Commit | Line | Data |
01dd4e4f |
1 | package SQL::Abstract::Tree; |
2 | |
3 | use strict; |
4 | use warnings; |
b3b79607 |
5 | no warnings 'qw'; |
01dd4e4f |
6 | use Carp; |
7 | |
0769ac0e |
8 | use Hash::Merge qw//; |
9 | |
10 | use base 'Class::Accessor::Grouped'; |
11 | |
ad591616 |
12 | __PACKAGE__->mk_group_accessors( simple => qw( |
0769ac0e |
13 | newline indent_string indent_amount colormap indentmap fill_in_placeholders |
14 | placeholder_surround |
ad591616 |
15 | )); |
2fed0b4b |
16 | |
bc482085 |
17 | my $merger = Hash::Merge->new; |
18 | |
19 | $merger->specify_behavior({ |
2fed0b4b |
20 | SCALAR => { |
21 | SCALAR => sub { $_[1] }, |
22 | ARRAY => sub { [ $_[0], @{$_[1]} ] }, |
23 | HASH => sub { $_[1] }, |
24 | }, |
25 | ARRAY => { |
26 | SCALAR => sub { $_[1] }, |
27 | ARRAY => sub { $_[1] }, |
28 | HASH => sub { $_[1] }, |
29 | }, |
30 | HASH => { |
31 | SCALAR => sub { $_[1] }, |
32 | ARRAY => sub { [ values %{$_[0]}, @{$_[1]} ] }, |
33 | HASH => sub { Hash::Merge::_merge_hashes( $_[0], $_[1] ) }, |
34 | }, |
0769ac0e |
35 | }, 'SQLA::Tree Behavior' ); |
1536de15 |
36 | |
0769ac0e |
37 | my $op_look_ahead = '(?: (?= [\s\)\(\;] ) | \z)'; |
b3b79607 |
38 | my $op_look_behind = '(?: (?<= [\,\s\)\(] ) | \A )'; |
39 | |
0769ac0e |
40 | my $quote_left = qr/[\`\'\"\[]/; |
41 | my $quote_right = qr/[\`\'\"\]]/; |
01dd4e4f |
42 | |
4e914a7c |
43 | my $placeholder_re = qr/(?: \? | \$\d+ )/x; |
44 | |
01dd4e4f |
45 | # These SQL keywords always signal end of the current expression (except inside |
46 | # of a parenthesized subexpression). |
0769ac0e |
47 | # Format: A list of strings that will be compiled to extended syntax ie. |
01dd4e4f |
48 | # /.../x) regexes, without capturing parentheses. They will be automatically |
0769ac0e |
49 | # anchored to op boundaries (excluding quotes) to match the whole token. |
50 | my @expression_start_keywords = ( |
01dd4e4f |
51 | 'SELECT', |
7853a177 |
52 | 'UPDATE', |
6c4d8eb8 |
53 | 'SET', |
7853a177 |
54 | 'INSERT \s+ INTO', |
55 | 'DELETE \s+ FROM', |
3d910890 |
56 | 'FROM', |
01dd4e4f |
57 | '(?: |
58 | (?: |
0769ac0e |
59 | (?: (?: LEFT | RIGHT | FULL ) \s+ )? |
60 | (?: (?: CROSS | INNER | OUTER ) \s+ )? |
01dd4e4f |
61 | )? |
62 | JOIN |
63 | )', |
64 | 'ON', |
65 | 'WHERE', |
efc991a0 |
66 | '(?: DEFAULT \s+ )? VALUES', |
01dd4e4f |
67 | 'GROUP \s+ BY', |
68 | 'HAVING', |
69 | 'ORDER \s+ BY', |
c0eaa9fd |
70 | 'SKIP', |
81b3e585 |
71 | 'FETCH', |
c0eaa9fd |
72 | 'FIRST', |
01dd4e4f |
73 | 'LIMIT', |
74 | 'OFFSET', |
75 | 'FOR', |
76 | 'UNION', |
77 | 'INTERSECT', |
78 | 'EXCEPT', |
820bb1f5 |
79 | 'BEGIN \s+ WORK', |
80 | 'COMMIT', |
81 | 'ROLLBACK \s+ TO \s+ SAVEPOINT', |
82 | 'ROLLBACK', |
83 | 'SAVEPOINT', |
84 | 'RELEASE \s+ SAVEPOINT', |
01dd4e4f |
85 | 'RETURNING', |
86 | ); |
87 | |
b3b79607 |
88 | my $expr_start_re = join ("\n\t|\n", @expression_start_keywords ); |
89 | $expr_start_re = qr/ $op_look_behind (?i: $expr_start_re ) $op_look_ahead /x; |
0769ac0e |
90 | |
01dd4e4f |
91 | # These are binary operator keywords always a single LHS and RHS |
92 | # * AND/OR are handled separately as they are N-ary |
93 | # * so is NOT as being unary |
3af02ccb |
94 | # * BETWEEN without parentheses around the ANDed arguments (which |
95 | # makes it a non-binary op) is detected and accommodated in |
01dd4e4f |
96 | # _recurse_parse() |
6c4d8eb8 |
97 | # * AS is not really an operator but is handled here as it's also LHS/RHS |
01dd4e4f |
98 | |
0769ac0e |
99 | # this will be included in the $binary_op_re, the distinction is interesting during |
1ec9b9e3 |
100 | # testing as one is tighter than the other, plus alphanum cmp ops have different |
101 | # look ahead/behind (e.g. "x"="y" ) |
102 | my @alphanum_cmp_op_keywords = (qw/< > != <> = <= >= /); |
103 | my $alphanum_cmp_op_re = join ("\n\t|\n", map |
0769ac0e |
104 | { "(?: (?<= [\\w\\s] | $quote_right ) | \\A )" . quotemeta ($_) . "(?: (?= [\\w\\s] | $quote_left ) | \\z )" } |
1ec9b9e3 |
105 | @alphanum_cmp_op_keywords |
01dd4e4f |
106 | ); |
1ec9b9e3 |
107 | $alphanum_cmp_op_re = qr/$alphanum_cmp_op_re/x; |
0769ac0e |
108 | |
