Commit | Line | Data |
01dd4e4f |
1 | package SQL::Abstract::Tree; |
2 | |
e9f56d7a |
3 | use Moo; |
b3b79607 |
4 | no warnings 'qw'; |
01dd4e4f |
5 | |
c54740ba |
6 | use Carp; |
e9f56d7a |
7 | use Sub::Quote 'quote_sub'; |
1536de15 |
8 | |
0769ac0e |
9 | my $op_look_ahead = '(?: (?= [\s\)\(\;] ) | \z)'; |
b3b79607 |
10 | my $op_look_behind = '(?: (?<= [\,\s\)\(] ) | \A )'; |
11 | |
0769ac0e |
12 | my $quote_left = qr/[\`\'\"\[]/; |
13 | my $quote_right = qr/[\`\'\"\]]/; |
01dd4e4f |
14 | |
4e914a7c |
15 | my $placeholder_re = qr/(?: \? | \$\d+ )/x; |
16 | |
01dd4e4f |
17 | # These SQL keywords always signal end of the current expression (except inside |
18 | # of a parenthesized subexpression). |
0769ac0e |
19 | # Format: A list of strings that will be compiled to extended syntax ie. |
01dd4e4f |
20 | # /.../x) regexes, without capturing parentheses. They will be automatically |
0769ac0e |
21 | # anchored to op boundaries (excluding quotes) to match the whole token. |
22 | my @expression_start_keywords = ( |
01dd4e4f |
23 | 'SELECT', |
7853a177 |
24 | 'UPDATE', |
6c4d8eb8 |
25 | 'SET', |
7853a177 |
26 | 'INSERT \s+ INTO', |
27 | 'DELETE \s+ FROM', |
3d910890 |
28 | 'FROM', |
01dd4e4f |
29 | '(?: |
30 | (?: |
0769ac0e |
31 | (?: (?: LEFT | RIGHT | FULL ) \s+ )? |
32 | (?: (?: CROSS | INNER | OUTER ) \s+ )? |
01dd4e4f |
33 | )? |
34 | JOIN |
35 | )', |
36 | 'ON', |
37 | 'WHERE', |
efc991a0 |
38 | '(?: DEFAULT \s+ )? VALUES', |
01dd4e4f |
39 | 'GROUP \s+ BY', |
40 | 'HAVING', |
41 | 'ORDER \s+ BY', |
c0eaa9fd |
42 | 'SKIP', |
81b3e585 |
43 | 'FETCH', |
c0eaa9fd |
44 | 'FIRST', |
01dd4e4f |
45 | 'LIMIT', |
46 | 'OFFSET', |
47 | 'FOR', |
48 | 'UNION', |
49 | 'INTERSECT', |
50 | 'EXCEPT', |
820bb1f5 |
51 | 'BEGIN \s+ WORK', |
52 | 'COMMIT', |
53 | 'ROLLBACK \s+ TO \s+ SAVEPOINT', |
54 | 'ROLLBACK', |
55 | 'SAVEPOINT', |
56 | 'RELEASE \s+ SAVEPOINT', |
01dd4e4f |
57 | 'RETURNING', |
58 | ); |
59 | |
b3b79607 |
60 | my $expr_start_re = join ("\n\t|\n", @expression_start_keywords ); |
61 | $expr_start_re = qr/ $op_look_behind (?i: $expr_start_re ) $op_look_ahead /x; |
0769ac0e |
62 | |
01dd4e4f |
63 | # These are binary operator keywords always a single LHS and RHS |
64 | # * AND/OR are handled separately as they are N-ary |
65 | # * so is NOT as being unary |
3af02ccb |
66 | # * BETWEEN without parentheses around the ANDed arguments (which |
67 | # makes it a non-binary op) is detected and accommodated in |
01dd4e4f |
68 | # _recurse_parse() |
6c4d8eb8 |
69 | # * AS is not really an operator but is handled here as it's also LHS/RHS |
01dd4e4f |
70 | |
0769ac0e |
71 | # this will be included in the $binary_op_re, the distinction is interesting during |
1ec9b9e3 |
72 | # testing as one is tighter than the other, plus alphanum cmp ops have different |
73 | # look ahead/behind (e.g. "x"="y" ) |
74 | my @alphanum_cmp_op_keywords = (qw/< > != <> = <= >= /); |
75 | my $alphanum_cmp_op_re = join ("\n\t|\n", map |
0769ac0e |
76 | { "(?: (?<= [\\w\\s] | $quote_right ) | \\A )" . quotemeta ($_) . "(?: (?= [\\w\\s] | $quote_left ) | \\z )" } |
1ec9b9e3 |
77 | @alphanum_cmp_op_keywords |
01dd4e4f |
78 | ); |
1ec9b9e3 |
79 | $alphanum_cmp_op_re = qr/$alphanum_cmp_op_re/x; |
0769ac0e |
80 | |
bf782d2e |
81 | my $binary_op_re = '(?: NOT \s+)? (?:' . join ('|', qw/IN BETWEEN [RI]?LIKE REGEXP/) . ')'; |
b3b79607 |
82 | $binary_op_re = join "\n\t|\n", |
6c4d8eb8 |
83 | "$op_look_behind (?i: $binary_op_re | AS ) $op_look_ahead", |
1ec9b9e3 |
84 | $alphanum_cmp_op_re, |
b3b79607 |
85 | $op_look_behind . 'IS (?:\s+ NOT)?' . "(?= \\s+ NULL \\b | $op_look_ahead )", |
86 | ; |
b7b0f832 |
87 | $binary_op_re = qr/$binary_op_re/x; |
0769ac0e |
88 | |
4247c384 |
89 | my $rno_re = qr/ROW_NUMBER \s* \( \s* \) \s+ OVER/ix; |
90 | |
91 | my $unary_op_re = 'NOT \s+ EXISTS | NOT | ' . $rno_re; |
6f2a5b66 |
92 | $unary_op_re = join "\n\t|\n", |
93 | "$op_look_behind (?i: $unary_op_re ) $op_look_ahead", |
94 | ; |
95 | $unary_op_re = qr/$unary_op_re/x; |
96 | |
ad591616 |
97 | my $asc_desc_re = qr/$op_look_behind (?i: ASC | DESC ) $op_look_ahead /x; |
