [SELECT]
COLUMNS
spplines.plate
sppparams.fehadop
TABLES
spplines
sppparams
WHERE COLUMNS
$VAR1 = {
'arg1' => {
'value' => 'fehadop',
'type' => 'column',
'fullorg' => 'fehadop'
},
'op' => '<',
'nots' => {},
'arg2' => {
'str' => '-?0?',
'fullorg' => '-3.5',
'name' => 'numeric_exp',
'value' => [
{
'fullorg' => '3.5',
'value' => '3.5',
'type' => 'number'
}
],
'type' => 'function'
},
'neg' => 0
};
The parser returns the name of columns (obtained through a call to $stmt->column_defs()) already renamed with the real tables names (e.g. spplines.plate instead of s1.plate) -- this is what I want.
I also want the names of the columns used in the WHERE clause.
I already know how to recursively parse the results of $stmt->where_hash() (didn't include the code to make the post clear), but even from dumping its contents I can see that the column names are not associated with the tables.
I would like to ensure that the columns names in the WHERE clause are also preceded by the tables name. After parsing the results of $stmt->where_hash() I would get sppparams.fehadop instead of fehadop.
Is this possible with SQL::Parser?
Thanks
(big edit -- tried to make the question clearer)

Answer 1

Since SQL::Statement has an eval_where, I suspect there might be a better way, but you can try a function like this:
get_column($stmt->column_defs(), $where_hash->{arg1});
sub get_column {
my ($columns, $arg) = #_;
return $arg->{fullorg} if ($arg->{type} ne 'column');
foreach my $col (#$columns) {
return $col->{value} if ($col->{fullorg} eq $arg->{fullorg});
my ($name) = ( $col->{fullorg} =~ /([^.]+)$/);
return $col->{value} if ($name eq $arg->{fullorg});
}
return $arg->{fullorg};
}