1SELECT(7)                PostgreSQL 14.3 Documentation               SELECT(7)
2
3
4

NAME

6       SELECT, TABLE, WITH - retrieve rows from a table or view
7

SYNOPSIS

9       [ WITH [ RECURSIVE ] with_query [, ...] ]
10       SELECT [ ALL | DISTINCT [ ON ( expression [, ...] ) ] ]
11           [ * | expression [ [ AS ] output_name ] [, ...] ]
12           [ FROM from_item [, ...] ]
13           [ WHERE condition ]
14           [ GROUP BY [ ALL | DISTINCT ] grouping_element [, ...] ]
15           [ HAVING condition ]
16           [ WINDOW window_name AS ( window_definition ) [, ...] ]
17           [ { UNION | INTERSECT | EXCEPT } [ ALL | DISTINCT ] select ]
18           [ ORDER BY expression [ ASC | DESC | USING operator ] [ NULLS { FIRST | LAST } ] [, ...] ]
19           [ LIMIT { count | ALL } ]
20           [ OFFSET start [ ROW | ROWS ] ]
21           [ FETCH { FIRST | NEXT } [ count ] { ROW | ROWS } { ONLY | WITH TIES } ]
22           [ FOR { UPDATE | NO KEY UPDATE | SHARE | KEY SHARE } [ OF table_name [, ...] ] [ NOWAIT | SKIP LOCKED ] [...] ]
23
24       where from_item can be one of:
25
26           [ ONLY ] table_name [ * ] [ [ AS ] alias [ ( column_alias [, ...] ) ] ]
27                       [ TABLESAMPLE sampling_method ( argument [, ...] ) [ REPEATABLE ( seed ) ] ]
28           [ LATERAL ] ( select ) [ AS ] alias [ ( column_alias [, ...] ) ]
29           with_query_name [ [ AS ] alias [ ( column_alias [, ...] ) ] ]
30           [ LATERAL ] function_name ( [ argument [, ...] ] )
31                       [ WITH ORDINALITY ] [ [ AS ] alias [ ( column_alias [, ...] ) ] ]
32           [ LATERAL ] function_name ( [ argument [, ...] ] ) [ AS ] alias ( column_definition [, ...] )
33           [ LATERAL ] function_name ( [ argument [, ...] ] ) AS ( column_definition [, ...] )
34           [ LATERAL ] ROWS FROM( function_name ( [ argument [, ...] ] ) [ AS ( column_definition [, ...] ) ] [, ...] )
35                       [ WITH ORDINALITY ] [ [ AS ] alias [ ( column_alias [, ...] ) ] ]
36           from_item [ NATURAL ] join_type from_item [ ON join_condition | USING ( join_column [, ...] ) [ AS join_using_alias ] ]
37
38       and grouping_element can be one of:
39
40           ( )
41           expression
42           ( expression [, ...] )
43           ROLLUP ( { expression | ( expression [, ...] ) } [, ...] )
44           CUBE ( { expression | ( expression [, ...] ) } [, ...] )
45           GROUPING SETS ( grouping_element [, ...] )
46
47       and with_query is:
48
49           with_query_name [ ( column_name [, ...] ) ] AS [ [ NOT ] MATERIALIZED ] ( select | values | insert | update | delete )
50               [ SEARCH { BREADTH | DEPTH } FIRST BY column_name [, ...] SET search_seq_col_name ]
51               [ CYCLE column_name [, ...] SET cycle_mark_col_name [ TO cycle_mark_value DEFAULT cycle_mark_default ] USING cycle_path_col_name ]
52
53       TABLE [ ONLY ] table_name [ * ]
54

DESCRIPTION

56       SELECT retrieves rows from zero or more tables. The general processing
57       of SELECT is as follows:
58
59        1. All queries in the WITH list are computed. These effectively serve
60           as temporary tables that can be referenced in the FROM list. A WITH
61           query that is referenced more than once in FROM is computed only
62           once, unless specified otherwise with NOT MATERIALIZED. (See WITH
63           Clause below.)
64
65        2. All elements in the FROM list are computed. (Each element in the
66           FROM list is a real or virtual table.) If more than one element is
67           specified in the FROM list, they are cross-joined together. (See
68           FROM Clause below.)
69
70        3. If the WHERE clause is specified, all rows that do not satisfy the
71           condition are eliminated from the output. (See WHERE Clause below.)
72
73        4. If the GROUP BY clause is specified, or if there are aggregate
74           function calls, the output is combined into groups of rows that
75           match on one or more values, and the results of aggregate functions
76           are computed. If the HAVING clause is present, it eliminates groups
77           that do not satisfy the given condition. (See GROUP BY Clause and
78           HAVING Clause below.)
79
80        5. The actual output rows are computed using the SELECT output
81           expressions for each selected row or row group. (See SELECT List
82           below.)