98 | my $and_or_re = qr/$op_look_behind (?i: AND | OR ) $op_look_ahead /x; |
6f2a5b66 |
99 | |
ad591616 |
100 | my $tokenizer_re = join("\n\t|\n", |
b3b79607 |
101 | $expr_start_re, |
0769ac0e |
102 | $binary_op_re, |
6f2a5b66 |
103 | $unary_op_re, |
ad591616 |
104 | $asc_desc_re, |
105 | $and_or_re, |
1ec9b9e3 |
106 | $op_look_behind . ' \* ' . $op_look_ahead, |
257ecc8a |
107 | (map { quotemeta $_ } qw/, ( )/), |
4e914a7c |
108 | $placeholder_re, |
0769ac0e |
109 | ); |
01dd4e4f |
110 | |
ad591616 |
111 | # this one *is* capturing for the split below |
b3b79607 |
112 | # splits on whitespace if all else fails |
3af02ccb |
113 | # has to happen before the composing qr's are anchored (below) |
ad591616 |
114 | $tokenizer_re = qr/ \s* ( $tokenizer_re ) \s* | \s+ /x; |
b3b79607 |
115 | |
116 | # Parser states for _recurse_parse() |
117 | use constant PARSE_TOP_LEVEL => 0; |
118 | use constant PARSE_IN_EXPR => 1; |
119 | use constant PARSE_IN_PARENS => 2; |
120 | use constant PARSE_IN_FUNC => 3; |
121 | use constant PARSE_RHS => 4; |
6f2a5b66 |
122 | use constant PARSE_LIST_ELT => 5; |
b3b79607 |
123 | |
ad591616 |
124 | my $expr_term_re = qr/$expr_start_re | \)/x; |
125 | my $rhs_term_re = qr/ $expr_term_re | $binary_op_re | $unary_op_re | $asc_desc_re | $and_or_re | \, /x; |
1ec9b9e3 |
126 | my $all_std_keywords_re = qr/ $rhs_term_re | \( | $placeholder_re /x; |
ad591616 |
127 | |
128 | # anchor everything - even though keywords are separated by the tokenizer, leakage may occur |
129 | for ( |
130 | $quote_left, |
131 | $quote_right, |
132 | $placeholder_re, |
133 | $expr_start_re, |
1ec9b9e3 |
134 | $alphanum_cmp_op_re, |
ad591616 |
135 | $binary_op_re, |
136 | $unary_op_re, |
137 | $asc_desc_re, |
138 | $and_or_re, |
139 | $expr_term_re, |
140 | $rhs_term_re, |
ad591616 |
141 | $all_std_keywords_re, |
142 | ) { |
143 | $_ = qr/ \A $_ \z /x; |
144 | } |
145 | |
b4085a1a |
146 | # what can be bunched together under one MISC in an AST |
147 | my $compressable_node_re = qr/^ \- (?: MISC | LITERAL | PLACEHOLDER ) $/x; |
01dd4e4f |
148 | |
7e5600e9 |
149 | my %indents = ( |
7853a177 |
150 | select => 0, |
151 | update => 0, |
152 | 'insert into' => 0, |
153 | 'delete from' => 0, |
3d910890 |
154 | from => 1, |
91916220 |
155 | where => 0, |
7853a177 |
156 | join => 1, |
157 | 'left join' => 1, |
158 | on => 2, |
2867f4f5 |
159 | having => 0, |
91916220 |
160 | 'group by' => 0, |
161 | 'order by' => 0, |
7853a177 |
162 | set => 1, |
163 | into => 1, |
91916220 |
164 | values => 1, |
c0eaa9fd |
165 | limit => 1, |
166 | offset => 1, |
167 | skip => 1, |
168 | first => 1, |
7e5600e9 |
169 | ); |
170 | |
c54740ba |
171 | |
172 | has [qw( |
173 | newline indent_string indent_amount fill_in_placeholders placeholder_surround |
174 | )] => (is => 'ro'); |
175 | |
176 | has [qw( indentmap colormap )] => ( is => 'ro', default => quote_sub('{}') ); |
177 | |
178 | # class global is in fact desired |
179 | my $merger; |
180 | |
181 | sub BUILDARGS { |
182 | my $class = shift; |
183 | my $args = ref $_[0] eq 'HASH' ? $_[0] : {@_}; |
184 | |
185 | if (my $p = delete $args->{profile}) { |
186 | my %extra_args; |
187 | if ($p eq 'console') { |
188 | %extra_args = ( |
189 | fill_in_placeholders => 1, |
190 | placeholder_surround => ['?/', ''], |
191 | indent_string => ' ', |
192 | indent_amount => 2, |
193 | newline => "\n", |
194 | colormap => {}, |
195 | indentmap => \%indents, |
196 | |
197 | ! ( eval { require Term::ANSIColor } ) ? () : do { |
aafbf833 |
198 | my $c = \&Term::ANSIColor::color; |
6d388c84 |
199 | |
200 | my $red = [$c->('red') , $c->('reset')]; |
201 | my $cyan = [$c->('cyan') , $c->('reset')]; |
202 | my $green = [$c->('green') , $c->('reset')]; |
203 | my $yellow = [$c->('yellow') , $c->('reset')]; |
204 | my $blue = [$c->('blue') , $c->('reset')]; |
205 | my $magenta = [$c->('magenta'), $c->('reset')]; |
206 | my $b_o_w = [$c->('black on_white'), $c->('reset')]; |
aafbf833 |
207 | ( |
fb98df48 |
208 | placeholder_surround => [$c->('black on_magenta'), $c->('reset')], |
aafbf833 |
209 | colormap => { |
6d388c84 |
210 | 'begin work' => $b_o_w, |
