1 package SQL::Abstract::Tree;
10 use base 'Class::Accessor::Grouped';
12 __PACKAGE__->mk_group_accessors( simple => $_ ) for qw(
13 newline indent_string indent_amount colormap indentmap fill_in_placeholders
17 my $merger = Hash::Merge->new;
19 $merger->specify_behavior({
21 SCALAR => sub { $_[1] },
22 ARRAY => sub { [ $_[0], @{$_[1]} ] },
23 HASH => sub { $_[1] },
26 SCALAR => sub { $_[1] },
27 ARRAY => sub { $_[1] },
28 HASH => sub { $_[1] },
31 SCALAR => sub { $_[1] },
32 ARRAY => sub { [ values %{$_[0]}, @{$_[1]} ] },
33 HASH => sub { Hash::Merge::_merge_hashes( $_[0], $_[1] ) },
35 }, 'SQLA::Tree Behavior' );
37 my $op_look_ahead = '(?: (?= [\s\)\(\;] ) | \z)';
38 my $op_look_behind = '(?: (?<= [\,\s\)\(] ) | \A )';
40 my $quote_left = qr/[\`\'\"\[]/;
41 my $quote_right = qr/[\`\'\"\]]/;
43 my $placeholder_re = qr/(?: \? | \$\d+ )/x;
45 # These SQL keywords always signal end of the current expression (except inside
46 # of a parenthesized subexpression).
47 # Format: A list of strings that will be compiled to extended syntax ie.
48 # /.../x) regexes, without capturing parentheses. They will be automatically
49 # anchored to op boundaries (excluding quotes) to match the whole token.
50 my @expression_start_keywords = (
59 (?: (?: LEFT | RIGHT | FULL ) \s+ )?
60 (?: (?: CROSS | INNER | OUTER ) \s+ )?
66 '(?: DEFAULT \s+ )? VALUES',
81 'ROLLBACK \s+ TO \s+ SAVEPOINT',
84 'RELEASE \s+ SAVEPOINT',
86 'ROW_NUMBER \s* \( \s* \) \s+ OVER',
89 my $expr_start_re = join ("\n\t|\n", @expression_start_keywords );
90 $expr_start_re = qr/ $op_look_behind (?i: $expr_start_re ) $op_look_ahead /x;
92 # These are binary operator keywords always a single LHS and RHS
93 # * AND/OR are handled separately as they are N-ary
94 # * so is NOT as being unary
95 # * BETWEEN without paranthesis around the ANDed arguments (which
96 # makes it a non-binary op) is detected and accomodated in
99 # this will be included in the $binary_op_re, the distinction is interesting during
100 # testing as one is tighter than the other, plus mathops have different look
101 # ahead/behind (e.g. "x"="y" )
102 my @math_op_keywords = (qw/ < > != <> = <= >= /);
103 my $math_re = join ("\n\t|\n", map
104 { "(?: (?<= [\\w\\s] | $quote_right ) | \\A )" . quotemeta ($_) . "(?: (?= [\\w\\s] | $quote_left ) | \\z )" }
107 $math_re = qr/$math_re/x;
109 sub _math_op_re { $math_re }
112 my $binary_op_re = '(?: NOT \s+)? (?:' . join ('|', qw/IN BETWEEN R?LIKE/) . ')';
113 $binary_op_re = join "\n\t|\n",
114 "$op_look_behind (?i: $binary_op_re ) $op_look_ahead",
116 $op_look_behind . 'IS (?:\s+ NOT)?' . "(?= \\s+ NULL \\b | $op_look_ahead )",
118 $binary_op_re = qr/$binary_op_re/x;
120 sub _binary_op_re { $binary_op_re }
122 my $all_known_re = join("\n\t|\n",
125 "$op_look_behind (?i: AND|OR|NOT ) $op_look_ahead",
126 (map { quotemeta $_ } qw/, ( ) */),
130 $all_known_re = qr/$all_known_re/x;
132 #this one *is* capturing for the split below
133 # splits on whitespace if all else fails
134 my $tokenizer_re = qr/ \s* ( $all_known_re ) \s* | \s+ /x;
136 # Parser states for _recurse_parse()
137 use constant PARSE_TOP_LEVEL => 0;
138 use constant PARSE_IN_EXPR => 1;
139 use constant PARSE_IN_PARENS => 2;
140 use constant PARSE_IN_FUNC => 3;
141 use constant PARSE_RHS => 4;
143 my $expr_term_re = qr/ ^ (?: $expr_start_re | \) ) $/x;
144 my $rhs_term_re = qr/ ^ (?: $expr_term_re | $binary_op_re | (?i: AND | OR | NOT | \, ) ) $/x;
145 my $func_start_re = qr/^ (?: \* | $placeholder_re | \( ) $/x;
170 fill_in_placeholders => 1,
171 placeholder_surround => ['?/', ''],
