$case_sensitive $sql_differ/;
our $case_sensitive = 0;
+our $parenthesis_significant = 0;
our $sql_differ; # keeps track of differing portion between SQLs
our $tb = __PACKAGE__->builder;
# Parser states for _recurse_parse()
+use constant PARSE_TOP_LEVEL => 0;
use constant PARSE_IN_EXPR => 1;
use constant PARSE_IN_PARENS => 2;
-use constant PARSE_TOP_LEVEL => 3;
+use constant PARSE_RHS => 3;
# These SQL keywords always signal end of the current expression (except inside
# of a parenthesized subexpression).
'EXCEPT',
);
-# All of these keywords allow their parameters to be wrapped in parenthesis without changing any semantics
-my @unrollable_sql_keywords = (
+# These are binary operator keywords always a single LHS and RHS
+# * AND/OR are handled separately as they are N-ary
+# * so is NOT as being unary
+# * BETWEEN without paranthesis around the ANDed arguments (which
+# makes it a non-binary op) is detected and accomodated in
+# _recurse_parse()
+my $stuff_around_mathops = qr/[\w\s\`\'\)]/;
+my @binary_op_keywords = (
+ ( map
+ { " (?<= $stuff_around_mathops) " . quotemeta $_ . "(?= $stuff_around_mathops )" }
+ (qw/< > != = <= >=/)
+ ),
+ ( map
+ { '\b (?: NOT \s+)?' . $_ . '\b' }
+ (qw/IN BETWEEN LIKE/)
+ ),
+);
+
+my $tokenizer_re_str = join("\n\t|\n",
+ ( map { '\b' . $_ . '\b' } @expression_terminator_sql_keywords, 'AND', 'OR', 'NOT'),
+ @binary_op_keywords,
+);
+
+my $tokenizer_re = qr/ \s* ( \( | \) | \? | $tokenizer_re_str ) \s* /xi;
+
+# All of these keywords allow their parameters to be specified with or without parenthesis without changing the semantics
+my @unrollable_ops = (
'ON',
'WHERE',
'GROUP \s+ BY',
'ORDER \s+ BY',
);
-my $tokenizer_re_str = join('|',
- map { '\b' . $_ . '\b' }
- @expression_terminator_sql_keywords, 'AND', 'OR'
-);
-
-my $tokenizer_re = qr/
- \s*
- (
- \(
- |
- \)
- |
- $tokenizer_re_str
- )
- \s*
-/xi;
-
-my $unrollable_re_str = join ('|', map { $_ } @unrollable_sql_keywords);
-my $unrollable_re = qr/^ (?: $unrollable_re_str ) $/ix;
-
-
sub is_same_sql_bind {
my ($sql1, $bind_ref1, $sql2, $bind_ref2, $msg) = @_;
# both are an op-list combo
else {
- for ($left, $right) {
-
- next unless (ref $_->[1]);
-
- # unroll parenthesis in an elaborate loop
- my $changes;
- do {
-
- my @children;
- $changes = 0;
-
- for my $child (@{$_->[1]}) {
- if (not ref $child or not $child->[0] eq 'PAREN') {
- push @children, $child;
- next;
- }
-
- # unroll nested parenthesis
- while ($child->[1][0][0] eq 'PAREN') {
- $child = $child->[1][0];
- $changes++;
- }
-
- # if the parens are wrapped around an AND/OR matching the parent AND/OR - open the parens up and merge the list
- if (
- ( $_->[0] eq 'AND' or $_->[0] eq 'OR')
- and
- $child->[1][0][0] eq $_->[0]
- ) {
- push @children, @{$child->[1][0][1]};
- $changes++;
- }
-
- # if the parent operator explcitly allows it, or if parents are wrapped around an expression just nuke them
- elsif ( $child->[1][0][0] eq 'EXPR' or $_->[0] =~ $unrollable_re ) {
- push @children, $child->[1][0];
- $changes++;
- }
-
- # otherwise no more mucking
- else {
- push @children, $child;
- }
- }
-
- $_->[1] = \@children;
- } while ($changes);
- }
+ # unroll parenthesis if possible/allowed
+ _parenthesis_unroll ($_) for ($left, $right);
# if operators are different
if ($left->[0] ne $right->[0]) {