211 | commit => $b_o_w, |
212 | rollback => $b_o_w, |
213 | savepoint => $b_o_w, |
214 | 'rollback to savepoint' => $b_o_w, |
215 | 'release savepoint' => $b_o_w, |
216 | |
217 | select => $red, |
218 | 'insert into' => $red, |
219 | update => $red, |
220 | 'delete from' => $red, |
221 | |
222 | set => $cyan, |
223 | from => $cyan, |
224 | |
225 | where => $green, |
226 | values => $yellow, |
227 | |
228 | join => $magenta, |
229 | 'left join' => $magenta, |
230 | on => $blue, |
231 | |
232 | 'group by' => $yellow, |
2867f4f5 |
233 | having => $yellow, |
6d388c84 |
234 | 'order by' => $yellow, |
235 | |
236 | skip => $green, |
237 | first => $green, |
238 | limit => $green, |
239 | offset => $green, |
aafbf833 |
240 | } |
241 | ); |
c54740ba |
242 | }, |
243 | ); |
244 | } |
245 | elsif ($p eq 'console_monochrome') { |
246 | %extra_args = ( |
247 | fill_in_placeholders => 1, |
248 | placeholder_surround => ['?/', ''], |
249 | indent_string => ' ', |
250 | indent_amount => 2, |
251 | newline => "\n", |
252 | indentmap => \%indents, |
253 | ); |
254 | } |
255 | elsif ($p eq 'html') { |
256 | %extra_args = ( |
257 | fill_in_placeholders => 1, |
258 | placeholder_surround => ['<span class="placeholder">', '</span>'], |
259 | indent_string => ' ', |
260 | indent_amount => 2, |
261 | newline => "<br />\n", |
262 | colormap => { map { |
263 | (my $class = $_) =~ s/\s+/-/g; |
264 | ( $_ => [ qq|<span class="$class">|, '</span>' ] ) |
265 | } ( |
266 | keys %indents, |
267 | qw(commit rollback savepoint), |
268 | 'begin work', 'rollback to savepoint', 'release savepoint', |
269 | ) }, |
270 | indentmap => \%indents, |
271 | ); |
272 | } |
273 | elsif ($p eq 'none') { |
274 | # nada |
275 | } |
276 | else { |
277 | croak "No such profile '$p'"; |
278 | } |
279 | |
280 | # see if we got any duplicates and merge if needed |
281 | if (scalar grep { exists $args->{$_} } keys %extra_args) { |
282 | # heavy-duty merge |
283 | $args = ($merger ||= do { |
284 | require Hash::Merge; |
285 | my $m = Hash::Merge->new; |
286 | |
287 | $m->specify_behavior({ |
288 | SCALAR => { |
289 | SCALAR => sub { $_[1] }, |
290 | ARRAY => sub { [ $_[0], @{$_[1]} ] }, |
291 | HASH => sub { $_[1] }, |
292 | }, |
293 | ARRAY => { |
294 | SCALAR => sub { $_[1] }, |
295 | ARRAY => sub { $_[1] }, |
296 | HASH => sub { $_[1] }, |
297 | }, |
298 | HASH => { |
299 | SCALAR => sub { $_[1] }, |
300 | ARRAY => sub { [ values %{$_[0]}, @{$_[1]} ] }, |
301 | HASH => sub { Hash::Merge::_merge_hashes( $_[0], $_[1] ) }, |
302 | }, |
303 | }, 'SQLA::Tree Behavior' ); |
304 | |
305 | $m; |
306 | })->merge(\%extra_args, $args ); |
307 | |
308 | } |
309 | else { |
310 | $args = { %extra_args, %$args }; |
311 | } |
312 | } |
313 | |
314 | $args; |
75c3a063 |
315 | } |
d695b0ad |
316 | |
01dd4e4f |
317 | sub parse { |
d695b0ad |
318 | my ($self, $s) = @_; |
01dd4e4f |
319 | |
70c6f0e9 |
320 | return [] unless defined $s; |
321 | |
01dd4e4f |
322 | # tokenize string, and remove all optional whitespace |
323 | my $tokens = []; |
324 | foreach my $token (split $tokenizer_re, $s) { |
b3b79607 |
325 | push @$tokens, $token if ( |
326 | defined $token |
327 | and |
328 | length $token |
09931431 |
329 | and |
b3b79607 |
330 | $token =~ /\S/ |
331 | ); |
01dd4e4f |
332 | } |
6f2a5b66 |
333 | |
334 | return [ $self->_recurse_parse($tokens, PARSE_TOP_LEVEL) ]; |
01dd4e4f |
335 | } |
336 | |
337 | sub _recurse_parse { |
d695b0ad |
338 | my ($self, $tokens, $state) = @_; |
01dd4e4f |
339 | |
6f2a5b66 |
340 | my @left; |
01dd4e4f |
341 | while (1) { # left-associative parsing |
342 | |
ca4f826a |
343 | if (! @$tokens |
01dd4e4f |
344 | or |
6f2a5b66 |
345 | ($state == PARSE_IN_PARENS && $tokens->[0] eq ')') |
01dd4e4f |
346 | or |
6f2a5b66 |
347 | ($state == PARSE_IN_EXPR && $tokens->[0] =~ $expr_term_re ) |
0769ac0e |
348 | or |
6f2a5b66 |
349 | ($state == PARSE_RHS && $tokens->[0] =~ $rhs_term_re ) |
01dd4e4f |
350 | or |
ad591616 |
351 | ($state == PARSE_LIST_ELT && ( $tokens->[0] eq ',' or $tokens->[0] =~ $expr_term_re ) ) |
01dd4e4f |
352 | ) { |
6f2a5b66 |
353 | return @left; |