109 | my $binary_op_re = '(?: NOT \s+)? (?:' . join ('|', qw/IN BETWEEN R?LIKE/) . ')'; |
b3b79607 |
110 | $binary_op_re = join "\n\t|\n", |
6c4d8eb8 |
111 | "$op_look_behind (?i: $binary_op_re | AS ) $op_look_ahead", |
1ec9b9e3 |
112 | $alphanum_cmp_op_re, |
b3b79607 |
113 | $op_look_behind . 'IS (?:\s+ NOT)?' . "(?= \\s+ NULL \\b | $op_look_ahead )", |
114 | ; |
b7b0f832 |
115 | $binary_op_re = qr/$binary_op_re/x; |
0769ac0e |
116 | |
4247c384 |
117 | my $rno_re = qr/ROW_NUMBER \s* \( \s* \) \s+ OVER/ix; |
118 | |
119 | my $unary_op_re = 'NOT \s+ EXISTS | NOT | ' . $rno_re; |
6f2a5b66 |
120 | $unary_op_re = join "\n\t|\n", |
121 | "$op_look_behind (?i: $unary_op_re ) $op_look_ahead", |
122 | ; |
123 | $unary_op_re = qr/$unary_op_re/x; |
124 | |
ad591616 |
125 | my $asc_desc_re = qr/$op_look_behind (?i: ASC | DESC ) $op_look_ahead /x; |
126 | my $and_or_re = qr/$op_look_behind (?i: AND | OR ) $op_look_ahead /x; |
6f2a5b66 |
127 | |
ad591616 |
128 | my $tokenizer_re = join("\n\t|\n", |
b3b79607 |
129 | $expr_start_re, |
0769ac0e |
130 | $binary_op_re, |
6f2a5b66 |
131 | $unary_op_re, |
ad591616 |
132 | $asc_desc_re, |
133 | $and_or_re, |
1ec9b9e3 |
134 | $op_look_behind . ' \* ' . $op_look_ahead, |
257ecc8a |
135 | (map { quotemeta $_ } qw/, ( )/), |
4e914a7c |
136 | $placeholder_re, |
0769ac0e |
137 | ); |
01dd4e4f |
138 | |
ad591616 |
139 | # this one *is* capturing for the split below |
b3b79607 |
140 | # splits on whitespace if all else fails |
3af02ccb |
141 | # has to happen before the composing qr's are anchored (below) |
ad591616 |
142 | $tokenizer_re = qr/ \s* ( $tokenizer_re ) \s* | \s+ /x; |
b3b79607 |
143 | |
144 | # Parser states for _recurse_parse() |
145 | use constant PARSE_TOP_LEVEL => 0; |
146 | use constant PARSE_IN_EXPR => 1; |
147 | use constant PARSE_IN_PARENS => 2; |
148 | use constant PARSE_IN_FUNC => 3; |
149 | use constant PARSE_RHS => 4; |
6f2a5b66 |
150 | use constant PARSE_LIST_ELT => 5; |
b3b79607 |
151 | |
ad591616 |
152 | my $expr_term_re = qr/$expr_start_re | \)/x; |
153 | my $rhs_term_re = qr/ $expr_term_re | $binary_op_re | $unary_op_re | $asc_desc_re | $and_or_re | \, /x; |
1ec9b9e3 |
154 | my $all_std_keywords_re = qr/ $rhs_term_re | \( | $placeholder_re /x; |
ad591616 |
155 | |
156 | # anchor everything - even though keywords are separated by the tokenizer, leakage may occur |
157 | for ( |
158 | $quote_left, |
159 | $quote_right, |
160 | $placeholder_re, |
161 | $expr_start_re, |
1ec9b9e3 |
162 | $alphanum_cmp_op_re, |
ad591616 |
163 | $binary_op_re, |
164 | $unary_op_re, |
165 | $asc_desc_re, |
166 | $and_or_re, |
167 | $expr_term_re, |
168 | $rhs_term_re, |
ad591616 |
169 | $all_std_keywords_re, |
170 | ) { |
171 | $_ = qr/ \A $_ \z /x; |
172 | } |
173 | |
b4085a1a |
174 | # what can be bunched together under one MISC in an AST |
175 | my $compressable_node_re = qr/^ \- (?: MISC | LITERAL | PLACEHOLDER ) $/x; |
01dd4e4f |
176 | |
7e5600e9 |
177 | my %indents = ( |
7853a177 |
178 | select => 0, |
179 | update => 0, |
180 | 'insert into' => 0, |
181 | 'delete from' => 0, |
3d910890 |
182 | from => 1, |
91916220 |
183 | where => 0, |
7853a177 |
184 | join => 1, |
185 | 'left join' => 1, |
186 | on => 2, |
2867f4f5 |
187 | having => 0, |
91916220 |
188 | 'group by' => 0, |
189 | 'order by' => 0, |
7853a177 |
190 | set => 1, |
191 | into => 1, |
91916220 |
192 | values => 1, |
c0eaa9fd |
193 | limit => 1, |
194 | offset => 1, |
195 | skip => 1, |
196 | first => 1, |
7e5600e9 |
197 | ); |
198 | |
75c3a063 |
199 | my %profiles = ( |
200 | console => { |
84c65032 |
201 | fill_in_placeholders => 1, |
9d11f0d4 |
202 | placeholder_surround => ['?/', ''], |
1536de15 |
203 | indent_string => ' ', |
75c3a063 |
204 | indent_amount => 2, |
1536de15 |
205 | newline => "\n", |
3be357b0 |
206 | colormap => {}, |
6d388c84 |
207 | indentmap => \%indents, |
aafbf833 |
208 | |
209 | eval { require Term::ANSIColor } |
210 | ? do { |
211 | my $c = \&Term::ANSIColor::color; |
6d388c84 |
212 | |
213 | my $red = [$c->('red') , $c->('reset')]; |
214 | my $cyan = [$c->('cyan') , $c->('reset')]; |
215 | my $green = [$c->('green') , $c->('reset')]; |
216 | my $yellow = [$c->('yellow') , $c->('reset')]; |
