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 | |
96449e8e |
8 | use strict; |
9 | use warnings; |
9d9d5bd6 |
10 | use Carp (); |
312d830b |
11 | use List::Util (); |
12 | use Scalar::Util (); |
96449e8e |
13 | |
14 | #====================================================================== |
15 | # GLOBALS |
16 | #====================================================================== |
17 | |
3cb8f017 |
18 | our $VERSION = '1.74'; |
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 = ( |
b8db59b8 |
28 | {regex => qr/^ (?: not \s )? between $/ix, handler => '_where_field_BETWEEN'}, |
29 | {regex => qr/^ (?: not \s )? in $/ix, handler => '_where_field_IN'}, |
cc422895 |
30 | {regex => qr/^ ident $/ix, handler => '_where_op_IDENT'}, |
31 | {regex => qr/^ value $/ix, handler => '_where_op_VALUE'}, |
96449e8e |
32 | ); |
33 | |
97a920ef |
34 | # unaryish operators - key maps to handler |
59f23b3d |
35 | my @BUILTIN_UNARY_OPS = ( |
a47b433a |
36 | # the digits are backcompat stuff |
b8db59b8 |
37 | { regex => qr/^ and (?: [_\s]? \d+ )? $/xi, handler => '_where_op_ANDOR' }, |
38 | { regex => qr/^ or (?: [_\s]? \d+ )? $/xi, handler => '_where_op_ANDOR' }, |
39 | { regex => qr/^ nest (?: [_\s]? \d+ )? $/xi, handler => '_where_op_NEST' }, |
40 | { regex => qr/^ (?: not \s )? bool $/xi, handler => '_where_op_BOOL' }, |
cc422895 |
41 | { regex => qr/^ ident $/xi, handler => '_where_op_IDENT' }, |
42 | { regex => qr/^ value $/ix, handler => '_where_op_VALUE' }, |
59f23b3d |
43 | ); |
97a920ef |
44 | |
96449e8e |
45 | #====================================================================== |
46 | # DEBUGGING AND ERROR REPORTING |
47 | #====================================================================== |
48 | |
49 | sub _debug { |
50 | return unless $_[0]->{debug}; shift; # a little faster |
51 | my $func = (caller(1))[3]; |
52 | warn "[$func] ", @_, "\n"; |
53 | } |
54 | |
55 | sub belch (@) { |
56 | my($func) = (caller(1))[3]; |
9d9d5bd6 |
57 | Carp::carp "[$func] Warning: ", @_; |
96449e8e |
58 | } |
59 | |
60 | sub puke (@) { |
61 | my($func) = (caller(1))[3]; |
9d9d5bd6 |
62 | Carp::croak "[$func] Fatal: ", @_; |
96449e8e |
63 | } |
64 | |
65 | |
66 | #====================================================================== |
67 | # NEW |
68 | #====================================================================== |
69 | |
70 | sub new { |
71 | my $self = shift; |
72 | my $class = ref($self) || $self; |
73 | my %opt = (ref $_[0] eq 'HASH') ? %{$_[0]} : @_; |
74 | |
75 | # choose our case by keeping an option around |
76 | delete $opt{case} if $opt{case} && $opt{case} ne 'lower'; |
77 | |
78 | # default logic for interpreting arrayrefs |
ef559da3 |
79 | $opt{logic} = $opt{logic} ? uc $opt{logic} : 'OR'; |
96449e8e |
80 | |
81 | # how to return bind vars |
82 | # LDNOTE: changed nwiger code : why this 'delete' ?? |
83 | # $opt{bindtype} ||= delete($opt{bind_type}) || 'normal'; |
84 | $opt{bindtype} ||= 'normal'; |
85 | |
86 | # default comparison is "=", but can be overridden |
87 | $opt{cmp} ||= '='; |
88 | |
3af02ccb |
89 | # try to recognize which are the 'equality' and 'inequality' ops |
96449e8e |
90 | # (temporary quickfix, should go through a more seasoned API) |
2281c758 |
91 | $opt{equality_op} = qr/^(\Q$opt{cmp}\E|is|(is\s+)?like)$/i; |
92 | $opt{inequality_op} = qr/^(!=|<>|(is\s+)?not(\s+like)?)$/i; |
96449e8e |
93 | |
94 | # SQL booleans |
95 | $opt{sqltrue} ||= '1=1'; |
96 | $opt{sqlfalse} ||= '0=1'; |
97 | |
9d48860e |
98 | # special operators |
96449e8e |
99 | $opt{special_ops} ||= []; |
b6251592 |
100 | # regexes are applied in order, thus push after user-defines |
96449e8e |
101 | push @{$opt{special_ops}}, @BUILTIN_SPECIAL_OPS; |
102 | |
9d48860e |
103 | # unary operators |
59f23b3d |
104 | $opt{unary_ops} ||= []; |
105 | push @{$opt{unary_ops}}, @BUILTIN_UNARY_OPS; |
106 | |
3af02ccb |
107 | # rudimentary sanity-check for user supplied bits treated as functions/operators |
b6251592 |
108 | # If a purported function matches this regular expression, an exception is thrown. |
109 | # Literal SQL is *NOT* subject to this check, only functions (and column names |
110 | # when quoting is not in effect) |
96449e8e |
111 | |
b6251592 |
112 | # FIXME |
113 | # need to guard against ()'s in column names too, but this will break tons of |
114 | # hacks... ideas anyone? |
115 | $opt{injection_guard} ||= qr/ |
116 | \; |
117 | | |
118 | ^ \s* go \s |
119 | /xmi; |
96449e8e |
120 | |
b6251592 |
121 | return bless \%opt, $class; |
122 | } |
96449e8e |
123 | |
170e6c33 |
124 | |
125 | sub _assert_pass_injection_guard { |
126 | if ($_[1] =~ $_[0]->{injection_guard}) { |
127 | my $class = ref $_[0]; |
128 | puke "Possible SQL injection attempt '$_[1]'. If this is indeed a part of the " |
129 | . "desired SQL use literal SQL ( \'...' or \[ '...' ] ) or supply your own " |
130 | . "{injection_guard} attribute to ${class}->new()" |
131 | } |
132 | } |
133 | |
134 | |
96449e8e |
135 | #====================================================================== |
136 | # INSERT methods |
137 | #====================================================================== |
138 | |
139 | sub insert { |
02288357 |
140 | my $self = shift; |
141 | my $table = $self->_table(shift); |
142 | my $data = shift || return; |
143 | my $options = shift; |
96449e8e |
144 | |
145 | my $method = $self->_METHOD_FOR_refkind("_insert", $data); |
02288357 |
146 | my ($sql, @bind) = $self->$method($data); |
96449e8e |
147 | $sql = join " ", $self->_sqlcase('insert into'), $table, $sql; |
02288357 |
148 | |
e82e648a |
149 | if ($options->{returning}) { |
150 | my ($s, @b) = $self->_insert_returning ($options); |
151 | $sql .= $s; |
152 | push @bind, @b; |
02288357 |
153 | } |
154 | |
96449e8e |
155 | return wantarray ? ($sql, @bind) : $sql; |
156 | } |
157 | |
6b1fe79d |
158 | sub _insert_returning { |
e82e648a |
159 | my ($self, $options) = @_; |
6b1fe79d |
160 | |
e82e648a |
161 | my $f = $options->{returning}; |
162 | |
163 | my $fieldlist = $self->_SWITCH_refkind($f, { |
164 | ARRAYREF => sub {join ', ', map { $self->_quote($_) } @$f;}, |
165 | SCALAR => sub {$self->_quote($f)}, |
166 | SCALARREF => sub {$$f}, |
6b1fe79d |
167 | }); |
e82e648a |
168 | return $self->_sqlcase(' returning ') . $fieldlist; |
6b1fe79d |
169 | } |
170 | |
96449e8e |
171 | sub _insert_HASHREF { # explicit list of fields and then values |
172 | my ($self, $data) = @_; |
173 | |
174 | my @fields = sort keys %$data; |
175 | |
fe3ae272 |
176 | my ($sql, @bind) = $self->_insert_values($data); |
96449e8e |
177 | |
178 | # assemble SQL |
179 | $_ = $self->_quote($_) foreach @fields; |
180 | $sql = "( ".join(", ", @fields).") ".$sql; |
181 | |
182 | return ($sql, @bind); |
183 | } |
184 | |
185 | sub _insert_ARRAYREF { # just generate values(?,?) part (no list of fields) |
186 | my ($self, $data) = @_; |
187 | |
188 | # no names (arrayref) so can't generate bindtype |
189 | $self->{bindtype} ne 'columns' |
190 | or belch "can't do 'columns' bindtype when called with arrayref"; |
191 | |
fe3ae272 |
192 | # fold the list of values into a hash of column name - value pairs |
193 | # (where the column names are artificially generated, and their |
194 | # lexicographical ordering keep the ordering of the original list) |
195 | my $i = "a"; # incremented values will be in lexicographical order |
196 | my $data_in_hash = { map { ($i++ => $_) } @$data }; |
197 | |
198 | return $self->_insert_values($data_in_hash); |
199 | } |
200 | |
201 | sub _insert_ARRAYREFREF { # literal SQL with bind |
202 | my ($self, $data) = @_; |
203 | |
204 | my ($sql, @bind) = @${$data}; |
205 | $self->_assert_bindval_matches_bindtype(@bind); |
206 | |
207 | return ($sql, @bind); |
208 | } |
209 | |
210 | |
211 | sub _insert_SCALARREF { # literal SQL without bind |
212 | my ($self, $data) = @_; |
213 | |
214 | return ($$data); |
215 | } |
216 | |
217 | sub _insert_values { |
218 | my ($self, $data) = @_; |
219 | |
96449e8e |
220 | my (@values, @all_bind); |
fe3ae272 |
221 | foreach my $column (sort keys %$data) { |
222 | my $v = $data->{$column}; |
96449e8e |
223 | |
224 | $self->_SWITCH_refkind($v, { |
225 | |
9d48860e |
226 | ARRAYREF => sub { |
96449e8e |
227 | if ($self->{array_datatypes}) { # if array datatype are activated |
228 | push @values, '?'; |
fe3ae272 |
229 | push @all_bind, $self->_bindtype($column, $v); |
96449e8e |
230 | } |
231 | else { # else literal SQL with bind |
232 | my ($sql, @bind) = @$v; |
fe3ae272 |
233 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
234 | push @values, $sql; |
235 | push @all_bind, @bind; |
236 | } |
237 | }, |
238 | |
239 | ARRAYREFREF => sub { # literal SQL with bind |
240 | my ($sql, @bind) = @${$v}; |
fe3ae272 |
241 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
242 | push @values, $sql; |
243 | push @all_bind, @bind; |
244 | }, |
245 | |
9d48860e |
246 | # THINK : anything useful to do with a HASHREF ? |
5db47f9f |
247 | HASHREF => sub { # (nothing, but old SQLA passed it through) |
248 | #TODO in SQLA >= 2.0 it will die instead |
249 | belch "HASH ref as bind value in insert is not supported"; |
250 | push @values, '?'; |
fe3ae272 |
251 | push @all_bind, $self->_bindtype($column, $v); |
5db47f9f |
252 | }, |
96449e8e |
253 | |
254 | SCALARREF => sub { # literal SQL without bind |
255 | push @values, $$v; |
256 | }, |
257 | |
258 | SCALAR_or_UNDEF => sub { |
259 | push @values, '?'; |
fe3ae272 |
260 | push @all_bind, $self->_bindtype($column, $v); |
96449e8e |
261 | }, |
262 | |
263 | }); |
264 | |
265 | } |
266 | |
267 | my $sql = $self->_sqlcase('values')." ( ".join(", ", @values)." )"; |
268 | return ($sql, @all_bind); |
269 | } |
270 | |
271 | |
96449e8e |
272 | |
273 | #====================================================================== |
274 | # UPDATE methods |
275 | #====================================================================== |
276 | |
277 | |
278 | sub update { |
279 | my $self = shift; |
280 | my $table = $self->_table(shift); |
281 | my $data = shift || return; |
282 | my $where = shift; |
283 | |
284 | # first build the 'SET' part of the sql statement |
285 | my (@set, @all_bind); |
286 | puke "Unsupported data type specified to \$sql->update" |
287 | unless ref $data eq 'HASH'; |
288 | |
289 | for my $k (sort keys %$data) { |
290 | my $v = $data->{$k}; |
291 | my $r = ref $v; |
292 | my $label = $self->_quote($k); |
293 | |
294 | $self->_SWITCH_refkind($v, { |
9d48860e |
295 | ARRAYREF => sub { |
96449e8e |
296 | if ($self->{array_datatypes}) { # array datatype |
297 | push @set, "$label = ?"; |
298 | push @all_bind, $self->_bindtype($k, $v); |
299 | } |
300 | else { # literal SQL with bind |
301 | my ($sql, @bind) = @$v; |
fe3ae272 |
302 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
303 | push @set, "$label = $sql"; |
fe3ae272 |
304 | push @all_bind, @bind; |
96449e8e |
305 | } |
306 | }, |
307 | ARRAYREFREF => sub { # literal SQL with bind |
308 | my ($sql, @bind) = @${$v}; |
fe3ae272 |
309 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
310 | push @set, "$label = $sql"; |
fe3ae272 |
311 | push @all_bind, @bind; |
96449e8e |
312 | }, |
313 | SCALARREF => sub { # literal SQL without bind |
314 | push @set, "$label = $$v"; |
0ec3aec7 |
315 | }, |
316 | HASHREF => sub { |
317 | my ($op, $arg, @rest) = %$v; |
318 | |
319 | puke 'Operator calls in update must be in the form { -op => $arg }' |
320 | if (@rest or not $op =~ /^\-(.+)/); |
321 | |
322 | local $self->{_nested_func_lhs} = $k; |
323 | my ($sql, @bind) = $self->_where_unary_op ($1, $arg); |
324 | |
325 | push @set, "$label = $sql"; |
326 | push @all_bind, @bind; |
327 | }, |
96449e8e |
328 | SCALAR_or_UNDEF => sub { |
329 | push @set, "$label = ?"; |
330 | push @all_bind, $self->_bindtype($k, $v); |
331 | }, |
332 | }); |
333 | } |
334 | |
335 | # generate sql |
336 | my $sql = $self->_sqlcase('update') . " $table " . $self->_sqlcase('set ') |
337 | . join ', ', @set; |
338 | |
339 | if ($where) { |
340 | my($where_sql, @where_bind) = $self->where($where); |
341 | $sql .= $where_sql; |
342 | push @all_bind, @where_bind; |
343 | } |
344 | |
345 | return wantarray ? ($sql, @all_bind) : $sql; |
346 | } |
347 | |
348 | |
349 | |
350 | |
351 | #====================================================================== |
352 | # SELECT |
353 | #====================================================================== |
354 | |
355 | |
356 | sub select { |
357 | my $self = shift; |
358 | my $table = $self->_table(shift); |
359 | my $fields = shift || '*'; |
360 | my $where = shift; |
361 | my $order = shift; |
362 | |
363 | my($where_sql, @bind) = $self->where($where, $order); |
364 | |
365 | my $f = (ref $fields eq 'ARRAY') ? join ', ', map { $self->_quote($_) } @$fields |
366 | : $fields; |
9d48860e |
367 | my $sql = join(' ', $self->_sqlcase('select'), $f, |
96449e8e |
368 | $self->_sqlcase('from'), $table) |
369 | . $where_sql; |
370 | |
9d48860e |
371 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
372 | } |
373 | |
374 | #====================================================================== |
375 | # DELETE |
376 | #====================================================================== |
377 | |
378 | |
379 | sub delete { |
380 | my $self = shift; |
381 | my $table = $self->_table(shift); |
382 | my $where = shift; |
383 | |
384 | |
385 | my($where_sql, @bind) = $self->where($where); |
386 | my $sql = $self->_sqlcase('delete from') . " $table" . $where_sql; |
387 | |
9d48860e |
388 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
389 | } |
390 | |
391 | |
392 | #====================================================================== |
393 | # WHERE: entry point |
394 | #====================================================================== |
395 | |
396 | |
397 | |
398 | # Finally, a separate routine just to handle WHERE clauses |
399 | sub where { |
400 | my ($self, $where, $order) = @_; |
401 | |