01dd4e4f |
354 | } |
355 | |
356 | my $token = shift @$tokens; |
357 | |
358 | # nested expression in () |
359 | if ($token eq '(' ) { |
6f2a5b66 |
360 | my @right = $self->_recurse_parse($tokens, PARSE_IN_PARENS); |
361 | $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse(\@right); |
362 | $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse(\@right); |
363 | |
364 | push @left, [ '-PAREN' => \@right ]; |
365 | } |
366 | |
367 | # AND/OR |
ad591616 |
368 | elsif ($token =~ $and_or_re) { |
6f2a5b66 |
369 | my $op = uc $token; |
01dd4e4f |
370 | |
6f2a5b66 |
371 | my @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
372 | |
373 | # Merge chunks if "logic" matches |
374 | @left = [ $op => [ @left, (@right and $op eq $right[0][0]) |
375 | ? @{ $right[0][1] } |
376 | : @right |
377 | ] ]; |
01dd4e4f |
378 | } |
b3b79607 |
379 | |
6f2a5b66 |
380 | # LIST (,) |
381 | elsif ($token eq ',') { |
382 | |
383 | my @right = $self->_recurse_parse($tokens, PARSE_LIST_ELT); |
384 | |
385 | # deal with malformed lists ( foo, bar, , baz ) |
386 | @right = [] unless @right; |
01dd4e4f |
387 | |
6f2a5b66 |
388 | @right = [ -MISC => [ @right ] ] if @right > 1; |
389 | |
390 | if (!@left) { |
391 | @left = [ -LIST => [ [], @right ] ]; |
392 | } |
393 | elsif ($left[0][0] eq '-LIST') { |
394 | push @{$left[0][1]}, (@{$right[0]} and $right[0][0] eq '-LIST') |
395 | ? @{$right[0][1]} |
396 | : @right |
397 | ; |
01dd4e4f |
398 | } |
399 | else { |
6f2a5b66 |
400 | @left = [ -LIST => [ @left, @right ] ]; |
01dd4e4f |
401 | } |
402 | } |
6f2a5b66 |
403 | |
01dd4e4f |
404 | # binary operator keywords |
6f2a5b66 |
405 | elsif ($token =~ $binary_op_re) { |
01dd4e4f |
406 | my $op = uc $token; |
6f2a5b66 |
407 | |
408 | my @right = $self->_recurse_parse($tokens, PARSE_RHS); |
01dd4e4f |
409 | |
410 | # A between with a simple LITERAL for a 1st RHS argument needs a |
411 | # rerun of the search to (hopefully) find the proper AND construct |
6f2a5b66 |
412 | if ($op eq 'BETWEEN' and $right[0] eq '-LITERAL') { |
413 | unshift @$tokens, $right[1][0]; |
414 | @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
415 | } |
416 | |
08e16360 |
417 | push @left, [$op => [ (@left ? pop @left : ''), @right ]]; |
01dd4e4f |
418 | } |
6f2a5b66 |
419 | |
420 | # unary op keywords |
ca4f826a |
421 | elsif ($token =~ $unary_op_re) { |
01dd4e4f |
422 | my $op = uc $token; |
4247c384 |
423 | |
424 | # normalize RNO explicitly |
425 | $op = 'ROW_NUMBER() OVER' if $op =~ /^$rno_re$/; |
426 | |
ca4f826a |
427 | my @right = $self->_recurse_parse($tokens, PARSE_RHS); |
6f2a5b66 |
428 | |
429 | push @left, [ $op => \@right ]; |
01dd4e4f |
430 | } |
6f2a5b66 |
431 | |
432 | # expression terminator keywords |
ca4f826a |
433 | elsif ($token =~ $expr_start_re) { |
01dd4e4f |
434 | my $op = uc $token; |
6f2a5b66 |
435 | my @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
436 | |
6f2a5b66 |
437 | push @left, [ $op => \@right ]; |
01dd4e4f |
438 | } |
6f2a5b66 |
439 | |
440 | # a '?' |
ca4f826a |
441 | elsif ($token =~ $placeholder_re) { |
6f2a5b66 |
442 | push @left, [ -PLACEHOLDER => [ $token ] ]; |
443 | } |
444 | |
445 | # check if the current token is an unknown op-start |
1ec9b9e3 |
446 | elsif (@$tokens and ($tokens->[0] eq '(' or $tokens->[0] =~ $placeholder_re ) ) { |
6f2a5b66 |
447 | push @left, [ $token => [ $self->_recurse_parse($tokens, PARSE_RHS) ] ]; |
4e914a7c |
448 | } |
6f2a5b66 |
449 | |
b3b79607 |
450 | # we're now in "unknown token" land - start eating tokens until |
6f30911f |
451 | # we see something familiar, OR in the case of RHS (binop) stop |
452 | # after the first token |
e2a120ce |
453 | # Also stop processing when we could end up with an unknown func |
01dd4e4f |
454 | else { |
6f2a5b66 |
455 | my @lits = [ -LITERAL => [$token] ]; |
b3b79607 |
456 | |
b4085a1a |
457 | unshift @lits, pop @left if @left == 1; |
458 | |
6f30911f |
459 | unless ( $state == PARSE_RHS ) { |
e2a120ce |
460 | while ( |
461 | @$tokens |
462 | and |
463 | $tokens->[0] !~ $all_std_keywords_re |
464 | and |
ca4f826a |
465 | ! (@$tokens > 1 and $tokens->[1] eq '(') |
