[Bast-commits] r9250 -
SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract
dhoss at dev.catalyst.perl.org
dhoss at dev.catalyst.perl.org
Tue Apr 27 21:43:25 GMT 2010
Author: dhoss
Date: 2010-04-27 22:43:25 +0100 (Tue, 27 Apr 2010)
New Revision: 9250
Modified:
SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Test.pm
SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Tree.pm
Log:
fixed up some syntax errors in ::Tree
Modified: SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Test.pm
===================================================================
--- SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Test.pm 2010-04-27 21:13:26 UTC (rev 9249)
+++ SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Test.pm 2010-04-27 21:43:25 UTC (rev 9250)
@@ -1,443 +1,181 @@
-package SQL::Abstract::Test; # see doc at end of file
-
+package SQL::Abstract::Test; # see doc at end of file
use strict;
use warnings;
use base qw/Test::Builder::Module Exporter/;
use Data::Dumper;
use Carp;
use Test::Builder;
-
+use SQL::Abstract::Tree;
our @EXPORT_OK = qw/&is_same_sql_bind &is_same_sql &is_same_bind
- &eq_sql_bind &eq_sql &eq_bind
- $case_sensitive $sql_differ/;
+ &eq_sql_bind &eq_sql &eq_bind
+ $case_sensitive $sql_differ/;
-our $case_sensitive = 0;
+our $case_sensitive = 0;
our $parenthesis_significant = 0;
-our $sql_differ; # keeps track of differing portion between SQLs
+our $sql_differ; # keeps track of differing portion between SQLs
our $tb = __PACKAGE__->builder;
+our $tree = SQL::Abstract::Tree->new;
-# 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_RHS => 3;
-
-# These SQL keywords always signal end of the current expression (except inside
-# of a parenthesized subexpression).
-# Format: A list of strings that will be compiled to extended syntax (ie.
-# /.../x) regexes, without capturing parentheses. They will be automatically
-# anchored to word boundaries to match the whole token).
-my @expression_terminator_sql_keywords = (
- 'SELECT',
- 'FROM',
- '(?:
- (?:
- (?: \b (?: LEFT | RIGHT | FULL ) \s+ )?
- (?: \b (?: CROSS | INNER | OUTER ) \s+ )?
- )?
- JOIN
- )',
- 'ON',
- 'WHERE',
- 'EXISTS',
- 'GROUP \s+ BY',
- 'HAVING',
- 'ORDER \s+ BY',
- 'LIMIT',
- 'OFFSET',
- 'FOR',
- 'UNION',
- 'INTERSECT',
- 'EXCEPT',
- 'RETURNING',
-);
-
-# 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
- {
- ' ^ ' . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ",
- " (?<= $stuff_around_mathops)" . quotemeta ($_) . "(?= \$ | $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',
- 'HAVING',
- 'ORDER \s+ BY',
-);
-
sub is_same_sql_bind {
- my ($sql1, $bind_ref1, $sql2, $bind_ref2, $msg) = @_;
+ my ( $sql1, $bind_ref1, $sql2, $bind_ref2, $msg ) = @_;
- # compare
- my $same_sql = eq_sql($sql1, $sql2);
- my $same_bind = eq_bind($bind_ref1, $bind_ref2);
+ # compare
+ my $same_sql = eq_sql( $sql1, $sql2 );
+ my $same_bind = eq_bind( $bind_ref1, $bind_ref2 );
- # call Test::Builder::ok
- my $ret = $tb->ok($same_sql && $same_bind, $msg);
+ # call Test::Builder::ok
+ my $ret = $tb->ok( $same_sql && $same_bind, $msg );
- # add debugging info
- if (!$same_sql) {
- _sql_differ_diag($sql1, $sql2);
- }
- if (!$same_bind) {
- _bind_differ_diag($bind_ref1, $bind_ref2);
- }
+ # add debugging info