217 | my $blue = [$c->('blue') , $c->('reset')]; |
218 | my $magenta = [$c->('magenta'), $c->('reset')]; |
219 | my $b_o_w = [$c->('black on_white'), $c->('reset')]; |
aafbf833 |
220 | ( |
fb98df48 |
221 | placeholder_surround => [$c->('black on_magenta'), $c->('reset')], |
aafbf833 |
222 | colormap => { |
6d388c84 |
223 | 'begin work' => $b_o_w, |
224 | commit => $b_o_w, |
225 | rollback => $b_o_w, |
226 | savepoint => $b_o_w, |
227 | 'rollback to savepoint' => $b_o_w, |
228 | 'release savepoint' => $b_o_w, |
229 | |
230 | select => $red, |
231 | 'insert into' => $red, |
232 | update => $red, |
233 | 'delete from' => $red, |
234 | |
235 | set => $cyan, |
236 | from => $cyan, |
237 | |
238 | where => $green, |
239 | values => $yellow, |
240 | |
241 | join => $magenta, |
242 | 'left join' => $magenta, |
243 | on => $blue, |
244 | |
245 | 'group by' => $yellow, |
2867f4f5 |
246 | having => $yellow, |
6d388c84 |
247 | 'order by' => $yellow, |
248 | |
249 | skip => $green, |
250 | first => $green, |
251 | limit => $green, |
252 | offset => $green, |
aafbf833 |
253 | } |
254 | ); |
255 | } : (), |
3be357b0 |
256 | }, |
257 | console_monochrome => { |
84c65032 |
258 | fill_in_placeholders => 1, |
9d11f0d4 |
259 | placeholder_surround => ['?/', ''], |
3be357b0 |
260 | indent_string => ' ', |
261 | indent_amount => 2, |
262 | newline => "\n", |
263 | colormap => {}, |
6d388c84 |
264 | indentmap => \%indents, |
7e5600e9 |
265 | }, |
266 | html => { |
84c65032 |
267 | fill_in_placeholders => 1, |
9d11f0d4 |
268 | placeholder_surround => ['<span class="placeholder">', '</span>'], |
7e5600e9 |
269 | indent_string => ' ', |
270 | indent_amount => 2, |
271 | newline => "<br />\n", |
272 | colormap => { |
7853a177 |
273 | select => ['<span class="select">' , '</span>'], |
274 | 'insert into' => ['<span class="insert-into">' , '</span>'], |
275 | update => ['<span class="select">' , '</span>'], |
276 | 'delete from' => ['<span class="delete-from">' , '</span>'], |
c0eaa9fd |
277 | |
278 | set => ['<span class="set">', '</span>'], |
7853a177 |
279 | from => ['<span class="from">' , '</span>'], |
c0eaa9fd |
280 | |
281 | where => ['<span class="where">' , '</span>'], |
282 | values => ['<span class="values">', '</span>'], |
283 | |
7853a177 |
284 | join => ['<span class="join">' , '</span>'], |
c0eaa9fd |
285 | 'left join' => ['<span class="left-join">','</span>'], |
7853a177 |
286 | on => ['<span class="on">' , '</span>'], |
c0eaa9fd |
287 | |
7853a177 |
288 | 'group by' => ['<span class="group-by">', '</span>'], |
2867f4f5 |
289 | having => ['<span class="having">', '</span>'], |
7853a177 |
290 | 'order by' => ['<span class="order-by">', '</span>'], |
c0eaa9fd |
291 | |
292 | skip => ['<span class="skip">', '</span>'], |
293 | first => ['<span class="first">', '</span>'], |
294 | limit => ['<span class="limit">', '</span>'], |
295 | offset => ['<span class="offset">', '</span>'], |
820bb1f5 |
296 | |
297 | 'begin work' => ['<span class="begin-work">', '</span>'], |
298 | commit => ['<span class="commit">', '</span>'], |
299 | rollback => ['<span class="rollback">', '</span>'], |
300 | savepoint => ['<span class="savepoint">', '</span>'], |
301 | 'rollback to savepoint' => ['<span class="rollback-to-savepoint">', '</span>'], |
302 | 'release savepoint' => ['<span class="release-savepoint">', '</span>'], |
1536de15 |
303 | }, |
6d388c84 |
304 | indentmap => \%indents, |
75c3a063 |
305 | }, |
306 | none => { |
1536de15 |
307 | colormap => {}, |
308 | indentmap => {}, |
75c3a063 |
309 | }, |
310 | ); |
311 | |
312 | sub new { |
2fed0b4b |
313 | my $class = shift; |
314 | my $args = shift || {}; |
75c3a063 |
315 | |
316 | my $profile = delete $args->{profile} || 'none'; |
1c33db5d |
317 | |
318 | die "No such profile '$profile'!" unless exists $profiles{$profile}; |
319 | |
bc482085 |
320 | my $data = $merger->merge( $profiles{$profile}, $args ); |
75c3a063 |
321 | |
322 | bless $data, $class |
323 | } |
d695b0ad |
324 | |
01dd4e4f |
325 | sub parse { |
d695b0ad |
326 | my ($self, $s) = @_; |
01dd4e4f |
327 | |
328 | # tokenize string, and remove all optional whitespace |
