Commit | Line | Data |
01dd4e4f |
1 | package SQL::Abstract::Tree; |
2 | |
3 | use strict; |
4 | use warnings; |
5 | use Carp; |
6 | |
a97eb57c |
7 | use List::Util; |
bc482085 |
8 | use Hash::Merge; |
2fed0b4b |
9 | |
bc482085 |
10 | my $merger = Hash::Merge->new; |
11 | |
12 | $merger->specify_behavior({ |
2fed0b4b |
13 | SCALAR => { |
14 | SCALAR => sub { $_[1] }, |
15 | ARRAY => sub { [ $_[0], @{$_[1]} ] }, |
16 | HASH => sub { $_[1] }, |
17 | }, |
18 | ARRAY => { |
19 | SCALAR => sub { $_[1] }, |
20 | ARRAY => sub { $_[1] }, |
21 | HASH => sub { $_[1] }, |
22 | }, |
23 | HASH => { |
24 | SCALAR => sub { $_[1] }, |
25 | ARRAY => sub { [ values %{$_[0]}, @{$_[1]} ] }, |
26 | HASH => sub { Hash::Merge::_merge_hashes( $_[0], $_[1] ) }, |
27 | }, |
28 | }, 'My Behavior' ); |
1536de15 |
29 | |
1536de15 |
30 | use base 'Class::Accessor::Grouped'; |
31 | |
32 | __PACKAGE__->mk_group_accessors( simple => $_ ) for qw( |
fb272e73 |
33 | newline indent_string indent_amount colormap indentmap fill_in_placeholders |
1536de15 |
34 | ); |
35 | |
01dd4e4f |
36 | # Parser states for _recurse_parse() |
37 | use constant PARSE_TOP_LEVEL => 0; |
38 | use constant PARSE_IN_EXPR => 1; |
39 | use constant PARSE_IN_PARENS => 2; |
40 | use constant PARSE_RHS => 3; |
41 | |
42 | # These SQL keywords always signal end of the current expression (except inside |
43 | # of a parenthesized subexpression). |
44 | # Format: A list of strings that will be compiled to extended syntax (ie. |
45 | # /.../x) regexes, without capturing parentheses. They will be automatically |
46 | # anchored to word boundaries to match the whole token). |
47 | my @expression_terminator_sql_keywords = ( |
48 | 'SELECT', |
7853a177 |
49 | 'UPDATE', |
50 | 'INSERT \s+ INTO', |
51 | 'DELETE \s+ FROM', |
3d910890 |
52 | 'FROM', |
7853a177 |
53 | 'SET', |
01dd4e4f |
54 | '(?: |
55 | (?: |
56 | (?: \b (?: LEFT | RIGHT | FULL ) \s+ )? |
57 | (?: \b (?: CROSS | INNER | OUTER ) \s+ )? |
58 | )? |
59 | JOIN |
60 | )', |
61 | 'ON', |
62 | 'WHERE', |
7853a177 |
63 | 'VALUES', |
01dd4e4f |
64 | 'EXISTS', |
65 | 'GROUP \s+ BY', |
66 | 'HAVING', |
67 | 'ORDER \s+ BY', |
68 | 'LIMIT', |
69 | 'OFFSET', |
70 | 'FOR', |
71 | 'UNION', |
72 | 'INTERSECT', |
73 | 'EXCEPT', |
74 | 'RETURNING', |
8d0dd7dc |
75 | 'ROW_NUMBER \s* \( \s* \) \s+ OVER', |
01dd4e4f |
76 | ); |
77 | |
78 | # These are binary operator keywords always a single LHS and RHS |
79 | # * AND/OR are handled separately as they are N-ary |
80 | # * so is NOT as being unary |
81 | # * BETWEEN without paranthesis around the ANDed arguments (which |
82 | # makes it a non-binary op) is detected and accomodated in |
83 | # _recurse_parse() |
84 | my $stuff_around_mathops = qr/[\w\s\`\'\"\)]/; |
85 | my @binary_op_keywords = ( |
86 | ( map |
87 | { |
88 | ' ^ ' . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ", |
89 | " (?<= $stuff_around_mathops)" . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ", |
90 | } |
91 | (qw/< > != <> = <= >=/) |
92 | ), |
93 | ( map |
94 | { '\b (?: NOT \s+)?' . $_ . '\b' } |
95 | (qw/IN BETWEEN LIKE/) |
96 | ), |
97 | ); |
98 | |
99 | my $tokenizer_re_str = join("\n\t|\n", |
100 | ( map { '\b' . $_ . '\b' } @expression_terminator_sql_keywords, 'AND', 'OR', 'NOT'), |
101 | @binary_op_keywords, |
102 | ); |
103 | |
104 | my $tokenizer_re = qr/ \s* ( $tokenizer_re_str | \( | \) | \? ) \s* /xi; |