+ if ( !$same_sql ) {
+ _sql_differ_diag( $sql1, $sql2 );
+ }
+ if ( !$same_bind ) {
+ _bind_differ_diag( $bind_ref1, $bind_ref2 );
+ }
- # pass ok() result further
- return $ret;
+ # pass ok() result further
+ return $ret;
}
sub is_same_sql {
- my ($sql1, $sql2, $msg) = @_;
+ my ( $sql1, $sql2, $msg ) = @_;
- # compare
- my $same_sql = eq_sql($sql1, $sql2);
+ # compare
+ my $same_sql = eq_sql( $sql1, $sql2 );
- # call Test::Builder::ok
- my $ret = $tb->ok($same_sql, $msg);
+ # call Test::Builder::ok
+ my $ret = $tb->ok( $same_sql, $msg );
- # add debugging info
- if (!$same_sql) {
- _sql_differ_diag($sql1, $sql2);
- }
+ # add debugging info
+ if ( !$same_sql ) {
+ _sql_differ_diag( $sql1, $sql2 );
+ }
- # pass ok() result further
- return $ret;
+ # pass ok() result further
+ return $ret;
}
sub is_same_bind {
- my ($bind_ref1, $bind_ref2, $msg) = @_;
+ my ( $bind_ref1, $bind_ref2, $msg ) = @_;
- # compare
- my $same_bind = eq_bind($bind_ref1, $bind_ref2);
+ # compare
+ my $same_bind = eq_bind( $bind_ref1, $bind_ref2 );
- # call Test::Builder::ok
- my $ret = $tb->ok($same_bind, $msg);
+ # call Test::Builder::ok
+ my $ret = $tb->ok( $same_bind, $msg );
- # add debugging info
- if (!$same_bind) {
- _bind_differ_diag($bind_ref1, $bind_ref2);
- }
+ # add debugging info
+ if ( !$same_bind ) {
+ _bind_differ_diag( $bind_ref1, $bind_ref2 );
+ }
- # pass ok() result further
- return $ret;
+ # pass ok() result further
+ return $ret;
}
sub _sql_differ_diag {
- my ($sql1, $sql2) = @_;
+ my ( $sql1, $sql2 ) = @_;
- $tb->diag("SQL expressions differ\n"
- ." got: $sql1\n"
- ."expected: $sql2\n"
- ."differing in :\n$sql_differ\n"
- );
+ $tb->diag( "SQL expressions differ\n"
+ . " got: $sql1\n"
+ . "expected: $sql2\n"
+ . "differing in :\n$sql_differ\n" );
}
sub _bind_differ_diag {
- my ($bind_ref1, $bind_ref2) = @_;
+ my ( $bind_ref1, $bind_ref2 ) = @_;
- $tb->diag("BIND values differ\n"
- ." got: " . Dumper($bind_ref1)
- ."expected: " . Dumper($bind_ref2)
- );
+ $tb->diag(
+ "BIND values differ\n" . " got: " . Dumper($bind_ref1) . "expected: " . Dumper($bind_ref2) );
}
sub eq_sql_bind {
- my ($sql1, $bind_ref1, $sql2, $bind_ref2) = @_;
+ my ( $sql1, $bind_ref1, $sql2, $bind_ref2 ) = @_;
- return eq_sql($sql1, $sql2) && eq_bind($bind_ref1, $bind_ref2);
+ return eq_sql( $sql1, $sql2 ) && eq_bind( $bind_ref1, $bind_ref2 );
}
-
sub eq_bind {
- my ($bind_ref1, $bind_ref2) = @_;
+ my ( $bind_ref1, $bind_ref2 ) = @_;
- local $Data::Dumper::Useqq = 1;
- local $Data::Dumper::Sortkeys = 1;
+ local $Data::Dumper::Useqq = 1;
+ local $Data::Dumper::Sortkeys = 1;
- return Dumper($bind_ref1) eq Dumper($bind_ref2);
+ return Dumper($bind_ref1) eq Dumper($bind_ref2);
}
sub eq_sql {
- my ($sql1, $sql2) = @_;
+ my ( $sql1, $sql2 ) = @_;
- # parse
- my $tree1 = parse($sql1);
- my $tree2 = parse($sql2);
+ # parse
+ my $tree1 = $tree->parse($sql1);
+ my $tree2 = $tree->parse($sql2);
- return 1 if _eq_sql($tree1, $tree2);
+ return 1 if _eq_sql( $tree1, $tree2 );
}
sub _eq_sql {
- my ($left, $right) = @_;
+ my ( $left, $right ) = @_;
- # one is defined the other not
- if ( (defined $left) xor (defined $right) ) {
- return 0;
- }
- # one is undefined, then so is the other
- elsif (not defined $left) {
- return 1;