402 | # where ? |
403 | my ($sql, @bind) = $self->_recurse_where($where); |
404 | $sql = $sql ? $self->_sqlcase(' where ') . "( $sql )" : ''; |
405 | |
406 | # order by? |
407 | if ($order) { |
408 | $sql .= $self->_order_by($order); |
409 | } |
410 | |
9d48860e |
411 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
412 | } |
413 | |
414 | |
415 | sub _recurse_where { |
416 | my ($self, $where, $logic) = @_; |
417 | |
418 | # dispatch on appropriate method according to refkind of $where |
419 | my $method = $self->_METHOD_FOR_refkind("_where", $where); |
311b2151 |
420 | |
9d48860e |
421 | my ($sql, @bind) = $self->$method($where, $logic); |
311b2151 |
422 | |
9d48860e |
423 | # DBIx::Class directly calls _recurse_where in scalar context, so |
311b2151 |
424 | # we must implement it, even if not in the official API |
9d48860e |
425 | return wantarray ? ($sql, @bind) : $sql; |
96449e8e |
426 | } |
427 | |
428 | |
429 | |
430 | #====================================================================== |
431 | # WHERE: top-level ARRAYREF |
432 | #====================================================================== |
433 | |
434 | |
435 | sub _where_ARRAYREF { |
5e1d09d5 |
436 | my ($self, $where, $logic) = @_; |
96449e8e |
437 | |
5e1d09d5 |
438 | $logic = uc($logic || $self->{logic}); |
96449e8e |
439 | $logic eq 'AND' or $logic eq 'OR' or puke "unknown logic: $logic"; |
440 | |
441 | my @clauses = @$where; |
442 | |
96449e8e |
443 | my (@sql_clauses, @all_bind); |
96449e8e |
444 | # need to use while() so can shift() for pairs |
9d48860e |
445 | while (my $el = shift @clauses) { |
96449e8e |
446 | |
447 | # switch according to kind of $el and get corresponding ($sql, @bind) |
448 | my ($sql, @bind) = $self->_SWITCH_refkind($el, { |
449 | |
450 | # skip empty elements, otherwise get invalid trailing AND stuff |
451 | ARRAYREF => sub {$self->_recurse_where($el) if @$el}, |
452 | |
c94a6c93 |
453 | ARRAYREFREF => sub { |
454 | my ($s, @b) = @$$el; |
455 | $self->_assert_bindval_matches_bindtype(@b); |
456 | ($s, @b); |
457 | }, |
474e3335 |
458 | |
96449e8e |
459 | HASHREF => sub {$self->_recurse_where($el, 'and') if %$el}, |
460 | # LDNOTE : previous SQLA code for hashrefs was creating a dirty |
461 | # side-effect: the first hashref within an array would change |
462 | # the global logic to 'AND'. So [ {cond1, cond2}, [cond3, cond4] ] |
9d48860e |
463 | # was interpreted as "(cond1 AND cond2) OR (cond3 AND cond4)", |
96449e8e |
464 | # whereas it should be "(cond1 AND cond2) OR (cond3 OR cond4)". |
465 | |
466 | SCALARREF => sub { ($$el); }, |
467 | |
468 | SCALAR => sub {# top-level arrayref with scalars, recurse in pairs |
469 | $self->_recurse_where({$el => shift(@clauses)})}, |
470 | |
471 | UNDEF => sub {puke "not supported : UNDEF in arrayref" }, |
472 | }); |
473 | |
4b7b6026 |
474 | if ($sql) { |
475 | push @sql_clauses, $sql; |
476 | push @all_bind, @bind; |
477 | } |
96449e8e |
478 | } |
479 | |
480 | return $self->_join_sql_clauses($logic, \@sql_clauses, \@all_bind); |
481 | } |
482 | |
474e3335 |
483 | #====================================================================== |
484 | # WHERE: top-level ARRAYREFREF |
485 | #====================================================================== |
96449e8e |
486 | |
474e3335 |
487 | sub _where_ARRAYREFREF { |
488 | my ($self, $where) = @_; |
c94a6c93 |
489 | my ($sql, @bind) = @$$where; |
490 | $self->_assert_bindval_matches_bindtype(@bind); |
474e3335 |
491 | return ($sql, @bind); |
492 | } |
96449e8e |
493 | |
494 | #====================================================================== |
495 | # WHERE: top-level HASHREF |
496 | #====================================================================== |
497 | |
498 | sub _where_HASHREF { |
499 | my ($self, $where) = @_; |
500 | my (@sql_clauses, @all_bind); |
501 | |
2281c758 |
502 | for my $k (sort keys %$where) { |
96449e8e |
503 | my $v = $where->{$k}; |
504 | |
2281c758 |
505 | # ($k => $v) is either a special unary op or a regular hashpair |
506 | my ($sql, @bind) = do { |
507 | if ($k =~ /^-./) { |
508 | # put the operator in canonical form |
509 | my $op = $k; |
b8db59b8 |
510 | $op = substr $op, 1; # remove initial dash |
2281c758 |
511 | $op =~ s/^\s+|\s+$//g;# remove leading/trailing space |
b8db59b8 |
512 | $op =~ s/\s+/ /g; # compress whitespace |
513 | |
514 | # so that -not_foo works correctly |
515 | $op =~ s/^not_/NOT /i; |
2281c758 |
516 | |
517 | $self->_debug("Unary OP(-$op) within hashref, recursing..."); |
0ec3aec7 |
518 | my ($s, @b) = $self->_where_unary_op ($op, $v); |
519 | |
520 | # top level vs nested |
521 | # we assume that handled unary ops will take care of their ()s |
522 | $s = "($s)" unless ( |
523 | List::Util::first {$op =~ $_->{regex}} @{$self->{unary_ops}} |
524 | or |
525 | defined($self->{_nested_func_lhs}) && ($self->{_nested_func_lhs} eq $k) |
526 | ); |
527 | ($s, @b); |
2281c758 |
528 | } |
529 | else { |
530 | my $method = $self->_METHOD_FOR_refkind("_where_hashpair", $v); |
531 | $self->$method($k, $v); |
532 | } |
533 | }; |
96449e8e |
534 | |
535 | push @sql_clauses, $sql; |
536 | push @all_bind, @bind; |
537 | } |
538 | |
539 | return $self->_join_sql_clauses('and', \@sql_clauses, \@all_bind); |
540 | } |
541 | |
0ec3aec7 |
542 | sub _where_unary_op { |
2281c758 |
543 | my ($self, $op, $rhs) = @_; |
96449e8e |
544 | |
0ec3aec7 |
545 | if (my $op_entry = List::Util::first {$op =~ $_->{regex}} @{$self->{unary_ops}}) { |
546 | my $handler = $op_entry->{handler}; |
547 | |
548 | if (not ref $handler) { |
549 | if ($op =~ s/ [_\s]? \d+ $//x ) { |
550 | belch 'Use of [and|or|nest]_N modifiers is deprecated and will be removed in SQLA v2.0. ' |
551 | . "You probably wanted ...-and => [ -$op => COND1, -$op => COND2 ... ]"; |
552 | } |
553 | return $self->$handler ($op, $rhs); |
554 | } |
555 | elsif (ref $handler eq 'CODE') { |
556 | return $handler->($self, $op, $rhs); |
557 | } |
558 | else { |
559 | puke "Illegal handler for operator $op - expecting a method name or a coderef"; |
560 | } |
561 | } |
562 | |
3d86e3b1 |
563 | $self->_debug("Generic unary OP: $op - recursing as function"); |
0ec3aec7 |
564 | |
170e6c33 |
565 | $self->_assert_pass_injection_guard($op); |
b6251592 |
566 | |
2281c758 |
567 | my ($sql, @bind) = $self->_SWITCH_refkind ($rhs, { |
568 | SCALAR => sub { |
a7661cfc |
569 | puke "Illegal use of top-level '$op'" |
570 | unless $self->{_nested_func_lhs}; |
571 | |
572 | return ( |
573 | $self->_convert('?'), |
574 | $self->_bindtype($self->{_nested_func_lhs}, $rhs) |
575 | ); |
2281c758 |
576 | }, |
577 | FALLBACK => sub { |
578 | $self->_recurse_where ($rhs) |
579 | }, |
580 | }); |
96449e8e |
581 | |
953d164e |
582 | $sql = sprintf ('%s %s', |
2281c758 |
583 | $self->_sqlcase($op), |
953d164e |
584 | $sql, |
2281c758 |
585 | ); |
96449e8e |
586 | |
2281c758 |
587 | return ($sql, @bind); |
97a920ef |
588 | } |
589 | |
590 | sub _where_op_ANDOR { |
2281c758 |
591 | my ($self, $op, $v) = @_; |
97a920ef |
592 | |
593 | $self->_SWITCH_refkind($v, { |
594 | ARRAYREF => sub { |
595 | return $self->_where_ARRAYREF($v, $op); |
596 | }, |
597 | |
598 | HASHREF => sub { |
59f23b3d |
599 | return ( $op =~ /^or/i ) |
97a920ef |
600 | ? $self->_where_ARRAYREF( [ map { $_ => $v->{$_} } ( sort keys %$v ) ], $op ) |
601 | : $self->_where_HASHREF($v); |
602 | }, |
603 | |
9d48860e |
604 | SCALARREF => sub { |
48d9f5f8 |
605 | puke "-$op => \\\$scalar makes little sense, use " . |
a0d6d323 |
606 | ($op =~ /^or/i |
48d9f5f8 |
607 | ? '[ \$scalar, \%rest_of_conditions ] instead' |
608 | : '-and => [ \$scalar, \%rest_of_conditions ] instead' |
609 | ); |
97a920ef |
610 | }, |
611 | |
612 | ARRAYREFREF => sub { |
48d9f5f8 |
613 | puke "-$op => \\[...] makes little sense, use " . |
a0d6d323 |
614 | ($op =~ /^or/i |
48d9f5f8 |
615 | ? '[ \[...], \%rest_of_conditions ] instead' |
616 | : '-and => [ \[...], \%rest_of_conditions ] instead' |
617 | ); |
97a920ef |
618 | }, |
619 | |
620 | SCALAR => sub { # permissively interpreted as SQL |
48d9f5f8 |
621 | puke "-$op => \$value makes little sense, use -bool => \$value instead"; |
97a920ef |
622 | }, |
623 | |
624 | UNDEF => sub { |
625 | puke "-$op => undef not supported"; |
626 | }, |
627 | }); |
628 | } |
629 | |
630 | sub _where_op_NEST { |
9d48860e |
631 | my ($self, $op, $v) = @_; |
97a920ef |
632 | |
96449e8e |
633 | $self->_SWITCH_refkind($v, { |
634 | |
96449e8e |
635 | SCALAR => sub { # permissively interpreted as SQL |
01a01e57 |
636 | belch "literal SQL should be -nest => \\'scalar' " |
637 | . "instead of -nest => 'scalar' "; |
9d48860e |
638 | return ($v); |
96449e8e |
639 | }, |
640 | |
641 | UNDEF => sub { |
642 | puke "-$op => undef not supported"; |
643 | }, |
e9501094 |
644 | |
645 | FALLBACK => sub { |
646 | $self->_recurse_where ($v); |
647 | }, |
648 | |
96449e8e |
649 | }); |
650 | } |
651 | |
652 | |
97a920ef |
653 | sub _where_op_BOOL { |
9d48860e |
654 | my ($self, $op, $v) = @_; |
97a920ef |
655 | |
b8db59b8 |
656 | my ($s, @b) = $self->_SWITCH_refkind($v, { |
657 | SCALAR => sub { # interpreted as SQL column |
658 | $self->_convert($self->_quote($v)); |
659 | }, |
ef03f1bc |
660 | |
b8db59b8 |
661 | UNDEF => sub { |
662 | puke "-$op => undef not supported"; |
663 | }, |
97a920ef |
664 | |
b8db59b8 |
665 | FALLBACK => sub { |
666 | $self->_recurse_where ($v); |
667 | }, |
668 | }); |
ef03f1bc |
669 | |
b8db59b8 |
670 | $s = "(NOT $s)" if $op =~ /^not/i; |
671 | ($s, @b); |
97a920ef |
672 | } |
673 | |
674 | |
cc422895 |
675 | sub _where_op_IDENT { |
676 | my $self = shift; |
677 | my ($op, $rhs) = splice @_, -2; |
678 | if (ref $rhs) { |
679 | puke "-$op takes a single scalar argument (a quotable identifier)"; |
680 | } |
681 | |
682 | # in case we are called as a top level special op (no '=') |
683 | my $lhs = shift; |
684 | |
685 | $_ = $self->_convert($self->_quote($_)) for ($lhs, $rhs); |
686 | |
687 | return $lhs |
688 | ? "$lhs = $rhs" |
689 | : $rhs |
690 | ; |
691 | } |
692 | |
693 | sub _where_op_VALUE { |
694 | my $self = shift; |
695 | my ($op, $rhs) = splice @_, -2; |
696 | |
697 | # in case we are called as a top level special op (no '=') |
698 | my $lhs = shift; |
699 | |
700 | my @bind = |
701 | $self->_bindtype ( |
702 | ($lhs || $self->{_nested_func_lhs}), |
703 | $rhs, |
704 | ) |
705 | ; |
706 | |
707 | return $lhs |
708 | ? ( |
709 | $self->_convert($self->_quote($lhs)) . ' = ' . $self->_convert('?'), |
710 | @bind |
711 | ) |
712 | : ( |
713 | $self->_convert('?'), |
714 | @bind, |
715 | ) |
716 | ; |
717 | } |
718 | |
96449e8e |
719 | sub _where_hashpair_ARRAYREF { |
720 | my ($self, $k, $v) = @_; |
721 | |
722 | if( @$v ) { |
723 | my @v = @$v; # need copy because of shift below |
724 | $self->_debug("ARRAY($k) means distribute over elements"); |
725 | |
726 | # put apart first element if it is an operator (-and, -or) |
e3cecb45 |
727 | my $op = ( |
728 | (defined $v[0] && $v[0] =~ /^ - (?: AND|OR ) $/ix) |
729 | ? shift @v |
730 | : '' |
04d940de |
731 | ); |
96449e8e |
732 | my @distributed = map { {$k => $_} } @v; |
04d940de |
733 | |
734 | if ($op) { |
735 | $self->_debug("OP($op) reinjected into the distributed array"); |
736 | unshift @distributed, $op; |
737 | } |
738 | |
f67591bf |
739 | my $logic = $op ? substr($op, 1) : ''; |
96449e8e |
740 | |
f67591bf |
741 | return $self->_recurse_where(\@distributed, $logic); |
9d48860e |
742 | } |
96449e8e |
743 | else { |
744 | # LDNOTE : not sure of this one. What does "distribute over nothing" mean? |
745 | $self->_debug("empty ARRAY($k) means 0=1"); |
746 | return ($self->{sqlfalse}); |
747 | } |
748 | } |
749 | |
750 | sub _where_hashpair_HASHREF { |
eb49170d |
751 | my ($self, $k, $v, $logic) = @_; |
752 | $logic ||= 'and'; |
96449e8e |
753 | |
a7661cfc |
754 | local $self->{_nested_func_lhs} = $self->{_nested_func_lhs}; |
a7661cfc |
755 | |
eb49170d |
756 | my ($all_sql, @all_bind); |
96449e8e |
757 | |
a47b433a |
758 | for my $orig_op (sort keys %$v) { |
759 | my $val = $v->{$orig_op}; |
96449e8e |
760 | |
761 | # put the operator in canonical form |
a47b433a |
762 | my $op = $orig_op; |
b8db59b8 |
763 | |
764 | # FIXME - we need to phase out dash-less ops |
765 | $op =~ s/^-//; # remove possible initial dash |
a47b433a |
766 | $op =~ s/^\s+|\s+$//g;# remove leading/trailing space |
b8db59b8 |
767 | $op =~ s/\s+/ /g; # compress whitespace |
768 | |
170e6c33 |
769 | $self->_assert_pass_injection_guard($op); |
b6251592 |
770 | |
b8db59b8 |
771 | # so that -not_foo works correctly |
772 | $op =~ s/^not_/NOT /i; |
96449e8e |
773 | |
774 | my ($sql, @bind); |
775 | |
2281c758 |
776 | # CASE: col-value logic modifiers |
777 | if ( $orig_op =~ /^ \- (and|or) $/xi ) { |
778 | ($sql, @bind) = $self->_where_hashpair_HASHREF($k, $val, $1); |
779 | } |
96449e8e |
780 | # CASE: special operators like -in or -between |
312d830b |
781 | elsif ( my $special_op = List::Util::first {$op =~ $_->{regex}} @{$self->{special_ops}} ) { |
3a2e1a5e |
782 | my $handler = $special_op->{handler}; |
783 | if (! $handler) { |
a47b433a |
784 | puke "No handler supplied for special operator $orig_op"; |
3a2e1a5e |
785 | } |
786 | elsif (not ref $handler) { |
787 | ($sql, @bind) = $self->$handler ($k, $op, $val); |
788 | } |
789 | elsif (ref $handler eq 'CODE') { |
790 | ($sql, @bind) = $handler->($self, $k, $op, $val); |
791 | } |
792 | else { |
a47b433a |
793 | puke "Illegal handler for special operator $orig_op - expecting a method name or a coderef"; |
3a2e1a5e |
794 | } |
96449e8e |
795 | } |
96449e8e |
796 | else { |
cf838930 |
797 | $self->_SWITCH_refkind($val, { |
798 | |
799 | ARRAYREF => sub { # CASE: col => {op => \@vals} |
800 | ($sql, @bind) = $self->_where_field_op_ARRAYREF($k, $op, $val); |
801 | }, |
802 | |
fe3ae272 |
803 | ARRAYREFREF => sub { # CASE: col => {op => \[$sql, @bind]} (literal SQL with bind) |
b3be7bd0 |
804 | my ($sub_sql, @sub_bind) = @$$val; |
fe3ae272 |