83
84        6. SELECT DISTINCT eliminates duplicate rows from the result.  SELECT
85           DISTINCT ON eliminates rows that match on all the specified
86           expressions.  SELECT ALL (the default) will return all candidate
87           rows, including duplicates. (See DISTINCT Clause below.)
88
89        7. Using the operators UNION, INTERSECT, and EXCEPT, the output of
90           more than one SELECT statement can be combined to form a single
91           result set. The UNION operator returns all rows that are in one or
92           both of the result sets. The INTERSECT operator returns all rows
93           that are strictly in both result sets. The EXCEPT operator returns
94           the rows that are in the first result set but not in the second. In
95           all three cases, duplicate rows are eliminated unless ALL is
96           specified. The noise word DISTINCT can be added to explicitly
97           specify eliminating duplicate rows. Notice that DISTINCT is the
98           default behavior here, even though ALL is the default for SELECT
99           itself. (See UNION Clause, INTERSECT Clause, and EXCEPT Clause
100           below.)
101
102        8. If the ORDER BY clause is specified, the returned rows are sorted
103           in the specified order. If ORDER BY is not given, the rows are
104           returned in whatever order the system finds fastest to produce.
105           (See ORDER BY Clause below.)
106
107        9. If the LIMIT (or FETCH FIRST) or OFFSET clause is specified, the
108           SELECT statement only returns a subset of the result rows. (See
109           LIMIT Clause below.)
110
111       10. If FOR UPDATE, FOR NO KEY UPDATE, FOR SHARE or FOR KEY SHARE is
112           specified, the SELECT statement locks the selected rows against
113           concurrent updates. (See The Locking Clause below.)
114
115       You must have SELECT privilege on each column used in a SELECT command.
116       The use of FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE or FOR KEY SHARE
117       requires UPDATE privilege as well (for at least one column of each
118       table so selected).
119

PARAMETERS

121   WITH Clause
122       The WITH clause allows you to specify one or more subqueries that can
123       be referenced by name in the primary query. The subqueries effectively
124       act as temporary tables or views for the duration of the primary query.
125       Each subquery can be a SELECT, TABLE, VALUES, INSERT, UPDATE or DELETE
126       statement. When writing a data-modifying statement (INSERT, UPDATE or
127       DELETE) in WITH, it is usual to include a RETURNING clause. It is the
128       output of RETURNING, not the underlying table that the statement
129       modifies, that forms the temporary table that is read by the primary
130       query. If RETURNING is omitted, the statement is still executed, but it
131       produces no output so it cannot be referenced as a table by the primary
132       query.
133
134       A name (without schema qualification) must be specified for each WITH
135       query. Optionally, a list of column names can be specified; if this is
136       omitted, the column names are inferred from the subquery.
137
138       If RECURSIVE is specified, it allows a SELECT subquery to reference
139       itself by name. Such a subquery must have the form
140
141           non_recursive_term UNION [ ALL | DISTINCT ] recursive_term
142
143       where the recursive self-reference must appear on the right-hand side
144       of the UNION. Only one recursive self-reference is permitted per query.
145       Recursive data-modifying statements are not supported, but you can use
146       the results of a recursive SELECT query in a data-modifying statement.
147       See Section 7.8 for an example.
148
149       Another effect of RECURSIVE is that WITH queries need not be ordered: a
150       query can reference another one that is later in the list. (However,
151       circular references, or mutual recursion, are not implemented.) Without
152       RECURSIVE, WITH queries can only reference sibling WITH queries that
153       are earlier in the WITH list.
154
155       When there are multiple queries in the WITH clause, RECURSIVE should be
156       written only once, immediately after WITH. It applies to all queries in
157       the WITH clause, though it has no effect on queries that do not use
158       recursion or forward references.
159
160       The optional SEARCH clause computes a search sequence column that can
161       be used for ordering the results of a recursive query in either
162       breadth-first or depth-first order. The supplied column name list
163       specifies the row key that is to be used for keeping track of visited
164       rows. A column named search_seq_col_name will be added to the result
165       column list of the WITH query. This column can be ordered by in the
166       outer query to achieve the respective ordering. See Section 7.8.2.1 for
167       examples.
168
169       The optional CYCLE clause is used to detect cycles in recursive
170       queries. The supplied column name list specifies the row key that is to
171       be used for keeping track of visited rows. A column named
172       cycle_mark_col_name will be added to the result column list of the WITH
173       query. This column will be set to cycle_mark_value when a cycle has
174       been detected, else to cycle_mark_default. Furthermore, processing of
175       the recursive union will stop when a cycle has been detected.
176       cycle_mark_value and cycle_mark_default must be constants and they must
177       be coercible to a common data type, and the data type must have an
178       inequality operator. (The SQL standard requires that they be Boolean
179       constants or character strings, but PostgreSQL does not require that.)