# elsif operators are identical, compare operands
else {
if ($left->[0] eq 'EXPR' ) { # unary operator
- (my $l = " $left->[1] " ) =~ s/\s+/ /g;
- (my $r = " $right->[1] ") =~ s/\s+/ /g;
+ (my $l = " $left->[1][0] " ) =~ s/\s+/ /g;
+ (my $r = " $right->[1][0] ") =~ s/\s+/ /g;
my $eq = $case_sensitive ? $l eq $r : uc($l) eq uc($r);
- $sql_differ = "[$left->[1]] != [$right->[1]]\n" if not $eq;
+ $sql_differ = "[$l] != [$r]\n" if not $eq;
return $eq;
}
else {
}
}
-
sub parse {
my $s = shift;
$token =~ s/\s+/ /g;
$token =~ s/\s+([^\w\s])/$1/g;
$token =~ s/([^\w\s])\s+/$1/g;
- push @$tokens, $token if $token !~ /^$/;
+ push @$tokens, $token if length $token;
}
my $tree = _recurse_parse($tokens, PARSE_TOP_LEVEL);
or
($state == PARSE_IN_PARENS && $lookahead eq ')')
or
- ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords) )
+ ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords ) )
+ or
+ ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords, @binary_op_keywords, 'AND', 'OR', 'NOT' ) )
) {
-
return $left;
- return ($state == PARSE_TOP_LEVEL
- ? $left->[0]
- : $left
- );
}
my $token = shift @$tokens;
# nested expression in ()
if ($token eq '(') {
my $right = _recurse_parse($tokens, PARSE_IN_PARENS);
- $token = shift @$tokens or croak "missing ')'";
- $token eq ')' or croak "unexpected token : $token";
+ $token = shift @$tokens or croak "missing closing ')' around block " . unparse ($right);
+ $token eq ')' or croak "unexpected token '$token' terminating block " . unparse ($right);
$left = $left ? [@$left, [PAREN => [$right] ]]
: [PAREN => [$right] ];
}
- # AND/OR
+ # AND/OR
elsif ($token =~ /^ (?: OR | AND ) $/xi ) {
my $op = uc $token;
my $right = _recurse_parse($tokens, PARSE_IN_EXPR);
$left = [$op => [$left, $right]];
}
}
+ # binary operator keywords
+ elsif (grep { $token =~ /^ $_ $/xi } @binary_op_keywords ) {
+ my $op = uc $token;
+ my $right = _recurse_parse($tokens, PARSE_RHS);
+
+ # A between with a simple EXPR for a 1st RHS argument needs a
+ # rerun of the search to (hopefully) find the proper AND construct
+ if ($op eq 'BETWEEN' and $right->[0] eq 'EXPR') {
+ unshift @$tokens, $right->[1][0];
+ $right = _recurse_parse($tokens, PARSE_IN_EXPR);
+ }
+
+ $left = [$op => [$left, $right] ];
+ }
# expression terminator keywords (as they start a new expression)
elsif (grep { $token =~ /^ $_ $/xi } @expression_terminator_sql_keywords ) {
my $op = uc $token;
my $right = _recurse_parse($tokens, PARSE_IN_EXPR);
$left = $left ? [@$left, [$op => [$right] ]]
- : [ [$op => [$right] ] ];
+ : [[ $op => [$right] ]];
+ }
+ # NOT (last as to allow all other NOT X pieces first)
+ elsif ( $token =~ /^ not $/ix ) {
+ my $op = uc $token;
+ my $right = _recurse_parse ($tokens, PARSE_RHS);
+ $left = $left ? [ @$left, [$op => [$right] ]]
+ : [[ $op => [$right] ]];
+
}
# leaf expression
else {
- $left = $left ? [@$left, [EXPR => $token] ]
- : [ EXPR => $token ];
+ $left = $left ? [@$left, [EXPR => [$token] ] ]
+ : [ EXPR => [$token] ];
}
}
}
+sub _parenthesis_unroll {
+ my $ast = shift;
+
+ return if $parenthesis_significant;
+ return unless (ref $ast and ref $ast->[1]);
+ my $changes;
+ do {
+ my @children;
+ $changes = 0;
+
+ for my $child (@{$ast->[1]}) {
+ if (not ref $child or not $child->[0] eq 'PAREN') {