805 | $self->_assert_bindval_matches_bindtype(@sub_bind); |
b3be7bd0 |
806 | $sql = join ' ', $self->_convert($self->_quote($k)), |
807 | $self->_sqlcase($op), |
808 | $sub_sql; |
fe3ae272 |
809 | @bind = @sub_bind; |
b3be7bd0 |
810 | }, |
811 | |
cf838930 |
812 | UNDEF => sub { # CASE: col => {op => undef} : sql "IS (NOT)? NULL" |
813 | my $is = ($op =~ $self->{equality_op}) ? 'is' : |
814 | ($op =~ $self->{inequality_op}) ? 'is not' : |
a47b433a |
815 | puke "unexpected operator '$orig_op' with undef operand"; |
cf838930 |
816 | $sql = $self->_quote($k) . $self->_sqlcase(" $is null"); |
817 | }, |
a47b433a |
818 | |
2281c758 |
819 | FALLBACK => sub { # CASE: col => {op/func => $stuff} |
07936978 |
820 | |
953d164e |
821 | # retain for proper column type bind |
822 | $self->{_nested_func_lhs} ||= $k; |
07936978 |
823 | |
0ec3aec7 |
824 | ($sql, @bind) = $self->_where_unary_op ($op, $val); |
953d164e |
825 | |
826 | $sql = join (' ', |
827 | $self->_convert($self->_quote($k)), |
828 | $self->{_nested_func_lhs} eq $k ? $sql : "($sql)", # top level vs nested |
829 | ); |
cf838930 |
830 | }, |
831 | }); |
96449e8e |
832 | } |
833 | |
eb49170d |
834 | ($all_sql) = (defined $all_sql and $all_sql) ? $self->_join_sql_clauses($logic, [$all_sql, $sql], []) : $sql; |
96449e8e |
835 | push @all_bind, @bind; |
836 | } |
eb49170d |
837 | return ($all_sql, @all_bind); |
96449e8e |
838 | } |
839 | |
840 | |
841 | |
842 | sub _where_field_op_ARRAYREF { |
843 | my ($self, $k, $op, $vals) = @_; |
844 | |
ce261791 |
845 | my @vals = @$vals; #always work on a copy |
846 | |
847 | if(@vals) { |
bd6a65ca |
848 | $self->_debug(sprintf '%s means multiple elements: [ %s ]', |
849 | $vals, |
850 | join (', ', map { defined $_ ? "'$_'" : 'NULL' } @vals ), |
851 | ); |
96449e8e |
852 | |
4030915f |
853 | # see if the first element is an -and/-or op |
854 | my $logic; |
bd6a65ca |
855 | if (defined $vals[0] && $vals[0] =~ /^ - ( AND|OR ) $/ix) { |
4030915f |
856 | $logic = uc $1; |
ce261791 |
857 | shift @vals; |
4030915f |
858 | } |
859 | |
ce261791 |
860 | # distribute $op over each remaining member of @vals, append logic if exists |
861 | return $self->_recurse_where([map { {$k => {$op, $_}} } @vals], $logic); |
4030915f |
862 | |
9d48860e |
863 | # LDNOTE : had planned to change the distribution logic when |
864 | # $op =~ $self->{inequality_op}, because of Morgan laws : |
96449e8e |
865 | # with {field => {'!=' => [22, 33]}}, it would be ridiculous to generate |
9d48860e |
866 | # WHERE field != 22 OR field != 33 : the user probably means |
96449e8e |
867 | # WHERE field != 22 AND field != 33. |
4030915f |
868 | # To do this, replace the above to roughly : |
f2d5020d |
869 | # my $logic = ($op =~ $self->{inequality_op}) ? 'AND' : 'OR'; |
ce261791 |
870 | # return $self->_recurse_where([map { {$k => {$op, $_}} } @vals], $logic); |
96449e8e |
871 | |
9d48860e |
872 | } |
96449e8e |
873 | else { |
9d48860e |
874 | # try to DWIM on equality operators |
96449e8e |
875 | # LDNOTE : not 100% sure this is the correct thing to do ... |
876 | return ($self->{sqlfalse}) if $op =~ $self->{equality_op}; |
877 | return ($self->{sqltrue}) if $op =~ $self->{inequality_op}; |
878 | |
879 | # otherwise |
880 | puke "operator '$op' applied on an empty array (field '$k')"; |
881 | } |
882 | } |
883 | |
884 | |
885 | sub _where_hashpair_SCALARREF { |
886 | my ($self, $k, $v) = @_; |
887 | $self->_debug("SCALAR($k) means literal SQL: $$v"); |
888 | my $sql = $self->_quote($k) . " " . $$v; |
889 | return ($sql); |
890 | } |
891 | |
fe3ae272 |
892 | # literal SQL with bind |
96449e8e |
893 | sub _where_hashpair_ARRAYREFREF { |
894 | my ($self, $k, $v) = @_; |
895 | $self->_debug("REF($k) means literal SQL: @${$v}"); |
c94a6c93 |
896 | my ($sql, @bind) = @$$v; |
fe3ae272 |
897 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
898 | $sql = $self->_quote($k) . " " . $sql; |
96449e8e |
899 | return ($sql, @bind ); |
900 | } |
901 | |
fe3ae272 |
902 | # literal SQL without bind |
96449e8e |
903 | sub _where_hashpair_SCALAR { |
904 | my ($self, $k, $v) = @_; |
905 | $self->_debug("NOREF($k) means simple key=val: $k $self->{cmp} $v"); |
9d48860e |
906 | my $sql = join ' ', $self->_convert($self->_quote($k)), |
907 | $self->_sqlcase($self->{cmp}), |
96449e8e |
908 | $self->_convert('?'); |
909 | my @bind = $self->_bindtype($k, $v); |
910 | return ( $sql, @bind); |
911 | } |
912 | |
913 | |
914 | sub _where_hashpair_UNDEF { |
915 | my ($self, $k, $v) = @_; |
916 | $self->_debug("UNDEF($k) means IS NULL"); |
917 | my $sql = $self->_quote($k) . $self->_sqlcase(' is null'); |
918 | return ($sql); |
919 | } |
920 | |
921 | #====================================================================== |
922 | # WHERE: TOP-LEVEL OTHERS (SCALARREF, SCALAR, UNDEF) |
923 | #====================================================================== |
924 | |
925 | |
926 | sub _where_SCALARREF { |
927 | my ($self, $where) = @_; |
928 | |
929 | # literal sql |
930 | $self->_debug("SCALAR(*top) means literal SQL: $$where"); |
931 | return ($$where); |
932 | } |
933 | |
934 | |
935 | sub _where_SCALAR { |
936 | my ($self, $where) = @_; |
937 | |
938 | # literal sql |
939 | $self->_debug("NOREF(*top) means literal SQL: $where"); |
940 | return ($where); |
941 | } |
942 | |
943 | |
944 | sub _where_UNDEF { |
945 | my ($self) = @_; |
946 | return (); |
947 | } |
948 | |
949 | |
950 | #====================================================================== |
951 | # WHERE: BUILTIN SPECIAL OPERATORS (-in, -between) |
952 | #====================================================================== |
953 | |
954 | |
955 | sub _where_field_BETWEEN { |
956 | my ($self, $k, $op, $vals) = @_; |
957 | |
4d8b3dc4 |
958 | my ($label, $and, $placeholder); |
cf02fc47 |
959 | $label = $self->_convert($self->_quote($k)); |
960 | $and = ' ' . $self->_sqlcase('and') . ' '; |
961 | $placeholder = $self->_convert('?'); |
96449e8e |
962 | $op = $self->_sqlcase($op); |
963 | |
4d8b3dc4 |
964 | my ($clause, @bind) = $self->_SWITCH_refkind($vals, { |
965 | ARRAYREFREF => sub { |
c94a6c93 |
966 | my ($s, @b) = @$$vals; |
967 | $self->_assert_bindval_matches_bindtype(@b); |
968 | ($s, @b); |
4d8b3dc4 |
969 | }, |
970 | SCALARREF => sub { |
971 | return $$vals; |
972 | }, |
973 | ARRAYREF => sub { |
974 | puke "special op 'between' accepts an arrayref with exactly two values" |
975 | if @$vals != 2; |
976 | |
977 | my (@all_sql, @all_bind); |
978 | foreach my $val (@$vals) { |
979 | my ($sql, @bind) = $self->_SWITCH_refkind($val, { |
980 | SCALAR => sub { |
5e5cbf51 |
981 | return ($placeholder, $self->_bindtype($k, $val) ); |
4d8b3dc4 |
982 | }, |
983 | SCALARREF => sub { |
0336eddb |
984 | return $$val; |
4d8b3dc4 |
985 | }, |
986 | ARRAYREFREF => sub { |
987 | my ($sql, @bind) = @$$val; |
c94a6c93 |
988 | $self->_assert_bindval_matches_bindtype(@bind); |
0336eddb |
989 | return ($sql, @bind); |
4d8b3dc4 |
990 | }, |
0336eddb |
991 | HASHREF => sub { |
992 | my ($func, $arg, @rest) = %$val; |
993 | puke ("Only simple { -func => arg } functions accepted as sub-arguments to BETWEEN") |
994 | if (@rest or $func !~ /^ \- (.+)/x); |
995 | local $self->{_nested_func_lhs} = $k; |
0ec3aec7 |
996 | $self->_where_unary_op ($1 => $arg); |
0336eddb |
997 | } |
4d8b3dc4 |
998 | }); |
999 | push @all_sql, $sql; |
1000 | push @all_bind, @bind; |
1001 | } |
1002 | |
1003 | return ( |
1004 | (join $and, @all_sql), |
5e5cbf51 |
1005 | @all_bind |
4d8b3dc4 |
1006 | ); |
1007 | }, |
1008 | FALLBACK => sub { |
1009 | puke "special op 'between' accepts an arrayref with two values, or a single literal scalarref/arrayref-ref"; |
1010 | }, |
1011 | }); |
cf02fc47 |
1012 | |
cf02fc47 |
1013 | my $sql = "( $label $op $clause )"; |
96449e8e |
1014 | return ($sql, @bind) |
1015 | } |
1016 | |
1017 | |
1018 | sub _where_field_IN { |
1019 | my ($self, $k, $op, $vals) = @_; |
1020 | |
1021 | # backwards compatibility : if scalar, force into an arrayref |
1022 | $vals = [$vals] if defined $vals && ! ref $vals; |
1023 | |
96449e8e |
1024 | my ($label) = $self->_convert($self->_quote($k)); |
1025 | my ($placeholder) = $self->_convert('?'); |
96449e8e |
1026 | $op = $self->_sqlcase($op); |
1027 | |
8a0d798a |
1028 | my ($sql, @bind) = $self->_SWITCH_refkind($vals, { |
1029 | ARRAYREF => sub { # list of choices |
1030 | if (@$vals) { # nonempty list |
0336eddb |
1031 | my (@all_sql, @all_bind); |
1032 | |
1033 | for my $val (@$vals) { |
1034 | my ($sql, @bind) = $self->_SWITCH_refkind($val, { |
1035 | SCALAR => sub { |
1036 | return ($placeholder, $val); |
1037 | }, |
1038 | SCALARREF => sub { |
1039 | return $$val; |
1040 | }, |
1041 | ARRAYREFREF => sub { |
1042 | my ($sql, @bind) = @$$val; |
1043 | $self->_assert_bindval_matches_bindtype(@bind); |
1044 | return ($sql, @bind); |
1045 | }, |
1046 | HASHREF => sub { |
1047 | my ($func, $arg, @rest) = %$val; |
1048 | puke ("Only simple { -func => arg } functions accepted as sub-arguments to IN") |
1049 | if (@rest or $func !~ /^ \- (.+)/x); |
1050 | local $self->{_nested_func_lhs} = $k; |
0ec3aec7 |
1051 | $self->_where_unary_op ($1 => $arg); |
279eb282 |
1052 | }, |
1053 | UNDEF => sub { |
032dfe20 |
1054 | puke( |
1055 | 'SQL::Abstract before v1.75 used to generate incorrect SQL when the ' |
1056 | . "-$op operator was given an undef-containing list: !!!AUDIT YOUR CODE " |
1057 | . 'AND DATA!!! (the upcoming Data::Query-based version of SQL::Abstract ' |
1058 | . 'will emit the logically correct SQL instead of raising this exception)' |
1059 | ); |
279eb282 |
1060 | }, |
0336eddb |
1061 | }); |
1062 | push @all_sql, $sql; |
1063 | push @all_bind, @bind; |
1064 | } |
96449e8e |
1065 | |
88a89939 |
1066 | return ( |
1067 | sprintf ('%s %s ( %s )', |
1068 | $label, |
1069 | $op, |
1070 | join (', ', @all_sql) |
1071 | ), |
1072 | $self->_bindtype($k, @all_bind), |
0336eddb |
1073 | ); |
8a0d798a |
1074 | } |
1075 | else { # empty list : some databases won't understand "IN ()", so DWIM |
1076 | my $sql = ($op =~ /\bnot\b/i) ? $self->{sqltrue} : $self->{sqlfalse}; |
1077 | return ($sql); |
1078 | } |
1079 | }, |
1080 | |
4a1f01a3 |
1081 | SCALARREF => sub { # literal SQL |
1082 | my $sql = $self->_open_outer_paren ($$vals); |
1083 | return ("$label $op ( $sql )"); |
1084 | }, |
8a0d798a |
1085 | ARRAYREFREF => sub { # literal SQL with bind |
1086 | my ($sql, @bind) = @$$vals; |
fe3ae272 |
1087 | $self->_assert_bindval_matches_bindtype(@bind); |
4a1f01a3 |
1088 | $sql = $self->_open_outer_paren ($sql); |
8a0d798a |
1089 | return ("$label $op ( $sql )", @bind); |
1090 | }, |
1091 | |
1092 | FALLBACK => sub { |
4a1f01a3 |
1093 | puke "special op 'in' requires an arrayref (or scalarref/arrayref-ref)"; |
8a0d798a |
1094 | }, |
1095 | }); |
1096 | |
1097 | return ($sql, @bind); |
96449e8e |
1098 | } |
1099 | |
4a1f01a3 |
1100 | # Some databases (SQLite) treat col IN (1, 2) different from |
1101 | # col IN ( (1, 2) ). Use this to strip all outer parens while |
1102 | # adding them back in the corresponding method |
1103 | sub _open_outer_paren { |
1104 | my ($self, $sql) = @_; |
171a709f |
1105 | $sql = $1 while $sql =~ /^ \s* \( (.*) \) \s* $/xs; |
4a1f01a3 |
1106 | return $sql; |
1107 | } |
1108 | |
96449e8e |
1109 | |
96449e8e |
1110 | #====================================================================== |
1111 | # ORDER BY |
1112 | #====================================================================== |
1113 | |
1114 | sub _order_by { |
1115 | my ($self, $arg) = @_; |
1116 | |
f267b646 |
1117 | my (@sql, @bind); |
1118 | for my $c ($self->_order_by_chunks ($arg) ) { |
1119 | $self->_SWITCH_refkind ($c, { |
1120 | SCALAR => sub { push @sql, $c }, |
1121 | ARRAYREF => sub { push @sql, shift @$c; push @bind, @$c }, |
1122 | }); |
1123 | } |
1124 | |
1125 | my $sql = @sql |
1126 | ? sprintf ('%s %s', |
1127 | $self->_sqlcase(' order by'), |
1128 | join (', ', @sql) |
1129 | ) |
1130 | : '' |
1131 | ; |
1132 | |
1133 | return wantarray ? ($sql, @bind) : $sql; |
1134 | } |
1135 | |
1136 | sub _order_by_chunks { |
1137 | my ($self, $arg) = @_; |
1138 | |
1139 | return $self->_SWITCH_refkind($arg, { |
96449e8e |
1140 | |
1141 | ARRAYREF => sub { |
f267b646 |
1142 | map { $self->_order_by_chunks ($_ ) } @$arg; |
96449e8e |
1143 | }, |
1144 | |
c94a6c93 |
1145 | ARRAYREFREF => sub { |
1146 | my ($s, @b) = @$$arg; |
1147 | $self->_assert_bindval_matches_bindtype(@b); |
1148 | [ $s, @b ]; |
1149 | }, |
f267b646 |
1150 | |
96449e8e |
1151 | SCALAR => sub {$self->_quote($arg)}, |
f267b646 |
1152 | |
1153 | UNDEF => sub {return () }, |
1154 | |
96449e8e |
1155 | SCALARREF => sub {$$arg}, # literal SQL, no quoting |
96449e8e |
1156 | |
f267b646 |
1157 | HASHREF => sub { |
5e436130 |
1158 | # get first pair in hash |
1159 | my ($key, $val, @rest) = %$arg; |
1160 | |
1161 | return () unless $key; |
1162 | |
1163 | if ( @rest or not $key =~ /^-(desc|asc)/i ) { |
1164 | puke "hash passed to _order_by must have exactly one key (-desc or -asc)"; |
f267b646 |
1165 | } |
5e436130 |
1166 | |
1167 | my $direction = $1; |
96449e8e |
1168 | |
e9bd3547 |
1169 | my @ret; |
f267b646 |
1170 | for my $c ($self->_order_by_chunks ($val)) { |
e9bd3547 |
1171 | my ($sql, @bind); |
96449e8e |
1172 | |
f267b646 |
1173 | $self->_SWITCH_refkind ($c, { |
1174 | SCALAR => sub { |
e9bd3547 |
1175 | $sql = $c; |
f267b646 |
1176 | }, |
1177 | ARRAYREF => sub { |
e9bd3547 |
1178 | ($sql, @bind) = @$c; |
f267b646 |
1179 | }, |
1180 | }); |
96449e8e |
1181 | |
5e436130 |
1182 | $sql = $sql . ' ' . $self->_sqlcase($direction); |
96449e8e |
1183 | |
e9bd3547 |
1184 | push @ret, [ $sql, @bind]; |
1185 | } |
96449e8e |
1186 | |
e9bd3547 |
1187 | return @ret; |
f267b646 |
1188 | }, |
1189 | }); |
96449e8e |
1190 | } |
1191 | |
1192 | |
96449e8e |
1193 | #====================================================================== |
1194 | # DATASOURCE (FOR NOW, JUST PLAIN TABLE OR LIST OF TABLES) |
1195 | #====================================================================== |