- }
- # one is a list, the other is an op with a list
- elsif (ref $left->[0] xor ref $right->[0]) {
- $sql_differ = sprintf ("left: %s\nright: %s\n", map { unparse ($_) } ($left, $right) );
- return 0;
- }
- # one is a list, so is the other
- elsif (ref $left->[0]) {
- for (my $i = 0; $i <= $#$left or $i <= $#$right; $i++ ) {
- return 0 if (not _eq_sql ($left->[$i], $right->[$i]) );
+ # one is defined the other not
+ if ( ( defined $left ) xor( defined $right ) ) {
+ return 0;
}
- return 1;
- }
- # both are an op-list combo
- else {
- # unroll parenthesis if possible/allowed
- _parenthesis_unroll ($_) for ($left, $right);
-
- # if operators are different
- if ( $left->[0] ne $right->[0] ) {
- $sql_differ = sprintf "OP [$left->[0]] != [$right->[0]] in\nleft: %s\nright: %s\n",
- unparse($left),
- unparse($right);
- return 0;
+ # one is undefined, then so is the other
+ elsif ( not defined $left ) {
+ return 1;
}
- # elsif operators are identical, compare operands
- else {
- if ($left->[0] eq 'LITERAL' ) { # unary
- (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);
- $sql_differ = "[$l] != [$r]\n" if not $eq;
- return $eq;
- }
- else {
- my $eq = _eq_sql($left->[1], $right->[1]);
- $sql_differ ||= sprintf ("left: %s\nright: %s\n", map { unparse ($_) } ($left, $right) ) if not $eq;
- return $eq;
- }
- }
- }
-}
-sub parse {
- my $s = shift;
-
- # tokenize string, and remove all optional whitespace
- my $tokens = [];
- foreach my $token (split $tokenizer_re, $s) {
- push @$tokens, $token if (length $token) && ($token =~ /\S/);
- }
-
- my $tree = _recurse_parse($tokens, PARSE_TOP_LEVEL);
- return $tree;
-}
-
-sub _recurse_parse {
- my ($tokens, $state) = @_;
-
- my $left;
- while (1) { # left-associative parsing
-
- my $lookahead = $tokens->[0];
- if ( not defined($lookahead)
- or
- ($state == PARSE_IN_PARENS && $lookahead eq ')')
- or
- ($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;
+ # one is a list, the other is an op with a list
+ elsif ( ref $left->[0] xor ref $right->[0] ) {
+ $sql_differ = sprintf( "left: %s\nright: %s\n", map { $tree->unparse($_) } ( $left, $right ) );
+ return 0;
}
- my $token = shift @$tokens;
-
- # nested expression in ()
- if ($token eq '(' ) {
- my $right = _recurse_parse($tokens, PARSE_IN_PARENS);
- $token = shift @$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] ];
+ # one is a list, so is the other
+ elsif ( ref $left->[0] ) {
+ for ( my $i = 0 ; $i <= $#$left or $i <= $#$right ; $i++ ) {
+ return 0 if ( not _eq_sql( $left->[$i], $right->[$i] ) );
+ }
+ return 1;
}
- # AND/OR
- elsif ($token =~ /^ (?: OR | AND ) $/xi ) {
- my $op = uc $token;
- my $right = _recurse_parse($tokens, PARSE_IN_EXPR);
- # Merge chunks if logic matches
- if (ref $right and $op eq $right->[0]) {
- $left = [ (shift @$right ), [$left, map { @$_ } @$right] ];
- }
- else {
- $left = [$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 LITERAL for a 1st RHS argument needs a
- # rerun of the search to (hopefully) find the proper AND construct
- if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') {
- unshift @$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] ]];
- }
- # 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] ]];
-
- }