172 indent_string => ' ',
176 indentmap => { %indents },
178 eval { require Term::ANSIColor }
180 my $c = \&Term::ANSIColor::color;
182 placeholder_surround => [q(') . $c->('black on_magenta'), $c->('reset') . q(')],
184 'begin work' => [$c->('black on_white'), $c->('reset')],
185 commit => [$c->('black on_white'), $c->('reset')],
186 rollback => [$c->('black on_white'), $c->('reset')],
187 savepoint => [$c->('black on_white'), $c->('reset')],
188 'rollback to savepoint' => [$c->('black on_white'), $c->('reset')],
189 'release savepoint' => [$c->('black on_white'), $c->('reset')],
191 select => [$c->('red'), $c->('reset')],
192 'insert into' => [$c->('red'), $c->('reset')],
193 update => [$c->('red'), $c->('reset')],
194 'delete from' => [$c->('red'), $c->('reset')],
196 set => [$c->('cyan'), $c->('reset')],
197 from => [$c->('cyan'), $c->('reset')],
199 where => [$c->('green'), $c->('reset')],
200 values => [$c->('yellow'), $c->('reset')],
202 join => [$c->('magenta'), $c->('reset')],
203 'left join' => [$c->('magenta'), $c->('reset')],
204 on => [$c->('blue'), $c->('reset')],
206 'group by' => [$c->('yellow'), $c->('reset')],
207 'order by' => [$c->('yellow'), $c->('reset')],
209 skip => [$c->('green'), $c->('reset')],
210 first => [$c->('green'), $c->('reset')],
211 limit => [$c->('green'), $c->('reset')],
212 offset => [$c->('green'), $c->('reset')],
217 console_monochrome => {
218 fill_in_placeholders => 1,
219 placeholder_surround => ['?/', ''],
220 indent_string => ' ',
224 indentmap => { %indents },
227 fill_in_placeholders => 1,
228 placeholder_surround => ['<span class="placeholder">', '</span>'],
229 indent_string => ' ',
231 newline => "<br />\n",
233 select => ['<span class="select">' , '</span>'],
234 'insert into' => ['<span class="insert-into">' , '</span>'],
235 update => ['<span class="select">' , '</span>'],
236 'delete from' => ['<span class="delete-from">' , '</span>'],
238 set => ['<span class="set">', '</span>'],
239 from => ['<span class="from">' , '</span>'],
241 where => ['<span class="where">' , '</span>'],
242 values => ['<span class="values">', '</span>'],
244 join => ['<span class="join">' , '</span>'],
245 'left join' => ['<span class="left-join">','</span>'],
246 on => ['<span class="on">' , '</span>'],
248 'group by' => ['<span class="group-by">', '</span>'],
249 'order by' => ['<span class="order-by">', '</span>'],
251 skip => ['<span class="skip">', '</span>'],
252 first => ['<span class="first">', '</span>'],
253 limit => ['<span class="limit">', '</span>'],
254 offset => ['<span class="offset">', '</span>'],
256 'begin work' => ['<span class="begin-work">', '</span>'],
257 commit => ['<span class="commit">', '</span>'],
258 rollback => ['<span class="rollback">', '</span>'],
259 savepoint => ['<span class="savepoint">', '</span>'],
260 'rollback to savepoint' => ['<span class="rollback-to-savepoint">', '</span>'],
261 'release savepoint' => ['<span class="release-savepoint">', '</span>'],
263 indentmap => { %indents },
273 my $args = shift || {};
275 my $profile = delete $args->{profile} || 'none';
276 my $data = $merger->merge( $profiles{$profile}, $args );
284 # tokenize string, and remove all optional whitespace
286 foreach my $token (split $tokenizer_re, $s) {
287 push @$tokens, $token if (
295 $self->_recurse_parse($tokens, PARSE_TOP_LEVEL);
299 my ($self, $tokens, $state) = @_;
302 while (1) { # left-associative parsing
304 my $lookahead = $tokens->[0];
305 if ( not defined($lookahead)
307 ($state == PARSE_IN_PARENS && $lookahead eq ')')