1196 | |
1197 | sub _table { |
1198 | my $self = shift; |
1199 | my $from = shift; |
1200 | $self->_SWITCH_refkind($from, { |
1201 | ARRAYREF => sub {join ', ', map { $self->_quote($_) } @$from;}, |
1202 | SCALAR => sub {$self->_quote($from)}, |
1203 | SCALARREF => sub {$$from}, |
96449e8e |
1204 | }); |
1205 | } |
1206 | |
1207 | |
1208 | #====================================================================== |
1209 | # UTILITY FUNCTIONS |
1210 | #====================================================================== |
1211 | |
955e77ca |
1212 | # highly optimized, as it's called way too often |
96449e8e |
1213 | sub _quote { |
955e77ca |
1214 | # my ($self, $label) = @_; |
96449e8e |
1215 | |
955e77ca |
1216 | return '' unless defined $_[1]; |
955e77ca |
1217 | return ${$_[1]} if ref($_[1]) eq 'SCALAR'; |
96449e8e |
1218 | |
b6251592 |
1219 | unless ($_[0]->{quote_char}) { |
170e6c33 |
1220 | $_[0]->_assert_pass_injection_guard($_[1]); |
b6251592 |
1221 | return $_[1]; |
1222 | } |
96449e8e |
1223 | |
07d7c35c |
1224 | my $qref = ref $_[0]->{quote_char}; |
955e77ca |
1225 | my ($l, $r); |
07d7c35c |
1226 | if (!$qref) { |
1227 | ($l, $r) = ( $_[0]->{quote_char}, $_[0]->{quote_char} ); |
955e77ca |
1228 | } |
07d7c35c |
1229 | elsif ($qref eq 'ARRAY') { |
1230 | ($l, $r) = @{$_[0]->{quote_char}}; |
955e77ca |
1231 | } |
1232 | else { |
1233 | puke "Unsupported quote_char format: $_[0]->{quote_char}"; |
1234 | } |
96449e8e |
1235 | |
07d7c35c |
1236 | # parts containing * are naturally unquoted |
1237 | return join( $_[0]->{name_sep}||'', map |
955e77ca |
1238 | { $_ eq '*' ? $_ : $l . $_ . $r } |
1239 | ( $_[0]->{name_sep} ? split (/\Q$_[0]->{name_sep}\E/, $_[1] ) : $_[1] ) |
1240 | ); |
96449e8e |
1241 | } |
1242 | |
1243 | |
1244 | # Conversion, if applicable |
1245 | sub _convert ($) { |
07d7c35c |
1246 | #my ($self, $arg) = @_; |
96449e8e |
1247 | |
1248 | # LDNOTE : modified the previous implementation below because |
1249 | # it was not consistent : the first "return" is always an array, |
1250 | # the second "return" is context-dependent. Anyway, _convert |
9d48860e |
1251 | # seems always used with just a single argument, so make it a |
96449e8e |
1252 | # scalar function. |
1253 | # return @_ unless $self->{convert}; |
1254 | # my $conv = $self->_sqlcase($self->{convert}); |
1255 | # my @ret = map { $conv.'('.$_.')' } @_; |
1256 | # return wantarray ? @ret : $ret[0]; |
07d7c35c |
1257 | if ($_[0]->{convert}) { |
1258 | return $_[0]->_sqlcase($_[0]->{convert}) .'(' . $_[1] . ')'; |
96449e8e |
1259 | } |
07d7c35c |
1260 | return $_[1]; |
96449e8e |
1261 | } |
1262 | |
1263 | # And bindtype |
1264 | sub _bindtype (@) { |
07d7c35c |
1265 | #my ($self, $col, @vals) = @_; |
96449e8e |
1266 | |
9d48860e |
1267 | #LDNOTE : changed original implementation below because it did not make |
96449e8e |
1268 | # sense when bindtype eq 'columns' and @vals > 1. |
1269 | # return $self->{bindtype} eq 'columns' ? [ $col, @vals ] : @vals; |
1270 | |
07d7c35c |
1271 | # called often - tighten code |
1272 | return $_[0]->{bindtype} eq 'columns' |
1273 | ? map {[$_[1], $_]} @_[2 .. $#_] |
1274 | : @_[2 .. $#_] |
1275 | ; |
96449e8e |
1276 | } |
1277 | |
fe3ae272 |
1278 | # Dies if any element of @bind is not in [colname => value] format |
1279 | # if bindtype is 'columns'. |
1280 | sub _assert_bindval_matches_bindtype { |
c94a6c93 |
1281 | # my ($self, @bind) = @_; |
1282 | my $self = shift; |
fe3ae272 |
1283 | if ($self->{bindtype} eq 'columns') { |
c94a6c93 |
1284 | for (@_) { |
1285 | if (!defined $_ || ref($_) ne 'ARRAY' || @$_ != 2) { |
3a06278c |
1286 | puke "bindtype 'columns' selected, you need to pass: [column_name => bind_value]" |
fe3ae272 |
1287 | } |
1288 | } |
1289 | } |
1290 | } |
1291 | |
96449e8e |
1292 | sub _join_sql_clauses { |
1293 | my ($self, $logic, $clauses_aref, $bind_aref) = @_; |
1294 | |
1295 | if (@$clauses_aref > 1) { |
1296 | my $join = " " . $self->_sqlcase($logic) . " "; |
1297 | my $sql = '( ' . join($join, @$clauses_aref) . ' )'; |
1298 | return ($sql, @$bind_aref); |
1299 | } |
1300 | elsif (@$clauses_aref) { |
1301 | return ($clauses_aref->[0], @$bind_aref); # no parentheses |
1302 | } |
1303 | else { |
1304 | return (); # if no SQL, ignore @$bind_aref |
1305 | } |
1306 | } |
1307 | |
1308 | |
1309 | # Fix SQL case, if so requested |
1310 | sub _sqlcase { |
96449e8e |
1311 | # LDNOTE: if $self->{case} is true, then it contains 'lower', so we |
1312 | # don't touch the argument ... crooked logic, but let's not change it! |
07d7c35c |
1313 | return $_[0]->{case} ? $_[1] : uc($_[1]); |
96449e8e |
1314 | } |
1315 | |
1316 | |
1317 | #====================================================================== |
1318 | # DISPATCHING FROM REFKIND |
1319 | #====================================================================== |
1320 | |
1321 | sub _refkind { |
1322 | my ($self, $data) = @_; |
96449e8e |
1323 | |
955e77ca |
1324 | return 'UNDEF' unless defined $data; |
1325 | |
1326 | # blessed objects are treated like scalars |
1327 | my $ref = (Scalar::Util::blessed $data) ? '' : ref $data; |
1328 | |
1329 | return 'SCALAR' unless $ref; |
1330 | |
1331 | my $n_steps = 1; |
1332 | while ($ref eq 'REF') { |
96449e8e |
1333 | $data = $$data; |
955e77ca |
1334 | $ref = (Scalar::Util::blessed $data) ? '' : ref $data; |
1335 | $n_steps++ if $ref; |
96449e8e |
1336 | } |
1337 | |
848556bc |
1338 | return ($ref||'SCALAR') . ('REF' x $n_steps); |
96449e8e |
1339 | } |
1340 | |
1341 | sub _try_refkind { |
1342 | my ($self, $data) = @_; |
1343 | my @try = ($self->_refkind($data)); |
1344 | push @try, 'SCALAR_or_UNDEF' if $try[0] eq 'SCALAR' || $try[0] eq 'UNDEF'; |
1345 | push @try, 'FALLBACK'; |
955e77ca |
1346 | return \@try; |
96449e8e |
1347 | } |
1348 | |
1349 | sub _METHOD_FOR_refkind { |
1350 | my ($self, $meth_prefix, $data) = @_; |
f39eaa60 |
1351 | |
1352 | my $method; |
955e77ca |
1353 | for (@{$self->_try_refkind($data)}) { |
f39eaa60 |
1354 | $method = $self->can($meth_prefix."_".$_) |
1355 | and last; |
1356 | } |
1357 | |
1358 | return $method || puke "cannot dispatch on '$meth_prefix' for ".$self->_refkind($data); |
96449e8e |
1359 | } |
1360 | |
1361 | |
1362 | sub _SWITCH_refkind { |
1363 | my ($self, $data, $dispatch_table) = @_; |
1364 | |
f39eaa60 |
1365 | my $coderef; |
955e77ca |
1366 | for (@{$self->_try_refkind($data)}) { |
f39eaa60 |
1367 | $coderef = $dispatch_table->{$_} |
1368 | and last; |
1369 | } |
1370 | |
1371 | puke "no dispatch entry for ".$self->_refkind($data) |
1372 | unless $coderef; |
1373 | |
96449e8e |
1374 | $coderef->(); |
1375 | } |
1376 | |
1377 | |
1378 | |
1379 | |
1380 | #====================================================================== |
1381 | # VALUES, GENERATE, AUTOLOAD |
1382 | #====================================================================== |
1383 | |
1384 | # LDNOTE: original code from nwiger, didn't touch code in that section |
1385 | # I feel the AUTOLOAD stuff should not be the default, it should |
1386 | # only be activated on explicit demand by user. |
1387 | |
1388 | sub values { |
1389 | my $self = shift; |
1390 | my $data = shift || return; |
1391 | puke "Argument to ", __PACKAGE__, "->values must be a \\%hash" |
1392 | unless ref $data eq 'HASH'; |
bab725ce |
1393 | |
1394 | my @all_bind; |
1395 | foreach my $k ( sort keys %$data ) { |
1396 | my $v = $data->{$k}; |
1397 | $self->_SWITCH_refkind($v, { |
9d48860e |
1398 | ARRAYREF => sub { |
bab725ce |
1399 | if ($self->{array_datatypes}) { # array datatype |
1400 | push @all_bind, $self->_bindtype($k, $v); |
1401 | } |
1402 | else { # literal SQL with bind |
1403 | my ($sql, @bind) = @$v; |
1404 | $self->_assert_bindval_matches_bindtype(@bind); |
1405 | push @all_bind, @bind; |
1406 | } |
1407 | }, |
1408 | ARRAYREFREF => sub { # literal SQL with bind |
1409 | my ($sql, @bind) = @${$v}; |
1410 | $self->_assert_bindval_matches_bindtype(@bind); |
1411 | push @all_bind, @bind; |
1412 | }, |
1413 | SCALARREF => sub { # literal SQL without bind |
1414 | }, |
1415 | SCALAR_or_UNDEF => sub { |
1416 | push @all_bind, $self->_bindtype($k, $v); |
1417 | }, |
1418 | }); |
1419 | } |
1420 | |
1421 | return @all_bind; |
96449e8e |
1422 | } |
1423 | |
1424 | sub generate { |
1425 | my $self = shift; |
1426 | |
1427 | my(@sql, @sqlq, @sqlv); |
1428 | |
1429 | for (@_) { |
1430 | my $ref = ref $_; |
1431 | if ($ref eq 'HASH') { |
1432 | for my $k (sort keys %$_) { |
1433 | my $v = $_->{$k}; |
1434 | my $r = ref $v; |
1435 | my $label = $self->_quote($k); |
1436 | if ($r eq 'ARRAY') { |
fe3ae272 |
1437 | # literal SQL with bind |
1438 | my ($sql, @bind) = @$v; |
1439 | $self->_assert_bindval_matches_bindtype(@bind); |
96449e8e |
1440 | push @sqlq, "$label = $sql"; |
fe3ae272 |
1441 | push @sqlv, @bind; |
96449e8e |
1442 | } elsif ($r eq 'SCALAR') { |
fe3ae272 |
1443 | # literal SQL without bind |
96449e8e |
1444 | push @sqlq, "$label = $$v"; |
9d48860e |
1445 | } else { |
96449e8e |
1446 | push @sqlq, "$label = ?"; |
1447 | push @sqlv, $self->_bindtype($k, $v); |
1448 | } |
1449 | } |
1450 | push @sql, $self->_sqlcase('set'), join ', ', @sqlq; |
1451 | } elsif ($ref eq 'ARRAY') { |
1452 | # unlike insert(), assume these are ONLY the column names, i.e. for SQL |
1453 | for my $v (@$_) { |
1454 | my $r = ref $v; |
fe3ae272 |
1455 | if ($r eq 'ARRAY') { # literal SQL with bind |
1456 | my ($sql, @bind) = @$v; |
1457 | $self->_assert_bindval_matches_bindtype(@bind); |
1458 | push @sqlq, $sql; |
1459 | push @sqlv, @bind; |
1460 | } elsif ($r eq 'SCALAR') { # literal SQL without bind |
96449e8e |
1461 | # embedded literal SQL |
1462 | push @sqlq, $$v; |
9d48860e |
1463 | } else { |
96449e8e |
1464 | push @sqlq, '?'; |
1465 | push @sqlv, $v; |
1466 | } |
1467 | } |
1468 | push @sql, '(' . join(', ', @sqlq) . ')'; |
1469 | } elsif ($ref eq 'SCALAR') { |
1470 | # literal SQL |
1471 | push @sql, $$_; |
1472 | } else { |
1473 | # strings get case twiddled |
1474 | push @sql, $self->_sqlcase($_); |
1475 | } |
1476 | } |
1477 | |
1478 | my $sql = join ' ', @sql; |
1479 | |
1480 | # this is pretty tricky |
1481 | # if ask for an array, return ($stmt, @bind) |
1482 | # otherwise, s/?/shift @sqlv/ to put it inline |
1483 | if (wantarray) { |
1484 | return ($sql, @sqlv); |
1485 | } else { |
1486 | 1 while $sql =~ s/\?/my $d = shift(@sqlv); |
1487 | ref $d ? $d->[1] : $d/e; |
1488 | return $sql; |
1489 | } |
1490 | } |
1491 | |
1492 | |
1493 | sub DESTROY { 1 } |
1494 | |
1495 | sub AUTOLOAD { |
1496 | # This allows us to check for a local, then _form, attr |
1497 | my $self = shift; |
1498 | my($name) = $AUTOLOAD =~ /.*::(.+)/; |
1499 | return $self->generate($name, @_); |
1500 | } |
1501 | |
1502 | 1; |
1503 | |
1504 | |
1505 | |
1506 | __END__ |
32eab2da |
1507 | |
1508 | =head1 NAME |
1509 | |
1510 | SQL::Abstract - Generate SQL from Perl data structures |
1511 | |
1512 | =head1 SYNOPSIS |
1513 | |
1514 | use SQL::Abstract; |
1515 | |
1516 | my $sql = SQL::Abstract->new; |
1517 | |
521647e7 |
1518 | my($stmt, @bind) = $sql->select($source, \@fields, \%where, \@order); |
32eab2da |
1519 | |
1520 | my($stmt, @bind) = $sql->insert($table, \%fieldvals || \@values); |
1521 | |
1522 | my($stmt, @bind) = $sql->update($table, \%fieldvals, \%where); |
1523 | |
1524 | my($stmt, @bind) = $sql->delete($table, \%where); |
1525 | |
1526 | # Then, use these in your DBI statements |
1527 | my $sth = $dbh->prepare($stmt); |
1528 | $sth->execute(@bind); |
1529 | |
1530 | # Just generate the WHERE clause |
abe72f94 |
1531 | my($stmt, @bind) = $sql->where(\%where, \@order); |
32eab2da |
1532 | |
1533 | # Return values in the same order, for hashed queries |
1534 | # See PERFORMANCE section for more details |
1535 | my @bind = $sql->values(\%fieldvals); |
1536 | |
1537 | =head1 DESCRIPTION |
1538 | |
1539 | This module was inspired by the excellent L<DBIx::Abstract>. |
1540 | However, in using that module I found that what I really wanted |
1541 | to do was generate SQL, but still retain complete control over my |
1542 | statement handles and use the DBI interface. So, I set out to |
1543 | create an abstract SQL generation module. |
1544 | |
1545 | While based on the concepts used by L<DBIx::Abstract>, there are |
1546 | several important differences, especially when it comes to WHERE |
1547 | clauses. I have modified the concepts used to make the SQL easier |
1548 | to generate from Perl data structures and, IMO, more intuitive. |
1549 | The underlying idea is for this module to do what you mean, based |
1550 | on the data structures you provide it. The big advantage is that |
1551 | you don't have to modify your code every time your data changes, |
1552 | as this module figures it out. |
1553 | |
1554 | To begin with, an SQL INSERT is as easy as just specifying a hash |
1555 | of C<key=value> pairs: |
1556 | |
1557 | my %data = ( |
1558 | name => 'Jimbo Bobson', |
1559 | phone => '123-456-7890', |
1560 | address => '42 Sister Lane', |
1561 | city => 'St. Louis', |
1562 | state => 'Louisiana', |
1563 | ); |
1564 | |
1565 | The SQL can then be generated with this: |
1566 | |
1567 | my($stmt, @bind) = $sql->insert('people', \%data); |
1568 | |
1569 | Which would give you something like this: |
1570 | |
1571 | $stmt = "INSERT INTO people |
1572 | (address, city, name, phone, state) |
1573 | VALUES (?, ?, ?, ?, ?)"; |
1574 | @bind = ('42 Sister Lane', 'St. Louis', 'Jimbo Bobson', |
1575 | '123-456-7890', 'Louisiana'); |
1576 | |
1577 | These are then used directly in your DBI code: |
1578 | |
1579 | my $sth = $dbh->prepare($stmt); |
1580 | $sth->execute(@bind); |
1581 | |
96449e8e |
1582 | =head2 Inserting and Updating Arrays |
1583 | |
1584 | If your database has array types (like for example Postgres), |
1585 | activate the special option C<< array_datatypes => 1 >> |
9d48860e |
1586 | when creating the C<SQL::Abstract> object. |
96449e8e |