e2a120ce |
466 | ) { |
6f30911f |
467 | push @lits, [ -LITERAL => [ shift @$tokens ] ]; |
b4085a1a |
468 | } |
6f30911f |
469 | } |
6f2a5b66 |
470 | |
6f2a5b66 |
471 | @lits = [ -MISC => [ @lits ] ] if @lits > 1; |
472 | |
473 | push @left, @lits; |
474 | } |
b3b79607 |
475 | |
b4085a1a |
476 | # compress -LITERAL -MISC and -PLACEHOLDER pieces into a single |
477 | # -MISC container |
478 | if (@left > 1) { |
479 | my $i = 0; |
480 | while ($#left > $i) { |
481 | if ($left[$i][0] =~ $compressable_node_re and $left[$i+1][0] =~ $compressable_node_re) { |
482 | splice @left, $i, 2, [ -MISC => [ |
483 | map { $_->[0] eq '-MISC' ? @{$_->[1]} : $_ } (@left[$i, $i+1]) |
484 | ]]; |
485 | } |
486 | else { |
487 | $i++; |
488 | } |
489 | } |
490 | } |
491 | |
492 | return @left if $state == PARSE_RHS; |
6f30911f |
493 | |
b4085a1a |
494 | # deal with post-fix operators |
495 | if (@$tokens) { |
496 | # asc/desc |
6f2a5b66 |
497 | if ($tokens->[0] =~ $asc_desc_re) { |
1ec9b9e3 |
498 | @left = [ ('-' . uc (shift @$tokens)) => [ @left ] ]; |
6f2a5b66 |
499 | } |
01dd4e4f |
500 | } |
501 | } |
502 | } |
503 | |
d695b0ad |
504 | sub format_keyword { |
505 | my ($self, $keyword) = @_; |
506 | |
1536de15 |
507 | if (my $around = $self->colormap->{lc $keyword}) { |
d695b0ad |
508 | $keyword = "$around->[0]$keyword$around->[1]"; |
509 | } |
510 | |
511 | return $keyword |
512 | } |
513 | |
728f26a2 |
514 | my %starters = ( |
515 | select => 1, |
516 | update => 1, |
517 | 'insert into' => 1, |
518 | 'delete from' => 1, |
519 | ); |
520 | |
f2ab166a |
521 | sub pad_keyword { |
a24cc3a0 |
522 | my ($self, $keyword, $depth) = @_; |
e171c446 |
523 | |
524 | my $before = ''; |
1536de15 |
525 | if (defined $self->indentmap->{lc $keyword}) { |
526 | $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword}); |
a24cc3a0 |
527 | } |
728f26a2 |
528 | $before = '' if $depth == 0 and defined $starters{lc $keyword}; |
e4570c8e |
529 | return [$before, '']; |
a24cc3a0 |
530 | } |
531 | |
1536de15 |
532 | sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) } |
a24cc3a0 |
533 | |
a97eb57c |
534 | sub _is_key { |
535 | my ($self, $tree) = @_; |
0569a14f |
536 | $tree = $tree->[0] while ref $tree; |
537 | |
a97eb57c |
538 | defined $tree && defined $self->indentmap->{lc $tree}; |
0569a14f |
539 | } |
540 | |
9d11f0d4 |
541 | sub fill_in_placeholder { |
fb272e73 |
542 | my ($self, $bindargs) = @_; |
543 | |
544 | if ($self->fill_in_placeholders) { |
ad46269d |
545 | my $val = shift @{$bindargs} || ''; |
4712657d |
546 | my $quoted = $val =~ s/^(['"])(.*)\1$/$2/; |
9d11f0d4 |
547 | my ($left, $right) = @{$self->placeholder_surround}; |
fb272e73 |
548 | $val =~ s/\\/\\\\/g; |
549 | $val =~ s/'/\\'/g; |
4712657d |
550 | $val = qq('$val') if $quoted; |
551 | return qq($left$val$right) |
fb272e73 |
552 | } |
553 | return '?' |
554 | } |
555 | |
3a247d23 |
556 | # FIXME - terrible name for a user facing API |
01dd4e4f |
557 | sub unparse { |
3a247d23 |
558 | my ($self, $tree, $bindargs) = @_; |
559 | $self->_unparse($tree, [@{$bindargs||[]}], 0); |
560 | } |
a24cc3a0 |
561 | |
3a247d23 |
562 | sub _unparse { |
563 | my ($self, $tree, $bindargs, $depth) = @_; |
01dd4e4f |
564 | |
0769ac0e |
565 | if (not $tree or not @$tree) { |
01dd4e4f |
566 | return ''; |
567 | } |
a24cc3a0 |
568 | |
007f0853 |
569 | # FIXME - needs a config switch to disable |
c01ac648 |
570 | $self->_parenthesis_unroll($tree); |
007f0853 |
571 | |
6f2a5b66 |
572 | my ($op, $args) = @{$tree}[0,1]; |
0769ac0e |
573 | |
6f2a5b66 |
574 | if (! defined $op or (! ref $op and ! defined $args) ) { |
0769ac0e |
575 | require Data::Dumper; |
576 | Carp::confess( sprintf ( "Internal error - malformed branch at depth $depth:\n%s", |
577 | Data::Dumper::Dumper($tree) |
578 | ) ); |
579 | } |
a24cc3a0 |
580 | |
6f2a5b66 |
581 | if (ref $op) { |
3a247d23 |
582 | return join (' ', map $self->_unparse($_, $bindargs, $depth), @$tree); |
01dd4e4f |
583 | } |
6f2a5b66 |
584 | elsif ($op eq '-LITERAL') { # literal has different sig |
585 | return $args->[0]; |
01dd4e4f |
586 | } |
6f2a5b66 |