309 ($state == PARSE_IN_EXPR && $lookahead =~ $expr_term_re )
311 ($state == PARSE_RHS && $lookahead =~ $rhs_term_re )
313 ($state == PARSE_IN_FUNC && $lookahead !~ $func_start_re) # if there are multiple values - the parenthesis will switch the $state
318 my $token = shift @$tokens;
320 # nested expression in ()
321 if ($token eq '(' ) {
322 my $right = $self->_recurse_parse($tokens, PARSE_IN_PARENS);
323 $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse($right);
324 $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse($right);
326 $left = $left ? [$left, [PAREN => [$right||()] ]]
327 : [PAREN => [$right||()] ];
329 # AND/OR and LIST (,)
330 elsif ($token =~ /^ (?: OR | AND | \, ) $/xi ) {
331 my $op = ($token eq ',') ? 'LIST' : uc $token;
333 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
335 # Merge chunks if logic matches
336 if (ref $right and $op eq $right->[0]) {
337 $left = [ (shift @$right ), [$left||(), map { @$_ } @$right] ];
340 $left = [$op => [ $left||(), $right||() ]];
343 # binary operator keywords
344 elsif ( $token =~ /^ $binary_op_re $ /x ) {
346 my $right = $self->_recurse_parse($tokens, PARSE_RHS);
348 # A between with a simple LITERAL for a 1st RHS argument needs a
349 # rerun of the search to (hopefully) find the proper AND construct
350 if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') {
351 unshift @$tokens, $right->[1][0];
352 $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
355 $left = [$op => [$left, $right] ];
357 # expression terminator keywords (as they start a new expression)
358 elsif ( $token =~ / ^ $expr_start_re $ /x ) {
360 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
361 $left = $left ? [ $left, [$op => [$right||()] ]]
362 : [ $op => [$right||()] ];
365 elsif ( $token =~ /^ NOT $/ix ) {
367 my $right = $self->_recurse_parse ($tokens, PARSE_RHS);
368 $left = $left ? [ @$left, [$op => [$right] ]]
369 : [ $op => [$right] ];
372 elsif ( $token =~ $placeholder_re) {
373 $left = $left ? [ $left, [ PLACEHOLDER => [ $token ] ] ]
374 : [ PLACEHOLDER => [ $token ] ];
376 # we're now in "unknown token" land - start eating tokens until
377 # we see something familiar
381 # check if the current token is an unknown op-start
382 if (@$tokens and $tokens->[0] =~ $func_start_re) {
383 $right = [ $token => [ $self->_recurse_parse($tokens, PARSE_IN_FUNC) || () ] ];
386 $right = [ LITERAL => [ $token ] ];
389 $left = $left ? [ $left, $right ]
396 my ($self, $keyword) = @_;
398 if (my $around = $self->colormap->{lc $keyword}) {
399 $keyword = "$around->[0]$keyword$around->[1]";
413 my ($self, $keyword, $depth) = @_;
416 if (defined $self->indentmap->{lc $keyword}) {
417 $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword});
419 $before = '' if $depth == 0 and defined $starters{lc $keyword};
420 return [$before, ''];
423 sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) }
426 my ($self, $tree) = @_;
427 $tree = $tree->[0] while ref $tree;
429 defined $tree && defined $self->indentmap->{lc $tree};
432 sub fill_in_placeholder {
433 my ($self, $bindargs) = @_;
435 if ($self->fill_in_placeholders) {
436 my $val = shift @{$bindargs} || '';
437 my ($left, $right) = @{$self->placeholder_surround};
440 return qq($left$val$right)
445 # FIXME - terrible name for a user facing API
447 my ($self, $tree, $bindargs) = @_;
448 $self->_unparse($tree, [@{$bindargs||[]}], 0);
452 my ($self, $tree, $bindargs, $depth) = @_;
454 if (not $tree or not @$tree) {
458 my ($car, $cdr) = @{$tree}[0,1];
460 if (! defined $car or (! ref $car and ! defined $cdr) ) {