1587 | Then you may use an arrayref to insert and update database array types: |
1588 | |
1589 | my $sql = SQL::Abstract->new(array_datatypes => 1); |
1590 | my %data = ( |
1591 | planets => [qw/Mercury Venus Earth Mars/] |
1592 | ); |
9d48860e |
1593 | |
96449e8e |
1594 | my($stmt, @bind) = $sql->insert('solar_system', \%data); |
1595 | |
1596 | This results in: |
1597 | |
1598 | $stmt = "INSERT INTO solar_system (planets) VALUES (?)" |
1599 | |
1600 | @bind = (['Mercury', 'Venus', 'Earth', 'Mars']); |
1601 | |
1602 | |
1603 | =head2 Inserting and Updating SQL |
1604 | |
1605 | In order to apply SQL functions to elements of your C<%data> you may |
1606 | specify a reference to an arrayref for the given hash value. For example, |
1607 | if you need to execute the Oracle C<to_date> function on a value, you can |
1608 | say something like this: |
32eab2da |
1609 | |
1610 | my %data = ( |
1611 | name => 'Bill', |
96449e8e |
1612 | date_entered => \["to_date(?,'MM/DD/YYYY')", "03/02/2003"], |
9d48860e |
1613 | ); |
32eab2da |
1614 | |
1615 | The first value in the array is the actual SQL. Any other values are |
1616 | optional and would be included in the bind values array. This gives |
1617 | you: |
1618 | |
1619 | my($stmt, @bind) = $sql->insert('people', \%data); |
1620 | |
9d48860e |
1621 | $stmt = "INSERT INTO people (name, date_entered) |
32eab2da |
1622 | VALUES (?, to_date(?,'MM/DD/YYYY'))"; |
1623 | @bind = ('Bill', '03/02/2003'); |
1624 | |
1625 | An UPDATE is just as easy, all you change is the name of the function: |
1626 | |
1627 | my($stmt, @bind) = $sql->update('people', \%data); |
1628 | |
1629 | Notice that your C<%data> isn't touched; the module will generate |
1630 | the appropriately quirky SQL for you automatically. Usually you'll |
1631 | want to specify a WHERE clause for your UPDATE, though, which is |
1632 | where handling C<%where> hashes comes in handy... |
1633 | |
96449e8e |
1634 | =head2 Complex where statements |
1635 | |
32eab2da |
1636 | This module can generate pretty complicated WHERE statements |
1637 | easily. For example, simple C<key=value> pairs are taken to mean |
1638 | equality, and if you want to see if a field is within a set |
1639 | of values, you can use an arrayref. Let's say we wanted to |
1640 | SELECT some data based on this criteria: |
1641 | |
1642 | my %where = ( |
1643 | requestor => 'inna', |
1644 | worker => ['nwiger', 'rcwe', 'sfz'], |
1645 | status => { '!=', 'completed' } |
1646 | ); |
1647 | |
1648 | my($stmt, @bind) = $sql->select('tickets', '*', \%where); |
1649 | |
1650 | The above would give you something like this: |
1651 | |
1652 | $stmt = "SELECT * FROM tickets WHERE |
1653 | ( requestor = ? ) AND ( status != ? ) |
1654 | AND ( worker = ? OR worker = ? OR worker = ? )"; |
1655 | @bind = ('inna', 'completed', 'nwiger', 'rcwe', 'sfz'); |
1656 | |
1657 | Which you could then use in DBI code like so: |
1658 | |
1659 | my $sth = $dbh->prepare($stmt); |
1660 | $sth->execute(@bind); |
1661 | |
1662 | Easy, eh? |
1663 | |
1664 | =head1 FUNCTIONS |
1665 | |
1666 | The functions are simple. There's one for each major SQL operation, |
1667 | and a constructor you use first. The arguments are specified in a |
9d48860e |
1668 | similar order to each function (table, then fields, then a where |
32eab2da |
1669 | clause) to try and simplify things. |
1670 | |
83cab70b |
1671 | |
83cab70b |
1672 | |
32eab2da |
1673 | |
1674 | =head2 new(option => 'value') |
1675 | |
1676 | The C<new()> function takes a list of options and values, and returns |
1677 | a new B<SQL::Abstract> object which can then be used to generate SQL |
1678 | through the methods below. The options accepted are: |
1679 | |
1680 | =over |
1681 | |
1682 | =item case |
1683 | |
1684 | If set to 'lower', then SQL will be generated in all lowercase. By |
1685 | default SQL is generated in "textbook" case meaning something like: |
1686 | |
1687 | SELECT a_field FROM a_table WHERE some_field LIKE '%someval%' |
1688 | |
96449e8e |
1689 | Any setting other than 'lower' is ignored. |
1690 | |
32eab2da |
1691 | =item cmp |
1692 | |
1693 | This determines what the default comparison operator is. By default |
1694 | it is C<=>, meaning that a hash like this: |
1695 | |
1696 | %where = (name => 'nwiger', email => 'nate@wiger.org'); |
1697 | |
1698 | Will generate SQL like this: |
1699 | |
1700 | WHERE name = 'nwiger' AND email = 'nate@wiger.org' |
1701 | |
1702 | However, you may want loose comparisons by default, so if you set |
1703 | C<cmp> to C<like> you would get SQL such as: |
1704 | |
1705 | WHERE name like 'nwiger' AND email like 'nate@wiger.org' |
1706 | |
3af02ccb |
1707 | You can also override the comparison on an individual basis - see |
32eab2da |
1708 | the huge section on L</"WHERE CLAUSES"> at the bottom. |
1709 | |
96449e8e |
1710 | =item sqltrue, sqlfalse |
1711 | |
1712 | Expressions for inserting boolean values within SQL statements. |
6e0c6552 |
1713 | By default these are C<1=1> and C<1=0>. They are used |
1714 | by the special operators C<-in> and C<-not_in> for generating |
1715 | correct SQL even when the argument is an empty array (see below). |
96449e8e |
1716 | |
32eab2da |
1717 | =item logic |
1718 | |
1719 | This determines the default logical operator for multiple WHERE |
7cac25e6 |
1720 | statements in arrays or hashes. If absent, the default logic is "or" |
1721 | for arrays, and "and" for hashes. This means that a WHERE |
32eab2da |
1722 | array of the form: |
1723 | |
1724 | @where = ( |
9d48860e |
1725 | event_date => {'>=', '2/13/99'}, |
1726 | event_date => {'<=', '4/24/03'}, |
32eab2da |
1727 | ); |
1728 | |
7cac25e6 |
1729 | will generate SQL like this: |
32eab2da |
1730 | |
1731 | WHERE event_date >= '2/13/99' OR event_date <= '4/24/03' |
1732 | |
1733 | This is probably not what you want given this query, though (look |
1734 | at the dates). To change the "OR" to an "AND", simply specify: |
1735 | |
1736 | my $sql = SQL::Abstract->new(logic => 'and'); |
1737 | |
1738 | Which will change the above C<WHERE> to: |
1739 | |
1740 | WHERE event_date >= '2/13/99' AND event_date <= '4/24/03' |
1741 | |
96449e8e |
1742 | The logic can also be changed locally by inserting |
7cac25e6 |
1743 | a modifier in front of an arrayref : |
96449e8e |
1744 | |
9d48860e |
1745 | @where = (-and => [event_date => {'>=', '2/13/99'}, |
7cac25e6 |
1746 | event_date => {'<=', '4/24/03'} ]); |
96449e8e |
1747 | |
1748 | See the L</"WHERE CLAUSES"> section for explanations. |
1749 | |
32eab2da |
1750 | =item convert |
1751 | |
1752 | This will automatically convert comparisons using the specified SQL |
1753 | function for both column and value. This is mostly used with an argument |
1754 | of C<upper> or C<lower>, so that the SQL will have the effect of |
1755 | case-insensitive "searches". For example, this: |
1756 | |
1757 | $sql = SQL::Abstract->new(convert => 'upper'); |
1758 | %where = (keywords => 'MaKe iT CAse inSeNSItive'); |
1759 | |
1760 | Will turn out the following SQL: |
1761 | |
1762 | WHERE upper(keywords) like upper('MaKe iT CAse inSeNSItive') |
1763 | |
1764 | The conversion can be C<upper()>, C<lower()>, or any other SQL function |
1765 | that can be applied symmetrically to fields (actually B<SQL::Abstract> does |
1766 | not validate this option; it will just pass through what you specify verbatim). |
1767 | |
1768 | =item bindtype |
1769 | |
1770 | This is a kludge because many databases suck. For example, you can't |
1771 | just bind values using DBI's C<execute()> for Oracle C<CLOB> or C<BLOB> fields. |
1772 | Instead, you have to use C<bind_param()>: |
1773 | |
1774 | $sth->bind_param(1, 'reg data'); |
1775 | $sth->bind_param(2, $lots, {ora_type => ORA_CLOB}); |
1776 | |
1777 | The problem is, B<SQL::Abstract> will normally just return a C<@bind> array, |
1778 | which loses track of which field each slot refers to. Fear not. |
1779 | |
1780 | If you specify C<bindtype> in new, you can determine how C<@bind> is returned. |
1781 | Currently, you can specify either C<normal> (default) or C<columns>. If you |
1782 | specify C<columns>, you will get an array that looks like this: |
1783 | |
1784 | my $sql = SQL::Abstract->new(bindtype => 'columns'); |
1785 | my($stmt, @bind) = $sql->insert(...); |
1786 | |
1787 | @bind = ( |
1788 | [ 'column1', 'value1' ], |
1789 | [ 'column2', 'value2' ], |
1790 | [ 'column3', 'value3' ], |
1791 | ); |
1792 | |
1793 | You can then iterate through this manually, using DBI's C<bind_param()>. |
e3f9dff4 |
1794 | |
32eab2da |
1795 | $sth->prepare($stmt); |
1796 | my $i = 1; |
1797 | for (@bind) { |
1798 | my($col, $data) = @$_; |
1799 | if ($col eq 'details' || $col eq 'comments') { |
1800 | $sth->bind_param($i, $data, {ora_type => ORA_CLOB}); |
1801 | } elsif ($col eq 'image') { |
1802 | $sth->bind_param($i, $data, {ora_type => ORA_BLOB}); |
1803 | } else { |
1804 | $sth->bind_param($i, $data); |
1805 | } |
1806 | $i++; |
1807 | } |
1808 | $sth->execute; # execute without @bind now |
1809 | |
1810 | Now, why would you still use B<SQL::Abstract> if you have to do this crap? |
1811 | Basically, the advantage is still that you don't have to care which fields |
1812 | are or are not included. You could wrap that above C<for> loop in a simple |
1813 | sub called C<bind_fields()> or something and reuse it repeatedly. You still |
1814 | get a layer of abstraction over manual SQL specification. |
1815 | |
deb148a2 |
1816 | Note that if you set L</bindtype> to C<columns>, the C<\[$sql, @bind]> |
1817 | construct (see L</Literal SQL with placeholders and bind values (subqueries)>) |
1818 | will expect the bind values in this format. |
1819 | |
32eab2da |
1820 | =item quote_char |
1821 | |
1822 | This is the character that a table or column name will be quoted |
9d48860e |
1823 | with. By default this is an empty string, but you could set it to |
32eab2da |
1824 | the character C<`>, to generate SQL like this: |
1825 | |
1826 | SELECT `a_field` FROM `a_table` WHERE `some_field` LIKE '%someval%' |
1827 | |
96449e8e |
1828 | Alternatively, you can supply an array ref of two items, the first being the left |
1829 | hand quote character, and the second the right hand quote character. For |
1830 | example, you could supply C<['[',']']> for SQL Server 2000 compliant quotes |
1831 | that generates SQL like this: |
1832 | |
1833 | SELECT [a_field] FROM [a_table] WHERE [some_field] LIKE '%someval%' |
1834 | |
9d48860e |
1835 | Quoting is useful if you have tables or columns names that are reserved |
96449e8e |
1836 | words in your database's SQL dialect. |
32eab2da |
1837 | |
1838 | =item name_sep |
1839 | |
1840 | This is the character that separates a table and column name. It is |
1841 | necessary to specify this when the C<quote_char> option is selected, |
1842 | so that tables and column names can be individually quoted like this: |
1843 | |
1844 | SELECT `table`.`one_field` FROM `table` WHERE `table`.`other_field` = 1 |
1845 | |
b6251592 |
1846 | =item injection_guard |
1847 | |
1848 | A regular expression C<qr/.../> that is applied to any C<-function> and unquoted |
1849 | column name specified in a query structure. This is a safety mechanism to avoid |
1850 | injection attacks when mishandling user input e.g.: |
1851 | |
1852 | my %condition_as_column_value_pairs = get_values_from_user(); |
1853 | $sqla->select( ... , \%condition_as_column_value_pairs ); |
1854 | |
1855 | If the expression matches an exception is thrown. Note that literal SQL |
1856 | supplied via C<\'...'> or C<\['...']> is B<not> checked in any way. |
1857 | |
1858 | Defaults to checking for C<;> and the C<GO> keyword (TransactSQL) |
1859 | |
96449e8e |
1860 | =item array_datatypes |
32eab2da |
1861 | |
9d48860e |
1862 | When this option is true, arrayrefs in INSERT or UPDATE are |
1863 | interpreted as array datatypes and are passed directly |
96449e8e |
1864 | to the DBI layer. |
1865 | When this option is false, arrayrefs are interpreted |
1866 | as literal SQL, just like refs to arrayrefs |
1867 | (but this behavior is for backwards compatibility; when writing |
1868 | new queries, use the "reference to arrayref" syntax |
1869 | for literal SQL). |
32eab2da |
1870 | |
32eab2da |
1871 | |
96449e8e |
1872 | =item special_ops |
32eab2da |
1873 | |
9d48860e |
1874 | Takes a reference to a list of "special operators" |
96449e8e |
1875 | to extend the syntax understood by L<SQL::Abstract>. |
1876 | See section L</"SPECIAL OPERATORS"> for details. |
32eab2da |
1877 | |
59f23b3d |
1878 | =item unary_ops |
1879 | |
9d48860e |
1880 | Takes a reference to a list of "unary operators" |
59f23b3d |
1881 | to extend the syntax understood by L<SQL::Abstract>. |
1882 | See section L</"UNARY OPERATORS"> for details. |
1883 | |
32eab2da |
1884 | |
32eab2da |
1885 | |
96449e8e |
1886 | =back |
32eab2da |
1887 | |
02288357 |
1888 | =head2 insert($table, \@values || \%fieldvals, \%options) |
32eab2da |
1889 | |
1890 | This is the simplest function. You simply give it a table name |
1891 | and either an arrayref of values or hashref of field/value pairs. |
1892 | It returns an SQL INSERT statement and a list of bind values. |
96449e8e |
1893 | See the sections on L</"Inserting and Updating Arrays"> and |
1894 | L</"Inserting and Updating SQL"> for information on how to insert |
1895 | with those data types. |
32eab2da |
1896 | |
02288357 |
1897 | The optional C<\%options> hash reference may contain additional |
1898 | options to generate the insert SQL. Currently supported options |
1899 | are: |
1900 | |
1901 | =over 4 |
1902 | |
1903 | =item returning |
1904 | |
1905 | Takes either a scalar of raw SQL fields, or an array reference of |
1906 | field names, and adds on an SQL C<RETURNING> statement at the end. |
1907 | This allows you to return data generated by the insert statement |
1908 | (such as row IDs) without performing another C<SELECT> statement. |
1909 | Note, however, this is not part of the SQL standard and may not |