329 | my $tokens = []; |
330 | foreach my $token (split $tokenizer_re, $s) { |
b3b79607 |
331 | push @$tokens, $token if ( |
332 | defined $token |
333 | and |
334 | length $token |
09931431 |
335 | and |
b3b79607 |
336 | $token =~ /\S/ |
337 | ); |
01dd4e4f |
338 | } |
6f2a5b66 |
339 | |
340 | return [ $self->_recurse_parse($tokens, PARSE_TOP_LEVEL) ]; |
01dd4e4f |
341 | } |
342 | |
343 | sub _recurse_parse { |
d695b0ad |
344 | my ($self, $tokens, $state) = @_; |
01dd4e4f |
345 | |
6f2a5b66 |
346 | my @left; |
01dd4e4f |
347 | while (1) { # left-associative parsing |
348 | |
6f2a5b66 |
349 | if ( ! @$tokens |
01dd4e4f |
350 | or |
6f2a5b66 |
351 | ($state == PARSE_IN_PARENS && $tokens->[0] eq ')') |
01dd4e4f |
352 | or |
6f2a5b66 |
353 | ($state == PARSE_IN_EXPR && $tokens->[0] =~ $expr_term_re ) |
0769ac0e |
354 | or |
6f2a5b66 |
355 | ($state == PARSE_RHS && $tokens->[0] =~ $rhs_term_re ) |
01dd4e4f |
356 | or |
ad591616 |
357 | ($state == PARSE_LIST_ELT && ( $tokens->[0] eq ',' or $tokens->[0] =~ $expr_term_re ) ) |
01dd4e4f |
358 | ) { |
6f2a5b66 |
359 | return @left; |
01dd4e4f |
360 | } |
361 | |
362 | my $token = shift @$tokens; |
363 | |
364 | # nested expression in () |
365 | if ($token eq '(' ) { |
6f2a5b66 |
366 | my @right = $self->_recurse_parse($tokens, PARSE_IN_PARENS); |
367 | $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse(\@right); |
368 | $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse(\@right); |
369 | |
370 | push @left, [ '-PAREN' => \@right ]; |
371 | } |
372 | |
373 | # AND/OR |
ad591616 |
374 | elsif ($token =~ $and_or_re) { |
6f2a5b66 |
375 | my $op = uc $token; |
01dd4e4f |
376 | |
6f2a5b66 |
377 | my @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
378 | |
379 | # Merge chunks if "logic" matches |
380 | @left = [ $op => [ @left, (@right and $op eq $right[0][0]) |
381 | ? @{ $right[0][1] } |
382 | : @right |
383 | ] ]; |
01dd4e4f |
384 | } |
b3b79607 |
385 | |
6f2a5b66 |
386 | # LIST (,) |
387 | elsif ($token eq ',') { |
388 | |
389 | my @right = $self->_recurse_parse($tokens, PARSE_LIST_ELT); |
390 | |
391 | # deal with malformed lists ( foo, bar, , baz ) |
392 | @right = [] unless @right; |
01dd4e4f |
393 | |
6f2a5b66 |
394 | @right = [ -MISC => [ @right ] ] if @right > 1; |
395 | |
396 | if (!@left) { |
397 | @left = [ -LIST => [ [], @right ] ]; |
398 | } |
399 | elsif ($left[0][0] eq '-LIST') { |
400 | push @{$left[0][1]}, (@{$right[0]} and $right[0][0] eq '-LIST') |
401 | ? @{$right[0][1]} |
402 | : @right |
403 | ; |
01dd4e4f |
404 | } |
405 | else { |
6f2a5b66 |
406 | @left = [ -LIST => [ @left, @right ] ]; |
01dd4e4f |
407 | } |
408 | } |
6f2a5b66 |
409 | |
01dd4e4f |
410 | # binary operator keywords |
6f2a5b66 |
411 | elsif ($token =~ $binary_op_re) { |
01dd4e4f |
412 | my $op = uc $token; |
6f2a5b66 |
413 | |
414 | my @right = $self->_recurse_parse($tokens, PARSE_RHS); |
01dd4e4f |
415 | |
416 | # A between with a simple LITERAL for a 1st RHS argument needs a |
417 | # rerun of the search to (hopefully) find the proper AND construct |
6f2a5b66 |
418 | if ($op eq 'BETWEEN' and $right[0] eq '-LITERAL') { |
419 | unshift @$tokens, $right[1][0]; |
420 | @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
421 | } |
422 | |
08e16360 |
423 | push @left, [$op => [ (@left ? pop @left : ''), @right ]]; |
01dd4e4f |
424 | } |
6f2a5b66 |
425 | |
426 | # unary op keywords |
427 | elsif ( $token =~ $unary_op_re ) { |
01dd4e4f |
428 | my $op = uc $token; |
4247c384 |
429 | |
430 | # normalize RNO explicitly |
431 | $op = 'ROW_NUMBER() OVER' if $op =~ /^$rno_re$/; |
432 | |
6f2a5b66 |
433 | my @right = $self->_recurse_parse ($tokens, PARSE_RHS); |
434 | |
435 | push @left, [ $op => \@right ]; |
01dd4e4f |
436 | } |
6f2a5b66 |
437 | |
438 | # expression terminator keywords |
439 | elsif ( $token =~ $expr_start_re ) { |
01dd4e4f |
440 | my $op = uc $token; |
6f2a5b66 |
441 | my @right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
442 | |
6f2a5b66 |
443 | push @left, [ $op => \@right ]; |
01dd4e4f |
444 | } |
6f2a5b66 |
445 | |
446 | # a '?' |
4e914a7c |
447 | elsif ( $token =~ $placeholder_re) { |
6f2a5b66 |
