s/// has apparent side effect on grep()

J

John E. Jardine

Hi,

Problem:
Executing 's///' has a side effect on grep null string matching.
If line 62, the substitution, is executed the last two values returned by
grep and printed on lines 68, 69 are different than the values returned and
printed when line 62 is commented out. Line 62 shouldn't have any impact on
lines 67,68 & 69.

Environment:
(1) Reproducable under Perl 5.6.1 running on Linux 2.4.5 (Slackware 8.0
installation) Pentium MMX/166MHz, 96MB Ram.
(2) Reproducable under Perl 5.8.0 running on Linux 2.6.5 (Slackware 9.1.0
installation) AMD Athlon 64 3000+, 1GB Ram

To reply by e-mail, remove the dot-baseball treat-dot from my e-mail address

Example Code:
#!/usr/bin/perl -w
my $item;
my $fltr_expr;
my $fltr_attr1;
my $fltr_op1;
my $fltr_val1;
my $fltr_attr2;
my $fltr_op2;
my $fltr_val2;
my $fltr_attr3;
my $fltr_op3;
my $fltr_val3;
my @test_data = ();
my %fltr_num_ops;
my %fltr_str_ops;
my %card_columns = (network => "network_srname",
node => "node_srname",
nodetype => "node_srtype",
area => "node_lata",
cardtype => "srcardtype");


$fltr_num_ops{EQUAL} = " == ";
$fltr_num_ops{NOTEQUAL} = " != ";
$fltr_num_ops{LIKE} = " =~ m// ";
$fltr_num_ops{NOTLIKE} = " !~ m// ";
$fltr_str_ops{EQUAL} = " eq ";
$fltr_str_ops{NOTEQUAL} = " ne ";
$fltr_str_ops{LIKE} = " =~ m// ";
$fltr_str_ops{NOTLIKE} = " !~ m// ";

#------------------------------------------------------------------------
# Test data normally retrieved from database. This accurately represents
# the problem though.
#------------------------------------------------------------------------
push(@test_data, "cardtype|LIKE|HUB|||||||");
push(@test_data, "cardtype|EQUAL|CC2|||||||");
foreach $item (@test_data) {
( $fltr_attr1, $fltr_op1, $fltr_val1, $fltr_attr2, $fltr_op2, $fltr_val2,
$fltr_attr3, $fltr_op3, $fltr_val3 ) = split(/\|/, $item);

$fltr_attr1 =~ tr/A-Z/a-z/;
$fltr_attr1 =~ s#^\s*(\S*)\s*$#$1#;
$fltr_op1 =~ s#^\s*(\S*)\s*$#$1#;
$fltr_val1 =~ s#^\s*(\S*)\s*$#$1#;

#---------------------------------------------------------------------
# $fltr_attr2, $fltr_op2, fltr_val2, $fltr_attr3, $fltr_op3, fltr_val3
# are not processed in this example - normally they would've been.
#---------------------------------------------------------------------
$fltr_expr = "";
if( $fltr_val1 =~ m#^\d+$# ) {
$fltr_expr = qq{$fltr_attr1 $fltr_num_ops{$fltr_op1} $fltr_val1}; }
else {
$fltr_expr = qq{$fltr_attr1 $fltr_str_ops{$fltr_op1} "$fltr_val1"}; }

#---------------------------------------------------------------------
# This is where the problem is. Try (un)commenting out the
# middle line (substitution). It changes the results of the grep()s
# in the next section.
#---------------------------------------------------------------------
print "Before Substitute: '$fltr_expr'\n";
# $fltr_expr =~ s#m// +"*([^"]+)"*#m/$1/#g;
print "After Substitute: '$fltr_expr'\n";

print
"$fltr_attr1|$fltr_op1|$fltr_val1|$fltr_attr2|$fltr_op2|$fltr_val2|$fltr_att
r3|$fltr_op3|$fltr_val3\n";

print sprintf("grep(/$fltr_attr1/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr1/, keys %card_columns)));
print sprintf("grep(/$fltr_attr2/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr2/, keys %card_columns)));
print sprintf("grep(/$fltr_attr3/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr3/, keys %card_columns)));
print "fltr_attr1 = '$fltr_attr1'\n";
print "fltr_op1 = '$fltr_op1'\n";
print "fltr_val1 = '$fltr_val1'\n";
print "fltr_attr2 = '$fltr_attr2'\n";
print "fltr_op2 = '$fltr_op2'\n";
print "fltr_val2 = '$fltr_val2'\n";
print "fltr_attr3 = '$fltr_attr3'\n";
print "fltr_op3 = '$fltr_op3'\n";
print "fltr_val3 = '$fltr_val3'\n";
foreach (keys %card_columns) {
print "'$_' = '$card_columns{$_}'\n"; }
print "\n";
}
 
J

Joe Smith

John said:
Executing 's///' has a side effect on grep null string matching.

That is correct. It is documented behavior.

If the PATTERN evaluates to the empty string, the last success-
fully matched regular expression is used instead.

You need to change lines 67-69 in your program to

printf("grep(/$fltr_attr1/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr1/, keys %card_columns))) if $fltr_addr1 ne "";
printf("grep(/$fltr_attr2/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr2/, keys %card_columns))) if $fltr_addr2 ne "";
printf("grep(/$fltr_attr3/, keys %%card_columns) = %d\n",
scalar(grep(/$fltr_attr3/, keys %card_columns))) if $fltr_addr3 ne "";

-Joe
 
J

John Jardine

Hi Joe,

My bad - I must have missed that and simply never hit the issue before.
Thanks for clearing it up for me.

Cheers,
J.J.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,755
Messages
2,569,535
Members
45,007
Latest member
obedient dusk

Latest Threads

Top