Fix over-eager parenthesis unrolling (only legal in AND/OR)

[dbsrgits/SQL-Abstract.git] / lib / SQL / Abstract / Tree.pm

diff --git a/lib/SQL/Abstract/Tree.pm b/lib/SQL/Abstract/Tree.pm
index 5b7b704..1a4560c 100644 (file)
--- a/lib/SQL/Abstract/Tree.pm
+++ b/lib/SQL/Abstract/Tree.pm
@@ -50,10 +50,10 @@ my $placeholder_re = qr/(?: \? | \$\d+ )/x;
 my @expression_start_keywords = (
   'SELECT',
   'UPDATE',
+  'SET',
   'INSERT \s+ INTO',
   'DELETE \s+ FROM',
   'FROM',
-  'SET',
   '(?:
     (?:
         (?: (?: LEFT | RIGHT | FULL ) \s+ )?
@@ -64,7 +64,7 @@ my @expression_start_keywords = (
   'ON',
   'WHERE',
   '(?: DEFAULT \s+ )? VALUES',
-  '(?:NOT \s+)? EXISTS',
+  '(?: NOT \s+)? EXISTS',
   'GROUP \s+ BY',
   'HAVING',
   'ORDER \s+ BY',
@@ -95,6 +95,7 @@ $expr_start_re = qr/ $op_look_behind (?i: $expr_start_re ) $op_look_ahead /x;
 # * BETWEEN without paranthesis around the ANDed arguments (which
 #   makes it a non-binary op) is detected and accomodated in
 #   _recurse_parse()
+# * AS is not really an operator but is handled here as it's also LHS/RHS
 
 # this will be included in the $binary_op_re, the distinction is interesting during
 # testing as one is tighter than the other, plus mathops have different look
@@ -111,7 +112,7 @@ sub _math_op_re { $math_re }
 
 my $binary_op_re = '(?: NOT \s+)? (?:' . join ('|', qw/IN BETWEEN R?LIKE/) . ')';
 $binary_op_re = join "\n\t|\n",
-  "$op_look_behind (?i: $binary_op_re ) $op_look_ahead",
+  "$op_look_behind (?i: $binary_op_re | AS ) $op_look_ahead",
   $math_re,
   $op_look_behind . 'IS (?:\s+ NOT)?' . "(?= \\s+ NULL \\b | $op_look_ahead )",
 ;
@@ -309,6 +310,10 @@ sub parse {
   $self->_recurse_parse($tokens, PARSE_TOP_LEVEL);
 }
 
+{
+# this is temporary, lists can be parsed *without* recursing, but
+# it requires a massive rewrite of the AST generator
+no warnings qw/recursion/;
 sub _recurse_parse {
   my ($self, $tokens, $state) = @_;
 
@@ -344,14 +349,14 @@ sub _recurse_parse {
     elsif ($token =~ /^ (?: OR | AND | \, ) $/xi )  {
       my $op = ($token eq ',') ? 'LIST' : uc $token;
 
-      my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
+      my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR) || [];
 
       # Merge chunks if logic matches
-      if (ref $right and $op eq $right->[0]) {
-        $left = [ (shift @$right ), [$left||(), map { @$_ } @$right] ];
+      if (ref $right and @$right and $op eq $right->[0]) {
+        $left = [ (shift @$right ), [$left||[], map { @$_ } @$right] ];
       }
       else {
-        $left = [$op => [ $left||(), $right||() ]];
+        $left = [$op => [ $left||[], $right ]];
       }
     }
     # binary operator keywords
@@ -405,6 +410,7 @@ sub _recurse_parse {
     }
   }
 }
+}
 
 sub format_keyword {
   my ($self, $keyword) = @_;
@@ -471,7 +477,9 @@ sub _unparse {
     return '';
   }
 
+  # FIXME - needs a config switch to disable
   $self->_parenthesis_unroll($tree);
+
   my ($car, $cdr) = @{$tree}[0,1];
 
   if (! defined $car or (! ref $car and ! defined $cdr) ) {
@@ -536,7 +544,6 @@ sub _parenthesis_unroll {
   my $self = shift;
   my $ast = shift;
 
-  #return if $self->parenthesis_significant;
   return unless (ref $ast and ref $ast->[1]);
 
   my $changes;
@@ -545,6 +552,7 @@ sub _parenthesis_unroll {
     $changes = 0;
 
     for my $child (@{$ast->[1]}) {
+
       # the current node in this loop is *always* a PAREN
       if (! ref $child or ! @$child or $child->[0] ne 'PAREN') {
         push @children, $child;
@@ -588,8 +596,8 @@ sub _parenthesis_unroll {
         $changes++;
       }
 
-      # only one element in the parenthesis which is a binary op
-      # and has exactly two grandchildren
+      # an AND/OR expression with only one binop in the parenthesis
+      # with exactly two grandchildren
       # the only time when we can *not* unroll this is when both
       # the parent and the child are mathops (in which case we'll
       # break precedence) or when the child is BETWEEN (special
@@ -597,6 +605,8 @@ sub _parenthesis_unroll {
       elsif (
         @{$child->[1]} == 1
           and
+        ($ast->[0] eq 'AND' or $ast->[0] eq 'OR')
+          and
         $child->[1][0][0] =~ SQL::Abstract::Tree::_binary_op_re()
           and
         $child->[1][0][0] ne 'BETWEEN'