448 | push @left, [ -PLACEHOLDER => [ $token ] ]; |
449 | } |
450 | |
451 | # check if the current token is an unknown op-start |
1ec9b9e3 |
452 | elsif (@$tokens and ($tokens->[0] eq '(' or $tokens->[0] =~ $placeholder_re ) ) { |
6f2a5b66 |
453 | push @left, [ $token => [ $self->_recurse_parse($tokens, PARSE_RHS) ] ]; |
4e914a7c |
454 | } |
6f2a5b66 |
455 | |
b3b79607 |
456 | # we're now in "unknown token" land - start eating tokens until |
6f30911f |
457 | # we see something familiar, OR in the case of RHS (binop) stop |
458 | # after the first token |
e2a120ce |
459 | # Also stop processing when we could end up with an unknown func |
01dd4e4f |
460 | else { |
6f2a5b66 |
461 | my @lits = [ -LITERAL => [$token] ]; |
b3b79607 |
462 | |
b4085a1a |
463 | unshift @lits, pop @left if @left == 1; |
464 | |
6f30911f |
465 | unless ( $state == PARSE_RHS ) { |
e2a120ce |
466 | while ( |
467 | @$tokens |
468 | and |
469 | $tokens->[0] !~ $all_std_keywords_re |
470 | and |
471 | ! ( @$tokens > 1 and $tokens->[1] eq '(' ) |
472 | ) { |
6f30911f |
473 | push @lits, [ -LITERAL => [ shift @$tokens ] ]; |
b4085a1a |
474 | } |
6f30911f |
475 | } |
6f2a5b66 |
476 | |
6f2a5b66 |
477 | @lits = [ -MISC => [ @lits ] ] if @lits > 1; |
478 | |
479 | push @left, @lits; |
480 | } |
b3b79607 |
481 | |
b4085a1a |
482 | # compress -LITERAL -MISC and -PLACEHOLDER pieces into a single |
483 | # -MISC container |
484 | if (@left > 1) { |
485 | my $i = 0; |
486 | while ($#left > $i) { |
487 | if ($left[$i][0] =~ $compressable_node_re and $left[$i+1][0] =~ $compressable_node_re) { |
488 | splice @left, $i, 2, [ -MISC => [ |
489 | map { $_->[0] eq '-MISC' ? @{$_->[1]} : $_ } (@left[$i, $i+1]) |
490 | ]]; |
491 | } |
492 | else { |
493 | $i++; |
494 | } |
495 | } |
496 | } |
497 | |
498 | return @left if $state == PARSE_RHS; |
6f30911f |
499 | |
b4085a1a |
500 | # deal with post-fix operators |
501 | if (@$tokens) { |
502 | # asc/desc |
6f2a5b66 |
503 | if ($tokens->[0] =~ $asc_desc_re) { |
1ec9b9e3 |
504 | @left = [ ('-' . uc (shift @$tokens)) => [ @left ] ]; |
6f2a5b66 |
505 | } |
01dd4e4f |
506 | } |
507 | } |
508 | } |
509 | |
d695b0ad |
510 | sub format_keyword { |
511 | my ($self, $keyword) = @_; |
512 | |
1536de15 |
513 | if (my $around = $self->colormap->{lc $keyword}) { |
d695b0ad |
514 | $keyword = "$around->[0]$keyword$around->[1]"; |
515 | } |
516 | |
517 | return $keyword |
518 | } |
519 | |
728f26a2 |
520 | my %starters = ( |
521 | select => 1, |
522 | update => 1, |
523 | 'insert into' => 1, |
524 | 'delete from' => 1, |
525 | ); |
526 | |
f2ab166a |
527 | sub pad_keyword { |
a24cc3a0 |
528 | my ($self, $keyword, $depth) = @_; |
e171c446 |
529 | |
530 | my $before = ''; |
1536de15 |
531 | if (defined $self->indentmap->{lc $keyword}) { |
532 | $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword}); |
a24cc3a0 |
533 | } |
728f26a2 |
534 | $before = '' if $depth == 0 and defined $starters{lc $keyword}; |
e4570c8e |
535 | return [$before, '']; |
a24cc3a0 |
536 | } |
537 | |
1536de15 |
538 | sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) } |
a24cc3a0 |
539 | |
a97eb57c |
540 | sub _is_key { |
541 | my ($self, $tree) = @_; |
0569a14f |
542 | $tree = $tree->[0] while ref $tree; |
543 | |
a97eb57c |
544 | defined $tree && defined $self->indentmap->{lc $tree}; |
0569a14f |
545 | } |
546 | |
9d11f0d4 |
547 | sub fill_in_placeholder { |
fb272e73 |
548 | my ($self, $bindargs) = @_; |
549 | |
550 | if ($self->fill_in_placeholders) { |
ad46269d |
551 | my $val = shift @{$bindargs} || ''; |
4712657d |
552 | my $quoted = $val =~ s/^(['"])(.*)\1$/$2/; |
9d11f0d4 |
553 | my ($left, $right) = @{$self->placeholder_surround}; |
fb272e73 |
554 | $val =~ s/\\/\\\\/g; |
555 | $val =~ s/'/\\'/g; |
4712657d |
556 | $val = qq('$val') if $quoted; |
557 | return qq($left$val$right) |
fb272e73 |
558 | } |
559 | return '?' |
560 | } |
561 | |
3a247d23 |
562 | # FIXME - terrible name for a user facing API |
01dd4e4f |
563 | sub unparse { |
3a247d23 |
564 | my ($self, $tree, $bindargs) = @_; |
565 | $self->_unparse($tree, [@{$bindargs||[]}], 0); |
566 | } |
a24cc3a0 |
567 | |
3a247d23 |
568 | sub _unparse { |