180       By default, TRUE and FALSE (of type boolean) are used. Furthermore, a
181       column named cycle_path_col_name will be added to the result column
182       list of the WITH query. This column is used internally for tracking
183       visited rows. See Section 7.8.2.2 for examples.
184
185       Both the SEARCH and the CYCLE clause are only valid for recursive WITH
186       queries. The with_query must be a UNION (or UNION ALL) of two SELECT
187       (or equivalent) commands (no nested UNIONs). If both clauses are used,
188       the column added by the SEARCH clause appears before the columns added
189       by the CYCLE clause.
190
191       The primary query and the WITH queries are all (notionally) executed at
192       the same time. This implies that the effects of a data-modifying
193       statement in WITH cannot be seen from other parts of the query, other
194       than by reading its RETURNING output. If two such data-modifying
195       statements attempt to modify the same row, the results are unspecified.
196
197       A key property of WITH queries is that they are normally evaluated only
198       once per execution of the primary query, even if the primary query
199       refers to them more than once. In particular, data-modifying statements
200       are guaranteed to be executed once and only once, regardless of whether
201       the primary query reads all or any of their output.
202
203       However, a WITH query can be marked NOT MATERIALIZED to remove this
204       guarantee. In that case, the WITH query can be folded into the primary
205       query much as though it were a simple sub-SELECT in the primary query's
206       FROM clause. This results in duplicate computations if the primary
207       query refers to that WITH query more than once; but if each such use
208       requires only a few rows of the WITH query's total output, NOT
209       MATERIALIZED can provide a net savings by allowing the queries to be
210       optimized jointly.  NOT MATERIALIZED is ignored if it is attached to a
211       WITH query that is recursive or is not side-effect-free (i.e., is not a
212       plain SELECT containing no volatile functions).
213
214       By default, a side-effect-free WITH query is folded into the primary
215       query if it is used exactly once in the primary query's FROM clause.
216       This allows joint optimization of the two query levels in situations
217       where that should be semantically invisible. However, such folding can
218       be prevented by marking the WITH query as MATERIALIZED. That might be
219       useful, for example, if the WITH query is being used as an optimization
220       fence to prevent the planner from choosing a bad plan.  PostgreSQL
221       versions before v12 never did such folding, so queries written for
222       older versions might rely on WITH to act as an optimization fence.
223
224       See Section 7.8 for additional information.
225
226   FROM Clause
227       The FROM clause specifies one or more source tables for the SELECT. If
228       multiple sources are specified, the result is the Cartesian product
229       (cross join) of all the sources. But usually qualification conditions
230       are added (via WHERE) to restrict the returned rows to a small subset
231       of the Cartesian product.
232
233       The FROM clause can contain the following elements:
234
235       table_name
236           The name (optionally schema-qualified) of an existing table or
237           view. If ONLY is specified before the table name, only that table
238           is scanned. If ONLY is not specified, the table and all its
239           descendant tables (if any) are scanned. Optionally, * can be
240           specified after the table name to explicitly indicate that
241           descendant tables are included.
242
243       alias
244           A substitute name for the FROM item containing the alias. An alias
245           is used for brevity or to eliminate ambiguity for self-joins (where
246           the same table is scanned multiple times). When an alias is
247           provided, it completely hides the actual name of the table or
248           function; for example given FROM foo AS f, the remainder of the
249           SELECT must refer to this FROM item as f not foo. If an alias is
250           written, a column alias list can also be written to provide
251           substitute names for one or more columns of the table.
252
253       TABLESAMPLE sampling_method ( argument [, ...] ) [ REPEATABLE ( seed )
254       ]
255           A TABLESAMPLE clause after a table_name indicates that the
256           specified sampling_method should be used to retrieve a subset of
257           the rows in that table. This sampling precedes the application of
258           any other filters such as WHERE clauses. The standard PostgreSQL
259           distribution includes two sampling methods, BERNOULLI and SYSTEM,
260           and other sampling methods can be installed in the database via
261           extensions.
262
263           The BERNOULLI and SYSTEM sampling methods each accept a single
264           argument which is the fraction of the table to sample, expressed as
265           a percentage between 0 and 100. This argument can be any
266           real-valued expression. (Other sampling methods might accept more
267           or different arguments.) These two methods each return a
268           randomly-chosen sample of the table that will contain approximately
269           the specified percentage of the table's rows. The BERNOULLI method
270           scans the whole table and selects or ignores individual rows
271           independently with the specified probability. The SYSTEM method
272           does block-level sampling with each block having the specified
273           chance of being selected; all rows in each selected block are
274           returned. The SYSTEM method is significantly faster than the
275           BERNOULLI method when small sampling percentages are specified, but
276           it may return a less-random sample of the table as a result of
277           clustering effects.