- # literal (eat everything on the right until RHS termination)
+ # both are an op-list combo
else {
- my $right = _recurse_parse ($tokens, PARSE_RHS);
- $left = $left ? [$left, [LITERAL => [join ' ', $token, unparse($right)||()] ] ]
- : [ LITERAL => [join ' ', $token, unparse($right)||()] ];
- }
- }
-}
-sub _parenthesis_unroll {
- my $ast = shift;
+ # unroll parenthesis if possible/allowed
+ $tree->_parenthesis_unroll($_) for ( $left, $right );
- return if $parenthesis_significant;
- return unless (ref $ast and ref $ast->[1]);
+ # if operators are different
+ if ( $left->[0] ne $right->[0] ) {
+ $sql_differ = sprintf "OP [$left->[0]] != [$right->[0]] in\nleft: %s\nright: %s\n",
+ $tree->unparse($left),
+ $tree->unparse($right);
+ return 0;
+ }
- 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];
- $changes++;
- }
-
- # only one LITERAL element in the parenthesis
- elsif (
- @{$child->[1]} == 1 && $child->[1][0][0] eq 'LITERAL'
- ) {
- push @children, $child->[1][0];
- $changes++;
- }
-
- # only one element in the parenthesis which is a binary op with two LITERAL sub-children
- elsif (
- @{$child->[1]} == 1
- and
- grep { $child->[1][0][0] =~ /^ $_ $/xi } (@binary_op_keywords)
- and
- $child->[1][0][1][0][0] eq 'LITERAL'
- and
- $child->[1][0][1][1][0] eq 'LITERAL'
- ) {
- push @children, $child->[1][0];
- $changes++;
- }
-
- # otherwise no more mucking for this pass
- else {
- push @children, $child;
- }
+ # elsif operators are identical, compare operands
+ else {
+ if ( $left->[0] eq 'LITERAL' ) { # unary
+ ( 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);
+ $sql_differ = "[$l] != [$r]\n" if not $eq;
+ return $eq;
+ } else {
+ my $eq = _eq_sql( $left->[1], $right->[1] );
+ $sql_differ ||= sprintf( "left: %s\nright: %s\n", map { $tree->unparse($_) } ( $left, $right ) )
+ if not $eq;
+ return $eq;
+ }
+ }
}
-
- $ast->[1] = \@children;
-
- } while ($changes);
-
}
-sub unparse {
- my $tree = shift;
-
- if (not $tree ) {
- return '';
- }
- elsif (ref $tree->[0]) {
- return join (" ", map { unparse ($_) } @$tree);
- }
- elsif ($tree->[0] eq 'LITERAL') {
- return $tree->[1][0];
- }
- elsif ($tree->[0] eq 'PAREN') {
- return sprintf '(%s)', join (" ", map {unparse($_)} @{$tree->[1]});
- }
- 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 {
- return sprintf '%s %s', $tree->[0], unparse ($tree->[1]);
- }
-}
-
-
1;
-
__END__
=head1 NAME
Modified: SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Tree.pm
===================================================================
--- SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Tree.pm 2010-04-27 21:13:26 UTC (rev 9249)
+++ SQL-Abstract/1.x/branches/sqla-tree/lib/SQL/Abstract/Tree.pm 2010-04-27 21:43:25 UTC (rev 9250)
@@ -5,7 +5,6 @@
use Data::Dumper;
use Carp;
-
has 'case_sensitive' => (
is => 'rw',
required => 1,
@@ -20,7 +19,6 @@
default => 0
);
-
has 'tb' => (
is => 'rw',
lazy_build => 1,
@@ -87,53 +85,57 @@
);
sub _build_binary_op_keyworkds {
- my $self = shift;
+ my $self = shift;
return (
map {
' ^ ' . quotemeta($_) . "(?= \$ | " . $self->stuff_around_mathops . " ) ",
- " (?<= " . $self->stuff_around_mathops . ")" . quotemeta($_) . "(?= \$ | " . $self->stuff_around_mathops . ") ",
+ " (?<= "
+ . $self->stuff_around_mathops . ")"