587 | elsif ($op eq '-PLACEHOLDER') { |
4e914a7c |
588 | return $self->fill_in_placeholder($bindargs); |
589 | } |
6f2a5b66 |
590 | elsif ($op eq '-PAREN') { |
c4d7cfcf |
591 | return sprintf ('( %s )', |
6f2a5b66 |
592 | join (' ', map { $self->_unparse($_, $bindargs, $depth + 2) } @{$args} ) |
e4570c8e |
593 | . |
6f2a5b66 |
594 | ($self->_is_key($args) |
e4570c8e |
595 | ? ( $self->newline||'' ) . $self->indent($depth + 1) |
596 | : '' |
597 | ) |
598 | ); |
01dd4e4f |
599 | } |
ad591616 |
600 | elsif ($op eq 'AND' or $op eq 'OR' or $op =~ $binary_op_re ) { |
6f2a5b66 |
601 | return join (" $op ", map $self->_unparse($_, $bindargs, $depth), @{$args}); |
602 | } |
603 | elsif ($op eq '-LIST' ) { |
604 | return join (', ', map $self->_unparse($_, $bindargs, $depth), @{$args}); |
01dd4e4f |
605 | } |
6f2a5b66 |
606 | elsif ($op eq '-MISC' ) { |
607 | return join (' ', map $self->_unparse($_, $bindargs, $depth), @{$args}); |
b3b79607 |
608 | } |
73835ff0 |
609 | elsif ($op =~ qr/^-(ASC|DESC)$/ ) { |
610 | my $dir = $1; |
611 | return join (' ', (map $self->_unparse($_, $bindargs, $depth), @{$args}), $dir); |
612 | } |
01dd4e4f |
613 | else { |
6f2a5b66 |
614 | my ($l, $r) = @{$self->pad_keyword($op, $depth)}; |
87abf9bc |
615 | |
616 | my $rhs = $self->_unparse($args, $bindargs, $depth); |
617 | |
618 | return sprintf "$l%s$r", join( |
6f2a5b66 |
619 | ( ref $args eq 'ARRAY' and @{$args} == 1 and $args->[0][0] eq '-PAREN' ) |
c4d7cfcf |
620 | ? '' # mysql-- |
621 | : ' ' |
622 | , |
87abf9bc |
623 | $self->format_keyword($op), |
624 | (length $rhs ? $rhs : () ), |
625 | ); |
01dd4e4f |
626 | } |
627 | } |
628 | |
bb54fcb4 |
629 | # All of these keywords allow their parameters to be specified with or without parenthesis without changing the semantics |
630 | my @unrollable_ops = ( |
631 | 'ON', |
632 | 'WHERE', |
633 | 'GROUP \s+ BY', |
634 | 'HAVING', |
635 | 'ORDER \s+ BY', |
6e9a377b |
636 | 'I?LIKE', |
bb54fcb4 |
637 | ); |
638 | my $unrollable_ops_re = join ' | ', @unrollable_ops; |
639 | $unrollable_ops_re = qr/$unrollable_ops_re/xi; |
640 | |
641 | sub _parenthesis_unroll { |
642 | my $self = shift; |
643 | my $ast = shift; |
644 | |
bb54fcb4 |
645 | return unless (ref $ast and ref $ast->[1]); |
646 | |
647 | my $changes; |
648 | do { |
649 | my @children; |
650 | $changes = 0; |
651 | |
652 | for my $child (@{$ast->[1]}) { |
007f0853 |
653 | |
bb54fcb4 |
654 | # the current node in this loop is *always* a PAREN |
6f2a5b66 |
655 | if (! ref $child or ! @$child or $child->[0] ne '-PAREN') { |
bb54fcb4 |
656 | push @children, $child; |
657 | next; |
658 | } |
659 | |
a4d17ff1 |
660 | my $parent_op = $ast->[0]; |
661 | |
bb54fcb4 |
662 | # unroll nested parenthesis |
a4d17ff1 |
663 | while ( $parent_op ne 'IN' and @{$child->[1]} == 1 and $child->[1][0][0] eq '-PAREN') { |
bb54fcb4 |
664 | $child = $child->[1][0]; |
665 | $changes++; |
666 | } |
667 | |
a4d17ff1 |
668 | # set to CHILD in the case of PARENT ( CHILD ) |
669 | # but NOT in the case of PARENT( CHILD1, CHILD2 ) |
670 | my $single_child_op = (@{$child->[1]} == 1) ? $child->[1][0][0] : ''; |
671 | |
672 | my $child_op_argc = $single_child_op ? scalar @{$child->[1][0][1]} : undef; |
673 | |
674 | my $single_grandchild_op |
675 | = ( $child_op_argc||0 == 1 and ref $child->[1][0][1][0] eq 'ARRAY' ) |
676 | ? $child->[1][0][1][0][0] |
677 | : '' |
678 | ; |
679 | |
0ec2e265 |
680 | # if the parent operator explicitly allows it AND the child isn't a subselect |
681 | # nuke the parenthesis |
a4d17ff1 |
682 | if ($parent_op =~ $unrollable_ops_re and $single_child_op ne 'SELECT') { |
6f2a5b66 |
683 | push @children, @{$child->[1]}; |
684 | $changes++; |
685 | } |
686 | |
bb54fcb4 |
687 | # if the parenthesis are wrapped around an AND/OR matching the parent AND/OR - open the parenthesis up and merge the list |
6f2a5b66 |
688 | elsif ( |
a4d17ff1 |
689 | $single_child_op eq $parent_op |
690 | and |
691 | ( $parent_op eq 'AND' or $parent_op eq 'OR') |
bb54fcb4 |
692 | ) { |
693 | push @children, @{$child->[1][0][1]}; |
694 | $changes++; |
695 | } |
696 | |
bb54fcb4 |
