Commit | Line | Data |
96449e8e |
1 | package SQL::Abstract; # see doc at end of file |
2 | |
3 | # LDNOTE : this code is heavy refactoring from original SQLA. |
4 | # Several design decisions will need discussion during |
5 | # the test / diffusion / acceptance phase; those are marked with flag |
6 | # 'LDNOTE' (note by laurent.dami AT free.fr) |
7 | |
8 | use Carp; |
9 | use strict; |
10 | use warnings; |
312d830b |
11 | use List::Util (); |
12 | use Scalar::Util (); |
96449e8e |
13 | |
14 | #====================================================================== |
15 | # GLOBALS |
16 | #====================================================================== |
17 | |
955e77ca |
18 | our $VERSION = '1.65_02'; |
7479e27e |
19 | |
22f1a437 |
20 | # This would confuse some packagers |
c520207b |
21 | $VERSION = eval $VERSION if $VERSION =~ /_/; # numify for warning-free dev releases |
96449e8e |
22 | |
23 | our $AUTOLOAD; |
24 | |
25 | # special operators (-in, -between). May be extended/overridden by user. |
26 | # See section WHERE: BUILTIN SPECIAL OPERATORS below for implementation |
27 | my @BUILTIN_SPECIAL_OPS = ( |
3a2e1a5e |
28 | {regex => qr/^(not )?between$/i, handler => '_where_field_BETWEEN'}, |
29 | {regex => qr/^(not )?in$/i, handler => '_where_field_IN'}, |
96449e8e |
30 | ); |
31 | |
97a920ef |
32 | # unaryish operators - key maps to handler |
59f23b3d |
33 | my @BUILTIN_UNARY_OPS = ( |
a47b433a |
34 | # the digits are backcompat stuff |
35 | { regex => qr/^and (?: \s? \d+ )? $/xi, handler => '_where_op_ANDOR' }, |
36 | { regex => qr/^or (?: \s? \d+ )? $/xi, handler => '_where_op_ANDOR' }, |
37 | { regex => qr/^nest (?: \s? \d+ )? $/xi, handler => '_where_op_NEST' }, |
38 | { regex => qr/^ (?: not \s )? bool $/xi, handler => '_where_op_BOOL' }, |
59f23b3d |
39 | ); |
97a920ef |
40 | |
96449e8e |
41 | #====================================================================== |
42 | # DEBUGGING AND ERROR REPORTING |
43 | #====================================================================== |
44 | |
45 | sub _debug { |
46 | return unless $_[0]->{debug}; shift; # a little faster |
47 | my $func = (caller(1))[3]; |
48 | warn "[$func] ", @_, "\n"; |
49 | } |
50 | |
51 | sub belch (@) { |
52 | my($func) = (caller(1))[3]; |
53 | carp "[$func] Warning: ", @_; |
54 | } |
55 | |
56 | sub puke (@) { |
57 | my($func) = (caller(1))[3]; |
58 | croak "[$func] Fatal: ", @_; |
59 | } |
60 | |
61 | |
62 | #====================================================================== |
63 | # NEW |
64 | #====================================================================== |
65 | |
66 | sub new { |
67 | my $self = shift; |
68 | my $class = ref($self) || $self; |
69 | my %opt = (ref $_[0] eq 'HASH') ? %{$_[0]} : @_; |
70 | |
71 | # choose our case by keeping an option around |
72 | delete $opt{case} if $opt{case} && $opt{case} ne 'lower'; |
73 | |
74 | # default logic for interpreting arrayrefs |
ef559da3 |
75 | $opt{logic} = $opt{logic} ? uc $opt{logic} : 'OR'; |
96449e8e |
76 | |
77 | # how to return bind vars |
78 | # LDNOTE: changed nwiger code : why this 'delete' ?? |
79 | # $opt{bindtype} ||= delete($opt{bind_type}) || 'normal'; |
80 | $opt{bindtype} ||= 'normal'; |
81 | |
82 | # default comparison is "=", but can be overridden |
83 | $opt{cmp} ||= '='; |
84 | |
85 | # try to recognize which are the 'equality' and 'unequality' ops |
86 | # (temporary quickfix, should go through a more seasoned API) |
2281c758 |
87 | $opt{equality_op} = qr/^(\Q$opt{cmp}\E|is|(is\s+)?like)$/i; |
88 | $opt{inequality_op} = qr/^(!=|<>|(is\s+)?not(\s+like)?)$/i; |
96449e8e |
89 | |
90 | # SQL booleans |
91 | $opt{sqltrue} ||= '1=1'; |
92 | $opt{sqlfalse} ||= '0=1'; |
93 | |
9d48860e |
94 | # special operators |
96449e8e |
95 | $opt{special_ops} ||= []; |
96 | push @{$opt{special_ops}}, @BUILTIN_SPECIAL_OPS; |
97 | |
9d48860e |
98 | # unary operators |
59f23b3d |
99 | $opt{unary_ops} ||= []; |
100 | push @{$opt{unary_ops}}, @BUILTIN_UNARY_OPS; |
101 | |
96449e8e |
102 | return bless \%opt, $class; |
103 | } |
104 | |
105 | |
106 | |
107 | #====================================================================== |
108 | # INSERT methods |
109 | #====================================================================== |
110 | |
111 | sub insert { |
02288357 |
112 | my $self = shift; |
113 | my $table = $self->_table(shift); |
114 | my $data = shift || return; |
115 | my $options = shift; |
96449e8e |
116 | |
117 | my $method = $self->_METHOD_FOR_refkind("_insert", $data); |
02288357 |
118 | my ($sql, @bind) = $self->$method($data); |
96449e8e |
119 | $sql = join " ", $self->_sqlcase('insert into'), $table, $sql; |
02288357 |
120 | |
6b1fe79d |
121 | if (my $ret = $options->{returning}) { |
122 | $sql .= $self->_insert_returning ($ret); |
02288357 |
123 | } |
124 | |
96449e8e |
125 | return wantarray ? ($sql, @bind) : $sql; |
126 | } |
127 | |
6b1fe79d |
128 | sub _insert_returning { |
129 | my ($self, $fields) = @_; |
130 | |
131 | my $f = $self->_SWITCH_refkind($fields, { |
132 | ARRAYREF => sub {join ', ', map { $self->_quote($_) } @$fields;}, |
133 | SCALAR => sub {$self->_quote($fields)}, |
134 | SCALARREF => sub {$$fields}, |
135 | }); |
136 | return join (' ', $self->_sqlcase(' returning'), $f); |
137 | } |
138 | |
96449e8e |
139 | sub _insert_HASHREF { # explicit list of fields and then values |
140 | my ($self, $data) = @_; |
141 | |
142 | my @fields = sort keys %$data; |
143 | |
fe3ae272 |
144 | my ($sql, @bind) = $self->_insert_values($data); |
96449e8e |
145 | |
146 | # assemble SQL |
147 | $_ = $self->_quote($_) foreach @fields; |
148 | $sql = "( ".join(", ", @fields).") ".$sql; |
149 | |
150 | return ($sql, @bind); |
151 | } |
152 | |
153 | sub _insert_ARRAYREF { # just generate values(?,?) part (no list of fields) |
154 | my ($self, $data) = @_; |
155 | |
156 | # no names (arrayref) so can't generate bindtype |
157 | $self->{bindtype} ne 'columns' |
158 | or belch "can't do 'columns' bindtype when called with arrayref"; |
159 | |
fe3ae272 |
160 | # fold the list of values into a hash of column name - value pairs |
161 | # (where the column names are artificially generated, and their |
162 | # lexicographical ordering keep the ordering of the original list) |
163 | my $i = "a"; # incremented values will be in lexicographical order |
164 | my $data_in_hash = { map { ($i++ => $_) } @$data }; |
165 | |
166 | return $self->_insert_values($data_in_hash); |
167 | } |
168 | |
169 | sub _insert_ARRAYREFREF { # literal SQL with bind |
170 | my ($self, $data) = @_; |
171 | |
172 | my ($sql, @bind) = @${$data}; |
173 | $self->_assert_bindval_matches_bindtype(@bind); |
174 | |
175 | return ($sql, @bind); |
176 | } |
177 | |
178 | |
179 | sub _insert_SCALARREF { # literal SQL without bind |
180 | my ($self, $data) = @_; |
181 | |
182 | return ($$data); |
183 | } |
184 | |
185 | sub _insert_values { |
186 | my ($self, $data) = @_; |
187 | |
96449e8e |
188 | my (@values, @all_bind); |
fe3ae272 |
189 | foreach my $column (sort keys %$data) { |
190 | my $v = $data->{$column}; |
96449e8e |
191 | |
192 | $self->_SWITCH_refkind($v, { |
193 | |
9d48860e |
194 | ARRAYREF => sub { |
96449e8e |
195 | if ($self->{array_datatypes}) { # if array datatype are activated |
196 | push @values, '?'; |
fe3ae272 |
197 | push @all_bind, $self->_bindtype($column, $v); |
96449e8e |
198 | } |
199 | else { # else literal SQL with bind |
200 | my ($sql, @bind) = @$v; |
fe3ae272 |
201 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
202 | push @values, $sql; |
203 | push @all_bind, @bind; |
204 | } |
205 | }, |
206 | |
207 | ARRAYREFREF => sub { # literal SQL with bind |
208 | my ($sql, @bind) = @${$v}; |
fe3ae272 |
209 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
210 | push @values, $sql; |
211 | push @all_bind, @bind; |
212 | }, |
213 | |
9d48860e |
214 | # THINK : anything useful to do with a HASHREF ? |
5db47f9f |
215 | HASHREF => sub { # (nothing, but old SQLA passed it through) |
216 | #TODO in SQLA >= 2.0 it will die instead |
217 | belch "HASH ref as bind value in insert is not supported"; |
218 | push @values, '?'; |
fe3ae272 |
219 | push @all_bind, $self->_bindtype($column, $v); |
5db47f9f |
220 | }, |
96449e8e |
221 | |
222 | SCALARREF => sub { # literal SQL without bind |
223 | push @values, $$v; |
224 | }, |
225 | |
226 | SCALAR_or_UNDEF => sub { |
227 | push @values, '?'; |
fe3ae272 |
228 | push @all_bind, $self->_bindtype($column, $v); |
96449e8e |
229 | }, |
230 | |
231 | }); |
232 | |
233 | } |
234 | |
235 | my $sql = $self->_sqlcase('values')." ( ".join(", ", @values)." )"; |
236 | return ($sql, @all_bind); |
237 | } |
238 | |
239 | |
96449e8e |
240 | |
241 | #====================================================================== |
242 | # UPDATE methods |
243 | #====================================================================== |
244 | |
245 | |
246 | sub update { |
247 | my $self = shift; |
248 | my $table = $self->_table(shift); |
249 | my $data = shift || return; |
250 | my $where = shift; |
251 | |
252 | # first build the 'SET' part of the sql statement |
253 | my (@set, @all_bind); |
254 | puke "Unsupported data type specified to \$sql->update" |
255 | unless ref $data eq 'HASH'; |
256 | |
257 | for my $k (sort keys %$data) { |
258 | my $v = $data->{$k}; |
259 | my $r = ref $v; |
260 | my $label = $self->_quote($k); |
261 | |
262 | $self->_SWITCH_refkind($v, { |
9d48860e |
263 | ARRAYREF => sub { |
96449e8e |
264 | if ($self->{array_datatypes}) { # array datatype |
265 | push @set, "$label = ?"; |
266 | push @all_bind, $self->_bindtype($k, $v); |
267 | } |
268 | else { # literal SQL with bind |
269 | my ($sql, @bind) = @$v; |
fe3ae272 |
270 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
271 | push @set, "$label = $sql"; |
fe3ae272 |
272 | push @all_bind, @bind; |
96449e8e |
273 | } |
274 | }, |
275 | ARRAYREFREF => sub { # literal SQL with bind |
276 | my ($sql, @bind) = @${$v}; |
fe3ae272 |
277 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
278 | push @set, "$label = $sql"; |
fe3ae272 |
279 | push @all_bind, @bind; |
96449e8e |
280 | }, |
281 | SCALARREF => sub { # literal SQL without bind |
282 | push @set, "$label = $$v"; |
283 | }, |
284 | SCALAR_or_UNDEF => sub { |
285 | push @set, "$label = ?"; |
286 | push @all_bind, $self->_bindtype($k, $v); |
287 | }, |
288 | }); |
289 | } |
290 | |
291 | # generate sql |
292 | my $sql = $self->_sqlcase('update') . " $table " . $self->_sqlcase('set ') |
293 | . join ', ', @set; |
294 | |
295 | if ($where) { |
296 | my($where_sql, @where_bind) = $self->where($where); |
297 | $sql .= $where_sql; |
298 | push @all_bind, @where_bind; |
299 | } |
300 | |
301 | return wantarray ? ($sql, @all_bind) : $sql; |
302 | } |
303 | |
304 | |
305 | |
306 | |
307 | #====================================================================== |
308 | # SELECT |
309 | #====================================================================== |
310 | |
311 | |
312 | sub select { |
313 | my $self = shift; |
314 | my $table = $self->_table(shift); |
315 | my $fields = shift || '*'; |
316 | my $where = shift; |
317 | my $order = shift; |
318 | |
319 | my($where_sql, @bind) = $self->where($where, $order); |
320 | |
321 | my $f = (ref $fields eq 'ARRAY') ? join ', ', map { $self->_quote($_) } @$fields |
322 | : $fields; |
9d48860e |
323 | my $sql = join(' ', $self->_sqlcase('select'), $f, |
96449e8e |
324 | $self->_sqlcase('from'), $table) |
325 | . $where_sql; |
326 | |
9d48860e |
327 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
328 | } |
329 | |
330 | #====================================================================== |
331 | # DELETE |
332 | #====================================================================== |
333 | |
334 | |
335 | sub delete { |
336 | my $self = shift; |
337 | my $table = $self->_table(shift); |
338 | my $where = shift; |
339 | |
340 | |
341 | my($where_sql, @bind) = $self->where($where); |
342 | my $sql = $self->_sqlcase('delete from') . " $table" . $where_sql; |
343 | |
9d48860e |
344 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
345 | } |
346 | |
347 | |
348 | #====================================================================== |
349 | # WHERE: entry point |
350 | #====================================================================== |
351 | |
352 | |
353 | |
354 | # Finally, a separate routine just to handle WHERE clauses |
355 | sub where { |
356 | my ($self, $where, $order) = @_; |
357 | |
358 | # where ? |
359 | my ($sql, @bind) = $self->_recurse_where($where); |
360 | $sql = $sql ? $self->_sqlcase(' where ') . "( $sql )" : ''; |
361 | |