569 | my ($self, $tree, $bindargs, $depth) = @_; |
01dd4e4f |
570 | |
0769ac0e |
571 | if (not $tree or not @$tree) { |
01dd4e4f |
572 | return ''; |
573 | } |
a24cc3a0 |
574 | |
007f0853 |
575 | # FIXME - needs a config switch to disable |
c01ac648 |
576 | $self->_parenthesis_unroll($tree); |
007f0853 |
577 | |
6f2a5b66 |
578 | my ($op, $args) = @{$tree}[0,1]; |
0769ac0e |
579 | |
6f2a5b66 |
580 | if (! defined $op or (! ref $op and ! defined $args) ) { |
0769ac0e |
581 | require Data::Dumper; |
582 | Carp::confess( sprintf ( "Internal error - malformed branch at depth $depth:\n%s", |
583 | Data::Dumper::Dumper($tree) |
584 | ) ); |
585 | } |
a24cc3a0 |
586 | |
6f2a5b66 |
587 | if (ref $op) { |
3a247d23 |
588 | return join (' ', map $self->_unparse($_, $bindargs, $depth), @$tree); |
01dd4e4f |
589 | } |
6f2a5b66 |
590 | elsif ($op eq '-LITERAL') { # literal has different sig |
591 | return $args->[0]; |
01dd4e4f |
592 | } |
6f2a5b66 |
593 | elsif ($op eq '-PLACEHOLDER') { |
4e914a7c |
594 | return $self->fill_in_placeholder($bindargs); |
595 | } |
6f2a5b66 |
596 | elsif ($op eq '-PAREN') { |
c4d7cfcf |
597 | return sprintf ('( %s )', |
6f2a5b66 |
598 | join (' ', map { $self->_unparse($_, $bindargs, $depth + 2) } @{$args} ) |
e4570c8e |
599 | . |
6f2a5b66 |
600 | ($self->_is_key($args) |
e4570c8e |
601 | ? ( $self->newline||'' ) . $self->indent($depth + 1) |
602 | : '' |
603 | ) |
604 | ); |
01dd4e4f |
605 | } |
ad591616 |
606 | elsif ($op eq 'AND' or $op eq 'OR' or $op =~ $binary_op_re ) { |
6f2a5b66 |
607 | return join (" $op ", map $self->_unparse($_, $bindargs, $depth), @{$args}); |
608 | } |
609 | elsif ($op eq '-LIST' ) { |
610 | return join (', ', map $self->_unparse($_, $bindargs, $depth), @{$args}); |
01dd4e4f |
611 | } |
6f2a5b66 |
612 | elsif ($op eq '-MISC' ) { |
613 | return join (' ', map $self->_unparse($_, $bindargs, $depth), @{$args}); |
b3b79607 |
614 | } |
73835ff0 |
615 | elsif ($op =~ qr/^-(ASC|DESC)$/ ) { |
616 | my $dir = $1; |
617 | return join (' ', (map $self->_unparse($_, $bindargs, $depth), @{$args}), $dir); |
618 | } |
01dd4e4f |
619 | else { |
6f2a5b66 |
620 | my ($l, $r) = @{$self->pad_keyword($op, $depth)}; |
87abf9bc |
621 | |
622 | my $rhs = $self->_unparse($args, $bindargs, $depth); |
623 | |
624 | return sprintf "$l%s$r", join( |
6f2a5b66 |
625 | ( ref $args eq 'ARRAY' and @{$args} == 1 and $args->[0][0] eq '-PAREN' ) |
c4d7cfcf |
626 | ? '' # mysql-- |
627 | : ' ' |
628 | , |
87abf9bc |
629 | $self->format_keyword($op), |
630 | (length $rhs ? $rhs : () ), |
631 | ); |
01dd4e4f |
632 | } |
633 | } |
634 | |
bb54fcb4 |
635 | # All of these keywords allow their parameters to be specified with or without parenthesis without changing the semantics |
636 | my @unrollable_ops = ( |
637 | 'ON', |
638 | 'WHERE', |
639 | 'GROUP \s+ BY', |
640 | 'HAVING', |
641 | 'ORDER \s+ BY', |
6e9a377b |
642 | 'I?LIKE', |
bb54fcb4 |
643 | ); |
644 | my $unrollable_ops_re = join ' | ', @unrollable_ops; |
645 | $unrollable_ops_re = qr/$unrollable_ops_re/xi; |
646 | |
647 | sub _parenthesis_unroll { |
648 | my $self = shift; |
649 | my $ast = shift; |
650 | |
bb54fcb4 |
651 | return unless (ref $ast and ref $ast->[1]); |
652 | |
653 | my $changes; |
654 | do { |
655 | my @children; |
656 | $changes = 0; |
657 | |
658 | for my $child (@{$ast->[1]}) { |
007f0853 |
659 | |
bb54fcb4 |
660 | # the current node in this loop is *always* a PAREN |
6f2a5b66 |
661 | if (! ref $child or ! @$child or $child->[0] ne '-PAREN') { |
bb54fcb4 |
662 | push @children, $child; |
663 | next; |
664 | } |
665 | |
666 | # unroll nested parenthesis |
7d273452 |
667 | while ( $ast->[0] ne 'IN' and @{$child->[1]} == 1 and $child->[1][0][0] eq '-PAREN') { |
bb54fcb4 |
668 | $child = $child->[1][0]; |
669 | $changes++; |
670 | } |
671 | |
0ec2e265 |
672 | # if the parent operator explicitly allows it AND the child isn't a subselect |
673 | # nuke the parenthesis |
674 | if ($ast->[0] =~ $unrollable_ops_re and $child->[1][0][0] ne 'SELECT') { |
6f2a5b66 |
675 | push @children, @{$child->[1]}; |
676 | $changes++; |
677 | } |
678 | |
bb54fcb4 |
