1SQL::Abstract::Plugin::UEsxetrraCColnatursiebsu(t3e)d PeSrQlL:D:oAcbusmternatcatt:i:oPnlugin::ExtraClauses(3)
2
3
4
6 SQL::Abstract::ExtraClauses - new/experimental additions to
7 SQL::Abstract
8
10 my $sqla = SQL::Abstract->new;
11 SQL::Abstract::ExtraClauses->apply_to($sqla);
12
14 This module is basically a nursery for things that seem like a good
15 idea to live in until we figure out if we were right about that.
16
18 apply_to
19 Applies the plugin to an SQL::Abstract object.
20
21 register_extensions
22 Registers the extensions described below
23
24 cb
25 For plugin authors, creates a callback to call a method on the plugin.
26
27 register
28 For plugin authors, registers callbacks more easily.
29
30 sqla
31 Available only during plugin callback executions, contains the
32 currently active SQL::Abstract object.
33
35 alias
36 Represents a table alias. Expands name and column names with ident as
37 default.
38
39 # expr
40 { -alias => [ 't', 'x', 'y', 'z' ] }
41
42 # aqt
43 { -alias => [
44 { -ident => [ 't' ] }, { -ident => [ 'x' ] },
45 { -ident => [ 'y' ] }, { -ident => [ 'z' ] },
46 ] }
47
48 # query
49 t(x, y, z)
50 []
51
52 as
53 Represents an sql AS. LHS is expanded with ident as default, RHS is
54 treated as a list of arguments for the alias node.
55
56 # expr
57 { foo => { -as => 'bar' } }
58
59 # aqt
60 { -as => [ { -ident => [ 'foo' ] }, { -ident => [ 'bar' ] } ] }
61
62 # query
63 foo AS bar
64 []
65
66 # expr
67 { -as => [ { -select => { _ => 'blah' } }, 't', 'blah' ] }
68
69 # aqt
70 { -as => [
71 { -select =>
72 { select => { -op => [ ',', { -ident => [ 'blah' ] } ] } }
73 },
74 { -alias => [ { -ident => [ 't' ] }, { -ident => [ 'blah' ] } ] },
75 ] }
76
77 # query
78 (SELECT blah) AS t(blah)
79 []
80
81 cast
82 # expr
83 { -cast => [ { -ident => 'birthday' }, 'date' ] }
84
85 # aqt
86 { -func => [
87 'cast', {
88 -as => [ { -ident => [ 'birthday' ] }, { -ident => [ 'date' ] } ]
89 },
90 ] }
91
92 # query
93 CAST(birthday AS date)
94 []
95
96 join
97 If given an arrayref, pretends it was given a hashref with the first
98 element of the arrayref as the value for 'to' and the remaining pairs
99 copied.
100
101 Given a hashref, the 'as' key is if presented expanded to wrap the
102 'to'.
103
104 If present the 'using' key is expanded as a list of idents.
105
106 Known keys are: 'from' (the left hand side), 'type' ('left', 'right',
107 or nothing), 'to' (the right hand side), 'on' and 'using'.