362 | # order by? |
363 | if ($order) { |
364 | $sql .= $self->_order_by($order); |
365 | } |
366 | |
9d48860e |
367 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
368 | } |
369 | |
370 | |
371 | sub _recurse_where { |
372 | my ($self, $where, $logic) = @_; |
373 | |
374 | # dispatch on appropriate method according to refkind of $where |
375 | my $method = $self->_METHOD_FOR_refkind("_where", $where); |
311b2151 |
376 | |
9d48860e |
377 | my ($sql, @bind) = $self->$method($where, $logic); |
311b2151 |
378 | |
9d48860e |
379 | # DBIx::Class directly calls _recurse_where in scalar context, so |
311b2151 |
380 | # we must implement it, even if not in the official API |
9d48860e |
381 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
382 | } |
383 | |
384 | |
385 | |
386 | #====================================================================== |
387 | # WHERE: top-level ARRAYREF |
388 | #====================================================================== |
389 | |
390 | |
391 | sub _where_ARRAYREF { |
5e1d09d5 |
392 | my ($self, $where, $logic) = @_; |
96449e8e |
393 | |
5e1d09d5 |
394 | $logic = uc($logic || $self->{logic}); |
96449e8e |
395 | $logic eq 'AND' or $logic eq 'OR' or puke "unknown logic: $logic"; |
396 | |
397 | my @clauses = @$where; |
398 | |
96449e8e |
399 | my (@sql_clauses, @all_bind); |
96449e8e |
400 | # need to use while() so can shift() for pairs |
9d48860e |
401 | while (my $el = shift @clauses) { |
96449e8e |
402 | |
403 | # switch according to kind of $el and get corresponding ($sql, @bind) |
404 | my ($sql, @bind) = $self->_SWITCH_refkind($el, { |
405 | |
406 | # skip empty elements, otherwise get invalid trailing AND stuff |
407 | ARRAYREF => sub {$self->_recurse_where($el) if @$el}, |
408 | |
474e3335 |
409 | ARRAYREFREF => sub { @{${$el}} if @{${$el}}}, |
410 | |
96449e8e |
411 | HASHREF => sub {$self->_recurse_where($el, 'and') if %$el}, |
412 | # LDNOTE : previous SQLA code for hashrefs was creating a dirty |
413 | # side-effect: the first hashref within an array would change |
414 | # the global logic to 'AND'. So [ {cond1, cond2}, [cond3, cond4] ] |
9d48860e |
415 | # was interpreted as "(cond1 AND cond2) OR (cond3 AND cond4)", |
96449e8e |
416 | # whereas it should be "(cond1 AND cond2) OR (cond3 OR cond4)". |
417 | |
418 | SCALARREF => sub { ($$el); }, |
419 | |
420 | SCALAR => sub {# top-level arrayref with scalars, recurse in pairs |
421 | $self->_recurse_where({$el => shift(@clauses)})}, |
422 | |
423 | UNDEF => sub {puke "not supported : UNDEF in arrayref" }, |
424 | }); |
425 | |
4b7b6026 |
426 | if ($sql) { |
427 | push @sql_clauses, $sql; |
428 | push @all_bind, @bind; |
429 | } |
96449e8e |
430 | } |
431 | |
432 | return $self->_join_sql_clauses($logic, \@sql_clauses, \@all_bind); |
433 | } |
434 | |
474e3335 |
435 | #====================================================================== |
436 | # WHERE: top-level ARRAYREFREF |
437 | #====================================================================== |
96449e8e |
438 | |
474e3335 |
439 | sub _where_ARRAYREFREF { |
440 | my ($self, $where) = @_; |
441 | my ($sql, @bind) = @{${$where}}; |
442 | |
443 | return ($sql, @bind); |
444 | } |
96449e8e |
445 | |
446 | #====================================================================== |
447 | # WHERE: top-level HASHREF |
448 | #====================================================================== |
449 | |
450 | sub _where_HASHREF { |
451 | my ($self, $where) = @_; |
452 | my (@sql_clauses, @all_bind); |
453 | |
2281c758 |
454 | for my $k (sort keys %$where) { |
96449e8e |
455 | my $v = $where->{$k}; |
456 | |
2281c758 |
457 | # ($k => $v) is either a special unary op or a regular hashpair |
458 | my ($sql, @bind) = do { |
459 | if ($k =~ /^-./) { |
460 | # put the operator in canonical form |
461 | my $op = $k; |
462 | $op =~ s/^-//; # remove initial dash |
463 | $op =~ s/[_\t ]+/ /g; # underscores and whitespace become single spaces |
464 | $op =~ s/^\s+|\s+$//g;# remove leading/trailing space |
465 | |
466 | $self->_debug("Unary OP(-$op) within hashref, recursing..."); |
467 | |
312d830b |
468 | my $op_entry = List::Util::first {$op =~ $_->{regex}} @{$self->{unary_ops}}; |
2281c758 |
469 | if (my $handler = $op_entry->{handler}) { |
470 | if (not ref $handler) { |
471 | if ($op =~ s/\s?\d+$//) { |
472 | belch 'Use of [and|or|nest]_N modifiers is deprecated and will be removed in SQLA v2.0. ' |
473 | . "You probably wanted ...-and => [ -$op => COND1, -$op => COND2 ... ]"; |
474 | } |
475 | $self->$handler ($op, $v); |
476 | } |
477 | elsif (ref $handler eq 'CODE') { |
478 | $handler->($self, $op, $v); |
479 | } |
480 | else { |
481 | puke "Illegal handler for operator $k - expecting a method name or a coderef"; |
482 | } |
483 | } |
484 | else { |
485 | $self->debug("Generic unary OP: $k - recursing as function"); |
953d164e |
486 | my ($sql, @bind) = $self->_where_func_generic ($op, $v); |
487 | $sql = "($sql)" unless $self->{_nested_func_lhs} eq $k; # top level vs nested |
488 | ($sql, @bind); |
2281c758 |
489 | } |
490 | } |
491 | else { |
492 | my $method = $self->_METHOD_FOR_refkind("_where_hashpair", $v); |
493 | $self->$method($k, $v); |
494 | } |
495 | }; |
96449e8e |
496 | |
497 | push @sql_clauses, $sql; |
498 | push @all_bind, @bind; |
499 | } |
500 | |
501 | return $self->_join_sql_clauses('and', \@sql_clauses, \@all_bind); |
502 | } |
503 | |
2281c758 |
504 | sub _where_func_generic { |
505 | my ($self, $op, $rhs) = @_; |
96449e8e |
506 | |
2281c758 |
507 | my ($sql, @bind) = $self->_SWITCH_refkind ($rhs, { |
508 | SCALAR => sub { |
a7661cfc |
509 | puke "Illegal use of top-level '$op'" |
510 | unless $self->{_nested_func_lhs}; |
511 | |
512 | return ( |
513 | $self->_convert('?'), |
514 | $self->_bindtype($self->{_nested_func_lhs}, $rhs) |
515 | ); |
2281c758 |
516 | }, |
517 | FALLBACK => sub { |
518 | $self->_recurse_where ($rhs) |
519 | }, |
520 | }); |
96449e8e |
521 | |
953d164e |
522 | $sql = sprintf ('%s %s', |
2281c758 |
523 | $self->_sqlcase($op), |
953d164e |
524 | $sql, |
2281c758 |
525 | ); |
96449e8e |
526 | |
2281c758 |
527 | return ($sql, @bind); |
97a920ef |
528 | } |
529 | |
530 | sub _where_op_ANDOR { |
2281c758 |
531 | my ($self, $op, $v) = @_; |
97a920ef |
532 | |
533 | $self->_SWITCH_refkind($v, { |
534 | ARRAYREF => sub { |
535 | return $self->_where_ARRAYREF($v, $op); |
536 | }, |
537 | |
538 | HASHREF => sub { |
59f23b3d |
539 | return ( $op =~ /^or/i ) |
97a920ef |
540 | ? $self->_where_ARRAYREF( [ map { $_ => $v->{$_} } ( sort keys %$v ) ], $op ) |
541 | : $self->_where_HASHREF($v); |
542 | }, |
543 | |
9d48860e |
544 | SCALARREF => sub { |
97a920ef |
545 | puke "-$op => \\\$scalar not supported, use -nest => ..."; |
546 | }, |
547 | |
548 | ARRAYREFREF => sub { |
549 | puke "-$op => \\[..] not supported, use -nest => ..."; |
550 | }, |
551 | |
552 | SCALAR => sub { # permissively interpreted as SQL |
553 | puke "-$op => 'scalar' not supported, use -nest => \\'scalar'"; |
554 | }, |
555 | |
556 | UNDEF => sub { |
557 | puke "-$op => undef not supported"; |
558 | }, |
559 | }); |
560 | } |
561 | |
562 | sub _where_op_NEST { |
9d48860e |
563 | my ($self, $op, $v) = @_; |
97a920ef |
564 | |
96449e8e |
565 | $self->_SWITCH_refkind($v, { |
566 | |
96449e8e |
567 | SCALAR => sub { # permissively interpreted as SQL |
01a01e57 |
568 | belch "literal SQL should be -nest => \\'scalar' " |
569 | . "instead of -nest => 'scalar' "; |
9d48860e |
570 | return ($v); |
96449e8e |
571 | }, |
572 | |
573 | UNDEF => sub { |
574 | puke "-$op => undef not supported"; |
575 | }, |
e9501094 |
576 | |
577 | FALLBACK => sub { |
578 | $self->_recurse_where ($v); |
579 | }, |
580 | |
96449e8e |
581 | }); |
582 | } |
583 | |
584 | |
97a920ef |
585 | sub _where_op_BOOL { |
9d48860e |
586 | my ($self, $op, $v) = @_; |
97a920ef |
587 | |
9d48860e |
588 | my ( $prefix, $suffix ) = ( $op =~ /\bnot\b/i ) |
589 | ? ( '(NOT ', ')' ) |
ef03f1bc |
590 | : ( '', '' ); |
ef03f1bc |
591 | |
2281c758 |
592 | my ($sql, @bind) = do { |
593 | $self->_SWITCH_refkind($v, { |
594 | SCALAR => sub { # interpreted as SQL column |
595 | $self->_convert($self->_quote($v)); |
596 | }, |
ef03f1bc |
597 | |
2281c758 |
598 | UNDEF => sub { |
599 | puke "-$op => undef not supported"; |
600 | }, |
97a920ef |
601 | |
2281c758 |
602 | FALLBACK => sub { |
603 | $self->_recurse_where ($v); |
604 | }, |
605 | }); |
606 | }; |
ef03f1bc |
607 | |
2281c758 |
608 | return ( |
609 | join ('', $prefix, $sql, $suffix), |
610 | @bind, |
611 | ); |
97a920ef |
612 | } |
613 | |
614 | |
96449e8e |
615 | sub _where_hashpair_ARRAYREF { |
616 | my ($self, $k, $v) = @_; |
617 | |
618 | if( @$v ) { |
619 | my @v = @$v; # need copy because of shift below |
620 | $self->_debug("ARRAY($k) means distribute over elements"); |
621 | |
622 | # put apart first element if it is an operator (-and, -or) |
e3cecb45 |
623 | my $op = ( |
624 | (defined $v[0] && $v[0] =~ /^ - (?: AND|OR ) $/ix) |
625 | ? shift @v |
626 | : '' |
04d940de |
627 | ); |
96449e8e |
628 | my @distributed = map { {$k => $_} } @v; |
04d940de |
629 | |
630 | if ($op) { |
631 | $self->_debug("OP($op) reinjected into the distributed array"); |
632 | unshift @distributed, $op; |
633 | } |
634 | |
f67591bf |
635 | my $logic = $op ? substr($op, 1) : ''; |
96449e8e |
636 | |
f67591bf |
637 | return $self->_recurse_where(\@distributed, $logic); |
9d48860e |
638 | } |
96449e8e |
639 | else { |
640 | # LDNOTE : not sure of this one. What does "distribute over nothing" mean? |
641 | $self->_debug("empty ARRAY($k) means 0=1"); |
642 | return ($self->{sqlfalse}); |
643 | } |
644 | } |
645 | |
646 | sub _where_hashpair_HASHREF { |
eb49170d |
647 | my ($self, $k, $v, $logic) = @_; |
648 | $logic ||= 'and'; |
96449e8e |
649 | |
a7661cfc |
650 | local $self->{_nested_func_lhs} = $self->{_nested_func_lhs}; |
a7661cfc |
651 | |
eb49170d |
652 | my ($all_sql, @all_bind); |
96449e8e |
653 | |
a47b433a |
654 | for my $orig_op (sort keys %$v) { |
655 | my $val = $v->{$orig_op}; |
96449e8e |
656 | |
657 | # put the operator in canonical form |
a47b433a |
658 | my $op = $orig_op; |
659 | $op =~ s/^-//; # remove initial dash |
660 | $op =~ s/[_\t ]+/ /g; # underscores and whitespace become single spaces |
661 | $op =~ s/^\s+|\s+$//g;# remove leading/trailing space |
96449e8e |
662 | |
663 | my ($sql, @bind); |
664 | |
2281c758 |
665 | # CASE: col-value logic modifiers |
666 | if ( $orig_op =~ /^ \- (and|or) $/xi ) { |
667 | ($sql, @bind) = $self->_where_hashpair_HASHREF($k, $val, $1); |
668 | } |
96449e8e |
669 | # CASE: special operators like -in or -between |
312d830b |
670 | elsif ( my $special_op = List::Util::first {$op =~ $_->{regex}} @{$self->{special_ops}} ) { |
3a2e1a5e |
671 | my $handler = $special_op->{handler}; |
672 | if (! $handler) { |
a47b433a |
673 | puke "No handler supplied for special operator $orig_op"; |
3a2e1a5e |
674 | } |
675 | elsif (not ref $handler) { |
676 | ($sql, @bind) = $self->$handler ($k, $op, $val); |
677 | } |
678 | elsif (ref $handler eq 'CODE') { |
679 | ($sql, @bind) = $handler->($self, $k, $op, $val); |
680 | } |
681 | else { |
a47b433a |
682 | puke "Illegal handler for special operator $orig_op - expecting a method name or a coderef"; |
3a2e1a5e |
683 | } |
96449e8e |
684 | } |
96449e8e |
685 | else { |
cf838930 |
686 | $self->_SWITCH_refkind($val, { |
687 | |
688 | ARRAYREF => sub { # CASE: col => {op => \@vals} |
689 | ($sql, @bind) = $self->_where_field_op_ARRAYREF($k, $op, $val); |
690 | }, |
691 | |
fe3ae272 |
692 | ARRAYREFREF => sub { # CASE: col => {op => \[$sql, @bind]} (literal SQL with bind) |
b3be7bd0 |
693 | my ($sub_sql, @sub_bind) = @$$val; |
fe3ae272 |
694 | $self->_assert_bindval_matches_bindtype(@sub_bind); |
b3be7bd0 |
695 | $sql = join ' ', $self->_convert($self->_quote($k)), |
696 | $self->_sqlcase($op), |
697 | $sub_sql; |
fe3ae272 |
698 | @bind = @sub_bind; |
b3be7bd0 |
699 | }, |
700 | |
cf838930 |
701 | UNDEF => sub { # CASE: col => {op => undef} : sql "IS (NOT)? NULL" |
702 | my $is = ($op =~ $self->{equality_op}) ? 'is' : |
703 | ($op =~ $self->{inequality_op}) ? 'is not' : |
a47b433a |
704 | puke "unexpected operator '$orig_op' with undef operand"; |
cf838930 |
705 | $sql = $self->_quote($k) . $self->_sqlcase(" $is null"); |
706 | }, |
a47b433a |
707 | |
2281c758 |
708 | FALLBACK => sub { # CASE: col => {op/func => $stuff} |
07936978 |
709 | |
953d164e |
710 | # retain for proper column type bind |
711 | $self->{_nested_func_lhs} ||= $k; |
07936978 |
712 | |
2281c758 |
713 | ($sql, @bind) = $self->_where_func_generic ($op, $val); |
953d164e |
714 | |
715 | $sql = join (' ', |