679 | # if the parenthesis are wrapped around an AND/OR matching the parent AND/OR - open the parenthesis up and merge the list |
6f2a5b66 |
680 | elsif ( |
681 | @{$child->[1]} == 1 |
682 | and |
bb54fcb4 |
683 | ( $ast->[0] eq 'AND' or $ast->[0] eq 'OR') |
684 | and |
6f2a5b66 |
685 | $child->[1][0][0] eq $ast->[0] |
bb54fcb4 |
686 | ) { |
687 | push @children, @{$child->[1][0][1]}; |
688 | $changes++; |
689 | } |
690 | |
bb54fcb4 |
691 | # only *ONE* LITERAL or placeholder element |
6e9a377b |
692 | # as an AND/OR/NOT argument |
bb54fcb4 |
693 | elsif ( |
694 | @{$child->[1]} == 1 && ( |
6f2a5b66 |
695 | $child->[1][0][0] eq '-LITERAL' |
bb54fcb4 |
696 | or |
6f2a5b66 |
697 | $child->[1][0][0] eq '-PLACEHOLDER' |
6e9a377b |
698 | ) && ( |
699 | $ast->[0] eq 'AND' or $ast->[0] eq 'OR' or $ast->[0] eq 'NOT' |
bb54fcb4 |
700 | ) |
701 | ) { |
6f2a5b66 |
702 | push @children, @{$child->[1]}; |
bb54fcb4 |
703 | $changes++; |
704 | } |
705 | |
007f0853 |
706 | # an AND/OR expression with only one binop in the parenthesis |
707 | # with exactly two grandchildren |
bb54fcb4 |
708 | # the only time when we can *not* unroll this is when both |
709 | # the parent and the child are mathops (in which case we'll |
710 | # break precedence) or when the child is BETWEEN (special |
711 | # case) |
712 | elsif ( |
713 | @{$child->[1]} == 1 |
714 | and |
007f0853 |
715 | ($ast->[0] eq 'AND' or $ast->[0] eq 'OR') |
716 | and |
ad591616 |
717 | $child->[1][0][0] =~ $binary_op_re |
bb54fcb4 |
718 | and |
719 | $child->[1][0][0] ne 'BETWEEN' |
720 | and |
721 | @{$child->[1][0][1]} == 2 |
722 | and |
723 | ! ( |
1ec9b9e3 |
724 | $child->[1][0][0] =~ $alphanum_cmp_op_re |
bb54fcb4 |
725 | and |
1ec9b9e3 |
726 | $ast->[0] =~ $alphanum_cmp_op_re |
bb54fcb4 |
727 | ) |
728 | ) { |
6f2a5b66 |
729 | push @children, @{$child->[1]}; |
bb54fcb4 |
730 | $changes++; |
731 | } |
732 | |
733 | # a function binds tighter than a mathop - see if our ancestor is a |
734 | # mathop, and our content is: |
735 | # a single non-mathop child with a single PAREN grandchild which |
736 | # would indicate mathop ( nonmathop ( ... ) ) |
737 | # or a single non-mathop with a single LITERAL ( nonmathop foo ) |
738 | # or a single non-mathop with a single PLACEHOLDER ( nonmathop ? ) |
739 | elsif ( |
740 | @{$child->[1]} == 1 |
741 | and |
742 | @{$child->[1][0][1]} == 1 |
743 | and |
1ec9b9e3 |
744 | $ast->[0] =~ $alphanum_cmp_op_re |
bb54fcb4 |
745 | and |
1ec9b9e3 |
746 | $child->[1][0][0] !~ $alphanum_cmp_op_re |
bb54fcb4 |
747 | and |
748 | ( |
6f2a5b66 |
749 | $child->[1][0][1][0][0] eq '-PAREN' |
bb54fcb4 |
750 | or |
6f2a5b66 |
751 | $child->[1][0][1][0][0] eq '-LITERAL' |
bb54fcb4 |
752 | or |
6f2a5b66 |
753 | $child->[1][0][1][0][0] eq '-PLACEHOLDER' |
bb54fcb4 |
754 | ) |
755 | ) { |
6f2a5b66 |
756 | push @children, @{$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 ( |
0ec2e265 |
764 | $ast->[0] ne 'ROW_NUMBER() OVER' |
765 | and |
1de1d085 |
766 | @{$child->[1]} == 1 |
767 | and |
768 | @{$child->[1][0][1]} == 1 |
769 | and |
770 | $child->[1][0][0] ne 'NOT' |
771 | and |
772 | ref $child->[1][0][1][0] eq 'ARRAY' |
773 | and |
774 | $child->[1][0][1][0][0] eq '-PAREN' |
775 | ) { |
776 | push @children, @{$child->[1]}; |
777 | $changes++; |
778 | } |
779 | |
bb54fcb4 |
780 | |
781 | # otherwise no more mucking for this pass |
782 | else { |
783 | push @children, $child; |
784 | } |
785 | } |
786 | |
787 | $ast->[1] = \@children; |
788 | |
789 | } while ($changes); |
bb54fcb4 |
790 | } |
791 | |
0c2de280 |
792 | sub _strip_asc_from_order_by { |
793 | my ($self, $ast) = @_; |
794 | |
795 | return $ast if ( |
796 | ref $ast ne 'ARRAY' |
797 | or |
798 | $ast->[0] ne 'ORDER BY' |
799 | ); |
800 | |
801 | |
802 | my $to_replace; |
803 | |
804 | if (@{$ast->[1]} == 1 and $ast->[1][0][0] eq '-ASC') { |
805 | $to_replace = [ $ast->[1][0] ]; |
806 | } |
807 | elsif (@{$ast->[1]} == 1 and $ast->[1][0][0] eq '-LIST') { |
808 | $to_replace = [ grep { $_->[0] eq '-ASC' } @{$ast->[1][0][1]} ]; |
809 | } |
810 | |
811 | @$_ = @{$_->[1][0]} for @$to_replace; |
812 | |
813 | $ast; |
814 | } |
815 | |
fb272e73 |
816 | sub format { my $self = shift; $self->unparse($self->parse($_[0]), $_[1]) } |
