1 package SQL::Abstract::Tree;
8 use base 'Class::Accessor::Grouped';
10 __PACKAGE__->mk_group_accessors( simple => $_ ) for qw(
11 newline indent_string indent_amount colormap indentmap
14 # Parser states for _recurse_parse()
15 use constant PARSE_TOP_LEVEL => 0;
16 use constant PARSE_IN_EXPR => 1;
17 use constant PARSE_IN_PARENS => 2;
18 use constant PARSE_RHS => 3;
20 # These SQL keywords always signal end of the current expression (except inside
21 # of a parenthesized subexpression).
22 # Format: A list of strings that will be compiled to extended syntax (ie.
23 # /.../x) regexes, without capturing parentheses. They will be automatically
24 # anchored to word boundaries to match the whole token).
25 my @expression_terminator_sql_keywords = (
30 (?: \b (?: LEFT | RIGHT | FULL ) \s+ )?
31 (?: \b (?: CROSS | INNER | OUTER ) \s+ )?
50 # These are binary operator keywords always a single LHS and RHS
51 # * AND/OR are handled separately as they are N-ary
52 # * so is NOT as being unary
53 # * BETWEEN without paranthesis around the ANDed arguments (which
54 # makes it a non-binary op) is detected and accomodated in
56 my $stuff_around_mathops = qr/[\w\s\`\'\"\)]/;
57 my @binary_op_keywords = (
60 ' ^ ' . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ",
61 " (?<= $stuff_around_mathops)" . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ",
63 (qw/< > != <> = <= >=/)
66 { '\b (?: NOT \s+)?' . $_ . '\b' }
71 my $tokenizer_re_str = join("\n\t|\n",
72 ( map { '\b' . $_ . '\b' } @expression_terminator_sql_keywords, 'AND', 'OR', 'NOT'),
76 my $tokenizer_re = qr/ \s* ( $tokenizer_re_str | \( | \) | \? ) \s* /xi;
78 sub _binary_op_keywords { @binary_op_keywords }
92 indentmap => { %indents },
94 console_monochrome => {
99 indentmap => { %indents },
102 indent_string => ' ',
104 newline => "<br />\n",
106 select => ['<span class="select">', '</span>'],
107 where => ['<span class="where">', '</span>'],
108 from => ['<span class="from">', '</span>'],
110 indentmap => { %indents },
119 require Term::ANSIColor;
120 $profiles{console}->{colormap} = {
121 select => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')],
122 where => [Term::ANSIColor::color('green'), Term::ANSIColor::color('reset')],
123 from => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')],
128 my ($class, $args) = @_;
130 my $profile = delete $args->{profile} || 'none';
131 my $data = {%{$profiles{$profile}}, %{$args||{}}};
139 # tokenize string, and remove all optional whitespace
141 foreach my $token (split $tokenizer_re, $s) {
142 push @$tokens, $token if (length $token) && ($token =~ /\S/);
145 my $tree = $self->_recurse_parse($tokens, PARSE_TOP_LEVEL);
150 my ($self, $tokens, $state) = @_;
153 while (1) { # left-associative parsing
155 my $lookahead = $tokens->[0];
156 if ( not defined($lookahead)
158 ($state == PARSE_IN_PARENS && $lookahead eq ')')
160 ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords ) )
162 ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords, @binary_op_keywords, 'AND', 'OR', 'NOT' ) )
167 my $token = shift @$tokens;
169 # nested expression in ()
170 if ($token eq '(' ) {
171 my $right = $self->_recurse_parse($tokens, PARSE_IN_PARENS);
172 $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse($right);
173 $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse($right);
175 $left = $left ? [@$left, [PAREN => [$right] ]]
176 : [PAREN => [$right] ];
179 elsif ($token =~ /^ (?: OR | AND ) $/xi ) {
181 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
183 # Merge chunks if logic matches
184 if (ref $right and $op eq $right->[0]) {
185 $left = [ (shift @$right ), [$left, map { @$_ } @$right] ];
188 $left = [$op => [$left, $right]];
191 # binary operator keywords
192 elsif (grep { $token =~ /^ $_ $/xi } @binary_op_keywords ) {
194 my $right = $self->_recurse_parse($tokens, PARSE_RHS);
196 # A between with a simple LITERAL for a 1st RHS argument needs a
197 # rerun of the search to (hopefully) find the proper AND construct
198 if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') {
199 unshift @$tokens, $right->[1][0];
200 $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
203 $left = [$op => [$left, $right] ];
205 # expression terminator keywords (as they start a new expression)
206 elsif (grep { $token =~ /^ $_ $/xi } @expression_terminator_sql_keywords ) {
208 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
209 $left = $left ? [ $left, [$op => [$right] ]]
210 : [ $op => [$right] ];
212 # NOT (last as to allow all other NOT X pieces first)
213 elsif ( $token =~ /^ not $/ix ) {
215 my $right = $self->_recurse_parse ($tokens, PARSE_RHS);
216 $left = $left ? [ @$left, [$op => [$right] ]]
217 : [ $op => [$right] ];
220 # literal (eat everything on the right until RHS termination)
222 my $right = $self->_recurse_parse ($tokens, PARSE_RHS);
223 $left = $left ? [ $left, [LITERAL => [join ' ', $token, $self->unparse($right)||()] ] ]
224 : [ LITERAL => [join ' ', $token, $self->unparse($right)||()] ];
230 my ($self, $keyword) = @_;
232 if (my $around = $self->colormap->{lc $keyword}) {
233 $keyword = "$around->[0]$keyword$around->[1]";
240 my ($self, $keyword, $depth) = @_;
244 if (defined $self->indentmap->{lc $keyword}) {
245 $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword});
247 $before = '' if $depth == 0 and lc $keyword eq 'select';
248 return [$before, $after];
251 sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) }
255 $tree = $tree->[0] while ref $tree;
257 defined $tree && lc $tree eq 'select';
261 my ($self, $tree, $depth) = @_;
269 my $car = $tree->[0];
270 my $cdr = $tree->[1];
273 return join ('', map $self->unparse($_, $depth), @$tree);
275 elsif ($car eq 'LITERAL') {
278 elsif ($car eq 'PAREN') {
281 map $self->unparse($_, $depth + 2), @{$cdr}) .
282 (_is_select($cdr)?( $self->newline||'' ).$self->indent($depth + 1):'') . ')';
284 elsif ($car eq 'OR' or $car eq 'AND' or (grep { $car =~ /^ $_ $/xi } @binary_op_keywords ) ) {
285 return join (" $car ", map $self->unparse($_, $depth), @{$cdr});
288 my ($l, $r) = @{$self->whitespace($car, $depth)};
289 return sprintf "$l%s %s$r", $self->format_keyword($car), $self->unparse($cdr, $depth);
293 sub format { my $self = shift; $self->unparse($self->parse(@_)) }
301 my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' });
303 print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2');