716 | $self->_convert($self->_quote($k)), |
717 | $self->{_nested_func_lhs} eq $k ? $sql : "($sql)", # top level vs nested |
718 | ); |
cf838930 |
719 | }, |
720 | }); |
96449e8e |
721 | } |
722 | |
eb49170d |
723 | ($all_sql) = (defined $all_sql and $all_sql) ? $self->_join_sql_clauses($logic, [$all_sql, $sql], []) : $sql; |
96449e8e |
724 | push @all_bind, @bind; |
725 | } |
eb49170d |
726 | return ($all_sql, @all_bind); |
96449e8e |
727 | } |
728 | |
729 | |
730 | |
731 | sub _where_field_op_ARRAYREF { |
732 | my ($self, $k, $op, $vals) = @_; |
733 | |
ce261791 |
734 | my @vals = @$vals; #always work on a copy |
735 | |
736 | if(@vals) { |
bd6a65ca |
737 | $self->_debug(sprintf '%s means multiple elements: [ %s ]', |
738 | $vals, |
739 | join (', ', map { defined $_ ? "'$_'" : 'NULL' } @vals ), |
740 | ); |
96449e8e |
741 | |
4030915f |
742 | # see if the first element is an -and/-or op |
743 | my $logic; |
bd6a65ca |
744 | if (defined $vals[0] && $vals[0] =~ /^ - ( AND|OR ) $/ix) { |
4030915f |
745 | $logic = uc $1; |
ce261791 |
746 | shift @vals; |
4030915f |
747 | } |
748 | |
ce261791 |
749 | # distribute $op over each remaining member of @vals, append logic if exists |
750 | return $self->_recurse_where([map { {$k => {$op, $_}} } @vals], $logic); |
4030915f |
751 | |
9d48860e |
752 | # LDNOTE : had planned to change the distribution logic when |
753 | # $op =~ $self->{inequality_op}, because of Morgan laws : |
96449e8e |
754 | # with {field => {'!=' => [22, 33]}}, it would be ridiculous to generate |
9d48860e |
755 | # WHERE field != 22 OR field != 33 : the user probably means |
96449e8e |
756 | # WHERE field != 22 AND field != 33. |
4030915f |
757 | # To do this, replace the above to roughly : |
f2d5020d |
758 | # my $logic = ($op =~ $self->{inequality_op}) ? 'AND' : 'OR'; |
ce261791 |
759 | # return $self->_recurse_where([map { {$k => {$op, $_}} } @vals], $logic); |
96449e8e |
760 | |
9d48860e |
761 | } |
96449e8e |
762 | else { |
9d48860e |
763 | # try to DWIM on equality operators |
96449e8e |
764 | # LDNOTE : not 100% sure this is the correct thing to do ... |
765 | return ($self->{sqlfalse}) if $op =~ $self->{equality_op}; |
766 | return ($self->{sqltrue}) if $op =~ $self->{inequality_op}; |
767 | |
768 | # otherwise |
769 | puke "operator '$op' applied on an empty array (field '$k')"; |
770 | } |
771 | } |
772 | |
773 | |
774 | sub _where_hashpair_SCALARREF { |
775 | my ($self, $k, $v) = @_; |
776 | $self->_debug("SCALAR($k) means literal SQL: $$v"); |
777 | my $sql = $self->_quote($k) . " " . $$v; |
778 | return ($sql); |
779 | } |
780 | |
fe3ae272 |
781 | # literal SQL with bind |
96449e8e |
782 | sub _where_hashpair_ARRAYREFREF { |
783 | my ($self, $k, $v) = @_; |
784 | $self->_debug("REF($k) means literal SQL: @${$v}"); |
785 | my ($sql, @bind) = @${$v}; |
fe3ae272 |
786 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
787 | $sql = $self->_quote($k) . " " . $sql; |
96449e8e |
788 | return ($sql, @bind ); |
789 | } |
790 | |
fe3ae272 |
791 | # literal SQL without bind |
96449e8e |
792 | sub _where_hashpair_SCALAR { |
793 | my ($self, $k, $v) = @_; |
794 | $self->_debug("NOREF($k) means simple key=val: $k $self->{cmp} $v"); |
9d48860e |
795 | my $sql = join ' ', $self->_convert($self->_quote($k)), |
796 | $self->_sqlcase($self->{cmp}), |
96449e8e |
797 | $self->_convert('?'); |
798 | my @bind = $self->_bindtype($k, $v); |
799 | return ( $sql, @bind); |
800 | } |
801 | |
802 | |
803 | sub _where_hashpair_UNDEF { |
804 | my ($self, $k, $v) = @_; |
805 | $self->_debug("UNDEF($k) means IS NULL"); |
806 | my $sql = $self->_quote($k) . $self->_sqlcase(' is null'); |
807 | return ($sql); |
808 | } |
809 | |
810 | #====================================================================== |
811 | # WHERE: TOP-LEVEL OTHERS (SCALARREF, SCALAR, UNDEF) |
812 | #====================================================================== |
813 | |
814 | |
815 | sub _where_SCALARREF { |
816 | my ($self, $where) = @_; |
817 | |
818 | # literal sql |
819 | $self->_debug("SCALAR(*top) means literal SQL: $$where"); |
820 | return ($$where); |
821 | } |
822 | |
823 | |
824 | sub _where_SCALAR { |
825 | my ($self, $where) = @_; |
826 | |
827 | # literal sql |
828 | $self->_debug("NOREF(*top) means literal SQL: $where"); |
829 | return ($where); |
830 | } |
831 | |
832 | |
833 | sub _where_UNDEF { |
834 | my ($self) = @_; |
835 | return (); |
836 | } |
837 | |
838 | |
839 | #====================================================================== |
840 | # WHERE: BUILTIN SPECIAL OPERATORS (-in, -between) |
841 | #====================================================================== |
842 | |
843 | |
844 | sub _where_field_BETWEEN { |
845 | my ($self, $k, $op, $vals) = @_; |
846 | |
4d8b3dc4 |
847 | my ($label, $and, $placeholder); |
cf02fc47 |
848 | $label = $self->_convert($self->_quote($k)); |
849 | $and = ' ' . $self->_sqlcase('and') . ' '; |
850 | $placeholder = $self->_convert('?'); |
96449e8e |
851 | $op = $self->_sqlcase($op); |
852 | |
4d8b3dc4 |
853 | my ($clause, @bind) = $self->_SWITCH_refkind($vals, { |
854 | ARRAYREFREF => sub { |
855 | return @$$vals; |
856 | }, |
857 | SCALARREF => sub { |
858 | return $$vals; |
859 | }, |
860 | ARRAYREF => sub { |
861 | puke "special op 'between' accepts an arrayref with exactly two values" |
862 | if @$vals != 2; |
863 | |
864 | my (@all_sql, @all_bind); |
865 | foreach my $val (@$vals) { |
866 | my ($sql, @bind) = $self->_SWITCH_refkind($val, { |
867 | SCALAR => sub { |
868 | return ($placeholder, ($val)); |
869 | }, |
870 | SCALARREF => sub { |
871 | return ($self->_convert($$val), ()); |
872 | }, |
873 | ARRAYREFREF => sub { |
874 | my ($sql, @bind) = @$$val; |
875 | return ($self->_convert($sql), @bind); |
876 | }, |
877 | }); |
878 | push @all_sql, $sql; |
879 | push @all_bind, @bind; |
880 | } |
881 | |
882 | return ( |
883 | (join $and, @all_sql), |
884 | $self->_bindtype($k, @all_bind), |
885 | ); |
886 | }, |
887 | FALLBACK => sub { |
888 | puke "special op 'between' accepts an arrayref with two values, or a single literal scalarref/arrayref-ref"; |
889 | }, |
890 | }); |
cf02fc47 |
891 | |
cf02fc47 |
892 | my $sql = "( $label $op $clause )"; |
96449e8e |
893 | return ($sql, @bind) |
894 | } |
895 | |
896 | |
897 | sub _where_field_IN { |
898 | my ($self, $k, $op, $vals) = @_; |
899 | |
900 | # backwards compatibility : if scalar, force into an arrayref |
901 | $vals = [$vals] if defined $vals && ! ref $vals; |
902 | |
96449e8e |
903 | my ($label) = $self->_convert($self->_quote($k)); |
904 | my ($placeholder) = $self->_convert('?'); |
96449e8e |
905 | $op = $self->_sqlcase($op); |
906 | |
8a0d798a |
907 | my ($sql, @bind) = $self->_SWITCH_refkind($vals, { |
908 | ARRAYREF => sub { # list of choices |
909 | if (@$vals) { # nonempty list |
910 | my $placeholders = join ", ", (($placeholder) x @$vals); |
911 | my $sql = "$label $op ( $placeholders )"; |
912 | my @bind = $self->_bindtype($k, @$vals); |
96449e8e |
913 | |
8a0d798a |
914 | return ($sql, @bind); |
915 | } |
916 | else { # empty list : some databases won't understand "IN ()", so DWIM |
917 | my $sql = ($op =~ /\bnot\b/i) ? $self->{sqltrue} : $self->{sqlfalse}; |
918 | return ($sql); |
919 | } |
920 | }, |
921 | |
4a1f01a3 |
922 | SCALARREF => sub { # literal SQL |
923 | my $sql = $self->_open_outer_paren ($$vals); |
924 | return ("$label $op ( $sql )"); |
925 | }, |
8a0d798a |
926 | ARRAYREFREF => sub { # literal SQL with bind |
927 | my ($sql, @bind) = @$$vals; |
fe3ae272 |
928 | $self->_assert_bindval_matches_bindtype(@bind); |
4a1f01a3 |
929 | $sql = $self->_open_outer_paren ($sql); |
8a0d798a |
930 | return ("$label $op ( $sql )", @bind); |
931 | }, |
932 | |
933 | FALLBACK => sub { |
4a1f01a3 |
934 | puke "special op 'in' requires an arrayref (or scalarref/arrayref-ref)"; |
8a0d798a |
935 | }, |
936 | }); |
937 | |
938 | return ($sql, @bind); |
96449e8e |
939 | } |
940 | |
4a1f01a3 |
941 | # Some databases (SQLite) treat col IN (1, 2) different from |
942 | # col IN ( (1, 2) ). Use this to strip all outer parens while |
943 | # adding them back in the corresponding method |
944 | sub _open_outer_paren { |
945 | my ($self, $sql) = @_; |
171a709f |
946 | $sql = $1 while $sql =~ /^ \s* \( (.*) \) \s* $/xs; |
4a1f01a3 |
947 | return $sql; |
948 | } |
949 | |
96449e8e |
950 | |
96449e8e |
951 | #====================================================================== |
952 | # ORDER BY |
953 | #====================================================================== |
954 | |
955 | sub _order_by { |
956 | my ($self, $arg) = @_; |
957 | |
f267b646 |
958 | my (@sql, @bind); |
959 | for my $c ($self->_order_by_chunks ($arg) ) { |
960 | $self->_SWITCH_refkind ($c, { |
961 | SCALAR => sub { push @sql, $c }, |
962 | ARRAYREF => sub { push @sql, shift @$c; push @bind, @$c }, |
963 | }); |
964 | } |
965 | |
966 | my $sql = @sql |
967 | ? sprintf ('%s %s', |
968 | $self->_sqlcase(' order by'), |
969 | join (', ', @sql) |
970 | ) |
971 | : '' |
972 | ; |
973 | |
974 | return wantarray ? ($sql, @bind) : $sql; |
975 | } |
976 | |
977 | sub _order_by_chunks { |
978 | my ($self, $arg) = @_; |
979 | |
980 | return $self->_SWITCH_refkind($arg, { |
96449e8e |
981 | |
982 | ARRAYREF => sub { |
f267b646 |
983 | map { $self->_order_by_chunks ($_ ) } @$arg; |
96449e8e |
984 | }, |
985 | |
f267b646 |
986 | ARRAYREFREF => sub { [ @$$arg ] }, |
987 | |
96449e8e |
988 | SCALAR => sub {$self->_quote($arg)}, |
f267b646 |
989 | |
990 | UNDEF => sub {return () }, |
991 | |
96449e8e |
992 | SCALARREF => sub {$$arg}, # literal SQL, no quoting |
96449e8e |
993 | |
f267b646 |
994 | HASHREF => sub { |
995 | # get first pair in hash |
996 | my ($key, $val) = each %$arg; |
96449e8e |
997 | |
f267b646 |
998 | return () unless $key; |
96449e8e |
999 | |
f267b646 |
1000 | if ( (keys %$arg) > 1 or not $key =~ /^-(desc|asc)/i ) { |
1001 | puke "hash passed to _order_by must have exactly one key (-desc or -asc)"; |
1002 | } |
96449e8e |
1003 | |
f267b646 |
1004 | my $direction = $1; |
96449e8e |
1005 | |
e9bd3547 |
1006 | my @ret; |
f267b646 |
1007 | for my $c ($self->_order_by_chunks ($val)) { |
e9bd3547 |
1008 | my ($sql, @bind); |
96449e8e |
1009 | |
f267b646 |
1010 | $self->_SWITCH_refkind ($c, { |
1011 | SCALAR => sub { |
e9bd3547 |
1012 | $sql = $c; |
f267b646 |
1013 | }, |
1014 | ARRAYREF => sub { |
e9bd3547 |
1015 | ($sql, @bind) = @$c; |
f267b646 |
1016 | }, |
1017 | }); |
96449e8e |
1018 | |
e9bd3547 |
1019 | $sql = $sql . ' ' . $self->_sqlcase($direction); |
96449e8e |
1020 | |
e9bd3547 |
1021 | push @ret, [ $sql, @bind]; |
1022 | } |
96449e8e |
1023 | |
e9bd3547 |
1024 | return @ret; |
f267b646 |
1025 | }, |
1026 | }); |
96449e8e |
1027 | } |
1028 | |
1029 | |
96449e8e |
1030 | #====================================================================== |
1031 | # DATASOURCE (FOR NOW, JUST PLAIN TABLE OR LIST OF TABLES) |
1032 | #====================================================================== |
1033 | |
1034 | sub _table { |
1035 | my $self = shift; |
1036 | my $from = shift; |
1037 | $self->_SWITCH_refkind($from, { |
1038 | ARRAYREF => sub {join ', ', map { $self->_quote($_) } @$from;}, |
1039 | SCALAR => sub {$self->_quote($from)}, |
1040 | SCALARREF => sub {$$from}, |
1041 | ARRAYREFREF => sub {join ', ', @$from;}, |
1042 | }); |
1043 | } |
1044 | |
1045 | |
1046 | #====================================================================== |
1047 | # UTILITY FUNCTIONS |
1048 | #====================================================================== |
1049 | |
955e77ca |
1050 | # highly optimized, as it's called way too often |
96449e8e |
1051 | sub _quote { |
955e77ca |
1052 | # my ($self, $label) = @_; |
96449e8e |
1053 | |
955e77ca |
1054 | return '' unless defined $_[1]; |
955e77ca |
1055 | return ${$_[1]} if ref($_[1]) eq 'SCALAR'; |
96449e8e |
1056 | |
955e77ca |
1057 | return $_[1] unless $_[0]->{quote_char}; |
96449e8e |
1058 | |
07d7c35c |
1059 | my $qref = ref $_[0]->{quote_char}; |
955e77ca |
1060 | my ($l, $r); |
07d7c35c |
1061 | if (!$qref) { |
1062 | ($l, $r) = ( $_[0]->{quote_char}, $_[0]->{quote_char} ); |
955e77ca |
1063 | } |
07d7c35c |
1064 | elsif ($qref eq 'ARRAY') { |
1065 | ($l, $r) = @{$_[0]->{quote_char}}; |
955e77ca |
1066 | } |
1067 | else { |
1068 | puke "Unsupported quote_char format: $_[0]->{quote_char}"; |
1069 | } |
96449e8e |
1070 | |
07d7c35c |
1071 | # parts containing * are naturally unquoted |
1072 | return join( $_[0]->{name_sep}||'', map |
955e77ca |
1073 | { $_ eq '*' ? $_ : $l . $_ . $r } |
1074 | ( $_[0]->{name_sep} ? split (/\Q$_[0]->{name_sep}\E/, $_[1] ) : $_[1] ) |
1075 | ); |