105 | |
106 | sub _binary_op_keywords { @binary_op_keywords } |
107 | |
7e5600e9 |
108 | my %indents = ( |
7853a177 |
109 | select => 0, |
110 | update => 0, |
111 | 'insert into' => 0, |
112 | 'delete from' => 0, |
3d910890 |
113 | from => 1, |
91916220 |
114 | where => 0, |
7853a177 |
115 | join => 1, |
116 | 'left join' => 1, |
117 | on => 2, |
91916220 |
118 | 'group by' => 0, |
119 | 'order by' => 0, |
7853a177 |
120 | set => 1, |
121 | into => 1, |
91916220 |
122 | values => 1, |
7e5600e9 |
123 | ); |
124 | |
75c3a063 |
125 | my %profiles = ( |
126 | console => { |
84c65032 |
127 | fill_in_placeholders => 1, |
1536de15 |
128 | indent_string => ' ', |
75c3a063 |
129 | indent_amount => 2, |
1536de15 |
130 | newline => "\n", |
3be357b0 |
131 | colormap => {}, |
7e5600e9 |
132 | indentmap => { %indents }, |
3be357b0 |
133 | }, |
134 | console_monochrome => { |
84c65032 |
135 | fill_in_placeholders => 1, |
3be357b0 |
136 | indent_string => ' ', |
137 | indent_amount => 2, |
138 | newline => "\n", |
139 | colormap => {}, |
7e5600e9 |
140 | indentmap => { %indents }, |
141 | }, |
142 | html => { |
84c65032 |
143 | fill_in_placeholders => 1, |
7e5600e9 |
144 | indent_string => ' ', |
145 | indent_amount => 2, |
146 | newline => "<br />\n", |
147 | colormap => { |
7853a177 |
148 | select => ['<span class="select">' , '</span>'], |
149 | 'insert into' => ['<span class="insert-into">' , '</span>'], |
150 | update => ['<span class="select">' , '</span>'], |
151 | 'delete from' => ['<span class="delete-from">' , '</span>'], |
152 | where => ['<span class="where">' , '</span>'], |
153 | from => ['<span class="from">' , '</span>'], |
154 | join => ['<span class="join">' , '</span>'], |
155 | on => ['<span class="on">' , '</span>'], |
156 | 'group by' => ['<span class="group-by">', '</span>'], |
157 | 'order by' => ['<span class="order-by">', '</span>'], |
158 | set => ['<span class="set">', '</span>'], |
159 | into => ['<span class="into">', '</span>'], |
160 | values => ['<span class="values">', '</span>'], |
1536de15 |
161 | }, |
7e5600e9 |
162 | indentmap => { %indents }, |
75c3a063 |
163 | }, |
164 | none => { |
1536de15 |
165 | colormap => {}, |
166 | indentmap => {}, |
75c3a063 |
167 | }, |
168 | ); |
169 | |
3be357b0 |
170 | eval { |
171 | require Term::ANSIColor; |
172 | $profiles{console}->{colormap} = { |
7853a177 |
173 | select => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')], |
174 | 'insert into' => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')], |
175 | update => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')], |
176 | 'delete from' => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')], |
177 | |
178 | set => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')], |
c1b89c4f |
179 | from => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')], |
7853a177 |
180 | |
181 | where => [Term::ANSIColor::color('green'), Term::ANSIColor::color('reset')], |
182 | values => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')], |
183 | |
184 | join => [Term::ANSIColor::color('magenta'), Term::ANSIColor::color('reset')], |
185 | 'left join' => [Term::ANSIColor::color('magenta'), Term::ANSIColor::color('reset')], |
186 | on => [Term::ANSIColor::color('blue'), Term::ANSIColor::color('reset')], |
187 | |