+ . quotemeta($_)
+ . "(?= \$ | "
+ . $self->stuff_around_mathops . ") ",
} (qw/< > != <> = <= >=/)
),
(
map {
'\b (?: NOT \s+)?' . $_ . '\b'
} (qw/IN BETWEEN LIKE/)
- ),
- );
+
+ );
}
has 'tokenizer_re_str' => (
- is => 'ro',
- required => 1,
- lazy_build => 1,
+ is => 'ro',
+ required => 1,
+ lazy_build => 1,
);
sub _build_tokenizer_re_str {
- my $self = shift;
- return join( "\n\t|\n",
- ( map { '\b' . $_ . '\b' } $self->expression_terminator_sql_keywords, 'AND', 'OR', 'NOT' ),
- $self->binary_op_keywords, );
+ my $self = shift;
+ return join( "\n\t|\n",
+ ( map { '\b' . $_ . '\b' } $self->expression_terminator_sql_keywords, 'AND', 'OR', 'NOT' ),
+ $self->binary_op_keywords, );
}
has 'tokenizer_re' => (
- is => 'ro',
- required => 1,
- lazy_build => 1,
+ is => 'ro',
+ required => 1,
+ lazy_build => 1,
);
sub _build_tokenizer_re {
- my $self = shift;
- return qr/ \s* ( $self->tokenizer_re_str | \( | \) | \? ) \s* /xi;
+ my $self = shift;
+ return qr/ \s* ( $self->tokenizer_re_str | \( | \) | \? ) \s* /xi;
}
has 'unrollable_ops' => (
- is => 'ro',
- required => 1,
- lazy_build => 1,
+ is => 'ro',
+ required => 1,
+ lazy_build => 1,
);
sub _build_unrollable_ops {
- return ( 'ON', 'WHERE', 'GROUP \s+ BY', 'HAVING', 'ORDER \s+ BY', );
+ return ( 'ON', 'WHERE', 'GROUP \s+ BY', 'HAVING', 'ORDER \s+ BY', );
}
# Parser states for _recurse_parse()
@@ -143,196 +145,208 @@
use constant PARSE_RHS => 3;
sub parse {
- my ($self, $s) = @_;
+ my ( $self, $s ) = @_;
- # tokenize string, and remove all optional whitespace
- my $tokens = [];
- foreach my $token (split $self->tokenizer_re, $s) {
- push @$tokens, $token if (length $token) && ($token =~ /\S/);
- }
+ # tokenize string, and remove all optional whitespace
+ my $tokens = [];
+ foreach my $token ( split $self->tokenizer_re, $s ) {
+ push @$tokens, $token if ( length $token ) && ( $token =~ /\S/ );
+ }
- my $tree = _recurse_parse($tokens, PARSE_TOP_LEVEL);
- return $tree;
+ my $tree = _recurse_parse( $tokens, PARSE_TOP_LEVEL );
+ return $tree;
}
sub _recurse_parse {
- my ($self, $tokens, $state) = @_;
+ my ( $self, $tokens, $state ) = @_;
- my $left;
- while (1) { # left-associative parsing
+ my $left;
+ while (1) { # left-associative parsing
- my $lookahead = $tokens->[0];
- if ( not defined($lookahead)
- or
- ($state == PARSE_IN_PARENS && $lookahead eq ')')
- or
- ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', $self->expression_terminator_sql_keywords ) )
- or
- ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', $self->expression_terminator_sql_keywords, $self->binary_op_keywords, 'AND', 'OR', 'NOT' ) )
- ) {
- return $left;
- }
+ my $lookahead = $tokens->[0];
+ if (
+ not defined($lookahead)
+ or ( $state == PARSE_IN_PARENS && $lookahead eq ')' )
+ or ( $state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi }
+ ( '\)', $self->expression_terminator_sql_keywords ) )
+ or (
+ $state == PARSE_RHS
+ && grep { $lookahead =~ /^ $_ $/xi } (
+ '\)', $self->expression_terminator_sql_keywords,
+ $self->binary_op_keywords, 'AND', 'OR', 'NOT'
+ )
+ )
+ )
+ {
+ return $left;
+ }
- my $token = shift @$tokens;
+ my $token = shift @$tokens;
- # nested expression in ()
- if ($token eq '(' ) {
- my $right = _recurse_parse($tokens, PARSE_IN_PARENS);