96449e8e |
1076 | } |
1077 | |
1078 | |
1079 | # Conversion, if applicable |
1080 | sub _convert ($) { |
07d7c35c |
1081 | #my ($self, $arg) = @_; |
96449e8e |
1082 | |
1083 | # LDNOTE : modified the previous implementation below because |
1084 | # it was not consistent : the first "return" is always an array, |
1085 | # the second "return" is context-dependent. Anyway, _convert |
9d48860e |
1086 | # seems always used with just a single argument, so make it a |
96449e8e |
1087 | # scalar function. |
1088 | # return @_ unless $self->{convert}; |
1089 | # my $conv = $self->_sqlcase($self->{convert}); |
1090 | # my @ret = map { $conv.'('.$_.')' } @_; |
1091 | # return wantarray ? @ret : $ret[0]; |
07d7c35c |
1092 | if ($_[0]->{convert}) { |
1093 | return $_[0]->_sqlcase($_[0]->{convert}) .'(' . $_[1] . ')'; |
96449e8e |
1094 | } |
07d7c35c |
1095 | return $_[1]; |
96449e8e |
1096 | } |
1097 | |
1098 | # And bindtype |
1099 | sub _bindtype (@) { |
07d7c35c |
1100 | #my ($self, $col, @vals) = @_; |
96449e8e |
1101 | |
9d48860e |
1102 | #LDNOTE : changed original implementation below because it did not make |
96449e8e |
1103 | # sense when bindtype eq 'columns' and @vals > 1. |
1104 | # return $self->{bindtype} eq 'columns' ? [ $col, @vals ] : @vals; |
1105 | |
07d7c35c |
1106 | # called often - tighten code |
1107 | return $_[0]->{bindtype} eq 'columns' |
1108 | ? map {[$_[1], $_]} @_[2 .. $#_] |
1109 | : @_[2 .. $#_] |
1110 | ; |
96449e8e |
1111 | } |
1112 | |
fe3ae272 |
1113 | # Dies if any element of @bind is not in [colname => value] format |
1114 | # if bindtype is 'columns'. |
1115 | sub _assert_bindval_matches_bindtype { |
1116 | my ($self, @bind) = @_; |
1117 | |
1118 | if ($self->{bindtype} eq 'columns') { |
1119 | foreach my $val (@bind) { |
1120 | if (!defined $val || ref($val) ne 'ARRAY' || @$val != 2) { |
1121 | die "bindtype 'columns' selected, you need to pass: [column_name => bind_value]" |
1122 | } |
1123 | } |
1124 | } |
1125 | } |
1126 | |
96449e8e |
1127 | sub _join_sql_clauses { |
1128 | my ($self, $logic, $clauses_aref, $bind_aref) = @_; |
1129 | |
1130 | if (@$clauses_aref > 1) { |
1131 | my $join = " " . $self->_sqlcase($logic) . " "; |
1132 | my $sql = '( ' . join($join, @$clauses_aref) . ' )'; |
1133 | return ($sql, @$bind_aref); |
1134 | } |
1135 | elsif (@$clauses_aref) { |
1136 | return ($clauses_aref->[0], @$bind_aref); # no parentheses |
1137 | } |
1138 | else { |
1139 | return (); # if no SQL, ignore @$bind_aref |
1140 | } |
1141 | } |
1142 | |
1143 | |
1144 | # Fix SQL case, if so requested |
1145 | sub _sqlcase { |
96449e8e |
1146 | # LDNOTE: if $self->{case} is true, then it contains 'lower', so we |
1147 | # don't touch the argument ... crooked logic, but let's not change it! |
07d7c35c |
1148 | return $_[0]->{case} ? $_[1] : uc($_[1]); |
96449e8e |
1149 | } |
1150 | |
1151 | |
1152 | #====================================================================== |
1153 | # DISPATCHING FROM REFKIND |
1154 | #====================================================================== |
1155 | |
1156 | sub _refkind { |
1157 | my ($self, $data) = @_; |
96449e8e |
1158 | |
955e77ca |
1159 | return 'UNDEF' unless defined $data; |
1160 | |
1161 | # blessed objects are treated like scalars |
1162 | my $ref = (Scalar::Util::blessed $data) ? '' : ref $data; |
1163 | |
1164 | return 'SCALAR' unless $ref; |
1165 | |
1166 | my $n_steps = 1; |
1167 | while ($ref eq 'REF') { |
96449e8e |
1168 | $data = $$data; |
955e77ca |
1169 | $ref = (Scalar::Util::blessed $data) ? '' : ref $data; |
1170 | $n_steps++ if $ref; |
96449e8e |
1171 | } |
1172 | |
848556bc |
1173 | return ($ref||'SCALAR') . ('REF' x $n_steps); |
96449e8e |
1174 | } |
1175 | |
1176 | sub _try_refkind { |
1177 | my ($self, $data) = @_; |
1178 | my @try = ($self->_refkind($data)); |
1179 | push @try, 'SCALAR_or_UNDEF' if $try[0] eq 'SCALAR' || $try[0] eq 'UNDEF'; |
1180 | push @try, 'FALLBACK'; |
955e77ca |
1181 | return \@try; |
96449e8e |
1182 | } |
1183 | |
1184 | sub _METHOD_FOR_refkind { |
1185 | my ($self, $meth_prefix, $data) = @_; |
f39eaa60 |
1186 | |
1187 | my $method; |
955e77ca |
1188 | for (@{$self->_try_refkind($data)}) { |
f39eaa60 |
1189 | $method = $self->can($meth_prefix."_".$_) |
1190 | and last; |
1191 | } |
1192 | |
1193 | return $method || puke "cannot dispatch on '$meth_prefix' for ".$self->_refkind($data); |
96449e8e |
1194 | } |
1195 | |
1196 | |
1197 | sub _SWITCH_refkind { |
1198 | my ($self, $data, $dispatch_table) = @_; |
1199 | |
f39eaa60 |
1200 | my $coderef; |
955e77ca |
1201 | for (@{$self->_try_refkind($data)}) { |
f39eaa60 |
1202 | $coderef = $dispatch_table->{$_} |
1203 | and last; |
1204 | } |
1205 | |
1206 | puke "no dispatch entry for ".$self->_refkind($data) |
1207 | unless $coderef; |
1208 | |
96449e8e |
1209 | $coderef->(); |
1210 | } |
1211 | |
1212 | |
1213 | |
1214 | |
1215 | #====================================================================== |
1216 | # VALUES, GENERATE, AUTOLOAD |
1217 | #====================================================================== |
1218 | |
1219 | # LDNOTE: original code from nwiger, didn't touch code in that section |
1220 | # I feel the AUTOLOAD stuff should not be the default, it should |
1221 | # only be activated on explicit demand by user. |
1222 | |
1223 | sub values { |
1224 | my $self = shift; |
1225 | my $data = shift || return; |
1226 | puke "Argument to ", __PACKAGE__, "->values must be a \\%hash" |
1227 | unless ref $data eq 'HASH'; |
bab725ce |
1228 | |
1229 | my @all_bind; |
1230 | foreach my $k ( sort keys %$data ) { |
1231 | my $v = $data->{$k}; |
1232 | $self->_SWITCH_refkind($v, { |
9d48860e |
1233 | ARRAYREF => sub { |
bab725ce |
1234 | if ($self->{array_datatypes}) { # array datatype |
1235 | push @all_bind, $self->_bindtype($k, $v); |
1236 | } |
1237 | else { # literal SQL with bind |
1238 | my ($sql, @bind) = @$v; |
1239 | $self->_assert_bindval_matches_bindtype(@bind); |
1240 | push @all_bind, @bind; |
1241 | } |
1242 | }, |
1243 | ARRAYREFREF => sub { # literal SQL with bind |
1244 | my ($sql, @bind) = @${$v}; |
1245 | $self->_assert_bindval_matches_bindtype(@bind); |
1246 | push @all_bind, @bind; |
1247 | }, |
1248 | SCALARREF => sub { # literal SQL without bind |
1249 | }, |
1250 | SCALAR_or_UNDEF => sub { |
1251 | push @all_bind, $self->_bindtype($k, $v); |
1252 | }, |
1253 | }); |
1254 | } |
1255 | |
1256 | return @all_bind; |
96449e8e |
1257 | } |
1258 | |
1259 | sub generate { |
1260 | my $self = shift; |
1261 | |
1262 | my(@sql, @sqlq, @sqlv); |
1263 | |
1264 | for (@_) { |
1265 | my $ref = ref $_; |
1266 | if ($ref eq 'HASH') { |
1267 | for my $k (sort keys %$_) { |
1268 | my $v = $_->{$k}; |
1269 | my $r = ref $v; |
1270 | my $label = $self->_quote($k); |
1271 | if ($r eq 'ARRAY') { |
fe3ae272 |
1272 | # literal SQL with bind |
1273 | my ($sql, @bind) = @$v; |
1274 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
1275 | push @sqlq, "$label = $sql"; |
fe3ae272 |
1276 | push @sqlv, @bind; |
96449e8e |
1277 | } elsif ($r eq 'SCALAR') { |
fe3ae272 |
1278 | # literal SQL without bind |
96449e8e |
1279 | push @sqlq, "$label = $$v"; |
9d48860e |
1280 | } else { |
96449e8e |
1281 | push @sqlq, "$label = ?"; |
1282 | push @sqlv, $self->_bindtype($k, $v); |
1283 | } |
1284 | } |
1285 | push @sql, $self->_sqlcase('set'), join ', ', @sqlq; |
1286 | } elsif ($ref eq 'ARRAY') { |
1287 | # unlike insert(), assume these are ONLY the column names, i.e. for SQL |
1288 | for my $v (@$_) { |
1289 | my $r = ref $v; |
fe3ae272 |
1290 | if ($r eq 'ARRAY') { # literal SQL with bind |
1291 | my ($sql, @bind) = @$v; |
1292 | $self->_assert_bindval_matches_bindtype(@bind); |
1293 | push @sqlq, $sql; |
1294 | push @sqlv, @bind; |
1295 | } elsif ($r eq 'SCALAR') { # literal SQL without bind |
96449e8e |
1296 | # embedded literal SQL |
1297 | push @sqlq, $$v; |
9d48860e |
1298 | } else { |
96449e8e |
1299 | push @sqlq, '?'; |
1300 | push @sqlv, $v; |
1301 | } |
1302 | } |
1303 | push @sql, '(' . join(', ', @sqlq) . ')'; |
1304 | } elsif ($ref eq 'SCALAR') { |
1305 | # literal SQL |
1306 | push @sql, $$_; |
1307 | } else { |
1308 | # strings get case twiddled |
1309 | push @sql, $self->_sqlcase($_); |
1310 | } |
1311 | } |
1312 | |
1313 | my $sql = join ' ', @sql; |
1314 | |
1315 | # this is pretty tricky |
1316 | # if ask for an array, return ($stmt, @bind) |
1317 | # otherwise, s/?/shift @sqlv/ to put it inline |
1318 | if (wantarray) { |
1319 | return ($sql, @sqlv); |
1320 | } else { |
1321 | 1 while $sql =~ s/\?/my $d = shift(@sqlv); |
1322 | ref $d ? $d->[1] : $d/e; |
1323 | return $sql; |
1324 | } |
1325 | } |
1326 | |
1327 | |
1328 | sub DESTROY { 1 } |
1329 | |
1330 | sub AUTOLOAD { |
1331 | # This allows us to check for a local, then _form, attr |
1332 | my $self = shift; |
1333 | my($name) = $AUTOLOAD =~ /.*::(.+)/; |
1334 | return $self->generate($name, @_); |
1335 | } |
1336 | |
1337 | 1; |
1338 | |
1339 | |
1340 | |
1341 | __END__ |
32eab2da |
1342 | |
1343 | =head1 NAME |
1344 | |
1345 | SQL::Abstract - Generate SQL from Perl data structures |
1346 | |
1347 | =head1 SYNOPSIS |
1348 | |
1349 | use SQL::Abstract; |
1350 | |
1351 | my $sql = SQL::Abstract->new; |
1352 | |
1353 | my($stmt, @bind) = $sql->select($table, \@fields, \%where, \@order); |
1354 | |
1355 | my($stmt, @bind) = $sql->insert($table, \%fieldvals || \@values); |
1356 | |
1357 | my($stmt, @bind) = $sql->update($table, \%fieldvals, \%where); |
1358 | |
1359 | my($stmt, @bind) = $sql->delete($table, \%where); |
1360 | |
1361 | # Then, use these in your DBI statements |
1362 | my $sth = $dbh->prepare($stmt); |
1363 | $sth->execute(@bind); |
1364 | |
1365 | # Just generate the WHERE clause |
abe72f94 |
1366 | my($stmt, @bind) = $sql->where(\%where, \@order); |
32eab2da |
1367 | |
1368 | # Return values in the same order, for hashed queries |
1369 | # See PERFORMANCE section for more details |
1370 | my @bind = $sql->values(\%fieldvals); |
1371 | |
1372 | =head1 DESCRIPTION |
1373 | |
1374 | This module was inspired by the excellent L<DBIx::Abstract>. |
1375 | However, in using that module I found that what I really wanted |
1376 | to do was generate SQL, but still retain complete control over my |
1377 | statement handles and use the DBI interface. So, I set out to |
1378 | create an abstract SQL generation module. |
1379 | |
1380 | While based on the concepts used by L<DBIx::Abstract>, there are |
1381 | several important differences, especially when it comes to WHERE |
1382 | clauses. I have modified the concepts used to make the SQL easier |
1383 | to generate from Perl data structures and, IMO, more intuitive. |
1384 | The underlying idea is for this module to do what you mean, based |
1385 | on the data structures you provide it. The big advantage is that |
1386 | you don't have to modify your code every time your data changes, |
1387 | as this module figures it out. |
1388 | |
1389 | To begin with, an SQL INSERT is as easy as just specifying a hash |
1390 | of C<key=value> pairs: |
1391 | |
1392 | my %data = ( |
1393 | name => 'Jimbo Bobson', |
1394 | phone => '123-456-7890', |
1395 | address => '42 Sister Lane', |
1396 | city => 'St. Louis', |
1397 | state => 'Louisiana', |
1398 | ); |
1399 | |
1400 | The SQL can then be generated with this: |
1401 | |
1402 | my($stmt, @bind) = $sql->insert('people', \%data); |
1403 | |
1404 | Which would give you something like this: |
1405 | |
1406 | $stmt = "INSERT INTO people |
1407 | (address, city, name, phone, state) |
1408 | VALUES (?, ?, ?, ?, ?)"; |
1409 | @bind = ('42 Sister Lane', 'St. Louis', 'Jimbo Bobson', |
1410 | '123-456-7890', 'Louisiana'); |
1411 | |
1412 | These are then used directly in your DBI code: |
1413 | |
1414 | my $sth = $dbh->prepare($stmt); |
1415 | $sth->execute(@bind); |
1416 | |
96449e8e |
1417 | =head2 Inserting and Updating Arrays |
1418 | |
1419 | If your database has array types (like for example Postgres), |
1420 | activate the special option C<< array_datatypes => 1 >> |
9d48860e |
1421 | when creating the C<SQL::Abstract> object. |
96449e8e |
1422 | Then you may use an arrayref to insert and update database array types: |
1423 | |
1424 | my $sql = SQL::Abstract->new(array_datatypes => 1); |
1425 | my %data = ( |
1426 | planets => [qw/Mercury Venus Earth Mars/] |
1427 | ); |
9d48860e |
1428 | |
96449e8e |
1429 | my($stmt, @bind) = $sql->insert('solar_system', \%data); |
1430 | |
1431 | This results in: |
1432 | |
1433 | $stmt = "INSERT INTO solar_system (planets) VALUES (?)" |
1434 | |
1435 | @bind = (['Mercury', 'Venus', 'Earth', 'Mars']); |
1436 | |
1437 | |
1438 | =head2 Inserting and Updating SQL |