278
279           The optional REPEATABLE clause specifies a seed number or
280           expression to use for generating random numbers within the sampling
281           method. The seed value can be any non-null floating-point value.
282           Two queries that specify the same seed and argument values will
283           select the same sample of the table, if the table has not been
284           changed meanwhile. But different seed values will usually produce
285           different samples. If REPEATABLE is not given then a new random
286           sample is selected for each query, based upon a system-generated
287           seed. Note that some add-on sampling methods do not accept
288           REPEATABLE, and will always produce new samples on each use.
289
290       select
291           A sub-SELECT can appear in the FROM clause. This acts as though its
292           output were created as a temporary table for the duration of this
293           single SELECT command. Note that the sub-SELECT must be surrounded
294           by parentheses, and an alias must be provided for it. A VALUES
295           command can also be used here.
296
297       with_query_name
298           A WITH query is referenced by writing its name, just as though the
299           query's name were a table name. (In fact, the WITH query hides any
300           real table of the same name for the purposes of the primary query.
301           If necessary, you can refer to a real table of the same name by
302           schema-qualifying the table's name.) An alias can be provided in
303           the same way as for a table.
304
305       function_name
306           Function calls can appear in the FROM clause. (This is especially
307           useful for functions that return result sets, but any function can
308           be used.) This acts as though the function's output were created as
309           a temporary table for the duration of this single SELECT command.
310           If the function's result type is composite (including the case of a
311           function with multiple OUT parameters), each attribute becomes a
312           separate column in the implicit table.
313
314           When the optional WITH ORDINALITY clause is added to the function
315           call, an additional column of type bigint will be appended to the
316           function's result column(s). This column numbers the rows of the
317           function's result set, starting from 1. By default, this column is
318           named ordinality.
319
320           An alias can be provided in the same way as for a table. If an
321           alias is written, a column alias list can also be written to
322           provide substitute names for one or more attributes of the
323           function's composite return type, including the ordinality column
324           if present.
325
326           Multiple function calls can be combined into a single FROM-clause
327           item by surrounding them with ROWS FROM( ... ). The output of such
328           an item is the concatenation of the first row from each function,
329           then the second row from each function, etc. If some of the
330           functions produce fewer rows than others, null values are
331           substituted for the missing data, so that the total number of rows
332           returned is always the same as for the function that produced the
333           most rows.
334
335           If the function has been defined as returning the record data type,
336           then an alias or the key word AS must be present, followed by a
337           column definition list in the form ( column_name data_type [, ...
338           ]). The column definition list must match the actual number and
339           types of columns returned by the function.
340
341           When using the ROWS FROM( ... ) syntax, if one of the functions
342           requires a column definition list, it's preferred to put the column
343           definition list after the function call inside ROWS FROM( ... ). A
344           column definition list can be placed after the ROWS FROM( ... )
345           construct only if there's just a single function and no WITH
346           ORDINALITY clause.
347
348           To use ORDINALITY together with a column definition list, you must
349           use the ROWS FROM( ... ) syntax and put the column definition list
350           inside ROWS FROM( ... ).
351
352       join_type
353           One of
354
355           •   [ INNER ] JOIN
356
357           •   LEFT [ OUTER ] JOIN
358
359           •   RIGHT [ OUTER ] JOIN
360
361           •   FULL [ OUTER ] JOIN
362
363           •   CROSS JOIN
364
365           For the INNER and OUTER join types, a join condition must be
366           specified, namely exactly one of NATURAL, ON join_condition, or
367           USING (join_column [, ...]). See below for the meaning. For CROSS
368           JOIN, none of these clauses can appear.
369
370           A JOIN clause combines two FROM items, which for convenience we
371           will refer to as “tables”, though in reality they can be any type
372           of FROM item. Use parentheses if necessary to determine the order
373           of nesting. In the absence of parentheses, JOINs nest
374           left-to-right. In any case JOIN binds more tightly than the commas
375           separating FROM-list items.
376
377           CROSS JOIN and INNER JOIN produce a simple Cartesian product, the
378           same result as you get from listing the two tables at the top level
379           of FROM, but restricted by the join condition (if any).  CROSS JOIN
380           is equivalent to INNER JOIN ON (TRUE), that is, no rows are removed
381           by qualification. These join types are just a notational
382           convenience, since they do nothing you couldn't do with plain FROM
383           and WHERE.
384
385           LEFT OUTER JOIN returns all rows in the qualified Cartesian product
386           (i.e., all combined rows that pass its join condition), plus one
387           copy of each row in the left-hand table for which there was no
388           right-hand row that passed the join condition. This left-hand row
389           is extended to the full width of the joined table by inserting null
390           values for the right-hand columns. Note that only the JOIN clause's
391           own condition is considered while deciding which rows have matches.