697 | # only *ONE* LITERAL or placeholder element |
6e9a377b |
698 | # as an AND/OR/NOT argument |
bb54fcb4 |
699 | elsif ( |
a4d17ff1 |
700 | ( $single_child_op eq '-LITERAL' or $single_child_op eq '-PLACEHOLDER' ) |
701 | and |
702 | ( $parent_op eq 'AND' or $parent_op eq 'OR' or $parent_op eq 'NOT' ) |
bb54fcb4 |
703 | ) { |
6f2a5b66 |
704 | push @children, @{$child->[1]}; |
bb54fcb4 |
705 | $changes++; |
706 | } |
707 | |
007f0853 |
708 | # an AND/OR expression with only one binop in the parenthesis |
709 | # with exactly two grandchildren |
bb54fcb4 |
710 | # the only time when we can *not* unroll this is when both |
711 | # the parent and the child are mathops (in which case we'll |
712 | # break precedence) or when the child is BETWEEN (special |
713 | # case) |
714 | elsif ( |
a4d17ff1 |
715 | ($parent_op eq 'AND' or $parent_op eq 'OR') |
007f0853 |
716 | and |
a4d17ff1 |
717 | $single_child_op =~ $binary_op_re |
bb54fcb4 |
718 | and |
a4d17ff1 |
719 | $single_child_op ne 'BETWEEN' |
bb54fcb4 |
720 | and |
a4d17ff1 |
721 | $child_op_argc == 2 |
bb54fcb4 |
722 | and |
723 | ! ( |
a4d17ff1 |
724 | $single_child_op =~ $alphanum_cmp_op_re |
bb54fcb4 |
725 | and |
a4d17ff1 |
726 | $parent_op =~ $alphanum_cmp_op_re |
bb54fcb4 |
727 | ) |
728 | ) { |
6f2a5b66 |
729 | push @children, @{$child->[1]}; |
bb54fcb4 |
730 | $changes++; |
731 | } |
732 | |
733 | # a function binds tighter than a mathop - see if our ancestor is a |
734 | # mathop, and our content is: |
735 | # a single non-mathop child with a single PAREN grandchild which |
736 | # would indicate mathop ( nonmathop ( ... ) ) |
737 | # or a single non-mathop with a single LITERAL ( nonmathop foo ) |
738 | # or a single non-mathop with a single PLACEHOLDER ( nonmathop ? ) |
739 | elsif ( |
a4d17ff1 |
740 | $single_child_op |
bb54fcb4 |
741 | and |
a4d17ff1 |
742 | $parent_op =~ $alphanum_cmp_op_re |
bb54fcb4 |
743 | and |
a4d17ff1 |
744 | $single_child_op !~ $alphanum_cmp_op_re |
bb54fcb4 |
745 | and |
a4d17ff1 |
746 | $child_op_argc == 1 |
bb54fcb4 |
747 | and |
748 | ( |
a4d17ff1 |
749 | $single_grandchild_op eq '-PAREN' |
bb54fcb4 |
750 | or |
a4d17ff1 |
751 | $single_grandchild_op eq '-LITERAL' |
bb54fcb4 |
752 | or |
a4d17ff1 |
753 | $single_grandchild_op eq '-PLACEHOLDER' |
bb54fcb4 |
754 | ) |
755 | ) { |
6f2a5b66 |
756 | push @children, @{$child->[1]}; |
bb54fcb4 |
757 | $changes++; |
758 | } |
759 | |
1de1d085 |
760 | # a construct of ... ( somefunc ( ... ) ) ... can safely lose the outer parens |
761 | # except for the case of ( NOT ( ... ) ) which has already been handled earlier |
0ec2e265 |
762 | # and except for the case of RNO, where the double are explicit syntax |
1de1d085 |
763 | elsif ( |
a4d17ff1 |
764 | $parent_op ne 'ROW_NUMBER() OVER' |
1de1d085 |
765 | and |
a4d17ff1 |
766 | $single_child_op |
1de1d085 |
767 | and |
a4d17ff1 |
768 | $single_child_op ne 'NOT' |
1de1d085 |
769 | and |
a4d17ff1 |
770 | $child_op_argc == 1 |
1de1d085 |
771 | and |
a4d17ff1 |
772 | $single_grandchild_op eq '-PAREN' |
1de1d085 |
773 | ) { |
774 | push @children, @{$child->[1]}; |
775 | $changes++; |
776 | } |
777 | |
bb54fcb4 |
778 | |
779 | # otherwise no more mucking for this pass |
780 | else { |
781 | push @children, $child; |
782 | } |
783 | } |
784 | |
785 | $ast->[1] = \@children; |
786 | |
787 | } while ($changes); |
bb54fcb4 |
788 | } |
789 | |
0c2de280 |
790 | sub _strip_asc_from_order_by { |
791 | my ($self, $ast) = @_; |
792 | |
793 | return $ast if ( |
794 | ref $ast ne 'ARRAY' |
795 | or |
796 | $ast->[0] ne 'ORDER BY' |
797 | ); |
798 | |
799 | |
800 | my $to_replace; |
801 | |
802 | if (@{$ast->[1]} == 1 and $ast->[1][0][0] eq '-ASC') { |
803 | $to_replace = [ $ast->[1][0] ]; |
804 | } |
805 | elsif (@{$ast->[1]} == 1 and $ast->[1][0][0] eq '-LIST') { |
806 | $to_replace = [ grep { $_->[0] eq '-ASC' } @{$ast->[1][0][1]} ]; |
807 | } |
808 | |
809 | @$_ = @{$_->[1][0]} for @$to_replace; |
810 | |
811 | $ast; |
812 | } |
813 | |
fb272e73 |
814 | sub format { my $self = shift; $self->unparse($self->parse($_[0]), $_[1]) } |
01dd4e4f |
815 | |
816 | 1; |