1439 | |
1440 | In order to apply SQL functions to elements of your C<%data> you may |
1441 | specify a reference to an arrayref for the given hash value. For example, |
1442 | if you need to execute the Oracle C<to_date> function on a value, you can |
1443 | say something like this: |
32eab2da |
1444 | |
1445 | my %data = ( |
1446 | name => 'Bill', |
96449e8e |
1447 | date_entered => \["to_date(?,'MM/DD/YYYY')", "03/02/2003"], |
9d48860e |
1448 | ); |
32eab2da |
1449 | |
1450 | The first value in the array is the actual SQL. Any other values are |
1451 | optional and would be included in the bind values array. This gives |
1452 | you: |
1453 | |
1454 | my($stmt, @bind) = $sql->insert('people', \%data); |
1455 | |
9d48860e |
1456 | $stmt = "INSERT INTO people (name, date_entered) |
32eab2da |
1457 | VALUES (?, to_date(?,'MM/DD/YYYY'))"; |
1458 | @bind = ('Bill', '03/02/2003'); |
1459 | |
1460 | An UPDATE is just as easy, all you change is the name of the function: |
1461 | |
1462 | my($stmt, @bind) = $sql->update('people', \%data); |
1463 | |
1464 | Notice that your C<%data> isn't touched; the module will generate |
1465 | the appropriately quirky SQL for you automatically. Usually you'll |
1466 | want to specify a WHERE clause for your UPDATE, though, which is |
1467 | where handling C<%where> hashes comes in handy... |
1468 | |
96449e8e |
1469 | =head2 Complex where statements |
1470 | |
32eab2da |
1471 | This module can generate pretty complicated WHERE statements |
1472 | easily. For example, simple C<key=value> pairs are taken to mean |
1473 | equality, and if you want to see if a field is within a set |
1474 | of values, you can use an arrayref. Let's say we wanted to |
1475 | SELECT some data based on this criteria: |
1476 | |
1477 | my %where = ( |
1478 | requestor => 'inna', |
1479 | worker => ['nwiger', 'rcwe', 'sfz'], |
1480 | status => { '!=', 'completed' } |
1481 | ); |
1482 | |
1483 | my($stmt, @bind) = $sql->select('tickets', '*', \%where); |
1484 | |
1485 | The above would give you something like this: |
1486 | |
1487 | $stmt = "SELECT * FROM tickets WHERE |
1488 | ( requestor = ? ) AND ( status != ? ) |
1489 | AND ( worker = ? OR worker = ? OR worker = ? )"; |
1490 | @bind = ('inna', 'completed', 'nwiger', 'rcwe', 'sfz'); |
1491 | |
1492 | Which you could then use in DBI code like so: |
1493 | |
1494 | my $sth = $dbh->prepare($stmt); |
1495 | $sth->execute(@bind); |
1496 | |
1497 | Easy, eh? |
1498 | |
1499 | =head1 FUNCTIONS |
1500 | |
1501 | The functions are simple. There's one for each major SQL operation, |
1502 | and a constructor you use first. The arguments are specified in a |
9d48860e |
1503 | similar order to each function (table, then fields, then a where |
32eab2da |
1504 | clause) to try and simplify things. |
1505 | |
83cab70b |
1506 | |
83cab70b |
1507 | |
32eab2da |
1508 | |
1509 | =head2 new(option => 'value') |
1510 | |
1511 | The C<new()> function takes a list of options and values, and returns |
1512 | a new B<SQL::Abstract> object which can then be used to generate SQL |
1513 | through the methods below. The options accepted are: |
1514 | |
1515 | =over |
1516 | |
1517 | =item case |
1518 | |
1519 | If set to 'lower', then SQL will be generated in all lowercase. By |
1520 | default SQL is generated in "textbook" case meaning something like: |
1521 | |
1522 | SELECT a_field FROM a_table WHERE some_field LIKE '%someval%' |
1523 | |
96449e8e |
1524 | Any setting other than 'lower' is ignored. |
1525 | |
32eab2da |
1526 | =item cmp |
1527 | |
1528 | This determines what the default comparison operator is. By default |
1529 | it is C<=>, meaning that a hash like this: |
1530 | |
1531 | %where = (name => 'nwiger', email => 'nate@wiger.org'); |
1532 | |
1533 | Will generate SQL like this: |
1534 | |
1535 | WHERE name = 'nwiger' AND email = 'nate@wiger.org' |
1536 | |
1537 | However, you may want loose comparisons by default, so if you set |
1538 | C<cmp> to C<like> you would get SQL such as: |
1539 | |
1540 | WHERE name like 'nwiger' AND email like 'nate@wiger.org' |
1541 | |
1542 | You can also override the comparsion on an individual basis - see |
1543 | the huge section on L</"WHERE CLAUSES"> at the bottom. |
1544 | |
96449e8e |
1545 | =item sqltrue, sqlfalse |
1546 | |
1547 | Expressions for inserting boolean values within SQL statements. |
6e0c6552 |
1548 | By default these are C<1=1> and C<1=0>. They are used |
1549 | by the special operators C<-in> and C<-not_in> for generating |
1550 | correct SQL even when the argument is an empty array (see below). |
96449e8e |
1551 | |
32eab2da |
1552 | =item logic |
1553 | |
1554 | This determines the default logical operator for multiple WHERE |
7cac25e6 |
1555 | statements in arrays or hashes. If absent, the default logic is "or" |
1556 | for arrays, and "and" for hashes. This means that a WHERE |
32eab2da |
1557 | array of the form: |
1558 | |
1559 | @where = ( |
9d48860e |
1560 | event_date => {'>=', '2/13/99'}, |
1561 | event_date => {'<=', '4/24/03'}, |
32eab2da |
1562 | ); |
1563 | |
7cac25e6 |
1564 | will generate SQL like this: |
32eab2da |
1565 | |
1566 | WHERE event_date >= '2/13/99' OR event_date <= '4/24/03' |
1567 | |
1568 | This is probably not what you want given this query, though (look |
1569 | at the dates). To change the "OR" to an "AND", simply specify: |
1570 | |
1571 | my $sql = SQL::Abstract->new(logic => 'and'); |
1572 | |
1573 | Which will change the above C<WHERE> to: |
1574 | |
1575 | WHERE event_date >= '2/13/99' AND event_date <= '4/24/03' |
1576 | |
96449e8e |
1577 | The logic can also be changed locally by inserting |
7cac25e6 |
1578 | a modifier in front of an arrayref : |
96449e8e |
1579 | |
9d48860e |
1580 | @where = (-and => [event_date => {'>=', '2/13/99'}, |
7cac25e6 |
1581 | event_date => {'<=', '4/24/03'} ]); |
96449e8e |
1582 | |
1583 | See the L</"WHERE CLAUSES"> section for explanations. |
1584 | |
32eab2da |
1585 | =item convert |
1586 | |
1587 | This will automatically convert comparisons using the specified SQL |
1588 | function for both column and value. This is mostly used with an argument |
1589 | of C<upper> or C<lower>, so that the SQL will have the effect of |
1590 | case-insensitive "searches". For example, this: |
1591 | |
1592 | $sql = SQL::Abstract->new(convert => 'upper'); |
1593 | %where = (keywords => 'MaKe iT CAse inSeNSItive'); |
1594 | |
1595 | Will turn out the following SQL: |
1596 | |
1597 | WHERE upper(keywords) like upper('MaKe iT CAse inSeNSItive') |
1598 | |
1599 | The conversion can be C<upper()>, C<lower()>, or any other SQL function |
1600 | that can be applied symmetrically to fields (actually B<SQL::Abstract> does |
1601 | not validate this option; it will just pass through what you specify verbatim). |
1602 | |
1603 | =item bindtype |
1604 | |
1605 | This is a kludge because many databases suck. For example, you can't |
1606 | just bind values using DBI's C<execute()> for Oracle C<CLOB> or C<BLOB> fields. |
1607 | Instead, you have to use C<bind_param()>: |
1608 | |
1609 | $sth->bind_param(1, 'reg data'); |
1610 | $sth->bind_param(2, $lots, {ora_type => ORA_CLOB}); |
1611 | |
1612 | The problem is, B<SQL::Abstract> will normally just return a C<@bind> array, |
1613 | which loses track of which field each slot refers to. Fear not. |
1614 | |
1615 | If you specify C<bindtype> in new, you can determine how C<@bind> is returned. |
1616 | Currently, you can specify either C<normal> (default) or C<columns>. If you |
1617 | specify C<columns>, you will get an array that looks like this: |
1618 | |
1619 | my $sql = SQL::Abstract->new(bindtype => 'columns'); |
1620 | my($stmt, @bind) = $sql->insert(...); |
1621 | |
1622 | @bind = ( |
1623 | [ 'column1', 'value1' ], |
1624 | [ 'column2', 'value2' ], |
1625 | [ 'column3', 'value3' ], |
1626 | ); |
1627 | |
1628 | You can then iterate through this manually, using DBI's C<bind_param()>. |
e3f9dff4 |
1629 | |
32eab2da |
1630 | $sth->prepare($stmt); |
1631 | my $i = 1; |
1632 | for (@bind) { |
1633 | my($col, $data) = @$_; |
1634 | if ($col eq 'details' || $col eq 'comments') { |
1635 | $sth->bind_param($i, $data, {ora_type => ORA_CLOB}); |
1636 | } elsif ($col eq 'image') { |
1637 | $sth->bind_param($i, $data, {ora_type => ORA_BLOB}); |
1638 | } else { |
1639 | $sth->bind_param($i, $data); |
1640 | } |
1641 | $i++; |
1642 | } |
1643 | $sth->execute; # execute without @bind now |
1644 | |
1645 | Now, why would you still use B<SQL::Abstract> if you have to do this crap? |
1646 | Basically, the advantage is still that you don't have to care which fields |
1647 | are or are not included. You could wrap that above C<for> loop in a simple |
1648 | sub called C<bind_fields()> or something and reuse it repeatedly. You still |
1649 | get a layer of abstraction over manual SQL specification. |
1650 | |
deb148a2 |
1651 | Note that if you set L</bindtype> to C<columns>, the C<\[$sql, @bind]> |
1652 | construct (see L</Literal SQL with placeholders and bind values (subqueries)>) |
1653 | will expect the bind values in this format. |
1654 | |
32eab2da |
1655 | =item quote_char |
1656 | |
1657 | This is the character that a table or column name will be quoted |
9d48860e |
1658 | with. By default this is an empty string, but you could set it to |
32eab2da |
1659 | the character C<`>, to generate SQL like this: |
1660 | |
1661 | SELECT `a_field` FROM `a_table` WHERE `some_field` LIKE '%someval%' |
1662 | |
96449e8e |
1663 | Alternatively, you can supply an array ref of two items, the first being the left |
1664 | hand quote character, and the second the right hand quote character. For |
1665 | example, you could supply C<['[',']']> for SQL Server 2000 compliant quotes |
1666 | that generates SQL like this: |
1667 | |
1668 | SELECT [a_field] FROM [a_table] WHERE [some_field] LIKE '%someval%' |
1669 | |
9d48860e |
1670 | Quoting is useful if you have tables or columns names that are reserved |
96449e8e |
1671 | words in your database's SQL dialect. |
32eab2da |
1672 | |
1673 | =item name_sep |
1674 | |
1675 | This is the character that separates a table and column name. It is |
1676 | necessary to specify this when the C<quote_char> option is selected, |
1677 | so that tables and column names can be individually quoted like this: |
1678 | |
1679 | SELECT `table`.`one_field` FROM `table` WHERE `table`.`other_field` = 1 |
1680 | |
96449e8e |
1681 | =item array_datatypes |
32eab2da |
1682 | |
9d48860e |
1683 | When this option is true, arrayrefs in INSERT or UPDATE are |
1684 | interpreted as array datatypes and are passed directly |
96449e8e |
1685 | to the DBI layer. |
1686 | When this option is false, arrayrefs are interpreted |
1687 | as literal SQL, just like refs to arrayrefs |
1688 | (but this behavior is for backwards compatibility; when writing |
1689 | new queries, use the "reference to arrayref" syntax |
1690 | for literal SQL). |
32eab2da |
1691 | |
32eab2da |
1692 | |
96449e8e |
1693 | =item special_ops |
32eab2da |
1694 | |
9d48860e |
1695 | Takes a reference to a list of "special operators" |
96449e8e |
1696 | to extend the syntax understood by L<SQL::Abstract>. |
1697 | See section L</"SPECIAL OPERATORS"> for details. |
32eab2da |
1698 | |
59f23b3d |
1699 | =item unary_ops |
1700 | |
9d48860e |
1701 | Takes a reference to a list of "unary operators" |
59f23b3d |
1702 | to extend the syntax understood by L<SQL::Abstract>. |
1703 | See section L</"UNARY OPERATORS"> for details. |
1704 | |
32eab2da |
1705 | |
32eab2da |
1706 | |
96449e8e |
1707 | =back |
32eab2da |
1708 | |
02288357 |
1709 | =head2 insert($table, \@values || \%fieldvals, \%options) |
32eab2da |
1710 | |
1711 | This is the simplest function. You simply give it a table name |
1712 | and either an arrayref of values or hashref of field/value pairs. |
1713 | It returns an SQL INSERT statement and a list of bind values. |
96449e8e |
1714 | See the sections on L</"Inserting and Updating Arrays"> and |
1715 | L</"Inserting and Updating SQL"> for information on how to insert |
1716 | with those data types. |
32eab2da |
1717 | |
02288357 |
1718 | The optional C<\%options> hash reference may contain additional |
1719 | options to generate the insert SQL. Currently supported options |
1720 | are: |
1721 | |
1722 | =over 4 |
1723 | |
1724 | =item returning |
1725 | |
1726 | Takes either a scalar of raw SQL fields, or an array reference of |
1727 | field names, and adds on an SQL C<RETURNING> statement at the end. |
1728 | This allows you to return data generated by the insert statement |
1729 | (such as row IDs) without performing another C<SELECT> statement. |