188 | 'group by' => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')], |
189 | 'order by' => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')], |
3be357b0 |
190 | }; |
191 | }; |
192 | |
75c3a063 |
193 | sub new { |
2fed0b4b |
194 | my $class = shift; |
195 | my $args = shift || {}; |
75c3a063 |
196 | |
197 | my $profile = delete $args->{profile} || 'none'; |
bc482085 |
198 | my $data = $merger->merge( $profiles{$profile}, $args ); |
75c3a063 |
199 | |
200 | bless $data, $class |
201 | } |
d695b0ad |
202 | |
01dd4e4f |
203 | sub parse { |
d695b0ad |
204 | my ($self, $s) = @_; |
01dd4e4f |
205 | |
206 | # tokenize string, and remove all optional whitespace |
207 | my $tokens = []; |
208 | foreach my $token (split $tokenizer_re, $s) { |
209 | push @$tokens, $token if (length $token) && ($token =~ /\S/); |
210 | } |
211 | |
d695b0ad |
212 | my $tree = $self->_recurse_parse($tokens, PARSE_TOP_LEVEL); |
01dd4e4f |
213 | return $tree; |
214 | } |
215 | |
216 | sub _recurse_parse { |
d695b0ad |
217 | my ($self, $tokens, $state) = @_; |
01dd4e4f |
218 | |
219 | my $left; |
220 | while (1) { # left-associative parsing |
221 | |
222 | my $lookahead = $tokens->[0]; |
223 | if ( not defined($lookahead) |
224 | or |
225 | ($state == PARSE_IN_PARENS && $lookahead eq ')') |
226 | or |
227 | ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords ) ) |
228 | or |
229 | ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords, @binary_op_keywords, 'AND', 'OR', 'NOT' ) ) |
230 | ) { |
231 | return $left; |
232 | } |
233 | |
234 | my $token = shift @$tokens; |
235 | |
236 | # nested expression in () |
237 | if ($token eq '(' ) { |
d695b0ad |
238 | my $right = $self->_recurse_parse($tokens, PARSE_IN_PARENS); |
239 | $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse($right); |
240 | $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse($right); |
01dd4e4f |
241 | |
242 | $left = $left ? [@$left, [PAREN => [$right] ]] |
243 | : [PAREN => [$right] ]; |
244 | } |
245 | # AND/OR |
246 | elsif ($token =~ /^ (?: OR | AND ) $/xi ) { |
247 | my $op = uc $token; |
d695b0ad |
248 | my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
249 | |
250 | # Merge chunks if logic matches |
251 | if (ref $right and $op eq $right->[0]) { |
252 | $left = [ (shift @$right ), [$left, map { @$_ } @$right] ]; |
253 | } |
254 | else { |
255 | $left = [$op => [$left, $right]]; |
256 | } |
257 | } |
258 | # binary operator keywords |
259 | elsif (grep { $token =~ /^ $_ $/xi } @binary_op_keywords ) { |
260 | my $op = uc $token; |
d695b0ad |
261 | my $right = $self->_recurse_parse($tokens, PARSE_RHS); |
01dd4e4f |
262 | |
263 | # A between with a simple LITERAL for a 1st RHS argument needs a |
264 | # rerun of the search to (hopefully) find the proper AND construct |
265 | if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') { |
266 | unshift @$tokens, $right->[1][0]; |
d695b0ad |
267 | $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
268 | } |
269 | |
270 | $left = [$op => [$left, $right] ]; |
271 | } |
272 | # expression terminator keywords (as they start a new expression) |
273 | elsif (grep { $token =~ /^ $_ $/xi } @expression_terminator_sql_keywords ) { |
274 | my $op = uc $token; |
d695b0ad |
275 | my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
276 | $left = $left ? [ $left, [$op => [$right] ]] |