01dd4e4f |
817 | |
818 | 1; |
819 | |
3be357b0 |
820 | =pod |
821 | |
b912ee1e |
822 | =head1 NAME |
823 | |
824 | SQL::Abstract::Tree - Represent SQL as an AST |
825 | |
3be357b0 |
826 | =head1 SYNOPSIS |
827 | |
828 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
829 | |
830 | print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2'); |
831 | |
832 | # SELECT * |
833 | # FROM foo |
834 | # WHERE foo.a > 2 |
835 | |
6b1bf9f8 |
836 | =head1 METHODS |
837 | |
838 | =head2 new |
839 | |
840 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
841 | |
c22f502d |
842 | $args = { |
843 | profile => 'console', # predefined profile to use (default: 'none') |
844 | fill_in_placeholders => 1, # true for placeholder population |
9d11f0d4 |
845 | placeholder_surround => # The strings that will be wrapped around |
846 | [GREEN, RESET], # populated placeholders if the above is set |
c22f502d |
847 | indent_string => ' ', # the string used when indenting |
848 | indent_amount => 2, # how many of above string to use for a single |
849 | # indent level |
850 | newline => "\n", # string for newline |
851 | colormap => { |
852 | select => [RED, RESET], # a pair of strings defining what to surround |
853 | # the keyword with for colorization |
854 | # ... |
855 | }, |
856 | indentmap => { |
857 | select => 0, # A zero means that the keyword will start on |
858 | # a new line |
859 | from => 1, # Any other positive integer means that after |
860 | on => 2, # said newline it will get that many indents |
861 | # ... |
862 | }, |
863 | } |
864 | |
865 | Returns a new SQL::Abstract::Tree object. All arguments are optional. |
866 | |
867 | =head3 profiles |
868 | |
869 | There are four predefined profiles, C<none>, C<console>, C<console_monochrome>, |
870 | and C<html>. Typically a user will probably just use C<console> or |
871 | C<console_monochrome>, but if something about a profile bothers you, merely |
872 | use the profile and override the parts that you don't like. |
873 | |
6b1bf9f8 |
874 | =head2 format |
875 | |
c22f502d |
876 | $sqlat->format('SELECT * FROM bar WHERE x = ?', [1]) |
877 | |
878 | Takes C<$sql> and C<\@bindargs>. |
6b1bf9f8 |
879 | |
1a3cc911 |
880 | Returns a formatting string based on the string passed in |
ee4227a7 |
881 | |
882 | =head2 parse |
883 | |
884 | $sqlat->parse('SELECT * FROM bar WHERE x = ?') |
885 | |
886 | Returns a "tree" representing passed in SQL. Please do not depend on the |
887 | structure of the returned tree. It may be stable at some point, but not yet. |
888 | |
889 | =head2 unparse |
890 | |
90d0250b |
891 | $sqlat->unparse($tree_structure, \@bindargs) |
ee4227a7 |
892 | |
893 | Transform "tree" into SQL, applying various transforms on the way. |
894 | |
895 | =head2 format_keyword |
896 | |
897 | $sqlat->format_keyword('SELECT') |
898 | |
899 | Currently this just takes a keyword and puts the C<colormap> stuff around it. |
900 | Later on it may do more and allow for coderef based transforms. |
901 | |
f2ab166a |
902 | =head2 pad_keyword |
ee4227a7 |
903 | |
f2ab166a |
904 | my ($before, $after) = @{$sqlat->pad_keyword('SELECT')}; |
ee4227a7 |
905 | |
906 | Returns whitespace to be inserted around a keyword. |
9d11f0d4 |
907 | |
908 | =head2 fill_in_placeholder |
909 | |
910 | my $value = $sqlat->fill_in_placeholder(\@bindargs) |
911 | |
912 | Removes last arg from passed arrayref and returns it, surrounded with |
913 | the values in placeholder_surround, and then surrounded with single quotes. |
f2ab166a |
914 | |
915 | =head2 indent |
916 | |
917 | Returns as many indent strings as indent amounts times the first argument. |
918 | |
919 | =head1 ACCESSORS |
920 | |
921 | =head2 colormap |
922 | |
923 | See L</new> |
924 | |
925 | =head2 fill_in_placeholders |
926 | |
927 | See L</new> |
928 | |
929 | =head2 indent_amount |
930 | |
931 | See L</new> |
932 | |
933 | =head2 indent_string |
934 | |
935 | See L</new> |
936 | |
937 | =head2 indentmap |
938 | |
939 | See L</new> |
940 | |
941 | =head2 newline |
942 | |
943 | See L</new> |
944 | |
945 | =head2 placeholder_surround |
946 | |
947 | See L</new> |
948 | |