1910 | be supported by all database engines. |
1911 | |
1912 | =back |
1913 | |
32eab2da |
1914 | =head2 update($table, \%fieldvals, \%where) |
1915 | |
1916 | This takes a table, hashref of field/value pairs, and an optional |
86298391 |
1917 | hashref L<WHERE clause|/WHERE CLAUSES>. It returns an SQL UPDATE function and a list |
32eab2da |
1918 | of bind values. |
96449e8e |
1919 | See the sections on L</"Inserting and Updating Arrays"> and |
1920 | L</"Inserting and Updating SQL"> for information on how to insert |
1921 | with those data types. |
32eab2da |
1922 | |
96449e8e |
1923 | =head2 select($source, $fields, $where, $order) |
32eab2da |
1924 | |
9d48860e |
1925 | This returns a SQL SELECT statement and associated list of bind values, as |
96449e8e |
1926 | specified by the arguments : |
32eab2da |
1927 | |
96449e8e |
1928 | =over |
32eab2da |
1929 | |
96449e8e |
1930 | =item $source |
32eab2da |
1931 | |
9d48860e |
1932 | Specification of the 'FROM' part of the statement. |
96449e8e |
1933 | The argument can be either a plain scalar (interpreted as a table |
1934 | name, will be quoted), or an arrayref (interpreted as a list |
1935 | of table names, joined by commas, quoted), or a scalarref |
1936 | (literal table name, not quoted), or a ref to an arrayref |
1937 | (list of literal table names, joined by commas, not quoted). |
32eab2da |
1938 | |
96449e8e |
1939 | =item $fields |
32eab2da |
1940 | |
9d48860e |
1941 | Specification of the list of fields to retrieve from |
96449e8e |
1942 | the source. |
1943 | The argument can be either an arrayref (interpreted as a list |
9d48860e |
1944 | of field names, will be joined by commas and quoted), or a |
96449e8e |
1945 | plain scalar (literal SQL, not quoted). |
521647e7 |
1946 | Please observe that this API is not as flexible as that of |
1947 | the first argument C<$source>, for backwards compatibility reasons. |
32eab2da |
1948 | |
96449e8e |
1949 | =item $where |
32eab2da |
1950 | |
96449e8e |
1951 | Optional argument to specify the WHERE part of the query. |
1952 | The argument is most often a hashref, but can also be |
9d48860e |
1953 | an arrayref or plain scalar -- |
96449e8e |
1954 | see section L<WHERE clause|/"WHERE CLAUSES"> for details. |
32eab2da |
1955 | |
96449e8e |
1956 | =item $order |
32eab2da |
1957 | |
96449e8e |
1958 | Optional argument to specify the ORDER BY part of the query. |
9d48860e |
1959 | The argument can be a scalar, a hashref or an arrayref |
96449e8e |
1960 | -- see section L<ORDER BY clause|/"ORDER BY CLAUSES"> |
1961 | for details. |
32eab2da |
1962 | |
96449e8e |
1963 | =back |
32eab2da |
1964 | |
32eab2da |
1965 | |
1966 | =head2 delete($table, \%where) |
1967 | |
86298391 |
1968 | This takes a table name and optional hashref L<WHERE clause|/WHERE CLAUSES>. |
32eab2da |
1969 | It returns an SQL DELETE statement and list of bind values. |
1970 | |
32eab2da |
1971 | =head2 where(\%where, \@order) |
1972 | |
1973 | This is used to generate just the WHERE clause. For example, |
1974 | if you have an arbitrary data structure and know what the |
1975 | rest of your SQL is going to look like, but want an easy way |
1976 | to produce a WHERE clause, use this. It returns an SQL WHERE |
1977 | clause and list of bind values. |
1978 | |
32eab2da |
1979 | |
1980 | =head2 values(\%data) |
1981 | |
1982 | This just returns the values from the hash C<%data>, in the same |
1983 | order that would be returned from any of the other above queries. |
1984 | Using this allows you to markedly speed up your queries if you |
1985 | are affecting lots of rows. See below under the L</"PERFORMANCE"> section. |
1986 | |
32eab2da |
1987 | =head2 generate($any, 'number', $of, \@data, $struct, \%types) |
1988 | |
1989 | Warning: This is an experimental method and subject to change. |
1990 | |
1991 | This returns arbitrarily generated SQL. It's a really basic shortcut. |
1992 | It will return two different things, depending on return context: |
1993 | |
1994 | my($stmt, @bind) = $sql->generate('create table', \$table, \@fields); |
1995 | my $stmt_and_val = $sql->generate('create table', \$table, \@fields); |
1996 | |
1997 | These would return the following: |
1998 | |
1999 | # First calling form |
2000 | $stmt = "CREATE TABLE test (?, ?)"; |
2001 | @bind = (field1, field2); |
2002 | |
2003 | # Second calling form |
2004 | $stmt_and_val = "CREATE TABLE test (field1, field2)"; |
2005 | |
2006 | Depending on what you're trying to do, it's up to you to choose the correct |
2007 | format. In this example, the second form is what you would want. |
2008 | |
2009 | By the same token: |
2010 | |
2011 | $sql->generate('alter session', { nls_date_format => 'MM/YY' }); |
2012 | |
2013 | Might give you: |
2014 | |
2015 | ALTER SESSION SET nls_date_format = 'MM/YY' |
2016 | |
2017 | You get the idea. Strings get their case twiddled, but everything |
2018 | else remains verbatim. |
2019 | |
32eab2da |
2020 | =head1 WHERE CLAUSES |
2021 | |
96449e8e |
2022 | =head2 Introduction |
2023 | |
32eab2da |
2024 | This module uses a variation on the idea from L<DBIx::Abstract>. It |
2025 | is B<NOT>, repeat I<not> 100% compatible. B<The main logic of this |
2026 | module is that things in arrays are OR'ed, and things in hashes |
2027 | are AND'ed.> |
2028 | |
2029 | The easiest way to explain is to show lots of examples. After |
2030 | each C<%where> hash shown, it is assumed you used: |
2031 | |
2032 | my($stmt, @bind) = $sql->where(\%where); |
2033 | |
2034 | However, note that the C<%where> hash can be used directly in any |
2035 | of the other functions as well, as described above. |
2036 | |
96449e8e |
2037 | =head2 Key-value pairs |
2038 | |
32eab2da |
2039 | So, let's get started. To begin, a simple hash: |
2040 | |
2041 | my %where = ( |
2042 | user => 'nwiger', |
2043 | status => 'completed' |
2044 | ); |
2045 | |
2046 | Is converted to SQL C<key = val> statements: |
2047 | |
2048 | $stmt = "WHERE user = ? AND status = ?"; |
2049 | @bind = ('nwiger', 'completed'); |
2050 | |
2051 | One common thing I end up doing is having a list of values that |
2052 | a field can be in. To do this, simply specify a list inside of |
2053 | an arrayref: |
2054 | |
2055 | my %where = ( |
2056 | user => 'nwiger', |
2057 | status => ['assigned', 'in-progress', 'pending']; |
2058 | ); |
2059 | |
2060 | This simple code will create the following: |
9d48860e |
2061 | |
32eab2da |
2062 | $stmt = "WHERE user = ? AND ( status = ? OR status = ? OR status = ? )"; |
2063 | @bind = ('nwiger', 'assigned', 'in-progress', 'pending'); |
2064 | |
9d48860e |
2065 | A field associated to an empty arrayref will be considered a |
7cac25e6 |
2066 | logical false and will generate 0=1. |
8a68b5be |
2067 | |
b864ba9b |
2068 | =head2 Tests for NULL values |
2069 | |
2070 | If the value part is C<undef> then this is converted to SQL <IS NULL> |
2071 | |
2072 | my %where = ( |
2073 | user => 'nwiger', |
2074 | status => undef, |
2075 | ); |
2076 | |
2077 | becomes: |
2078 | |
2079 | $stmt = "WHERE user = ? AND status IS NULL"; |
2080 | @bind = ('nwiger'); |
2081 | |
e9614080 |
2082 | To test if a column IS NOT NULL: |
2083 | |
2084 | my %where = ( |
2085 | user => 'nwiger', |
2086 | status => { '!=', undef }, |
2087 | ); |
cc422895 |
2088 | |
6e0c6552 |
2089 | =head2 Specific comparison operators |
96449e8e |
2090 | |
32eab2da |
2091 | If you want to specify a different type of operator for your comparison, |
2092 | you can use a hashref for a given column: |
2093 | |
2094 | my %where = ( |
2095 | user => 'nwiger', |
2096 | status => { '!=', 'completed' } |
2097 | ); |
2098 | |
2099 | Which would generate: |
2100 | |
2101 | $stmt = "WHERE user = ? AND status != ?"; |
2102 | @bind = ('nwiger', 'completed'); |
2103 | |
2104 | To test against multiple values, just enclose the values in an arrayref: |
2105 | |
96449e8e |
2106 | status => { '=', ['assigned', 'in-progress', 'pending'] }; |
2107 | |
f2d5020d |
2108 | Which would give you: |
96449e8e |
2109 | |
2110 | "WHERE status = ? OR status = ? OR status = ?" |
2111 | |
2112 | |
2113 | The hashref can also contain multiple pairs, in which case it is expanded |
32eab2da |
2114 | into an C<AND> of its elements: |
2115 | |
2116 | my %where = ( |
2117 | user => 'nwiger', |
2118 | status => { '!=', 'completed', -not_like => 'pending%' } |
2119 | ); |
2120 | |
2121 | # Or more dynamically, like from a form |
2122 | $where{user} = 'nwiger'; |
2123 | $where{status}{'!='} = 'completed'; |
2124 | $where{status}{'-not_like'} = 'pending%'; |
2125 | |
2126 | # Both generate this |
2127 | $stmt = "WHERE user = ? AND status != ? AND status NOT LIKE ?"; |
2128 | @bind = ('nwiger', 'completed', 'pending%'); |
2129 | |
96449e8e |
2130 | |
32eab2da |
2131 | To get an OR instead, you can combine it with the arrayref idea: |
2132 | |
2133 | my %where => ( |
2134 | user => 'nwiger', |
1a6f2a03 |
2135 | priority => [ { '=', 2 }, { '>', 5 } ] |
32eab2da |
2136 | ); |
2137 | |
2138 | Which would generate: |
2139 | |
1a6f2a03 |
2140 | $stmt = "WHERE ( priority = ? OR priority > ? ) AND user = ?"; |
2141 | @bind = ('2', '5', 'nwiger'); |
32eab2da |
2142 | |
44b9e502 |
2143 | If you want to include literal SQL (with or without bind values), just use a |
2144 | scalar reference or array reference as the value: |
2145 | |
2146 | my %where = ( |
2147 | date_entered => { '>' => \["to_date(?, 'MM/DD/YYYY')", "11/26/2008"] }, |
2148 | date_expires => { '<' => \"now()" } |
2149 | ); |
2150 | |
2151 | Which would generate: |
2152 | |
2153 | $stmt = "WHERE date_entered > "to_date(?, 'MM/DD/YYYY') AND date_expires < now()"; |
2154 | @bind = ('11/26/2008'); |
2155 | |
96449e8e |
2156 | |
2157 | =head2 Logic and nesting operators |
2158 | |
2159 | In the example above, |
2160 | there is a subtle trap if you want to say something like |
32eab2da |
2161 | this (notice the C<AND>): |
2162 | |
2163 | WHERE priority != ? AND priority != ? |
2164 | |
2165 | Because, in Perl you I<can't> do this: |
2166 | |
2167 | priority => { '!=', 2, '!=', 1 } |
2168 | |
2169 | As the second C<!=> key will obliterate the first. The solution |
2170 | is to use the special C<-modifier> form inside an arrayref: |
2171 | |
9d48860e |
2172 | priority => [ -and => {'!=', 2}, |
96449e8e |
2173 | {'!=', 1} ] |
2174 | |
32eab2da |
2175 | |
2176 | Normally, these would be joined by C<OR>, but the modifier tells it |
2177 | to use C<AND> instead. (Hint: You can use this in conjunction with the |
2178 | C<logic> option to C<new()> in order to change the way your queries |
2179 | work by default.) B<Important:> Note that the C<-modifier> goes |
2180 | B<INSIDE> the arrayref, as an extra first element. This will |
2181 | B<NOT> do what you think it might: |
2182 | |
2183 | priority => -and => [{'!=', 2}, {'!=', 1}] # WRONG! |
2184 | |
2185 | Here is a quick list of equivalencies, since there is some overlap: |
2186 | |
2187 | # Same |
2188 | status => {'!=', 'completed', 'not like', 'pending%' } |
2189 | status => [ -and => {'!=', 'completed'}, {'not like', 'pending%'}] |
2190 | |
2191 | # Same |
2192 | status => {'=', ['assigned', 'in-progress']} |
2193 | status => [ -or => {'=', 'assigned'}, {'=', 'in-progress'}] |
2194 | status => [ {'=', 'assigned'}, {'=', 'in-progress'} ] |
2195 | |
e3f9dff4 |
2196 | |
2197 | |
96449e8e |
2198 | =head2 Special operators : IN, BETWEEN, etc. |
2199 | |
32eab2da |
2200 | You can also use the hashref format to compare a list of fields using the |
2201 | C<IN> comparison operator, by specifying the list as an arrayref: |
2202 | |
2203 | my %where = ( |
2204 | status => 'completed', |
2205 | reportid => { -in => [567, 2335, 2] } |
2206 | ); |
2207 | |
2208 | Which would generate: |
2209 | |
2210 | $stmt = "WHERE status = ? AND reportid IN (?,?,?)"; |
2211 | @bind = ('completed', '567', '2335', '2'); |
2212 | |
9d48860e |
2213 | The reverse operator C<-not_in> generates SQL C<NOT IN> and is used in |
96449e8e |
2214 | the same way. |
2215 | |
6e0c6552 |
2216 | If the argument to C<-in> is an empty array, 'sqlfalse' is generated |
2217 | (by default : C<1=0>). Similarly, C<< -not_in => [] >> generates |
2218 | 'sqltrue' (by default : C<1=1>). |
2219 | |
e41c3bdd |
2220 | In addition to the array you can supply a chunk of literal sql or |
2221 | literal sql with bind: |
6e0c6552 |
2222 | |
e41c3bdd |
2223 | my %where = { |
2224 | customer => { -in => \[ |
2225 | 'SELECT cust_id FROM cust WHERE balance > ?', |
2226 | 2000, |
2227 | ], |
2228 | status => { -in => \'SELECT status_codes FROM states' }, |
2229 | }; |
6e0c6552 |
2230 | |
e41c3bdd |
2231 | would generate: |
2232 | |
2233 | $stmt = "WHERE ( |
2234 | customer IN ( SELECT cust_id FROM cust WHERE balance > ? ) |
2235 | AND status IN ( SELECT status_codes FROM states ) |
2236 | )"; |
2237 | @bind = ('2000'); |
2238 | |
0dfd2442 |
2239 | Finally, if the argument to C<-in> is not a reference, it will be |
2240 | treated as a single-element array. |
e41c3bdd |
2241 | |
2242 | Another pair of operators is C<-between> and C<-not_between>, |
96449e8e |
2243 | used with an arrayref of two values: |
32eab2da |
2244 | |
2245 | my %where = ( |
2246 | user => 'nwiger', |
2247 | completion_date => { |
2248 | -not_between => ['2002-10-01', '2003-02-06'] |
2249 | } |
2250 | ); |
2251 | |
2252 | Would give you: |
2253 | |
2254 | WHERE user = ? AND completion_date NOT BETWEEN ( ? AND ? ) |
2255 | |
e41c3bdd |
2256 | Just like with C<-in> all plausible combinations of literal SQL |
2257 | are possible: |
2258 | |
2259 | my %where = { |
2260 | start0 => { -between => [ 1, 2 ] }, |
2261 | start1 => { -between => \["? AND ?", 1, 2] }, |
2262 | start2 => { -between => \"lower(x) AND upper(y)" }, |
9d48860e |
2263 | start3 => { -between => [ |
e41c3bdd |
2264 | \"lower(x)", |
2265 | \["upper(?)", 'stuff' ], |
2266 | ] }, |
2267 | }; |
2268 | |
2269 | Would give you: |
2270 | |
2271 | $stmt = "WHERE ( |