277 | : [ $op => [$right] ]; |
278 | } |
279 | # NOT (last as to allow all other NOT X pieces first) |
280 | elsif ( $token =~ /^ not $/ix ) { |
281 | my $op = uc $token; |
d695b0ad |
282 | my $right = $self->_recurse_parse ($tokens, PARSE_RHS); |
01dd4e4f |
283 | $left = $left ? [ @$left, [$op => [$right] ]] |
284 | : [ $op => [$right] ]; |
285 | |
286 | } |
287 | # literal (eat everything on the right until RHS termination) |
288 | else { |
d695b0ad |
289 | my $right = $self->_recurse_parse ($tokens, PARSE_RHS); |
290 | $left = $left ? [ $left, [LITERAL => [join ' ', $token, $self->unparse($right)||()] ] ] |
291 | : [ LITERAL => [join ' ', $token, $self->unparse($right)||()] ]; |
01dd4e4f |
292 | } |
293 | } |
294 | } |
295 | |
d695b0ad |
296 | sub format_keyword { |
297 | my ($self, $keyword) = @_; |
298 | |
1536de15 |
299 | if (my $around = $self->colormap->{lc $keyword}) { |
d695b0ad |
300 | $keyword = "$around->[0]$keyword$around->[1]"; |
301 | } |
302 | |
303 | return $keyword |
304 | } |
305 | |
728f26a2 |
306 | my %starters = ( |
307 | select => 1, |
308 | update => 1, |
309 | 'insert into' => 1, |
310 | 'delete from' => 1, |
311 | ); |
312 | |
ee4227a7 |
313 | sub whitespace_keyword { |
a24cc3a0 |
314 | my ($self, $keyword, $depth) = @_; |
e171c446 |
315 | |
316 | my $before = ''; |
1536de15 |
317 | if (defined $self->indentmap->{lc $keyword}) { |
318 | $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword}); |
a24cc3a0 |
319 | } |
728f26a2 |
320 | $before = '' if $depth == 0 and defined $starters{lc $keyword}; |
b4e0e260 |
321 | return [$before, ' ']; |
a24cc3a0 |
322 | } |
323 | |
1536de15 |
324 | sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) } |
a24cc3a0 |
325 | |
a97eb57c |
326 | sub _is_key { |
327 | my ($self, $tree) = @_; |
0569a14f |
328 | $tree = $tree->[0] while ref $tree; |
329 | |
a97eb57c |
330 | defined $tree && defined $self->indentmap->{lc $tree}; |
0569a14f |
331 | } |
332 | |
fb272e73 |
333 | sub _fill_in_placeholder { |
334 | my ($self, $bindargs) = @_; |
335 | |
336 | if ($self->fill_in_placeholders) { |
637bb22c |
337 | my $val = pop @{$bindargs} || ''; |
fb272e73 |
338 | $val =~ s/\\/\\\\/g; |
339 | $val =~ s/'/\\'/g; |
340 | return qq('$val') |
341 | } |
342 | return '?' |
343 | } |
344 | |
01dd4e4f |
345 | sub unparse { |
1a3cc911 |
346 | my ($self, $tree, $bindargs, $depth) = @_; |
a24cc3a0 |
347 | |
1a3cc911 |
348 | $depth ||= 0; |
01dd4e4f |
349 | |
350 | if (not $tree ) { |
351 | return ''; |
352 | } |
a24cc3a0 |
353 | |
354 | my $car = $tree->[0]; |
355 | my $cdr = $tree->[1]; |
356 | |
357 | if (ref $car) { |
1a3cc911 |
358 | return join ('', map $self->unparse($_, $bindargs, $depth), @$tree); |
01dd4e4f |
359 | } |
a24cc3a0 |
360 | elsif ($car eq 'LITERAL') { |
fb272e73 |
361 | if ($cdr->[0] eq '?') { |
362 | return $self->_fill_in_placeholder($bindargs) |
363 | } |
a24cc3a0 |
364 | return $cdr->[0]; |
01dd4e4f |
365 | } |
a24cc3a0 |
366 | elsif ($car eq 'PAREN') { |
e171c446 |
367 | return '(' . |
a24cc3a0 |
368 | join(' ', |
1a3cc911 |
369 | map $self->unparse($_, $bindargs, $depth + 2), @{$cdr}) . |
370 | ($self->_is_key($cdr)?( $self->newline||'' ).$self->indent($depth + 1):'') . ') '; |
01dd4e4f |
371 | } |
a24cc3a0 |
372 | elsif ($car eq 'OR' or $car eq 'AND' or (grep { $car =~ /^ $_ $/xi } @binary_op_keywords ) ) { |