+ push @children, $child;
+ next;
+ }
+
+ # unroll nested parenthesis
+ while ($child->[1][0][0] eq 'PAREN') {
+ $child = $child->[1][0];
+ $changes++;
+ }
+
+ # if the parenthesis are wrapped around an AND/OR matching the parent AND/OR - open the parenthesis up and merge the list
+ if (
+ ( $ast->[0] eq 'AND' or $ast->[0] eq 'OR')
+ and
+ $child->[1][0][0] eq $ast->[0]
+ ) {
+ push @children, @{$child->[1][0][1]};
+ $changes++;
+ }
+
+ # if the parent operator explcitly allows it nuke the parenthesis
+ elsif ( grep { $ast->[0] =~ /^ $_ $/xi } @unrollable_ops ) {
+ push @children, $child->[1][0];
+ $changes++;
+ }
+
+ # only one EXPR element in the parenthesis
+ elsif (
+ @{$child->[1]} == 1 && $child->[1][0][0] eq 'EXPR'
+ ) {
+ push @children, $child->[1][0];
+ $changes++;
+ }
+
+ # only one element in the parenthesis which is a binary op with two EXPR sub-children
+ elsif (
+ @{$child->[1]} == 1
+ and
+ grep { $child->[1][0][0] =~ /^ $_ $/xi } (@binary_op_keywords)
+ and
+ $child->[1][0][1][0][0] eq 'EXPR'
+ and
+ $child->[1][0][1][1][0] eq 'EXPR'
+ ) {
+ push @children, $child->[1][0];
+ $changes++;
+ }
+
+ # otherwise no more mucking for this pass
+ else {
+ push @children, $child;
+ }
+ }
+
+ $ast->[1] = \@children;
+
+ } while ($changes);
+
+}
sub unparse {
my $tree = shift;
return join (" ", map { unparse ($_) } @$tree);
}
elsif ($tree->[0] eq 'EXPR') {
- return $tree->[1];
+ return $tree->[1][0];
}
elsif ($tree->[0] eq 'PAREN') {
- return sprintf '( %s )', join (" ", map {unparse($_)} @{$tree->[1]});
+ return sprintf '(%s)', join (" ", map {unparse($_)} @{$tree->[1]});
}
- elsif ($tree->[0] eq 'OR' or $tree->[0] eq 'AND') {
+ elsif ($tree->[0] eq 'OR' or $tree->[0] eq 'AND' or (grep { $tree->[0] =~ /^ $_ $/xi } @binary_op_keywords ) ) {
return join (" $tree->[0] ", map {unparse($_)} @{$tree->[1]});
}
else {
is_same_sql_bind is_same_sql is_same_bind
eq_sql_bind eq_sql eq_bind
/];
-
+
my ($sql, @bind) = SQL::Abstract->new->select(%args);
is_same_sql_bind($given_sql, \@given_bind,
Therefore the tests will pass as long as the semantics
is preserved, even if the surface syntax has changed.
-B<Disclaimer> : this is only a half-cooked semantic equivalence;
-parsing is simple-minded, and comparison of SQL abstract syntax trees
-ignores commutativity or associativity of AND/OR operators, Morgan
-laws, etc.
+B<Disclaimer> : the semantic equivalence handling is pretty limited.
+A lot of effort goes into distinguishing significant from
+non-significant parenthesis, including AND/OR operator associativity.
+Currently this module does not support commutativity and more
+intelligent transformations like Morgan laws, etc.
+
+For a good overview of what this test framework is capable of refer
+to C<t/10test.t>
=head1 FUNCTIONS
If true, SQL comparisons will be case-sensitive. Default is false;
+=head2 $parenthesis_significant
+
+If true, SQL comparison will preserve and report difference in nested
+parenthesis. Useful for testing the C<-nest> modifier. Defaults to false;
+
=head2 $sql_differ
When L</eq_sql> returns false, the global variable
Norbert Buchmuller <norbi@nix.hu>
+Peter Rabbitson <ribasushi@cpan.org>
+
=head1 COPYRIGHT AND LICENSE
Copyright 2008 by Laurent Dami.
projects / scpubgit/Q-Branch.git / blobdiff