817 | |
3be357b0 |
818 | =pod |
819 | |
b912ee1e |
820 | =head1 NAME |
821 | |
822 | SQL::Abstract::Tree - Represent SQL as an AST |
823 | |
3be357b0 |
824 | =head1 SYNOPSIS |
825 | |
826 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
827 | |
828 | print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2'); |
829 | |
830 | # SELECT * |
831 | # FROM foo |
832 | # WHERE foo.a > 2 |
833 | |
6b1bf9f8 |
834 | =head1 METHODS |
835 | |
836 | =head2 new |
837 | |
838 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
839 | |
c22f502d |
840 | $args = { |
841 | profile => 'console', # predefined profile to use (default: 'none') |
842 | fill_in_placeholders => 1, # true for placeholder population |
9d11f0d4 |
843 | placeholder_surround => # The strings that will be wrapped around |
844 | [GREEN, RESET], # populated placeholders if the above is set |
c22f502d |
845 | indent_string => ' ', # the string used when indenting |
846 | indent_amount => 2, # how many of above string to use for a single |
847 | # indent level |
848 | newline => "\n", # string for newline |
849 | colormap => { |
850 | select => [RED, RESET], # a pair of strings defining what to surround |
851 | # the keyword with for colorization |
852 | # ... |
853 | }, |
854 | indentmap => { |
855 | select => 0, # A zero means that the keyword will start on |
856 | # a new line |
857 | from => 1, # Any other positive integer means that after |
858 | on => 2, # said newline it will get that many indents |
859 | # ... |
860 | }, |
861 | } |
862 | |
863 | Returns a new SQL::Abstract::Tree object. All arguments are optional. |
864 | |
865 | =head3 profiles |
866 | |
867 | There are four predefined profiles, C<none>, C<console>, C<console_monochrome>, |
868 | and C<html>. Typically a user will probably just use C<console> or |
869 | C<console_monochrome>, but if something about a profile bothers you, merely |
870 | use the profile and override the parts that you don't like. |
871 | |
6b1bf9f8 |
872 | =head2 format |
873 | |
c22f502d |
874 | $sqlat->format('SELECT * FROM bar WHERE x = ?', [1]) |
875 | |
876 | Takes C<$sql> and C<\@bindargs>. |
6b1bf9f8 |
877 | |
1a3cc911 |
878 | Returns a formatting string based on the string passed in |
ee4227a7 |
879 | |
880 | =head2 parse |
881 | |
882 | $sqlat->parse('SELECT * FROM bar WHERE x = ?') |
883 | |
884 | Returns a "tree" representing passed in SQL. Please do not depend on the |
885 | structure of the returned tree. It may be stable at some point, but not yet. |
886 | |
887 | =head2 unparse |
888 | |
90d0250b |
889 | $sqlat->unparse($tree_structure, \@bindargs) |
ee4227a7 |
890 | |
891 | Transform "tree" into SQL, applying various transforms on the way. |
892 | |
893 | =head2 format_keyword |
894 | |
895 | $sqlat->format_keyword('SELECT') |
896 | |
897 | Currently this just takes a keyword and puts the C<colormap> stuff around it. |
898 | Later on it may do more and allow for coderef based transforms. |
899 | |
f2ab166a |
900 | =head2 pad_keyword |
ee4227a7 |
901 | |
f2ab166a |
902 | my ($before, $after) = @{$sqlat->pad_keyword('SELECT')}; |
ee4227a7 |
903 | |
904 | Returns whitespace to be inserted around a keyword. |
9d11f0d4 |
905 | |
906 | =head2 fill_in_placeholder |
907 | |
908 | my $value = $sqlat->fill_in_placeholder(\@bindargs) |
909 | |
910 | Removes last arg from passed arrayref and returns it, surrounded with |
911 | the values in placeholder_surround, and then surrounded with single quotes. |
f2ab166a |
912 | |
913 | =head2 indent |
914 | |
915 | Returns as many indent strings as indent amounts times the first argument. |
916 | |
917 | =head1 ACCESSORS |
918 | |
919 | =head2 colormap |
920 | |
921 | See L</new> |
922 | |
923 | =head2 fill_in_placeholders |
924 | |
925 | See L</new> |
926 | |
927 | =head2 indent_amount |
928 | |
929 | See L</new> |
930 | |
931 | =head2 indent_string |
932 | |
933 | See L</new> |
934 | |
935 | =head2 indentmap |
936 | |
937 | See L</new> |
938 | |
939 | =head2 newline |
940 | |
941 | See L</new> |
942 | |
943 | =head2 placeholder_surround |
944 | |
945 | See L</new> |
946 | |