461 require Data::Dumper;
462 Carp::confess( sprintf ( "Internal error - malformed branch at depth $depth:\n%s",
463 Data::Dumper::Dumper($tree)
468 return join (' ', map $self->_unparse($_, $bindargs, $depth), @$tree);
470 elsif ($car eq 'LITERAL') {
473 elsif ($car eq 'PLACEHOLDER') {
474 return $self->fill_in_placeholder($bindargs);
476 elsif ($car eq 'PAREN') {
477 return sprintf ('(%s)',
478 join (' ', map { $self->_unparse($_, $bindargs, $depth + 2) } @{$cdr} )
480 ($self->_is_key($cdr)
481 ? ( $self->newline||'' ) . $self->indent($depth + 1)
486 elsif ($car eq 'AND' or $car eq 'OR' or $car =~ / ^ $binary_op_re $ /x ) {
487 return join (" $car ", map $self->_unparse($_, $bindargs, $depth), @{$cdr});
489 elsif ($car eq 'LIST' ) {
490 return join (', ', map $self->_unparse($_, $bindargs, $depth), @{$cdr});
493 my ($l, $r) = @{$self->pad_keyword($car, $depth)};
494 return sprintf "$l%s %s$r", $self->format_keyword($car), $self->_unparse($cdr, $bindargs, $depth);
498 sub format { my $self = shift; $self->unparse($self->parse($_[0]), $_[1]) }
506 my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' });
508 print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2');
518 my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' });
521 profile => 'console', # predefined profile to use (default: 'none')
522 fill_in_placeholders => 1, # true for placeholder population
523 placeholder_surround => # The strings that will be wrapped around
524 [GREEN, RESET], # populated placeholders if the above is set
525 indent_string => ' ', # the string used when indenting
526 indent_amount => 2, # how many of above string to use for a single
528 newline => "\n", # string for newline
530 select => [RED, RESET], # a pair of strings defining what to surround
531 # the keyword with for colorization
535 select => 0, # A zero means that the keyword will start on
537 from => 1, # Any other positive integer means that after
538 on => 2, # said newline it will get that many indents
543 Returns a new SQL::Abstract::Tree object. All arguments are optional.
547 There are four predefined profiles, C<none>, C<console>, C<console_monochrome>,
548 and C<html>. Typically a user will probably just use C<console> or
549 C<console_monochrome>, but if something about a profile bothers you, merely
550 use the profile and override the parts that you don't like.
554 $sqlat->format('SELECT * FROM bar WHERE x = ?', [1])
556 Takes C<$sql> and C<\@bindargs>.
558 Returns a formatting string based on the string passed in
562 $sqlat->parse('SELECT * FROM bar WHERE x = ?')
564 Returns a "tree" representing passed in SQL. Please do not depend on the
565 structure of the returned tree. It may be stable at some point, but not yet.
569 $sqlat->parse($tree_structure, \@bindargs)
571 Transform "tree" into SQL, applying various transforms on the way.
573 =head2 format_keyword
575 $sqlat->format_keyword('SELECT')
577 Currently this just takes a keyword and puts the C<colormap> stuff around it.
578 Later on it may do more and allow for coderef based transforms.
582 my ($before, $after) = @{$sqlat->pad_keyword('SELECT')};
584 Returns whitespace to be inserted around a keyword.
586 =head2 fill_in_placeholder
588 my $value = $sqlat->fill_in_placeholder(\@bindargs)
590 Removes last arg from passed arrayref and returns it, surrounded with
591 the values in placeholder_surround, and then surrounded with single quotes.
595 Returns as many indent strings as indent amounts times the first argument.
603 =head2 fill_in_placeholders
623 =head2 placeholder_surround