108
109 # expr
110 { -join => {
111 from => 'lft',
112 on => { 'lft.bloo' => { '>' => 'rgt.blee' } },
113 to => 'rgt',
114 type => 'left',
115 } }
116
117 # aqt
118 { -join => {
119 from => { -ident => [ 'lft' ] },
120 on => { -op => [
121 '>', { -ident => [ 'lft', 'bloo' ] },
122 { -ident => [ 'rgt', 'blee' ] },
123 ] },
124 to => { -ident => [ 'rgt' ] },
125 type => 'left',
126 } }
127
128 # query
129 lft LEFT JOIN rgt ON lft.bloo > rgt.blee
130 []
131
132 from_list
133 List of components of the FROM clause; -foo type elements indicate a
134 pair with the next element; this is easiest if I show you:
135
136 # expr
137 { -from_list => [
138 't1', -as => 'table_one', -join =>
139 [ 't2', 'on', { 'table_one.x' => 't2.x' } ],
140 ] }
141
142 # aqt
143 { -join => {
144 from =>
145 {
146 -as => [ { -ident => [ 't1' ] }, { -ident => [ 'table_one' ] } ]
147 },
148 on => { -op => [
149 '=', { -ident => [ 'table_one', 'x' ] },
150 { -ident => [ 't2', 'x' ] },
151 ] },
152 to => { -ident => [ 't2' ] },
153 type => undef,
154 } }
155
156 # query
157 t1 AS table_one JOIN t2 ON table_one.x = t2.x
158 []
159
160 Or with using:
161
162 # expr
163 { -from_list =>
164 [ 't1', -as => 'table_one', -join => [ 't2', 'using', [ 'x' ] ] ]
165 }
166
167 # aqt
168 { -join => {
169 from =>
170 {
171 -as => [ { -ident => [ 't1' ] }, { -ident => [ 'table_one' ] } ]
172 },
173 to => { -ident => [ 't2' ] },
174 type => undef,
175 using =>
176 { -op => [ 'or', { -op => [ 'or', { -ident => [ 'x' ] } ] } ] },
177 } }
178
179 # query
180 t1 AS table_one JOIN t2 USING ( x )
181 []
182
183 With oddities:
184
185 # expr
186 { -from_list => [
187 'x', -join =>
188 [ [ 'y', -join => [ 'z', 'type', 'left' ] ], 'type', 'left' ],
189 ] }
190
191 # aqt
192 { -join => {
193 from => { -ident => [ 'x' ] },
194 to => { -join => {
195 from => { -ident => [ 'y' ] },
196 to => { -ident => [ 'z' ] },
197 type => 'left',
198 } },
199 type => 'left',
200 } }
201
202 # query
203 x LEFT JOIN ( y LEFT JOIN z )
204 []
205
206 setops
207 Expanders are provided for union, union_all, intersect, intersect_all,
208 except and except_all, and each takes an arrayref of queries:
209
210 # expr
211 { -union => [
212 { -select => { _ => { -value => 1 } } },
213 { -select => { _ => { -value => 2 } } },
214 ] }
215
216 # aqt
217 { -union => { queries => [
218 { -select =>
219 { select => { -op => [ ',', { -bind => [ undef, 1 ] } ] } }
220 },
221 { -select =>
222 { select => { -op => [ ',', { -bind => [ undef, 2 ] } ] } }
223 },
224 ] } }
225
226 # query
227 (SELECT ?) UNION (SELECT ?)
228 [ 1, 2 ]
229
230 # expr
231 { -union_all => [
232 { -select => { _ => { -value => 1 } } },
233 { -select => { _ => { -value => 2 } } },
234 { -select => { _ => { -value => 1 } } },
235 ] }
236
237 # aqt
238 { -union => {
239 queries => [
240 { -select =>
241 { select => { -op => [ ',', { -bind => [ undef, 1 ] } ] } }
242 },
243 { -select =>
244 { select => { -op => [ ',', { -bind => [ undef, 2 ] } ] } }
245 },
246 { -select =>
247 { select => { -op => [ ',', { -bind => [ undef, 1 ] } ] } }
248 },
249 ],
250 type => 'all',
251 } }
252
253 # query
254 (SELECT ?) UNION ALL (SELECT ?) UNION ALL (SELECT ?)
255 [ 1, 2, 1 ]
256
258 group by clause for select
259 Expanded as a list with an ident default:
260
261 # expr
262 { -select => { group_by => [ 'foo', 'bar' ] } }
263
264 # aqt
265 { -select => { group_by =>
266 {
267 -op => [ ',', { -ident => [ 'foo' ] }, { -ident => [ 'bar' ] } ]
268 }
269 } }
270
271 # query
272 GROUP BY foo, bar
273 []
274
275 having clause for select
276 Basic expr, just like where, given having is pretty much post-group-by
277 where clause:
278
279 # expr
280 { -select =>
281 { having => { '>' => [ { -count => { -ident => 'foo' } }, 3 ] } }
282 }
283
284 # aqt
285 { -select => { having => { -op => [
286 '>', { -func => [ 'count', { -ident => [ 'foo' ] } ] },
287 { -bind => [ undef, 3 ] },
288 ] } } }
289
290 # query
291 HAVING COUNT(foo) > ?
292 [ 3 ]
293
294 setop clauses
295 If a select query contains a clause matching any of the setop node
296 types, clauses that appear before the setop would in the resulting
297 query are gathered together and moved into an inner select node:
298
299 # expr
300 { -select => {
301 _ => '*',
302 from => 'foo',
303 order_by => 'baz',
304 union =>
305 {
306 -select => { _ => '*', from => 'bar', where => { thing => 1 } }
307 },
308 where => { thing => 1 },
309 } }
310
311 # aqt
312 { -select => {
313 order_by => { -op => [ ',', { -ident => [ 'baz' ] } ] },
314 setop => { -union => { queries => [
315 { -select => {
316 from => { -ident => [ 'foo' ] },
317 select => { -op => [ ',', { -ident => [ '*' ] } ] },
318 where => { -op => [
319 '=', { -ident => [ 'thing' ] },
320 { -bind => [ 'thing', 1 ] },
321 ] },
322 } }, ] },
323 { -select => {
324 from => { -ident => [ 'bar' ] },
325 select => { -op => [ ',', { -ident => [ '*' ] } ] },
326 where => { -op => [
327 '=', { -ident => [ 'thing' ] },
328 { -bind => [ 'thing', 1 ] },
329 } },
330 ] } },
331 } }
332
333 # query
334 (SELECT * FROM foo WHERE thing = ?) UNION (
335 SELECT * FROM bar WHERE thing = ?
336 )
337 ORDER BY baz
338 [ 1, 1 ]
339
340 update from clause
341 Some databases allow an additional FROM clause to reference other
342 tables for the data to update; this clause is expanded as a normal from
343 list, check your database for what is and isn't allowed in practice.
344
345 # expr
346 { -update => {
347 _ => 'employees',
348 from => 'accounts',
349 set => { sales_count => { sales_count => { '+' => \1 } } },
350 where => {
351 'accounts.name' => { '=' => \"'Acme Corporation'" },
352 'employees.id' => { -ident => 'accounts.sales_person' },
353 },
354 } }
355
356 # aqt
357 { -update => {
358 from => { -ident => [ 'accounts' ] },
359 set => { -op => [
360 ',', { -op => [
361 '=', { -ident => [ 'sales_count' ] }, { -op => [
362 '+', { -ident => [ 'sales_count' ] },
363 { -literal => [ 1 ] },
364 ] },
365 ] },
366 ] },
367 target => { -ident => [ 'employees' ] },
368 where => { -op => [
369 'and', { -op => [
370 '=', { -ident => [ 'accounts', 'name' ] },
371 { -literal => [ "'Acme Corporation'" ] },
372 ] }, { -op => [
373 '=', { -ident => [ 'employees', 'id' ] },
374 { -ident => [ 'accounts', 'sales_person' ] },
375 ] },
376 ] },
377 } }
378
379 # query
380 UPDATE employees SET sales_count = sales_count + 1 FROM accounts
381 WHERE (
382 accounts.name = 'Acme Corporation'
383 AND employees.id = accounts.sales_person
384 )
385 []
386
387 delete using clause
388 Some databases allow an additional USING clause to reference other
389 tables for the data to update; this clause is expanded as a normal from
390 list, check your database for what is and isn't allowed in practice.
391
392 # expr
393 { -delete => {
394 from => 'x',
395 using => 'y',
396 where => { 'x.id' => { -ident => 'y.x_id' } },
397 } }
398
399 # aqt
400 { -delete => {
401 target => { -op => [ ',', { -ident => [ 'x' ] } ] },
402 using => { -ident => [ 'y' ] },
403 where => { -op => [
404 '=', { -ident => [ 'x', 'id' ] },
405 { -ident => [ 'y', 'x_id' ] },
406 ] },
407 } }
408
409 # query
410 DELETE FROM x USING y WHERE x.id = y.x_id
411 []
412
413 insert rowvalues and select clauses
414 rowvalues and select are shorthand for
415
416 { from => { -select ... } }
417
418 and
419
420 { from => { -values ... } }
421
422 respectively:
423
424 # expr
425 { -insert =>
426 { into => 'numbers', rowvalues => [ [ 1, 2 ], [ 3, 4 ], [ 5, 6 ] ] }
427 }
428
429 # aqt
430 { -insert => {
431 from => { -values => [
432 { -row =>
433 [ { -bind => [ undef, 1 ] }, { -bind => [ undef, 2 ] } ]
434 },
435 { -row =>
436 [ { -bind => [ undef, 3 ] }, { -bind => [ undef, 4 ] } ]
437 },
438 { -row =>
439 [ { -bind => [ undef, 5 ] }, { -bind => [ undef, 6 ] } ]