392           Outer conditions are applied afterwards.
393
394           Conversely, RIGHT OUTER JOIN returns all the joined rows, plus one
395           row for each unmatched right-hand row (extended with nulls on the
396           left). This is just a notational convenience, since you could
397           convert it to a LEFT OUTER JOIN by switching the left and right
398           tables.
399
400           FULL OUTER JOIN returns all the joined rows, plus one row for each
401           unmatched left-hand row (extended with nulls on the right), plus
402           one row for each unmatched right-hand row (extended with nulls on
403           the left).
404
405       ON join_condition
406           join_condition is an expression resulting in a value of type
407           boolean (similar to a WHERE clause) that specifies which rows in a
408           join are considered to match.
409
410       USING ( join_column [, ...] ) [ AS join_using_alias ]
411           A clause of the form USING ( a, b, ... ) is shorthand for ON
412           left_table.a = right_table.a AND left_table.b = right_table.b ....
413           Also, USING implies that only one of each pair of equivalent
414           columns will be included in the join output, not both.
415
416           If a join_using_alias name is specified, it provides a table alias
417           for the join columns. Only the join columns listed in the USING
418           clause are addressable by this name. Unlike a regular alias, this
419           does not hide the names of the joined tables from the rest of the
420           query. Also unlike a regular alias, you cannot write a column alias
421           list — the output names of the join columns are the same as they
422           appear in the USING list.
423
424       NATURAL
425           NATURAL is shorthand for a USING list that mentions all columns in
426           the two tables that have matching names. If there are no common
427           column names, NATURAL is equivalent to ON TRUE.
428
429       LATERAL
430           The LATERAL key word can precede a sub-SELECT FROM item. This
431           allows the sub-SELECT to refer to columns of FROM items that appear
432           before it in the FROM list. (Without LATERAL, each sub-SELECT is
433           evaluated independently and so cannot cross-reference any other
434           FROM item.)
435
436           LATERAL can also precede a function-call FROM item, but in this
437           case it is a noise word, because the function expression can refer
438           to earlier FROM items in any case.
439
440           A LATERAL item can appear at top level in the FROM list, or within
441           a JOIN tree. In the latter case it can also refer to any items that
442           are on the left-hand side of a JOIN that it is on the right-hand
443           side of.
444
445           When a FROM item contains LATERAL cross-references, evaluation
446           proceeds as follows: for each row of the FROM item providing the
447           cross-referenced column(s), or set of rows of multiple FROM items
448           providing the columns, the LATERAL item is evaluated using that row
449           or row set's values of the columns. The resulting row(s) are joined
450           as usual with the rows they were computed from. This is repeated
451           for each row or set of rows from the column source table(s).
452
453           The column source table(s) must be INNER or LEFT joined to the
454           LATERAL item, else there would not be a well-defined set of rows
455           from which to compute each set of rows for the LATERAL item. Thus,
456           although a construct such as X RIGHT JOIN LATERAL Y is
457           syntactically valid, it is not actually allowed for Y to reference
458           X.
459
460   WHERE Clause
461       The optional WHERE clause has the general form
462
463           WHERE condition
464
465       where condition is any expression that evaluates to a result of type
466       boolean. Any row that does not satisfy this condition will be
467       eliminated from the output. A row satisfies the condition if it returns
468       true when the actual row values are substituted for any variable
469       references.
470
471   GROUP BY Clause
472       The optional GROUP BY clause has the general form
473
474           GROUP BY [ ALL | DISTINCT ] grouping_element [, ...]
475
476       GROUP BY will condense into a single row all selected rows that share
477       the same values for the grouped expressions. An expression used inside
478       a grouping_element can be an input column name, or the name or ordinal
479       number of an output column (SELECT list item), or an arbitrary
480       expression formed from input-column values. In case of ambiguity, a
481       GROUP BY name will be interpreted as an input-column name rather than
482       an output column name.
483
484       If any of GROUPING SETS, ROLLUP or CUBE are present as grouping
485       elements, then the GROUP BY clause as a whole defines some number of
486       independent grouping sets. The effect of this is equivalent to
487       constructing a UNION ALL between subqueries with the individual
488       grouping sets as their GROUP BY clauses. The optional DISTINCT clause
489       removes duplicate sets before processing; it does not transform the
490       UNION ALL into a UNION DISTINCT. For further details on the handling of
491       grouping sets see Section 7.2.4.
492
493       Aggregate functions, if any are used, are computed across all rows
494       making up each group, producing a separate value for each group. (If
495       there are aggregate functions but no GROUP BY clause, the query is
496       treated as having a single group comprising all the selected rows.) The
497       set of rows fed to each aggregate function can be further filtered by
498       attaching a FILTER clause to the aggregate function call; see
499       Section 4.2.7 for more information. When a FILTER clause is present,
500       only those rows matching it are included in the input to that aggregate
501       function.