- $token = shift @$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
- elsif ($token =~ /^ (?: OR | AND ) $/xi ) {
- my $op = uc $token;
- my $right = _recurse_parse($tokens, PARSE_IN_EXPR);
+ # nested expression in ()
+ if ( $token eq '(' ) {
+ my $right = _recurse_parse( $tokens, PARSE_IN_PARENS );
+ $token = shift @$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] ];
+ }
- # Merge chunks if logic matches
- if (ref $right and $op eq $right->[0]) {
- $left = [ (shift @$right ), [$left, map { @$_ } @$right] ];
- }
- else {
- $left = [$op => [$left, $right]];
- }
- }
- # binary operator keywords
- elsif (grep { $token =~ /^ $_ $/xi } $self->binary_op_keywords ) {
- my $op = uc $token;
- my $right = _recurse_parse($tokens, PARSE_RHS);
+ # AND/OR
+ elsif ( $token =~ /^ (?: OR | AND ) $/xi ) {
+ my $op = uc $token;
+ my $right = _recurse_parse( $tokens, PARSE_IN_EXPR );
- # A between with a simple LITERAL for a 1st RHS argument needs a
- # rerun of the search to (hopefully) find the proper AND construct
- if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') {
- unshift @$tokens, $right->[1][0];
- $right = _recurse_parse($tokens, PARSE_IN_EXPR);
- }
+ # Merge chunks if logic matches
+ if ( ref $right and $op eq $right->[0] ) {
+ $left = [ ( shift @$right ), [ $left, map { @$_ } @$right ] ];
+ } else {
+ $left = [ $op => [ $left, $right ] ];
+ }
+ }
- $left = [$op => [$left, $right] ];
- }
- # expression terminator keywords (as they start a new expression)
- elsif (grep { $token =~ /^ $_ $/xi } $self->expression_terminator_sql_keywords ) {
- my $op = uc $token;
- my $right = _recurse_parse($tokens, PARSE_IN_EXPR);
- $left = $left ? [@$left, [$op => [$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] ]];
+ # binary operator keywords
+ elsif ( grep { $token =~ /^ $_ $/xi } $self->binary_op_keywords ) {
+ my $op = uc $token;
+ my $right = _recurse_parse( $tokens, PARSE_RHS );
+ # A between with a simple LITERAL for a 1st RHS argument needs a
+ # rerun of the search to (hopefully) find the proper AND construct
+ if ( $op eq 'BETWEEN' and $right->[0] eq 'LITERAL' ) {
+ unshift @$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 } $self->expression_terminator_sql_keywords ) {
+ my $op = uc $token;
+ my $right = _recurse_parse( $tokens, PARSE_IN_EXPR );
+ $left =
+ $left
+ ? [ @$left, [ $op => [$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] ] ];
+
+ }
+
+ # literal (eat everything on the right until RHS termination)
+ else {
+ my $right = _recurse_parse( $tokens, PARSE_RHS );
+ $left =
+ $left
+ ? [ $left, [ LITERAL => [ join ' ', $token, unparse($right) || () ] ] ]
+ : [ LITERAL => [ join ' ', $token, unparse($right) || () ] ];
+ }
}
- # literal (eat everything on the right until RHS termination)
- else {
- my $right = _recurse_parse ($tokens, PARSE_RHS);
- $left = $left ? [$left, [LITERAL => [join ' ', $token, unparse($right)||()] ] ]
- : [ LITERAL => [join ' ', $token, unparse($right)||()] ];
- }
- }
}
sub _parenthesis_unroll {
- my ($self, $ast) = @_;
+ my ( $self, $ast ) = @_;
- return if $parenthesis_significant;
- return unless (ref $ast and ref $ast->[1]);
+ return if $self->parenthesis_significant;
+ return unless ( ref $ast and ref $ast->[1] );