440 },
441 ] },
442 target => { -ident => [ 'numbers' ] },
443 } }
444
445 # query
446 INSERT INTO numbers VALUES (?, ?), (?, ?), (?, ?)
447 [ 1, 2, 3, 4, 5, 6 ]
448
449 # expr
450 { -insert =>
451 { into => 'numbers', select => { _ => '*', from => 'old_numbers' } }
452 }
453
454 # aqt
455 { -insert => {
456 from => { -select => {
457 from => { -ident => [ 'old_numbers' ] },
458 select => { -op => [ ',', { -ident => [ '*' ] } ] },
459 } },
460 target => { -ident => [ 'numbers' ] },
461 } }
462
463 # query
464 INSERT INTO numbers SELECT * FROM old_numbers
465 []
466
467 with and with_recursive clauses
468 These clauses are available on select/insert/update/delete queries;
469 check your database for applicability (e.g. mysql supports all four but
470 mariadb only select).
471
472 The value should be an arrayref of name/query pairs:
473
474 # expr
475 { -select => {
476 from => 'foo',
477 select => '*',
478 with => [ 'foo', { -select => { select => \1 } } ],
479 } }
480
481 # aqt
482 { -select => {
483 from => { -ident => [ 'foo' ] },
484 select => { -op => [ ',', { -ident => [ '*' ] } ] },
485 with => { queries => [ [
486 { -ident => [ 'foo' ] }, { -select =>
487 { select => { -op => [ ',', { -literal => [ 1 ] } ] } }
488 },
489 ] ] },
490 } }
491
492 # query
493 WITH foo AS (SELECT 1) SELECT * FROM foo
494 []
495
496 A more complete example (designed for mariadb, (ab)using the fact that
497 mysqloids materialise subselects in FROM into an unindexed temp table
498 to circumvent the restriction that you can't select from the table
499 you're currently updating:
500
501 # expr
502 { -update => {
503 _ => [
504 'tree_table', -join => {
505 as => 'tree',
506 on => { 'tree.id' => 'tree_with_path.id' },
507 to => { -select => {
508 from => 'tree_with_path',
509 select => '*',
510 with_recursive => [
511 [ 'tree_with_path', 'id', 'parent_id', 'path' ],
512 { -select => {
513 _ => [
514 'id', 'parent_id', { -as => [
515 { -cast => { -as => [ 'id', 'char', 255 ] } },
516 'path',
517 ] } ],
518 from => 'tree_table',
519 union_all => { -select => {
520 _ => [
521 't.id', 't.parent_id', { -as => [
522 { -concat => [ 'r.path', \"'/'", 't.id' ] },
523 'path',
524 ] },
525 ],
526 from => [
527 'tree_table', -as => 't', -join => {
528 as => 'r',
529 on => { 't.parent_id' => 'r.id' },
530 to => 'tree_with_path',
531 },
532 ],
533 } },
534 where => { parent_id => undef },
535 } },
536 ],
537 } },
538 },
539 ],
540 set => { path => { -ident => [ 'tree', 'path' ] } },
541 } }
542
543 # query
544 UPDATE
545 tree_table JOIN
546 (
547 WITH RECURSIVE
548 tree_with_path(id, parent_id, path) AS (
549 (
550 SELECT id, parent_id, CAST(id AS char(255)) AS path
551 FROM tree_table WHERE parent_id IS NULL
552 ) UNION ALL (
553 SELECT t.id, t.parent_id, CONCAT(r.path, '/', t.id) AS path
554 FROM
555 tree_table AS t JOIN tree_with_path AS r ON
556 t.parent_id = r.id
557 )
558 )
559 SELECT * FROM tree_with_path
560 ) AS tree
561 ON tree.id = tree_with_path.id
562 SET path = tree.path
563 []
564
565
566
567perl v5.34.0 2021-07S-Q2L2::Abstract::Plugin::ExtraClauses(3)