502
503       When GROUP BY is present, or any aggregate functions are present, it is
504       not valid for the SELECT list expressions to refer to ungrouped columns
505       except within aggregate functions or when the ungrouped column is
506       functionally dependent on the grouped columns, since there would
507       otherwise be more than one possible value to return for an ungrouped
508       column. A functional dependency exists if the grouped columns (or a
509       subset thereof) are the primary key of the table containing the
510       ungrouped column.
511
512       Keep in mind that all aggregate functions are evaluated before
513       evaluating any “scalar” expressions in the HAVING clause or SELECT
514       list. This means that, for example, a CASE expression cannot be used to
515       skip evaluation of an aggregate function; see Section 4.2.14.
516
517       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
518       cannot be specified with GROUP BY.
519
520   HAVING Clause
521       The optional HAVING clause has the general form
522
523           HAVING condition
524
525       where condition is the same as specified for the WHERE clause.
526
527       HAVING eliminates group rows that do not satisfy the condition.  HAVING
528       is different from WHERE: WHERE filters individual rows before the
529       application of GROUP BY, while HAVING filters group rows created by
530       GROUP BY. Each column referenced in condition must unambiguously
531       reference a grouping column, unless the reference appears within an
532       aggregate function or the ungrouped column is functionally dependent on
533       the grouping columns.
534
535       The presence of HAVING turns a query into a grouped query even if there
536       is no GROUP BY clause. This is the same as what happens when the query
537       contains aggregate functions but no GROUP BY clause. All the selected
538       rows are considered to form a single group, and the SELECT list and
539       HAVING clause can only reference table columns from within aggregate
540       functions. Such a query will emit a single row if the HAVING condition
541       is true, zero rows if it is not true.
542
543       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
544       cannot be specified with HAVING.
545
546   WINDOW Clause
547       The optional WINDOW clause has the general form
548
549           WINDOW window_name AS ( window_definition ) [, ...]
550
551       where window_name is a name that can be referenced from OVER clauses or
552       subsequent window definitions, and window_definition is
553
554           [ existing_window_name ]
555           [ PARTITION BY expression [, ...] ]
556           [ ORDER BY expression [ ASC | DESC | USING operator ] [ NULLS { FIRST | LAST } ] [, ...] ]
557           [ frame_clause ]
558
559       If an existing_window_name is specified it must refer to an earlier
560       entry in the WINDOW list; the new window copies its partitioning clause
561       from that entry, as well as its ordering clause if any. In this case
562       the new window cannot specify its own PARTITION BY clause, and it can
563       specify ORDER BY only if the copied window does not have one. The new
564       window always uses its own frame clause; the copied window must not
565       specify a frame clause.
566
567       The elements of the PARTITION BY list are interpreted in much the same
568       fashion as elements of a GROUP BY clause, except that they are always
569       simple expressions and never the name or number of an output column.
570       Another difference is that these expressions can contain aggregate
571       function calls, which are not allowed in a regular GROUP BY clause.
572       They are allowed here because windowing occurs after grouping and
573       aggregation.
574
575       Similarly, the elements of the ORDER BY list are interpreted in much
576       the same fashion as elements of a statement-level ORDER BY clause,
577       except that the expressions are always taken as simple expressions and
578       never the name or number of an output column.
579
580       The optional frame_clause defines the window frame for window functions
581       that depend on the frame (not all do). The window frame is a set of
582       related rows for each row of the query (called the current row). The
583       frame_clause can be one of
584
585           { RANGE | ROWS | GROUPS } frame_start [ frame_exclusion ]
586           { RANGE | ROWS | GROUPS } BETWEEN frame_start AND frame_end [ frame_exclusion ]
587
588       where frame_start and frame_end can be one of
589
590           UNBOUNDED PRECEDING
591           offset PRECEDING
592           CURRENT ROW
593           offset FOLLOWING
594           UNBOUNDED FOLLOWING
595
596       and frame_exclusion can be one of
597
598           EXCLUDE CURRENT ROW
599           EXCLUDE GROUP
600           EXCLUDE TIES
601           EXCLUDE NO OTHERS
602
603       If frame_end is omitted it defaults to CURRENT ROW. Restrictions are
604       that frame_start cannot be UNBOUNDED FOLLOWING, frame_end cannot be
605       UNBOUNDED PRECEDING, and the frame_end choice cannot appear earlier in
606       the above list of frame_start and frame_end options than the
607       frame_start choice does — for example RANGE BETWEEN CURRENT ROW AND
608       offset PRECEDING is not allowed.