- my $changes;
- do {
- my @children;
- $changes = 0;
+ 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;
- }
+ 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++;
- }
+ # 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 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 } $self->unrollable_ops ) {
- push @children, $child->[1][0];
- $changes++;
- }
+ # if the parent operator explcitly allows it nuke the parenthesis
+ elsif ( grep { $ast->[0] =~ /^ $_ $/xi } $self->unrollable_ops ) {
+ push @children, $child->[1][0];
+ $changes++;
+ }
- # only one LITERAL element in the parenthesis
- elsif (
- @{$child->[1]} == 1 && $child->[1][0][0] eq 'LITERAL'
- ) {
- push @children, $child->[1][0];
- $changes++;
- }
+ # only one LITERAL element in the parenthesis
+ elsif ( @{ $child->[1] } == 1 && $child->[1][0][0] eq 'LITERAL' ) {
+ push @children, $child->[1][0];
+ $changes++;
+ }
- # only one element in the parenthesis which is a binary op with two LITERAL sub-children
- elsif (
- @{$child->[1]} == 1
- and
- grep { $child->[1][0][0] =~ /^ $_ $/xi } ($self->binary_op_keywords)
- and
- $child->[1][0][1][0][0] eq 'LITERAL'
- and
- $child->[1][0][1][1][0] eq 'LITERAL'
- ) {
- push @children, $child->[1][0];
- $changes++;
- }
+ # only one element in the parenthesis which is a binary op with two LITERAL sub-children
+ elsif ( @{ $child->[1] } == 1
+ and grep { $child->[1][0][0] =~ /^ $_ $/xi } ( $self->binary_op_keywords )
+ and $child->[1][0][1][0][0] eq 'LITERAL'
+ and $child->[1][0][1][1][0] eq 'LITERAL' )
+ {
+ push @children, $child->[1][0];
+ $changes++;
+ }
- # otherwise no more mucking for this pass
- else {
- push @children, $child;
- }
- }
+ # otherwise no more mucking for this pass
+ else {
+ push @children, $child;
+ }
+ }
- $ast->[1] = \@children;
+ $ast->[1] = \@children;
- } while ($changes);
+ } while ($changes);
}
sub unparse {
- my ($self, $tree) = shift;
+ my ( $self, $tree ) = shift;
- if (not $tree ) {
- return '';
- }
- elsif (ref $tree->[0]) {
- return join (" ", map { unparse ($_) } @$tree);
- }
- elsif ($tree->[0] eq 'LITERAL') {
- return $tree->[1][0];
- }
- elsif ($tree->[0] eq 'PAREN') {
- return sprintf '(%s)', join (" ", map {unparse($_)} @{$tree->[1]});
- }
- elsif ($tree->[0] eq 'OR' or $tree->[0] eq 'AND' or (grep { $tree->[0] =~ /^ $_ $/xi } $self->binary_op_keywords ) ) {
- return join (" $tree->[0] ", map {unparse($_)} @{$tree->[1]});
- }
- else {
- return sprintf '%s %s', $tree->[0], unparse ($tree->[1]);
- }
+ if ( not $tree ) {
+ return '';
+ } elsif ( ref $tree->[0] ) {
+ return join( " ", map { unparse($_) } @$tree );
+ } elsif ( $tree->[0] eq 'LITERAL' ) {
+ return $tree->[1][0];
+ } elsif ( $tree->[0] eq 'PAREN' ) {
+ return sprintf '(%s)', join( " ", map { unparse($_) } @{ $tree->[1] } );
+ } elsif (
+ $tree->[0] eq 'OR' or $tree->[0] eq 'AND' or (
+ grep {
+ $tree->[0] =~ /^ $_ $/xi
+ } $self->binary_op_keywords
+ )
+ )
+ {
+ return join( " $tree->[0] ", map { unparse($_) } @{ $tree->[1] } );
+ } else {
+ return sprintf '%s %s', $tree->[0], unparse( $tree->[1] );
+ }
}
## DO THE DAMN THANG
sub pretty {
- my $self = shift;
+ my $self = shift;
}
__PACKAGE__->meta->make_immutable;
More information about the Bast-commits
mailing list