1730 | Note, however, this is not part of the SQL standard and may not |
1731 | be supported by all database engines. |
1732 | |
1733 | =back |
1734 | |
32eab2da |
1735 | =head2 update($table, \%fieldvals, \%where) |
1736 | |
1737 | This takes a table, hashref of field/value pairs, and an optional |
86298391 |
1738 | hashref L<WHERE clause|/WHERE CLAUSES>. It returns an SQL UPDATE function and a list |
32eab2da |
1739 | of bind values. |
96449e8e |
1740 | See the sections on L</"Inserting and Updating Arrays"> and |
1741 | L</"Inserting and Updating SQL"> for information on how to insert |
1742 | with those data types. |
32eab2da |
1743 | |
96449e8e |
1744 | =head2 select($source, $fields, $where, $order) |
32eab2da |
1745 | |
9d48860e |
1746 | This returns a SQL SELECT statement and associated list of bind values, as |
96449e8e |
1747 | specified by the arguments : |
32eab2da |
1748 | |
96449e8e |
1749 | =over |
32eab2da |
1750 | |
96449e8e |
1751 | =item $source |
32eab2da |
1752 | |
9d48860e |
1753 | Specification of the 'FROM' part of the statement. |
96449e8e |
1754 | The argument can be either a plain scalar (interpreted as a table |
1755 | name, will be quoted), or an arrayref (interpreted as a list |
1756 | of table names, joined by commas, quoted), or a scalarref |
1757 | (literal table name, not quoted), or a ref to an arrayref |
1758 | (list of literal table names, joined by commas, not quoted). |
32eab2da |
1759 | |
96449e8e |
1760 | =item $fields |
32eab2da |
1761 | |
9d48860e |
1762 | Specification of the list of fields to retrieve from |
96449e8e |
1763 | the source. |
1764 | The argument can be either an arrayref (interpreted as a list |
9d48860e |
1765 | of field names, will be joined by commas and quoted), or a |
96449e8e |
1766 | plain scalar (literal SQL, not quoted). |
1767 | Please observe that this API is not as flexible as for |
e3f9dff4 |
1768 | the first argument C<$table>, for backwards compatibility reasons. |
32eab2da |
1769 | |
96449e8e |
1770 | =item $where |
32eab2da |
1771 | |
96449e8e |
1772 | Optional argument to specify the WHERE part of the query. |
1773 | The argument is most often a hashref, but can also be |
9d48860e |
1774 | an arrayref or plain scalar -- |
96449e8e |
1775 | see section L<WHERE clause|/"WHERE CLAUSES"> for details. |
32eab2da |
1776 | |
96449e8e |
1777 | =item $order |
32eab2da |
1778 | |
96449e8e |
1779 | Optional argument to specify the ORDER BY part of the query. |
9d48860e |
1780 | The argument can be a scalar, a hashref or an arrayref |
96449e8e |
1781 | -- see section L<ORDER BY clause|/"ORDER BY CLAUSES"> |
1782 | for details. |
32eab2da |
1783 | |
96449e8e |
1784 | =back |
32eab2da |
1785 | |
32eab2da |
1786 | |
1787 | =head2 delete($table, \%where) |
1788 | |
86298391 |
1789 | This takes a table name and optional hashref L<WHERE clause|/WHERE CLAUSES>. |
32eab2da |
1790 | It returns an SQL DELETE statement and list of bind values. |
1791 | |
32eab2da |
1792 | =head2 where(\%where, \@order) |
1793 | |
1794 | This is used to generate just the WHERE clause. For example, |
1795 | if you have an arbitrary data structure and know what the |
1796 | rest of your SQL is going to look like, but want an easy way |
1797 | to produce a WHERE clause, use this. It returns an SQL WHERE |
1798 | clause and list of bind values. |
1799 | |
32eab2da |
1800 | |
1801 | =head2 values(\%data) |
1802 | |
1803 | This just returns the values from the hash C<%data>, in the same |
1804 | order that would be returned from any of the other above queries. |
1805 | Using this allows you to markedly speed up your queries if you |
1806 | are affecting lots of rows. See below under the L</"PERFORMANCE"> section. |
1807 | |
32eab2da |
1808 | =head2 generate($any, 'number', $of, \@data, $struct, \%types) |
1809 | |
1810 | Warning: This is an experimental method and subject to change. |
1811 | |
1812 | This returns arbitrarily generated SQL. It's a really basic shortcut. |
1813 | It will return two different things, depending on return context: |
1814 | |
1815 | my($stmt, @bind) = $sql->generate('create table', \$table, \@fields); |
1816 | my $stmt_and_val = $sql->generate('create table', \$table, \@fields); |
1817 | |
1818 | These would return the following: |
1819 | |
1820 | # First calling form |
1821 | $stmt = "CREATE TABLE test (?, ?)"; |
1822 | @bind = (field1, field2); |
1823 | |
1824 | # Second calling form |
1825 | $stmt_and_val = "CREATE TABLE test (field1, field2)"; |
1826 | |
1827 | Depending on what you're trying to do, it's up to you to choose the correct |
1828 | format. In this example, the second form is what you would want. |
1829 | |
1830 | By the same token: |
1831 | |
1832 | $sql->generate('alter session', { nls_date_format => 'MM/YY' }); |
1833 | |
1834 | Might give you: |
1835 | |
1836 | ALTER SESSION SET nls_date_format = 'MM/YY' |
1837 | |
1838 | You get the idea. Strings get their case twiddled, but everything |
1839 | else remains verbatim. |
1840 | |
32eab2da |
1841 | |
32eab2da |
1842 | |
32eab2da |
1843 | |
1844 | =head1 WHERE CLAUSES |
1845 | |
96449e8e |
1846 | =head2 Introduction |
1847 | |
32eab2da |
1848 | This module uses a variation on the idea from L<DBIx::Abstract>. It |
1849 | is B<NOT>, repeat I<not> 100% compatible. B<The main logic of this |
1850 | module is that things in arrays are OR'ed, and things in hashes |
1851 | are AND'ed.> |
1852 | |
1853 | The easiest way to explain is to show lots of examples. After |
1854 | each C<%where> hash shown, it is assumed you used: |
1855 | |
1856 | my($stmt, @bind) = $sql->where(\%where); |
1857 | |
1858 | However, note that the C<%where> hash can be used directly in any |
1859 | of the other functions as well, as described above. |
1860 | |
96449e8e |
1861 | =head2 Key-value pairs |
1862 | |
32eab2da |
1863 | So, let's get started. To begin, a simple hash: |
1864 | |
1865 | my %where = ( |
1866 | user => 'nwiger', |
1867 | status => 'completed' |
1868 | ); |
1869 | |
1870 | Is converted to SQL C<key = val> statements: |
1871 | |
1872 | $stmt = "WHERE user = ? AND status = ?"; |
1873 | @bind = ('nwiger', 'completed'); |
1874 | |
1875 | One common thing I end up doing is having a list of values that |
1876 | a field can be in. To do this, simply specify a list inside of |
1877 | an arrayref: |
1878 | |
1879 | my %where = ( |
1880 | user => 'nwiger', |
1881 | status => ['assigned', 'in-progress', 'pending']; |
1882 | ); |
1883 | |
1884 | This simple code will create the following: |
9d48860e |
1885 | |
32eab2da |
1886 | $stmt = "WHERE user = ? AND ( status = ? OR status = ? OR status = ? )"; |
1887 | @bind = ('nwiger', 'assigned', 'in-progress', 'pending'); |
1888 | |
9d48860e |
1889 | A field associated to an empty arrayref will be considered a |
7cac25e6 |
1890 | logical false and will generate 0=1. |
8a68b5be |
1891 | |
6e0c6552 |
1892 | =head2 Specific comparison operators |
96449e8e |
1893 | |
32eab2da |
1894 | If you want to specify a different type of operator for your comparison, |
1895 | you can use a hashref for a given column: |
1896 | |
1897 | my %where = ( |
1898 | user => 'nwiger', |
1899 | status => { '!=', 'completed' } |
1900 | ); |
1901 | |
1902 | Which would generate: |
1903 | |
1904 | $stmt = "WHERE user = ? AND status != ?"; |
1905 | @bind = ('nwiger', 'completed'); |
1906 | |
1907 | To test against multiple values, just enclose the values in an arrayref: |
1908 | |
96449e8e |
1909 | status => { '=', ['assigned', 'in-progress', 'pending'] }; |
1910 | |
f2d5020d |
1911 | Which would give you: |
96449e8e |
1912 | |
1913 | "WHERE status = ? OR status = ? OR status = ?" |
1914 | |
1915 | |
1916 | The hashref can also contain multiple pairs, in which case it is expanded |
32eab2da |
1917 | into an C<AND> of its elements: |
1918 | |
1919 | my %where = ( |
1920 | user => 'nwiger', |
1921 | status => { '!=', 'completed', -not_like => 'pending%' } |
1922 | ); |
1923 | |
1924 | # Or more dynamically, like from a form |
1925 | $where{user} = 'nwiger'; |
1926 | $where{status}{'!='} = 'completed'; |
1927 | $where{status}{'-not_like'} = 'pending%'; |
1928 | |
1929 | # Both generate this |
1930 | $stmt = "WHERE user = ? AND status != ? AND status NOT LIKE ?"; |
1931 | @bind = ('nwiger', 'completed', 'pending%'); |
1932 | |
96449e8e |
1933 | |
32eab2da |
1934 | To get an OR instead, you can combine it with the arrayref idea: |
1935 | |
1936 | my %where => ( |
1937 | user => 'nwiger', |
1938 | priority => [ {'=', 2}, {'!=', 1} ] |
1939 | ); |
1940 | |
1941 | Which would generate: |
1942 | |
1943 | $stmt = "WHERE user = ? AND priority = ? OR priority != ?"; |
1944 | @bind = ('nwiger', '2', '1'); |
1945 | |
44b9e502 |
1946 | If you want to include literal SQL (with or without bind values), just use a |
1947 | scalar reference or array reference as the value: |
1948 | |
1949 | my %where = ( |
1950 | date_entered => { '>' => \["to_date(?, 'MM/DD/YYYY')", "11/26/2008"] }, |
1951 | date_expires => { '<' => \"now()" } |
1952 | ); |
1953 | |
1954 | Which would generate: |
1955 | |
1956 | $stmt = "WHERE date_entered > "to_date(?, 'MM/DD/YYYY') AND date_expires < now()"; |
1957 | @bind = ('11/26/2008'); |
1958 | |
96449e8e |
1959 | |
1960 | =head2 Logic and nesting operators |
1961 | |
1962 | In the example above, |
1963 | there is a subtle trap if you want to say something like |
32eab2da |
1964 | this (notice the C<AND>): |
1965 | |
1966 | WHERE priority != ? AND priority != ? |
1967 | |
1968 | Because, in Perl you I<can't> do this: |
1969 | |
1970 | priority => { '!=', 2, '!=', 1 } |
1971 | |
1972 | As the second C<!=> key will obliterate the first. The solution |
1973 | is to use the special C<-modifier> form inside an arrayref: |
1974 | |
9d48860e |
1975 | priority => [ -and => {'!=', 2}, |
96449e8e |
1976 | {'!=', 1} ] |
1977 | |
32eab2da |
1978 | |
1979 | Normally, these would be joined by C<OR>, but the modifier tells it |
1980 | to use C<AND> instead. (Hint: You can use this in conjunction with the |
1981 | C<logic> option to C<new()> in order to change the way your queries |
1982 | work by default.) B<Important:> Note that the C<-modifier> goes |
1983 | B<INSIDE> the arrayref, as an extra first element. This will |
1984 | B<NOT> do what you think it might: |
1985 | |
1986 | priority => -and => [{'!=', 2}, {'!=', 1}] # WRONG! |
1987 | |
1988 | Here is a quick list of equivalencies, since there is some overlap: |
1989 | |
1990 | # Same |
1991 | status => {'!=', 'completed', 'not like', 'pending%' } |
1992 | status => [ -and => {'!=', 'completed'}, {'not like', 'pending%'}] |
1993 | |
1994 | # Same |
1995 | status => {'=', ['assigned', 'in-progress']} |
1996 | status => [ -or => {'=', 'assigned'}, {'=', 'in-progress'}] |
1997 | status => [ {'=', 'assigned'}, {'=', 'in-progress'} ] |
1998 | |
e3f9dff4 |
1999 | |
2000 | |
96449e8e |
2001 | =head2 Special operators : IN, BETWEEN, etc. |
2002 | |
32eab2da |
2003 | You can also use the hashref format to compare a list of fields using the |
2004 | C<IN> comparison operator, by specifying the list as an arrayref: |
2005 | |
2006 | my %where = ( |
2007 | status => 'completed', |
2008 | reportid => { -in => [567, 2335, 2] } |
2009 | ); |
2010 | |
2011 | Which would generate: |
2012 | |
2013 | $stmt = "WHERE status = ? AND reportid IN (?,?,?)"; |
2014 | @bind = ('completed', '567', '2335', '2'); |
2015 | |
9d48860e |
2016 | The reverse operator C<-not_in> generates SQL C<NOT IN> and is used in |
96449e8e |
2017 | the same way. |
2018 | |
6e0c6552 |
2019 | If the argument to C<-in> is an empty array, 'sqlfalse' is generated |
2020 | (by default : C<1=0>). Similarly, C<< -not_in => [] >> generates |
2021 | 'sqltrue' (by default : C<1=1>). |
2022 | |
e41c3bdd |
2023 | In addition to the array you can supply a chunk of literal sql or |
2024 | literal sql with bind: |
6e0c6552 |
2025 | |
e41c3bdd |
2026 | my %where = { |
2027 | customer => { -in => \[ |
2028 | 'SELECT cust_id FROM cust WHERE balance > ?', |
2029 | 2000, |
2030 | ], |
2031 | status => { -in => \'SELECT status_codes FROM states' }, |
2032 | }; |
6e0c6552 |
2033 | |
e41c3bdd |
2034 | would generate: |
2035 | |
2036 | $stmt = "WHERE ( |
2037 | customer IN ( SELECT cust_id FROM cust WHERE balance > ? ) |
2038 | AND status IN ( SELECT status_codes FROM states ) |
2039 | )"; |
2040 | @bind = ('2000'); |
2041 | |
2042 | |
2043 | |
2044 | Another pair of operators is C<-between> and C<-not_between>, |
96449e8e |
2045 | used with an arrayref of two values: |
32eab2da |
2046 | |
2047 | my %where = ( |
2048 | user => 'nwiger', |
2049 | completion_date => { |
2050 | -not_between => ['2002-10-01', '2003-02-06'] |