609
610       The default framing option is RANGE UNBOUNDED PRECEDING, which is the
611       same as RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW; it sets the
612       frame to be all rows from the partition start up through the current
613       row's last peer (a row that the window's ORDER BY clause considers
614       equivalent to the current row; all rows are peers if there is no ORDER
615       BY). In general, UNBOUNDED PRECEDING means that the frame starts with
616       the first row of the partition, and similarly UNBOUNDED FOLLOWING means
617       that the frame ends with the last row of the partition, regardless of
618       RANGE, ROWS or GROUPS mode. In ROWS mode, CURRENT ROW means that the
619       frame starts or ends with the current row; but in RANGE or GROUPS mode
620       it means that the frame starts or ends with the current row's first or
621       last peer in the ORDER BY ordering. The offset PRECEDING and offset
622       FOLLOWING options vary in meaning depending on the frame mode. In ROWS
623       mode, the offset is an integer indicating that the frame starts or ends
624       that many rows before or after the current row. In GROUPS mode, the
625       offset is an integer indicating that the frame starts or ends that many
626       peer groups before or after the current row's peer group, where a peer
627       group is a group of rows that are equivalent according to the window's
628       ORDER BY clause. In RANGE mode, use of an offset option requires that
629       there be exactly one ORDER BY column in the window definition. Then the
630       frame contains those rows whose ordering column value is no more than
631       offset less than (for PRECEDING) or more than (for FOLLOWING) the
632       current row's ordering column value. In these cases the data type of
633       the offset expression depends on the data type of the ordering column.
634       For numeric ordering columns it is typically of the same type as the
635       ordering column, but for datetime ordering columns it is an interval.
636       In all these cases, the value of the offset must be non-null and
637       non-negative. Also, while the offset does not have to be a simple
638       constant, it cannot contain variables, aggregate functions, or window
639       functions.
640
641       The frame_exclusion option allows rows around the current row to be
642       excluded from the frame, even if they would be included according to
643       the frame start and frame end options.  EXCLUDE CURRENT ROW excludes
644       the current row from the frame.  EXCLUDE GROUP excludes the current row
645       and its ordering peers from the frame.  EXCLUDE TIES excludes any peers
646       of the current row from the frame, but not the current row itself.
647       EXCLUDE NO OTHERS simply specifies explicitly the default behavior of
648       not excluding the current row or its peers.
649
650       Beware that the ROWS mode can produce unpredictable results if the
651       ORDER BY ordering does not order the rows uniquely. The RANGE and
652       GROUPS modes are designed to ensure that rows that are peers in the
653       ORDER BY ordering are treated alike: all rows of a given peer group
654       will be in the frame or excluded from it.
655
656       The purpose of a WINDOW clause is to specify the behavior of window
657       functions appearing in the query's SELECT list or ORDER BY clause.
658       These functions can reference the WINDOW clause entries by name in
659       their OVER clauses. A WINDOW clause entry does not have to be
660       referenced anywhere, however; if it is not used in the query it is
661       simply ignored. It is possible to use window functions without any
662       WINDOW clause at all, since a window function call can specify its
663       window definition directly in its OVER clause. However, the WINDOW
664       clause saves typing when the same window definition is needed for more
665       than one window function.
666
667       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
668       cannot be specified with WINDOW.
669
670       Window functions are described in detail in Section 3.5, Section 4.2.8,
671       and Section 7.2.5.
672
673   SELECT List
674       The SELECT list (between the key words SELECT and FROM) specifies
675       expressions that form the output rows of the SELECT statement. The
676       expressions can (and usually do) refer to columns computed in the FROM
677       clause.
678
679       Just as in a table, every output column of a SELECT has a name. In a
680       simple SELECT this name is just used to label the column for display,
681       but when the SELECT is a sub-query of a larger query, the name is seen
682       by the larger query as the column name of the virtual table produced by
683       the sub-query. To specify the name to use for an output column, write
684       AS output_name after the column's expression. (You can omit AS, but
685       only if the desired output name does not match any PostgreSQL keyword
686       (see Appendix C). For protection against possible future keyword
687       additions, it is recommended that you always either write AS or
688       double-quote the output name.) If you do not specify a column name, a
689       name is chosen automatically by PostgreSQL. If the column's expression
690       is a simple column reference then the chosen name is the same as that
691       column's name. In more complex cases a function or type name may be
692       used, or the system may fall back on a generated name such as ?column?.
693
694       An output column's name can be used to refer to the column's value in
695       ORDER BY and GROUP BY clauses, but not in the WHERE or HAVING clauses;
696       there you must write out the expression instead.
697
698       Instead of an expression, * can be written in the output list as a
699       shorthand for all the columns of the selected rows. Also, you can write
700       table_name.*  as a shorthand for the columns coming from just that
701       table. In these cases it is not possible to specify new names with AS;
702       the output column names will be the same as the table columns' names.