1a3cc911 |
373 | return join (" $car ", map $self->unparse($_, $bindargs, $depth), @{$cdr}); |
01dd4e4f |
374 | } |
375 | else { |
ee4227a7 |
376 | my ($l, $r) = @{$self->whitespace_keyword($car, $depth)}; |
1a3cc911 |
377 | return sprintf "$l%s %s$r", $self->format_keyword($car), $self->unparse($cdr, $bindargs, $depth); |
01dd4e4f |
378 | } |
379 | } |
380 | |
fb272e73 |
381 | sub format { my $self = shift; $self->unparse($self->parse($_[0]), $_[1]) } |
01dd4e4f |
382 | |
383 | 1; |
384 | |
3be357b0 |
385 | =pod |
386 | |
387 | =head1 SYNOPSIS |
388 | |
389 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
390 | |
391 | print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2'); |
392 | |
393 | # SELECT * |
394 | # FROM foo |
395 | # WHERE foo.a > 2 |
396 | |
6b1bf9f8 |
397 | =head1 METHODS |
398 | |
399 | =head2 new |
400 | |
401 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
402 | |
c22f502d |
403 | $args = { |
404 | profile => 'console', # predefined profile to use (default: 'none') |
405 | fill_in_placeholders => 1, # true for placeholder population |
406 | indent_string => ' ', # the string used when indenting |
407 | indent_amount => 2, # how many of above string to use for a single |
408 | # indent level |
409 | newline => "\n", # string for newline |
410 | colormap => { |
411 | select => [RED, RESET], # a pair of strings defining what to surround |
412 | # the keyword with for colorization |
413 | # ... |
414 | }, |
415 | indentmap => { |
416 | select => 0, # A zero means that the keyword will start on |
417 | # a new line |
418 | from => 1, # Any other positive integer means that after |
419 | on => 2, # said newline it will get that many indents |
420 | # ... |
421 | }, |
422 | } |
423 | |
424 | Returns a new SQL::Abstract::Tree object. All arguments are optional. |
425 | |
426 | =head3 profiles |
427 | |
428 | There are four predefined profiles, C<none>, C<console>, C<console_monochrome>, |
429 | and C<html>. Typically a user will probably just use C<console> or |
430 | C<console_monochrome>, but if something about a profile bothers you, merely |
431 | use the profile and override the parts that you don't like. |
432 | |
6b1bf9f8 |
433 | =head2 format |
434 | |
c22f502d |
435 | $sqlat->format('SELECT * FROM bar WHERE x = ?', [1]) |
436 | |
437 | Takes C<$sql> and C<\@bindargs>. |
6b1bf9f8 |
438 | |
1a3cc911 |
439 | Returns a formatting string based on the string passed in |
ee4227a7 |
440 | |
441 | =head2 parse |
442 | |
443 | $sqlat->parse('SELECT * FROM bar WHERE x = ?') |
444 | |
445 | Returns a "tree" representing passed in SQL. Please do not depend on the |
446 | structure of the returned tree. It may be stable at some point, but not yet. |
447 | |
448 | =head2 unparse |
449 | |
450 | $sqlat->parse($tree_structure, \@bindargs) |
451 | |
452 | Transform "tree" into SQL, applying various transforms on the way. |
453 | |
454 | =head2 format_keyword |
455 | |
456 | $sqlat->format_keyword('SELECT') |
457 | |
458 | Currently this just takes a keyword and puts the C<colormap> stuff around it. |
459 | Later on it may do more and allow for coderef based transforms. |
460 | |
461 | =head2 whitespace_keyword |
462 | |
463 | my ($before, $after) = @{$sqlat->whitespace_keyword('SELECT')}; |
464 | |
465 | Returns whitespace to be inserted around a keyword. |