2051 | } |
2052 | ); |
2053 | |
2054 | Would give you: |
2055 | |
2056 | WHERE user = ? AND completion_date NOT BETWEEN ( ? AND ? ) |
2057 | |
e41c3bdd |
2058 | Just like with C<-in> all plausible combinations of literal SQL |
2059 | are possible: |
2060 | |
2061 | my %where = { |
2062 | start0 => { -between => [ 1, 2 ] }, |
2063 | start1 => { -between => \["? AND ?", 1, 2] }, |
2064 | start2 => { -between => \"lower(x) AND upper(y)" }, |
9d48860e |
2065 | start3 => { -between => [ |
e41c3bdd |
2066 | \"lower(x)", |
2067 | \["upper(?)", 'stuff' ], |
2068 | ] }, |
2069 | }; |
2070 | |
2071 | Would give you: |
2072 | |
2073 | $stmt = "WHERE ( |
2074 | ( start0 BETWEEN ? AND ? ) |
2075 | AND ( start1 BETWEEN ? AND ? ) |
2076 | AND ( start2 BETWEEN lower(x) AND upper(y) ) |
2077 | AND ( start3 BETWEEN lower(x) AND upper(?) ) |
2078 | )"; |
2079 | @bind = (1, 2, 1, 2, 'stuff'); |
2080 | |
2081 | |
9d48860e |
2082 | These are the two builtin "special operators"; but the |
96449e8e |
2083 | list can be expanded : see section L</"SPECIAL OPERATORS"> below. |
2084 | |
59f23b3d |
2085 | =head2 Unary operators: bool |
97a920ef |
2086 | |
2087 | If you wish to test against boolean columns or functions within your |
2088 | database you can use the C<-bool> and C<-not_bool> operators. For |
2089 | example to test the column C<is_user> being true and the column |
2090 | <is_enabled> being false you would use:- |
2091 | |
2092 | my %where = ( |
2093 | -bool => 'is_user', |
2094 | -not_bool => 'is_enabled', |
2095 | ); |
2096 | |
2097 | Would give you: |
2098 | |
277b5d3f |
2099 | WHERE is_user AND NOT is_enabled |
97a920ef |
2100 | |
0b604e9d |
2101 | If a more complex combination is required, testing more conditions, |
2102 | then you should use the and/or operators:- |
2103 | |
2104 | my %where = ( |
2105 | -and => [ |
2106 | -bool => 'one', |
2107 | -bool => 'two', |
2108 | -bool => 'three', |
2109 | -not_bool => 'four', |
2110 | ], |
2111 | ); |
2112 | |
2113 | Would give you: |
2114 | |
2115 | WHERE one AND two AND three AND NOT four |
97a920ef |
2116 | |
2117 | |
107b72f1 |
2118 | =head2 Nested conditions, -and/-or prefixes |
96449e8e |
2119 | |
32eab2da |
2120 | So far, we've seen how multiple conditions are joined with a top-level |
2121 | C<AND>. We can change this by putting the different conditions we want in |
2122 | hashes and then putting those hashes in an array. For example: |
2123 | |
2124 | my @where = ( |
2125 | { |
2126 | user => 'nwiger', |
2127 | status => { -like => ['pending%', 'dispatched'] }, |
2128 | }, |
2129 | { |
2130 | user => 'robot', |
2131 | status => 'unassigned', |
2132 | } |
2133 | ); |
2134 | |
2135 | This data structure would create the following: |
2136 | |
2137 | $stmt = "WHERE ( user = ? AND ( status LIKE ? OR status LIKE ? ) ) |
2138 | OR ( user = ? AND status = ? ) )"; |
2139 | @bind = ('nwiger', 'pending', 'dispatched', 'robot', 'unassigned'); |
2140 | |
107b72f1 |
2141 | |
01a01e57 |
2142 | There is also a special C<-nest> |
107b72f1 |
2143 | operator which adds an additional set of parens, to create a subquery. |
2144 | For example, to get something like this: |
2145 | |
2146 | $stmt = "WHERE user = ? AND ( workhrs > ? OR geo = ? )"; |
2147 | @bind = ('nwiger', '20', 'ASIA'); |
2148 | |
2149 | You would do: |
2150 | |
2151 | my %where = ( |
2152 | user => 'nwiger', |
01a01e57 |
2153 | -nest => [ workhrs => {'>', 20}, geo => 'ASIA' ], |
107b72f1 |
2154 | ); |
2155 | |
2156 | |
2157 | Finally, clauses in hashrefs or arrayrefs can be |
7cac25e6 |
2158 | prefixed with an C<-and> or C<-or> to change the logic |
2159 | inside : |
32eab2da |
2160 | |
2161 | my @where = ( |
2162 | -and => [ |
2163 | user => 'nwiger', |
01a01e57 |
2164 | -nest => [ |
7cac25e6 |
2165 | -and => [workhrs => {'>', 20}, geo => 'ASIA' ], |
2166 | -and => [workhrs => {'<', 50}, geo => 'EURO' ] |
32eab2da |
2167 | ], |
2168 | ], |
2169 | ); |
2170 | |
2171 | That would yield: |
2172 | |
9d48860e |
2173 | WHERE ( user = ? AND |
32eab2da |
2174 | ( ( workhrs > ? AND geo = ? ) |
2175 | OR ( workhrs < ? AND geo = ? ) ) ) |
2176 | |
107b72f1 |
2177 | |
2178 | =head2 Algebraic inconsistency, for historical reasons |
2179 | |
7cac25e6 |
2180 | C<Important note>: when connecting several conditions, the C<-and->|C<-or> |
2181 | operator goes C<outside> of the nested structure; whereas when connecting |
2182 | several constraints on one column, the C<-and> operator goes |
2183 | C<inside> the arrayref. Here is an example combining both features : |
2184 | |
2185 | my @where = ( |
2186 | -and => [a => 1, b => 2], |
2187 | -or => [c => 3, d => 4], |
2188 | e => [-and => {-like => 'foo%'}, {-like => '%bar'} ] |
2189 | ) |
2190 | |
2191 | yielding |
2192 | |
9d48860e |
2193 | WHERE ( ( ( a = ? AND b = ? ) |
2194 | OR ( c = ? OR d = ? ) |
7cac25e6 |
2195 | OR ( e LIKE ? AND e LIKE ? ) ) ) |
2196 | |
107b72f1 |
2197 | This difference in syntax is unfortunate but must be preserved for |
2198 | historical reasons. So be careful : the two examples below would |
2199 | seem algebraically equivalent, but they are not |
2200 | |
9d48860e |
2201 | {col => [-and => {-like => 'foo%'}, {-like => '%bar'}]} |
107b72f1 |
2202 | # yields : WHERE ( ( col LIKE ? AND col LIKE ? ) ) |
2203 | |
9d48860e |
2204 | [-and => {col => {-like => 'foo%'}, {col => {-like => '%bar'}}]] |
107b72f1 |
2205 | # yields : WHERE ( ( col LIKE ? OR col LIKE ? ) ) |
2206 | |
7cac25e6 |
2207 | |
96449e8e |
2208 | =head2 Literal SQL |
2209 | |
32eab2da |
2210 | Finally, sometimes only literal SQL will do. If you want to include |
2211 | literal SQL verbatim, you can specify it as a scalar reference, namely: |
2212 | |
2213 | my $inn = 'is Not Null'; |
2214 | my %where = ( |
2215 | priority => { '<', 2 }, |
2216 | requestor => \$inn |
2217 | ); |
2218 | |
2219 | This would create: |
2220 | |
2221 | $stmt = "WHERE priority < ? AND requestor is Not Null"; |
2222 | @bind = ('2'); |
2223 | |
2224 | Note that in this example, you only get one bind parameter back, since |
2225 | the verbatim SQL is passed as part of the statement. |
2226 | |
2227 | Of course, just to prove a point, the above can also be accomplished |
2228 | with this: |
2229 | |
2230 | my %where = ( |
2231 | priority => { '<', 2 }, |
2232 | requestor => { '!=', undef }, |
2233 | ); |
2234 | |
96449e8e |
2235 | |
e41c3bdd |
2236 | TMTOWTDI |
32eab2da |
2237 | |
0b604e9d |
2238 | Conditions on boolean columns can be expressed in the same way, passing |
2239 | a reference to an empty string, however using liternal SQL in this way |
2240 | is deprecated - the preferred method is to use the boolean operators - |
2241 | see L</"Unary operators: bool"> : |
96449e8e |
2242 | |
2243 | my %where = ( |
2244 | priority => { '<', 2 }, |
2245 | is_ready => \""; |
2246 | ); |
2247 | |
2248 | which yields |
2249 | |
2250 | $stmt = "WHERE priority < ? AND is_ready"; |
2251 | @bind = ('2'); |
2252 | |
2253 | |
2254 | =head2 Literal SQL with placeholders and bind values (subqueries) |
2255 | |
2256 | If the literal SQL to be inserted has placeholders and bind values, |
2257 | use a reference to an arrayref (yes this is a double reference -- |
2258 | not so common, but perfectly legal Perl). For example, to find a date |
2259 | in Postgres you can use something like this: |
2260 | |
2261 | my %where = ( |
2262 | date_column => \[q/= date '2008-09-30' - ?::integer/, 10/] |
2263 | ) |
2264 | |
2265 | This would create: |
2266 | |
d2a8fe1a |
2267 | $stmt = "WHERE ( date_column = date '2008-09-30' - ?::integer )" |
96449e8e |
2268 | @bind = ('10'); |
2269 | |
deb148a2 |
2270 | Note that you must pass the bind values in the same format as they are returned |
62552e7d |
2271 | by L</where>. That means that if you set L</bindtype> to C<columns>, you must |
26f2dca5 |
2272 | provide the bind values in the C<< [ column_meta => value ] >> format, where |
2273 | C<column_meta> is an opaque scalar value; most commonly the column name, but |
62552e7d |
2274 | you can use any scalar value (including references and blessed references), |
2275 | L<SQL::Abstract> will simply pass it through intact. So if C<bindtype> is set |
2276 | to C<columns> the above example will look like: |
deb148a2 |
2277 | |
2278 | my %where = ( |
2279 | date_column => \[q/= date '2008-09-30' - ?::integer/, [ dummy => 10 ]/] |
2280 | ) |
96449e8e |
2281 | |
2282 | Literal SQL is especially useful for nesting parenthesized clauses in the |
2283 | main SQL query. Here is a first example : |
2284 | |
2285 | my ($sub_stmt, @sub_bind) = ("SELECT c1 FROM t1 WHERE c2 < ? AND c3 LIKE ?", |
2286 | 100, "foo%"); |
2287 | my %where = ( |
2288 | foo => 1234, |
2289 | bar => \["IN ($sub_stmt)" => @sub_bind], |
2290 | ); |
2291 | |
2292 | This yields : |
2293 | |
9d48860e |
2294 | $stmt = "WHERE (foo = ? AND bar IN (SELECT c1 FROM t1 |
96449e8e |
2295 | WHERE c2 < ? AND c3 LIKE ?))"; |
2296 | @bind = (1234, 100, "foo%"); |
2297 | |
9d48860e |
2298 | Other subquery operators, like for example C<"E<gt> ALL"> or C<"NOT IN">, |
96449e8e |
2299 | are expressed in the same way. Of course the C<$sub_stmt> and |
9d48860e |
2300 | its associated bind values can be generated through a former call |
96449e8e |
2301 | to C<select()> : |
2302 | |
2303 | my ($sub_stmt, @sub_bind) |
9d48860e |
2304 | = $sql->select("t1", "c1", {c2 => {"<" => 100}, |
96449e8e |
2305 | c3 => {-like => "foo%"}}); |
2306 | my %where = ( |
2307 | foo => 1234, |
2308 | bar => \["> ALL ($sub_stmt)" => @sub_bind], |
2309 | ); |
2310 | |
2311 | In the examples above, the subquery was used as an operator on a column; |
9d48860e |
2312 | but the same principle also applies for a clause within the main C<%where> |
96449e8e |
2313 | hash, like an EXISTS subquery : |
2314 | |
9d48860e |
2315 | my ($sub_stmt, @sub_bind) |
96449e8e |
2316 | = $sql->select("t1", "*", {c1 => 1, c2 => \"> t0.c0"}); |
2317 | my %where = ( |
2318 | foo => 1234, |
01a01e57 |
2319 | -nest => \["EXISTS ($sub_stmt)" => @sub_bind], |
96449e8e |
2320 | ); |
2321 | |
2322 | which yields |
2323 | |
9d48860e |
2324 | $stmt = "WHERE (foo = ? AND EXISTS (SELECT * FROM t1 |
96449e8e |
2325 | WHERE c1 = ? AND c2 > t0.c0))"; |
2326 | @bind = (1234, 1); |
2327 | |
2328 | |
9d48860e |
2329 | Observe that the condition on C<c2> in the subquery refers to |
2330 | column C<t0.c0> of the main query : this is I<not> a bind |
2331 | value, so we have to express it through a scalar ref. |
96449e8e |
2332 | Writing C<< c2 => {">" => "t0.c0"} >> would have generated |
2333 | C<< c2 > ? >> with bind value C<"t0.c0"> ... not exactly |
2334 | what we wanted here. |
2335 | |
2336 | Another use of the subquery technique is when some SQL clauses need |
2337 | parentheses, as it often occurs with some proprietary SQL extensions |
9d48860e |
2338 | like for example fulltext expressions, geospatial expressions, |
96449e8e |
2339 | NATIVE clauses, etc. Here is an example of a fulltext query in MySQL : |
2340 | |
2341 | my %where = ( |
01a01e57 |
2342 | -nest => \["MATCH (col1, col2) AGAINST (?)" => qw/apples/] |
96449e8e |
2343 | ); |
2344 | |
2345 | Finally, here is an example where a subquery is used |
2346 | for expressing unary negation: |
2347 | |
9d48860e |
2348 | my ($sub_stmt, @sub_bind) |
96449e8e |
2349 | = $sql->where({age => [{"<" => 10}, {">" => 20}]}); |
2350 | $sub_stmt =~ s/^ where //i; # don't want "WHERE" in the subclause |
2351 | my %where = ( |
2352 | lname => {like => '%son%'}, |
01a01e57 |
2353 | -nest => \["NOT ($sub_stmt)" => @sub_bind], |
96449e8e |
2354 | ); |
2355 | |
2356 | This yields |
2357 | |
2358 | $stmt = "lname LIKE ? AND NOT ( age < ? OR age > ? )" |
2359 | @bind = ('%son%', 10, 20) |
2360 | |
2361 | |
2362 | |
2363 | =head2 Conclusion |
2364 | |
32eab2da |
2365 | These pages could go on for a while, since the nesting of the data |
2366 | structures this module can handle are pretty much unlimited (the |
2367 | module implements the C<WHERE> expansion as a recursive function |
2368 | internally). Your best bet is to "play around" with the module a |
2369 | little to see how the data structures behave, and choose the best |
2370 | format for your data based on that. |
2371 | |
2372 | And of course, all the values above will probably be replaced with |
2373 | variables gotten from forms or the command line. After all, if you |
2374 | knew everything ahead of time, you wouldn't have to worry about |
2375 | dynamically-generating SQL and could just hardwire it into your |
2376 | script. |
2377 | |
96449e8e |
2378 | |
2379 | |
2380 | |
86298391 |
2381 | =head1 ORDER BY CLAUSES |
2382 | |
9d48860e |