2272 | ( start0 BETWEEN ? AND ? ) |
2273 | AND ( start1 BETWEEN ? AND ? ) |
2274 | AND ( start2 BETWEEN lower(x) AND upper(y) ) |
2275 | AND ( start3 BETWEEN lower(x) AND upper(?) ) |
2276 | )"; |
2277 | @bind = (1, 2, 1, 2, 'stuff'); |
2278 | |
2279 | |
9d48860e |
2280 | These are the two builtin "special operators"; but the |
96449e8e |
2281 | list can be expanded : see section L</"SPECIAL OPERATORS"> below. |
2282 | |
59f23b3d |
2283 | =head2 Unary operators: bool |
97a920ef |
2284 | |
2285 | If you wish to test against boolean columns or functions within your |
2286 | database you can use the C<-bool> and C<-not_bool> operators. For |
2287 | example to test the column C<is_user> being true and the column |
827bb0eb |
2288 | C<is_enabled> being false you would use:- |
97a920ef |
2289 | |
2290 | my %where = ( |
2291 | -bool => 'is_user', |
2292 | -not_bool => 'is_enabled', |
2293 | ); |
2294 | |
2295 | Would give you: |
2296 | |
277b5d3f |
2297 | WHERE is_user AND NOT is_enabled |
97a920ef |
2298 | |
0b604e9d |
2299 | If a more complex combination is required, testing more conditions, |
2300 | then you should use the and/or operators:- |
2301 | |
2302 | my %where = ( |
2303 | -and => [ |
2304 | -bool => 'one', |
2305 | -bool => 'two', |
2306 | -bool => 'three', |
2307 | -not_bool => 'four', |
2308 | ], |
2309 | ); |
2310 | |
2311 | Would give you: |
2312 | |
2313 | WHERE one AND two AND three AND NOT four |
97a920ef |
2314 | |
2315 | |
107b72f1 |
2316 | =head2 Nested conditions, -and/-or prefixes |
96449e8e |
2317 | |
32eab2da |
2318 | So far, we've seen how multiple conditions are joined with a top-level |
2319 | C<AND>. We can change this by putting the different conditions we want in |
2320 | hashes and then putting those hashes in an array. For example: |
2321 | |
2322 | my @where = ( |
2323 | { |
2324 | user => 'nwiger', |
2325 | status => { -like => ['pending%', 'dispatched'] }, |
2326 | }, |
2327 | { |
2328 | user => 'robot', |
2329 | status => 'unassigned', |
2330 | } |
2331 | ); |
2332 | |
2333 | This data structure would create the following: |
2334 | |
2335 | $stmt = "WHERE ( user = ? AND ( status LIKE ? OR status LIKE ? ) ) |
2336 | OR ( user = ? AND status = ? ) )"; |
2337 | @bind = ('nwiger', 'pending', 'dispatched', 'robot', 'unassigned'); |
2338 | |
107b72f1 |
2339 | |
48d9f5f8 |
2340 | Clauses in hashrefs or arrayrefs can be prefixed with an C<-and> or C<-or> |
2341 | to change the logic inside : |
32eab2da |
2342 | |
2343 | my @where = ( |
2344 | -and => [ |
2345 | user => 'nwiger', |
48d9f5f8 |
2346 | [ |
2347 | -and => [ workhrs => {'>', 20}, geo => 'ASIA' ], |
2348 | -or => { workhrs => {'<', 50}, geo => 'EURO' }, |
32eab2da |
2349 | ], |
2350 | ], |
2351 | ); |
2352 | |
2353 | That would yield: |
2354 | |
48d9f5f8 |
2355 | WHERE ( user = ? AND ( |
2356 | ( workhrs > ? AND geo = ? ) |
2357 | OR ( workhrs < ? OR geo = ? ) |
2358 | ) ) |
107b72f1 |
2359 | |
cc422895 |
2360 | =head3 Algebraic inconsistency, for historical reasons |
107b72f1 |
2361 | |
7cac25e6 |
2362 | C<Important note>: when connecting several conditions, the C<-and->|C<-or> |
2363 | operator goes C<outside> of the nested structure; whereas when connecting |
2364 | several constraints on one column, the C<-and> operator goes |
2365 | C<inside> the arrayref. Here is an example combining both features : |
2366 | |
2367 | my @where = ( |
2368 | -and => [a => 1, b => 2], |
2369 | -or => [c => 3, d => 4], |
2370 | e => [-and => {-like => 'foo%'}, {-like => '%bar'} ] |
2371 | ) |
2372 | |
2373 | yielding |
2374 | |
9d48860e |
2375 | WHERE ( ( ( a = ? AND b = ? ) |
2376 | OR ( c = ? OR d = ? ) |
7cac25e6 |
2377 | OR ( e LIKE ? AND e LIKE ? ) ) ) |
2378 | |
107b72f1 |
2379 | This difference in syntax is unfortunate but must be preserved for |
2380 | historical reasons. So be careful : the two examples below would |
2381 | seem algebraically equivalent, but they are not |
2382 | |
9d48860e |
2383 | {col => [-and => {-like => 'foo%'}, {-like => '%bar'}]} |
107b72f1 |
2384 | # yields : WHERE ( ( col LIKE ? AND col LIKE ? ) ) |
2385 | |
9d48860e |
2386 | [-and => {col => {-like => 'foo%'}, {col => {-like => '%bar'}}]] |
107b72f1 |
2387 | # yields : WHERE ( ( col LIKE ? OR col LIKE ? ) ) |
2388 | |
7cac25e6 |
2389 | |
cc422895 |
2390 | =head2 Literal SQL and value type operators |
96449e8e |
2391 | |
cc422895 |
2392 | The basic premise of SQL::Abstract is that in WHERE specifications the "left |
2393 | side" is a column name and the "right side" is a value (normally rendered as |
2394 | a placeholder). This holds true for both hashrefs and arrayref pairs as you |
2395 | see in the L</WHERE CLAUSES> examples above. Sometimes it is necessary to |
2396 | alter this behavior. There are several ways of doing so. |
e9614080 |
2397 | |
cc422895 |
2398 | =head3 -ident |
2399 | |
2400 | This is a virtual operator that signals the string to its right side is an |
2401 | identifier (a column name) and not a value. For example to compare two |
2402 | columns you would write: |
32eab2da |
2403 | |
e9614080 |
2404 | my %where = ( |
2405 | priority => { '<', 2 }, |
cc422895 |
2406 | requestor => { -ident => 'submitter' }, |
e9614080 |
2407 | ); |
2408 | |
2409 | which creates: |
2410 | |
2411 | $stmt = "WHERE priority < ? AND requestor = submitter"; |
2412 | @bind = ('2'); |
2413 | |
cc422895 |
2414 | If you are maintaining legacy code you may see a different construct as |
2415 | described in L</Deprecated usage of Literal SQL>, please use C<-ident> in new |
2416 | code. |
2417 | |
2418 | =head3 -value |
e9614080 |
2419 | |
cc422895 |
2420 | This is a virtual operator that signals that the construct to its right side |
2421 | is a value to be passed to DBI. This is for example necessary when you want |
2422 | to write a where clause against an array (for RDBMS that support such |
2423 | datatypes). For example: |
e9614080 |
2424 | |
32eab2da |
2425 | my %where = ( |
cc422895 |
2426 | array => { -value => [1, 2, 3] } |
32eab2da |
2427 | ); |
2428 | |
cc422895 |
2429 | will result in: |
32eab2da |
2430 | |
cc422895 |
2431 | $stmt = 'WHERE array = ?'; |
2432 | @bind = ([1, 2, 3]); |
32eab2da |
2433 | |
cc422895 |
2434 | Note that if you were to simply say: |
32eab2da |
2435 | |
2436 | my %where = ( |
cc422895 |
2437 | array => [1, 2, 3] |
32eab2da |
2438 | ); |
2439 | |
3af02ccb |
2440 | the result would probably not be what you wanted: |
cc422895 |
2441 | |
2442 | $stmt = 'WHERE array = ? OR array = ? OR array = ?'; |
2443 | @bind = (1, 2, 3); |
2444 | |
2445 | =head3 Literal SQL |
96449e8e |
2446 | |
cc422895 |
2447 | Finally, sometimes only literal SQL will do. To include a random snippet |
2448 | of SQL verbatim, you specify it as a scalar reference. Consider this only |
2449 | as a last resort. Usually there is a better way. For example: |
96449e8e |
2450 | |
2451 | my %where = ( |
cc422895 |
2452 | priority => { '<', 2 }, |
2453 | requestor => { -in => \'(SELECT name FROM hitmen)' }, |
96449e8e |
2454 | ); |
2455 | |
cc422895 |
2456 | Would create: |
96449e8e |
2457 | |
cc422895 |
2458 | $stmt = "WHERE priority < ? AND requestor IN (SELECT name FROM hitmen)" |
2459 | @bind = (2); |
2460 | |
2461 | Note that in this example, you only get one bind parameter back, since |
2462 | the verbatim SQL is passed as part of the statement. |
2463 | |
2464 | =head4 CAVEAT |
2465 | |
2466 | Never use untrusted input as a literal SQL argument - this is a massive |
2467 | security risk (there is no way to check literal snippets for SQL |
2468 | injections and other nastyness). If you need to deal with untrusted input |
2469 | use literal SQL with placeholders as described next. |
96449e8e |
2470 | |
cc422895 |
2471 | =head3 Literal SQL with placeholders and bind values (subqueries) |
96449e8e |
2472 | |
2473 | If the literal SQL to be inserted has placeholders and bind values, |
2474 | use a reference to an arrayref (yes this is a double reference -- |
2475 | not so common, but perfectly legal Perl). For example, to find a date |
2476 | in Postgres you can use something like this: |
2477 | |
2478 | my %where = ( |
2479 | date_column => \[q/= date '2008-09-30' - ?::integer/, 10/] |
2480 | ) |
2481 | |
2482 | This would create: |
2483 | |
d2a8fe1a |
2484 | $stmt = "WHERE ( date_column = date '2008-09-30' - ?::integer )" |
96449e8e |
2485 | @bind = ('10'); |
2486 | |
deb148a2 |
2487 | Note that you must pass the bind values in the same format as they are returned |
62552e7d |
2488 | by L</where>. That means that if you set L</bindtype> to C<columns>, you must |
26f2dca5 |
2489 | provide the bind values in the C<< [ column_meta => value ] >> format, where |
2490 | C<column_meta> is an opaque scalar value; most commonly the column name, but |
62552e7d |
2491 | you can use any scalar value (including references and blessed references), |
2492 | L<SQL::Abstract> will simply pass it through intact. So if C<bindtype> is set |
2493 | to C<columns> the above example will look like: |
deb148a2 |
2494 | |
2495 | my %where = ( |
2496 | date_column => \[q/= date '2008-09-30' - ?::integer/, [ dummy => 10 ]/] |
2497 | ) |
96449e8e |
2498 | |
2499 | Literal SQL is especially useful for nesting parenthesized clauses in the |
2500 | main SQL query. Here is a first example : |
2501 | |
2502 | my ($sub_stmt, @sub_bind) = ("SELECT c1 FROM t1 WHERE c2 < ? AND c3 LIKE ?", |
2503 | 100, "foo%"); |
2504 | my %where = ( |
2505 | foo => 1234, |
2506 | bar => \["IN ($sub_stmt)" => @sub_bind], |
2507 | ); |
2508 | |
2509 | This yields : |
2510 | |
9d48860e |
2511 | $stmt = "WHERE (foo = ? AND bar IN (SELECT c1 FROM t1 |
96449e8e |
2512 | WHERE c2 < ? AND c3 LIKE ?))"; |
2513 | @bind = (1234, 100, "foo%"); |
2514 | |
9d48860e |
2515 | Other subquery operators, like for example C<"E<gt> ALL"> or C<"NOT IN">, |
96449e8e |
2516 | are expressed in the same way. Of course the C<$sub_stmt> and |
9d48860e |
2517 | its associated bind values can be generated through a former call |
96449e8e |
2518 | to C<select()> : |
2519 | |
2520 | my ($sub_stmt, @sub_bind) |
9d48860e |
2521 | = $sql->select("t1", "c1", {c2 => {"<" => 100}, |
96449e8e |
2522 | c3 => {-like => "foo%"}}); |
2523 | my %where = ( |
2524 | foo => 1234, |
2525 | bar => \["> ALL ($sub_stmt)" => @sub_bind], |
2526 | ); |
2527 | |
2528 | In the examples above, the subquery was used as an operator on a column; |
9d48860e |
2529 | but the same principle also applies for a clause within the main C<%where> |
96449e8e |
2530 | hash, like an EXISTS subquery : |
2531 | |
9d48860e |
2532 | my ($sub_stmt, @sub_bind) |
96449e8e |
2533 | = $sql->select("t1", "*", {c1 => 1, c2 => \"> t0.c0"}); |
48d9f5f8 |
2534 | my %where = ( -and => [ |
96449e8e |
2535 | foo => 1234, |
48d9f5f8 |
2536 | \["EXISTS ($sub_stmt)" => @sub_bind], |
2537 | ]); |
96449e8e |
2538 | |
2539 | which yields |
2540 | |
9d48860e |
2541 | $stmt = "WHERE (foo = ? AND EXISTS (SELECT * FROM t1 |
96449e8e |
2542 | WHERE c1 = ? AND c2 > t0.c0))"; |
2543 | @bind = (1234, 1); |
2544 | |
2545 | |
9d48860e |
2546 | Observe that the condition on C<c2> in the subquery refers to |
2547 | column C<t0.c0> of the main query : this is I<not> a bind |
2548 | value, so we have to express it through a scalar ref. |
96449e8e |
2549 | Writing C<< c2 => {">" => "t0.c0"} >> would have generated |
2550 | C<< c2 > ? >> with bind value C<"t0.c0"> ... not exactly |
2551 | what we wanted here. |
2552 | |
96449e8e |
2553 | Finally, here is an example where a subquery is used |
2554 | for expressing unary negation: |
2555 | |
9d48860e |
2556 | my ($sub_stmt, @sub_bind) |
96449e8e |
2557 | = $sql->where({age => [{"<" => 10}, {">" => 20}]}); |
2558 | $sub_stmt =~ s/^ where //i; # don't want "WHERE" in the subclause |
2559 | my %where = ( |
2560 | lname => {like => '%son%'}, |
48d9f5f8 |
2561 | \["NOT ($sub_stmt)" => @sub_bind], |
96449e8e |
2562 | ); |
2563 | |
2564 | This yields |
2565 | |
2566 | $stmt = "lname LIKE ? AND NOT ( age < ? OR age > ? )" |
2567 | @bind = ('%son%', 10, 20) |
2568 | |
cc422895 |
2569 | =head3 Deprecated usage of Literal SQL |
2570 | |
2571 | Below are some examples of archaic use of literal SQL. It is shown only as |
2572 | reference for those who deal with legacy code. Each example has a much |
2573 | better, cleaner and safer alternative that users should opt for in new code. |
2574 | |
2575 | =over |
2576 | |
2577 | =item * |
2578 | |
2579 | my %where = ( requestor => \'IS NOT NULL' ) |
2580 | |
2581 | $stmt = "WHERE requestor IS NOT NULL" |
2582 | |
2583 | This used to be the way of generating NULL comparisons, before the handling |
2584 | of C<undef> got formalized. For new code please use the superior syntax as |
2585 | described in L</Tests for NULL values>. |
96449e8e |
2586 | |
cc422895 |
2587 | =item * |
2588 | |
2589 | my %where = ( requestor => \'= submitter' ) |
2590 | |
2591 | $stmt = "WHERE requestor = submitter" |
2592 | |
2593 | This used to be the only way to compare columns. Use the superior L</-ident> |
2594 | method for all new code. For example an identifier declared in such a way |
2595 | will be properly quoted if L</quote_char> is properly set, while the legacy |
2596 | form will remain as supplied. |
2597 | |
2598 | =item * |
2599 | |
2600 | my %where = ( is_ready => \"", completed => { '>', '2012-12-21' } ) |
2601 | |
2602 | $stmt = "WHERE completed > ? AND is_ready" |
2603 | @bind = ('2012-12-21') |
2604 | |
2605 | Using an empty string literal used to be the only way to express a boolean. |
2606 | For all new code please use the much more readable |
2607 | L<-bool|/Unary operators: bool> operator. |
2608 | |
2609 | =back |
96449e8e |
2610 | |
2611 | =head2 Conclusion |
2612 | |
32eab2da |
2613 | These pages could go on for a while, since the nesting of the data |
2614 | structures this module can handle are pretty much unlimited (the |
2615 | module implements the C<WHERE> expansion as a recursive function |