2383 | Some functions take an order by clause. This can either be a scalar (just a |
86298391 |
2384 | column name,) a hash of C<< { -desc => 'col' } >> or C<< { -asc => 'col' } >>, |
1cfa1db3 |
2385 | or an array of either of the two previous forms. Examples: |
2386 | |
952f9e2d |
2387 | Given | Will Generate |
1cfa1db3 |
2388 | ---------------------------------------------------------- |
952f9e2d |
2389 | | |
2390 | \'colA DESC' | ORDER BY colA DESC |
2391 | | |
2392 | 'colA' | ORDER BY colA |
2393 | | |
2394 | [qw/colA colB/] | ORDER BY colA, colB |
2395 | | |
2396 | {-asc => 'colA'} | ORDER BY colA ASC |
2397 | | |
2398 | {-desc => 'colB'} | ORDER BY colB DESC |
2399 | | |
2400 | ['colA', {-asc => 'colB'}] | ORDER BY colA, colB ASC |
2401 | | |
855e6047 |
2402 | { -asc => [qw/colA colB/] } | ORDER BY colA ASC, colB ASC |
952f9e2d |
2403 | | |
2404 | [ | |
2405 | { -asc => 'colA' }, | ORDER BY colA ASC, colB DESC, |
2406 | { -desc => [qw/colB/], | colC ASC, colD ASC |
2407 | { -asc => [qw/colC colD/],| |
2408 | ] | |
2409 | =========================================================== |
86298391 |
2410 | |
96449e8e |
2411 | |
2412 | |
2413 | =head1 SPECIAL OPERATORS |
2414 | |
e3f9dff4 |
2415 | my $sqlmaker = SQL::Abstract->new(special_ops => [ |
3a2e1a5e |
2416 | { |
2417 | regex => qr/.../, |
e3f9dff4 |
2418 | handler => sub { |
2419 | my ($self, $field, $op, $arg) = @_; |
2420 | ... |
3a2e1a5e |
2421 | }, |
2422 | }, |
2423 | { |
2424 | regex => qr/.../, |
2425 | handler => 'method_name', |
e3f9dff4 |
2426 | }, |
2427 | ]); |
2428 | |
9d48860e |
2429 | A "special operator" is a SQL syntactic clause that can be |
e3f9dff4 |
2430 | applied to a field, instead of a usual binary operator. |
9d48860e |
2431 | For example : |
e3f9dff4 |
2432 | |
2433 | WHERE field IN (?, ?, ?) |
2434 | WHERE field BETWEEN ? AND ? |
2435 | WHERE MATCH(field) AGAINST (?, ?) |
96449e8e |
2436 | |
e3f9dff4 |
2437 | Special operators IN and BETWEEN are fairly standard and therefore |
3a2e1a5e |
2438 | are builtin within C<SQL::Abstract> (as the overridable methods |
2439 | C<_where_field_IN> and C<_where_field_BETWEEN>). For other operators, |
2440 | like the MATCH .. AGAINST example above which is specific to MySQL, |
2441 | you can write your own operator handlers - supply a C<special_ops> |
2442 | argument to the C<new> method. That argument takes an arrayref of |
2443 | operator definitions; each operator definition is a hashref with two |
2444 | entries: |
96449e8e |
2445 | |
e3f9dff4 |
2446 | =over |
2447 | |
2448 | =item regex |
2449 | |
2450 | the regular expression to match the operator |
96449e8e |
2451 | |
e3f9dff4 |
2452 | =item handler |
2453 | |
3a2e1a5e |
2454 | Either a coderef or a plain scalar method name. In both cases |
2455 | the expected return is C<< ($sql, @bind) >>. |
2456 | |
2457 | When supplied with a method name, it is simply called on the |
2458 | L<SQL::Abstract/> object as: |
2459 | |
2460 | $self->$method_name ($field, $op, $arg) |
2461 | |
2462 | Where: |
2463 | |
2464 | $op is the part that matched the handler regex |
2465 | $field is the LHS of the operator |
2466 | $arg is the RHS |
2467 | |
2468 | When supplied with a coderef, it is called as: |
2469 | |
2470 | $coderef->($self, $field, $op, $arg) |
2471 | |
e3f9dff4 |
2472 | |
2473 | =back |
2474 | |
9d48860e |
2475 | For example, here is an implementation |
e3f9dff4 |
2476 | of the MATCH .. AGAINST syntax for MySQL |
2477 | |
2478 | my $sqlmaker = SQL::Abstract->new(special_ops => [ |
9d48860e |
2479 | |
e3f9dff4 |
2480 | # special op for MySql MATCH (field) AGAINST(word1, word2, ...) |
9d48860e |
2481 | {regex => qr/^match$/i, |
e3f9dff4 |
2482 | handler => sub { |
2483 | my ($self, $field, $op, $arg) = @_; |
2484 | $arg = [$arg] if not ref $arg; |
2485 | my $label = $self->_quote($field); |
2486 | my ($placeholder) = $self->_convert('?'); |
2487 | my $placeholders = join ", ", (($placeholder) x @$arg); |
2488 | my $sql = $self->_sqlcase('match') . " ($label) " |
2489 | . $self->_sqlcase('against') . " ($placeholders) "; |
2490 | my @bind = $self->_bindtype($field, @$arg); |
2491 | return ($sql, @bind); |
2492 | } |
2493 | }, |
9d48860e |
2494 | |
e3f9dff4 |
2495 | ]); |
96449e8e |
2496 | |
2497 | |
59f23b3d |
2498 | =head1 UNARY OPERATORS |
2499 | |
112b5232 |
2500 | my $sqlmaker = SQL::Abstract->new(unary_ops => [ |
59f23b3d |
2501 | { |
2502 | regex => qr/.../, |
2503 | handler => sub { |
2504 | my ($self, $op, $arg) = @_; |
2505 | ... |
2506 | }, |
2507 | }, |
2508 | { |
2509 | regex => qr/.../, |
2510 | handler => 'method_name', |
2511 | }, |
2512 | ]); |
2513 | |
9d48860e |
2514 | A "unary operator" is a SQL syntactic clause that can be |
59f23b3d |
2515 | applied to a field - the operator goes before the field |
2516 | |
2517 | You can write your own operator handlers - supply a C<unary_ops> |
2518 | argument to the C<new> method. That argument takes an arrayref of |
2519 | operator definitions; each operator definition is a hashref with two |
2520 | entries: |
2521 | |
2522 | =over |
2523 | |
2524 | =item regex |
2525 | |
2526 | the regular expression to match the operator |
2527 | |
2528 | =item handler |
2529 | |
2530 | Either a coderef or a plain scalar method name. In both cases |
2531 | the expected return is C<< $sql >>. |
2532 | |
2533 | When supplied with a method name, it is simply called on the |
2534 | L<SQL::Abstract/> object as: |
2535 | |
2536 | $self->$method_name ($op, $arg) |
2537 | |
2538 | Where: |
2539 | |
2540 | $op is the part that matched the handler regex |
2541 | $arg is the RHS or argument of the operator |
2542 | |
2543 | When supplied with a coderef, it is called as: |
2544 | |
2545 | $coderef->($self, $op, $arg) |
2546 | |
2547 | |
2548 | =back |
2549 | |
2550 | |
32eab2da |
2551 | =head1 PERFORMANCE |
2552 | |
2553 | Thanks to some benchmarking by Mark Stosberg, it turns out that |
2554 | this module is many orders of magnitude faster than using C<DBIx::Abstract>. |
2555 | I must admit this wasn't an intentional design issue, but it's a |
2556 | byproduct of the fact that you get to control your C<DBI> handles |
2557 | yourself. |
2558 | |
2559 | To maximize performance, use a code snippet like the following: |
2560 | |
2561 | # prepare a statement handle using the first row |
2562 | # and then reuse it for the rest of the rows |
2563 | my($sth, $stmt); |
2564 | for my $href (@array_of_hashrefs) { |
2565 | $stmt ||= $sql->insert('table', $href); |
2566 | $sth ||= $dbh->prepare($stmt); |
2567 | $sth->execute($sql->values($href)); |
2568 | } |
2569 | |
2570 | The reason this works is because the keys in your C<$href> are sorted |
2571 | internally by B<SQL::Abstract>. Thus, as long as your data retains |
2572 | the same structure, you only have to generate the SQL the first time |
2573 | around. On subsequent queries, simply use the C<values> function provided |
2574 | by this module to return your values in the correct order. |
2575 | |
96449e8e |
2576 | |
32eab2da |
2577 | =head1 FORMBUILDER |
2578 | |
2579 | If you use my C<CGI::FormBuilder> module at all, you'll hopefully |
2580 | really like this part (I do, at least). Building up a complex query |
2581 | can be as simple as the following: |
2582 | |
2583 | #!/usr/bin/perl |
2584 | |
2585 | use CGI::FormBuilder; |
2586 | use SQL::Abstract; |
2587 | |
2588 | my $form = CGI::FormBuilder->new(...); |
2589 | my $sql = SQL::Abstract->new; |
2590 | |
2591 | if ($form->submitted) { |
2592 | my $field = $form->field; |
2593 | my $id = delete $field->{id}; |
2594 | my($stmt, @bind) = $sql->update('table', $field, {id => $id}); |
2595 | } |
2596 | |
2597 | Of course, you would still have to connect using C<DBI> to run the |
2598 | query, but the point is that if you make your form look like your |
2599 | table, the actual query script can be extremely simplistic. |
2600 | |
2601 | If you're B<REALLY> lazy (I am), check out C<HTML::QuickTable> for |
9d48860e |
2602 | a fast interface to returning and formatting data. I frequently |
32eab2da |
2603 | use these three modules together to write complex database query |
2604 | apps in under 50 lines. |
2605 | |
32eab2da |
2606 | |
96449e8e |
2607 | =head1 CHANGES |
2608 | |
2609 | Version 1.50 was a major internal refactoring of C<SQL::Abstract>. |
2610 | Great care has been taken to preserve the I<published> behavior |
2611 | documented in previous versions in the 1.* family; however, |
9d48860e |
2612 | some features that were previously undocumented, or behaved |
96449e8e |
2613 | differently from the documentation, had to be changed in order |
2614 | to clarify the semantics. Hence, client code that was relying |
9d48860e |
2615 | on some dark areas of C<SQL::Abstract> v1.* |
96449e8e |
2616 | B<might behave differently> in v1.50. |
32eab2da |
2617 | |
d2a8fe1a |
2618 | The main changes are : |
2619 | |
96449e8e |
2620 | =over |
32eab2da |
2621 | |
9d48860e |
2622 | =item * |
32eab2da |
2623 | |
96449e8e |
2624 | support for literal SQL through the C<< \ [$sql, bind] >> syntax. |
2625 | |
2626 | =item * |
2627 | |
145fbfc8 |
2628 | support for the { operator => \"..." } construct (to embed literal SQL) |
2629 | |
2630 | =item * |
2631 | |
9c37b9c0 |
2632 | support for the { operator => \["...", @bind] } construct (to embed literal SQL with bind values) |
2633 | |
2634 | =item * |
2635 | |
96449e8e |
2636 | optional support for L<array datatypes|/"Inserting and Updating Arrays"> |
2637 | |
9d48860e |
2638 | =item * |
96449e8e |
2639 | |
2640 | defensive programming : check arguments |
2641 | |
2642 | =item * |
2643 | |
2644 | fixed bug with global logic, which was previously implemented |
7cac25e6 |
2645 | through global variables yielding side-effects. Prior versions would |
96449e8e |
2646 | interpret C<< [ {cond1, cond2}, [cond3, cond4] ] >> |
2647 | as C<< "(cond1 AND cond2) OR (cond3 AND cond4)" >>. |
2648 | Now this is interpreted |
2649 | as C<< "(cond1 AND cond2) OR (cond3 OR cond4)" >>. |
2650 | |
96449e8e |
2651 | |
2652 | =item * |
2653 | |
2654 | fixed semantics of _bindtype on array args |
2655 | |
9d48860e |
2656 | =item * |
96449e8e |
2657 | |
2658 | dropped the C<_anoncopy> of the %where tree. No longer necessary, |
2659 | we just avoid shifting arrays within that tree. |
2660 | |
2661 | =item * |
2662 | |
2663 | dropped the C<_modlogic> function |
2664 | |
2665 | =back |
32eab2da |
2666 | |
32eab2da |
2667 | |
32eab2da |
2668 | |
2669 | =head1 ACKNOWLEDGEMENTS |
2670 | |
2671 | There are a number of individuals that have really helped out with |
2672 | this module. Unfortunately, most of them submitted bugs via CPAN |
2673 | so I have no idea who they are! But the people I do know are: |
2674 | |
9d48860e |
2675 | Ash Berlin (order_by hash term support) |
b643abe1 |
2676 | Matt Trout (DBIx::Class support) |
32eab2da |
2677 | Mark Stosberg (benchmarking) |
2678 | Chas Owens (initial "IN" operator support) |
2679 | Philip Collins (per-field SQL functions) |
2680 | Eric Kolve (hashref "AND" support) |
2681 | Mike Fragassi (enhancements to "BETWEEN" and "LIKE") |
2682 | Dan Kubb (support for "quote_char" and "name_sep") |
f5aab26e |
2683 | Guillermo Roditi (patch to cleanup "IN" and "BETWEEN", fix and tests for _order_by) |
01a01e57 |
2684 | Laurent Dami (internal refactoring, multiple -nest, extensible list of special operators, literal SQL) |
dbdf7648 |
2685 | Norbert Buchmuller (support for literal SQL in hashpair, misc. fixes & tests) |
e96c510a |
2686 | Peter Rabbitson (rewrite of SQLA::Test, misc. fixes & tests) |
02288357 |
2687 | Oliver Charles (support for "RETURNING" after "INSERT") |
32eab2da |
2688 | |
2689 | Thanks! |
2690 | |
32eab2da |
2691 | =head1 SEE ALSO |
2692 | |
86298391 |
2693 | L<DBIx::Class>, L<DBIx::Abstract>, L<CGI::FormBuilder>, L<HTML::QuickTable>. |
32eab2da |
2694 | |
32eab2da |
2695 | =head1 AUTHOR |
2696 | |
b643abe1 |
2697 | Copyright (c) 2001-2007 Nathan Wiger <nwiger@cpan.org>. All Rights Reserved. |
2698 | |
2699 | This module is actively maintained by Matt Trout <mst@shadowcatsystems.co.uk> |
32eab2da |
2700 | |
abe72f94 |
2701 | For support, your best bet is to try the C<DBIx::Class> users mailing list. |
2702 | While not an official support venue, C<DBIx::Class> makes heavy use of |
2703 | C<SQL::Abstract>, and as such list members there are very familiar with |
2704 | how to create queries. |
2705 | |
0d067ded |
2706 | =head1 LICENSE |
2707 | |
d988ab87 |
2708 | This module is free software; you may copy this under the same |
2709 | terms as perl itself (either the GNU General Public License or |
2710 | the Artistic License) |
32eab2da |
2711 | |
2712 | =cut |
2713 | |