2616 | internally). Your best bet is to "play around" with the module a |
2617 | little to see how the data structures behave, and choose the best |
2618 | format for your data based on that. |
2619 | |
2620 | And of course, all the values above will probably be replaced with |
2621 | variables gotten from forms or the command line. After all, if you |
2622 | knew everything ahead of time, you wouldn't have to worry about |
2623 | dynamically-generating SQL and could just hardwire it into your |
2624 | script. |
2625 | |
86298391 |
2626 | =head1 ORDER BY CLAUSES |
2627 | |
9d48860e |
2628 | Some functions take an order by clause. This can either be a scalar (just a |
86298391 |
2629 | column name,) a hash of C<< { -desc => 'col' } >> or C<< { -asc => 'col' } >>, |
1cfa1db3 |
2630 | or an array of either of the two previous forms. Examples: |
2631 | |
952f9e2d |
2632 | Given | Will Generate |
1cfa1db3 |
2633 | ---------------------------------------------------------- |
952f9e2d |
2634 | | |
2635 | \'colA DESC' | ORDER BY colA DESC |
2636 | | |
2637 | 'colA' | ORDER BY colA |
2638 | | |
2639 | [qw/colA colB/] | ORDER BY colA, colB |
2640 | | |
2641 | {-asc => 'colA'} | ORDER BY colA ASC |
2642 | | |
2643 | {-desc => 'colB'} | ORDER BY colB DESC |
2644 | | |
2645 | ['colA', {-asc => 'colB'}] | ORDER BY colA, colB ASC |
2646 | | |
855e6047 |
2647 | { -asc => [qw/colA colB/] } | ORDER BY colA ASC, colB ASC |
952f9e2d |
2648 | | |
2649 | [ | |
2650 | { -asc => 'colA' }, | ORDER BY colA ASC, colB DESC, |
2651 | { -desc => [qw/colB/], | colC ASC, colD ASC |
2652 | { -asc => [qw/colC colD/],| |
2653 | ] | |
2654 | =========================================================== |
86298391 |
2655 | |
96449e8e |
2656 | |
2657 | |
2658 | =head1 SPECIAL OPERATORS |
2659 | |
e3f9dff4 |
2660 | my $sqlmaker = SQL::Abstract->new(special_ops => [ |
3a2e1a5e |
2661 | { |
2662 | regex => qr/.../, |
e3f9dff4 |
2663 | handler => sub { |
2664 | my ($self, $field, $op, $arg) = @_; |
2665 | ... |
3a2e1a5e |
2666 | }, |
2667 | }, |
2668 | { |
2669 | regex => qr/.../, |
2670 | handler => 'method_name', |
e3f9dff4 |
2671 | }, |
2672 | ]); |
2673 | |
9d48860e |
2674 | A "special operator" is a SQL syntactic clause that can be |
e3f9dff4 |
2675 | applied to a field, instead of a usual binary operator. |
9d48860e |
2676 | For example : |
e3f9dff4 |
2677 | |
2678 | WHERE field IN (?, ?, ?) |
2679 | WHERE field BETWEEN ? AND ? |
2680 | WHERE MATCH(field) AGAINST (?, ?) |
96449e8e |
2681 | |
e3f9dff4 |
2682 | Special operators IN and BETWEEN are fairly standard and therefore |
3a2e1a5e |
2683 | are builtin within C<SQL::Abstract> (as the overridable methods |
2684 | C<_where_field_IN> and C<_where_field_BETWEEN>). For other operators, |
2685 | like the MATCH .. AGAINST example above which is specific to MySQL, |
2686 | you can write your own operator handlers - supply a C<special_ops> |
2687 | argument to the C<new> method. That argument takes an arrayref of |
2688 | operator definitions; each operator definition is a hashref with two |
2689 | entries: |
96449e8e |
2690 | |
e3f9dff4 |
2691 | =over |
2692 | |
2693 | =item regex |
2694 | |
2695 | the regular expression to match the operator |
96449e8e |
2696 | |
e3f9dff4 |
2697 | =item handler |
2698 | |
3a2e1a5e |
2699 | Either a coderef or a plain scalar method name. In both cases |
2700 | the expected return is C<< ($sql, @bind) >>. |
2701 | |
2702 | When supplied with a method name, it is simply called on the |
2703 | L<SQL::Abstract/> object as: |
2704 | |
2705 | $self->$method_name ($field, $op, $arg) |
2706 | |
2707 | Where: |
2708 | |
2709 | $op is the part that matched the handler regex |
2710 | $field is the LHS of the operator |
2711 | $arg is the RHS |
2712 | |
2713 | When supplied with a coderef, it is called as: |
2714 | |
2715 | $coderef->($self, $field, $op, $arg) |
2716 | |
e3f9dff4 |
2717 | |
2718 | =back |
2719 | |
9d48860e |
2720 | For example, here is an implementation |
e3f9dff4 |
2721 | of the MATCH .. AGAINST syntax for MySQL |
2722 | |
2723 | my $sqlmaker = SQL::Abstract->new(special_ops => [ |
9d48860e |
2724 | |
e3f9dff4 |
2725 | # special op for MySql MATCH (field) AGAINST(word1, word2, ...) |
9d48860e |
2726 | {regex => qr/^match$/i, |
e3f9dff4 |
2727 | handler => sub { |
2728 | my ($self, $field, $op, $arg) = @_; |
2729 | $arg = [$arg] if not ref $arg; |
2730 | my $label = $self->_quote($field); |
2731 | my ($placeholder) = $self->_convert('?'); |
2732 | my $placeholders = join ", ", (($placeholder) x @$arg); |
2733 | my $sql = $self->_sqlcase('match') . " ($label) " |
2734 | . $self->_sqlcase('against') . " ($placeholders) "; |
2735 | my @bind = $self->_bindtype($field, @$arg); |
2736 | return ($sql, @bind); |
2737 | } |
2738 | }, |
9d48860e |
2739 | |
e3f9dff4 |
2740 | ]); |
96449e8e |
2741 | |
2742 | |
59f23b3d |
2743 | =head1 UNARY OPERATORS |
2744 | |
112b5232 |
2745 | my $sqlmaker = SQL::Abstract->new(unary_ops => [ |
59f23b3d |
2746 | { |
2747 | regex => qr/.../, |
2748 | handler => sub { |
2749 | my ($self, $op, $arg) = @_; |
2750 | ... |
2751 | }, |
2752 | }, |
2753 | { |
2754 | regex => qr/.../, |
2755 | handler => 'method_name', |
2756 | }, |
2757 | ]); |
2758 | |
9d48860e |
2759 | A "unary operator" is a SQL syntactic clause that can be |
59f23b3d |
2760 | applied to a field - the operator goes before the field |
2761 | |
2762 | You can write your own operator handlers - supply a C<unary_ops> |
2763 | argument to the C<new> method. That argument takes an arrayref of |
2764 | operator definitions; each operator definition is a hashref with two |
2765 | entries: |
2766 | |
2767 | =over |
2768 | |
2769 | =item regex |
2770 | |
2771 | the regular expression to match the operator |
2772 | |
2773 | =item handler |
2774 | |
2775 | Either a coderef or a plain scalar method name. In both cases |
2776 | the expected return is C<< $sql >>. |
2777 | |
2778 | When supplied with a method name, it is simply called on the |
2779 | L<SQL::Abstract/> object as: |
2780 | |
2781 | $self->$method_name ($op, $arg) |
2782 | |
2783 | Where: |
2784 | |
2785 | $op is the part that matched the handler regex |
2786 | $arg is the RHS or argument of the operator |
2787 | |
2788 | When supplied with a coderef, it is called as: |
2789 | |
2790 | $coderef->($self, $op, $arg) |
2791 | |
2792 | |
2793 | =back |
2794 | |
2795 | |
32eab2da |
2796 | =head1 PERFORMANCE |
2797 | |
2798 | Thanks to some benchmarking by Mark Stosberg, it turns out that |
2799 | this module is many orders of magnitude faster than using C<DBIx::Abstract>. |
2800 | I must admit this wasn't an intentional design issue, but it's a |
2801 | byproduct of the fact that you get to control your C<DBI> handles |
2802 | yourself. |
2803 | |
2804 | To maximize performance, use a code snippet like the following: |
2805 | |
2806 | # prepare a statement handle using the first row |
2807 | # and then reuse it for the rest of the rows |
2808 | my($sth, $stmt); |
2809 | for my $href (@array_of_hashrefs) { |
2810 | $stmt ||= $sql->insert('table', $href); |
2811 | $sth ||= $dbh->prepare($stmt); |
2812 | $sth->execute($sql->values($href)); |
2813 | } |
2814 | |
2815 | The reason this works is because the keys in your C<$href> are sorted |
2816 | internally by B<SQL::Abstract>. Thus, as long as your data retains |
2817 | the same structure, you only have to generate the SQL the first time |
2818 | around. On subsequent queries, simply use the C<values> function provided |
2819 | by this module to return your values in the correct order. |
2820 | |
b864ba9b |
2821 | However this depends on the values having the same type - if, for |
2822 | example, the values of a where clause may either have values |
2823 | (resulting in sql of the form C<column = ?> with a single bind |
2824 | value), or alternatively the values might be C<undef> (resulting in |
2825 | sql of the form C<column IS NULL> with no bind value) then the |
2826 | caching technique suggested will not work. |
96449e8e |
2827 | |
32eab2da |
2828 | =head1 FORMBUILDER |
2829 | |
2830 | If you use my C<CGI::FormBuilder> module at all, you'll hopefully |
2831 | really like this part (I do, at least). Building up a complex query |
2832 | can be as simple as the following: |
2833 | |
2834 | #!/usr/bin/perl |
2835 | |
2836 | use CGI::FormBuilder; |
2837 | use SQL::Abstract; |
2838 | |
2839 | my $form = CGI::FormBuilder->new(...); |
2840 | my $sql = SQL::Abstract->new; |
2841 | |
2842 | if ($form->submitted) { |
2843 | my $field = $form->field; |
2844 | my $id = delete $field->{id}; |
2845 | my($stmt, @bind) = $sql->update('table', $field, {id => $id}); |
2846 | } |
2847 | |
2848 | Of course, you would still have to connect using C<DBI> to run the |
2849 | query, but the point is that if you make your form look like your |
2850 | table, the actual query script can be extremely simplistic. |
2851 | |
2852 | If you're B<REALLY> lazy (I am), check out C<HTML::QuickTable> for |
9d48860e |
2853 | a fast interface to returning and formatting data. I frequently |
32eab2da |
2854 | use these three modules together to write complex database query |
2855 | apps in under 50 lines. |
2856 | |
d8cc1792 |
2857 | =head1 REPO |
2858 | |
2859 | =over |
2860 | |
6d19fbf9 |
2861 | =item * gitweb: L<http://git.shadowcat.co.uk/gitweb/gitweb.cgi?p=dbsrgits/SQL-Abstract.git> |
d8cc1792 |
2862 | |
6d19fbf9 |
2863 | =item * git: L<git://git.shadowcat.co.uk/dbsrgits/SQL-Abstract.git> |
d8cc1792 |
2864 | |
2865 | =back |
32eab2da |
2866 | |
96449e8e |
2867 | =head1 CHANGES |
2868 | |
2869 | Version 1.50 was a major internal refactoring of C<SQL::Abstract>. |
2870 | Great care has been taken to preserve the I<published> behavior |
2871 | documented in previous versions in the 1.* family; however, |
9d48860e |
2872 | some features that were previously undocumented, or behaved |
96449e8e |
2873 | differently from the documentation, had to be changed in order |
2874 | to clarify the semantics. Hence, client code that was relying |
9d48860e |
2875 | on some dark areas of C<SQL::Abstract> v1.* |
96449e8e |
2876 | B<might behave differently> in v1.50. |
32eab2da |
2877 | |
d2a8fe1a |
2878 | The main changes are : |
2879 | |
96449e8e |
2880 | =over |
32eab2da |
2881 | |
9d48860e |
2882 | =item * |
32eab2da |
2883 | |
96449e8e |
2884 | support for literal SQL through the C<< \ [$sql, bind] >> syntax. |
2885 | |
2886 | =item * |
2887 | |
145fbfc8 |
2888 | support for the { operator => \"..." } construct (to embed literal SQL) |
2889 | |
2890 | =item * |
2891 | |
9c37b9c0 |
2892 | support for the { operator => \["...", @bind] } construct (to embed literal SQL with bind values) |
2893 | |
2894 | =item * |
2895 | |
96449e8e |
2896 | optional support for L<array datatypes|/"Inserting and Updating Arrays"> |
2897 | |
9d48860e |
2898 | =item * |
96449e8e |
2899 | |
2900 | defensive programming : check arguments |
2901 | |
2902 | =item * |
2903 | |
2904 | fixed bug with global logic, which was previously implemented |
7cac25e6 |
2905 | through global variables yielding side-effects. Prior versions would |
96449e8e |
2906 | interpret C<< [ {cond1, cond2}, [cond3, cond4] ] >> |
2907 | as C<< "(cond1 AND cond2) OR (cond3 AND cond4)" >>. |
2908 | Now this is interpreted |
2909 | as C<< "(cond1 AND cond2) OR (cond3 OR cond4)" >>. |
2910 | |
96449e8e |
2911 | |
2912 | =item * |
2913 | |
2914 | fixed semantics of _bindtype on array args |
2915 | |
9d48860e |
2916 | =item * |
96449e8e |
2917 | |
2918 | dropped the C<_anoncopy> of the %where tree. No longer necessary, |
2919 | we just avoid shifting arrays within that tree. |
2920 | |
2921 | =item * |
2922 | |
2923 | dropped the C<_modlogic> function |
2924 | |
2925 | =back |
32eab2da |
2926 | |
32eab2da |
2927 | =head1 ACKNOWLEDGEMENTS |
2928 | |
2929 | There are a number of individuals that have really helped out with |
2930 | this module. Unfortunately, most of them submitted bugs via CPAN |
2931 | so I have no idea who they are! But the people I do know are: |
2932 | |
9d48860e |
2933 | Ash Berlin (order_by hash term support) |
b643abe1 |
2934 | Matt Trout (DBIx::Class support) |
32eab2da |
2935 | Mark Stosberg (benchmarking) |
2936 | Chas Owens (initial "IN" operator support) |
2937 | Philip Collins (per-field SQL functions) |
2938 | Eric Kolve (hashref "AND" support) |
2939 | Mike Fragassi (enhancements to "BETWEEN" and "LIKE") |
2940 | Dan Kubb (support for "quote_char" and "name_sep") |
f5aab26e |
2941 | Guillermo Roditi (patch to cleanup "IN" and "BETWEEN", fix and tests for _order_by) |
48d9f5f8 |
2942 | Laurent Dami (internal refactoring, extensible list of special operators, literal SQL) |
dbdf7648 |
2943 | Norbert Buchmuller (support for literal SQL in hashpair, misc. fixes & tests) |
e96c510a |
2944 | Peter Rabbitson (rewrite of SQLA::Test, misc. fixes & tests) |
02288357 |
2945 | Oliver Charles (support for "RETURNING" after "INSERT") |
32eab2da |
2946 | |
2947 | Thanks! |
2948 | |
32eab2da |
2949 | =head1 SEE ALSO |
2950 | |
86298391 |
2951 | L<DBIx::Class>, L<DBIx::Abstract>, L<CGI::FormBuilder>, L<HTML::QuickTable>. |
32eab2da |
2952 | |
32eab2da |
2953 | =head1 AUTHOR |
2954 | |
b643abe1 |
2955 | Copyright (c) 2001-2007 Nathan Wiger <nwiger@cpan.org>. All Rights Reserved. |
2956 | |
2957 | This module is actively maintained by Matt Trout <mst@shadowcatsystems.co.uk> |
32eab2da |
2958 | |
abe72f94 |
2959 | For support, your best bet is to try the C<DBIx::Class> users mailing list. |
2960 | While not an official support venue, C<DBIx::Class> makes heavy use of |
2961 | C<SQL::Abstract>, and as such list members there are very familiar with |
2962 | how to create queries. |
2963 | |
0d067ded |
2964 | =head1 LICENSE |
2965 | |
d988ab87 |
2966 | This module is free software; you may copy this under the same |
2967 | terms as perl itself (either the GNU General Public License or |
2968 | the Artistic License) |
32eab2da |
2969 | |
2970 | =cut |
2971 | |