703
704       According to the SQL standard, the expressions in the output list
705       should be computed before applying DISTINCT, ORDER BY, or LIMIT. This
706       is obviously necessary when using DISTINCT, since otherwise it's not
707       clear what values are being made distinct. However, in many cases it is
708       convenient if output expressions are computed after ORDER BY and LIMIT;
709       particularly if the output list contains any volatile or expensive
710       functions. With that behavior, the order of function evaluations is
711       more intuitive and there will not be evaluations corresponding to rows
712       that never appear in the output.  PostgreSQL will effectively evaluate
713       output expressions after sorting and limiting, so long as those
714       expressions are not referenced in DISTINCT, ORDER BY or GROUP BY. (As a
715       counterexample, SELECT f(x) FROM tab ORDER BY 1 clearly must evaluate
716       f(x) before sorting.) Output expressions that contain set-returning
717       functions are effectively evaluated after sorting and before limiting,
718       so that LIMIT will act to cut off the output from a set-returning
719       function.
720
721           Note
722           PostgreSQL versions before 9.6 did not provide any guarantees about
723           the timing of evaluation of output expressions versus sorting and
724           limiting; it depended on the form of the chosen query plan.
725
726   DISTINCT Clause
727       If SELECT DISTINCT is specified, all duplicate rows are removed from
728       the result set (one row is kept from each group of duplicates).  SELECT
729       ALL specifies the opposite: all rows are kept; that is the default.
730
731       SELECT DISTINCT ON ( expression [, ...] ) keeps only the first row of
732       each set of rows where the given expressions evaluate to equal. The
733       DISTINCT ON expressions are interpreted using the same rules as for
734       ORDER BY (see above). Note that the “first row” of each set is
735       unpredictable unless ORDER BY is used to ensure that the desired row
736       appears first. For example:
737
738           SELECT DISTINCT ON (location) location, time, report
739               FROM weather_reports
740               ORDER BY location, time DESC;
741
742       retrieves the most recent weather report for each location. But if we
743       had not used ORDER BY to force descending order of time values for each
744       location, we'd have gotten a report from an unpredictable time for each
745       location.
746
747       The DISTINCT ON expression(s) must match the leftmost ORDER BY
748       expression(s). The ORDER BY clause will normally contain additional
749       expression(s) that determine the desired precedence of rows within each
750       DISTINCT ON group.
751
752       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
753       cannot be specified with DISTINCT.
754
755   UNION Clause
756       The UNION clause has this general form:
757
758           select_statement UNION [ ALL | DISTINCT ] select_statement
759
760       select_statement is any SELECT statement without an ORDER BY, LIMIT,
761       FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE, or FOR KEY SHARE clause.
762       (ORDER BY and LIMIT can be attached to a subexpression if it is
763       enclosed in parentheses. Without parentheses, these clauses will be
764       taken to apply to the result of the UNION, not to its right-hand input
765       expression.)
766
767       The UNION operator computes the set union of the rows returned by the
768       involved SELECT statements. A row is in the set union of two result
769       sets if it appears in at least one of the result sets. The two SELECT
770       statements that represent the direct operands of the UNION must produce
771       the same number of columns, and corresponding columns must be of
772       compatible data types.
773
774       The result of UNION does not contain any duplicate rows unless the ALL
775       option is specified.  ALL prevents elimination of duplicates.
776       (Therefore, UNION ALL is usually significantly quicker than UNION; use
777       ALL when you can.)  DISTINCT can be written to explicitly specify the
778       default behavior of eliminating duplicate rows.
779
780       Multiple UNION operators in the same SELECT statement are evaluated
781       left to right, unless otherwise indicated by parentheses.
782
783       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
784       cannot be specified either for a UNION result or for any input of a
785       UNION.
786
787   INTERSECT Clause
788       The INTERSECT clause has this general form:
789
790           select_statement INTERSECT [ ALL | DISTINCT ] select_statement
791
792       select_statement is any SELECT statement without an ORDER BY, LIMIT,
793       FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE, or FOR KEY SHARE clause.
794
795       The INTERSECT operator computes the set intersection of the rows
796       returned by the involved SELECT statements. A row is in the
797       intersection of two result sets if it appears in both result sets.
798
799       The result of INTERSECT does not contain any duplicate rows unless the
800       ALL option is specified. With ALL, a row that has m duplicates in the
801       left table and n duplicates in the right table will appear min(m,n)
802       times in the result set.  DISTINCT can be written to explicitly specify
803       the default behavior of eliminating duplicate rows.
804
805       Multiple INTERSECT operators in the same SELECT statement are evaluated
806       left to right, unless parentheses dictate otherwise.  INTERSECT binds
807       more tightly than UNION. That is, A UNION B INTERSECT C will be read as
808       A UNION (B INTERSECT C).
809
810       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
811       cannot be specified either for an INTERSECT result or for any input of
812       an INTERSECT.
813
814   EXCEPT Clause
815       The EXCEPT clause has this general form:
816
817           select_statement EXCEPT [ ALL | DISTINCT ] select_statement
818
819       select_statement is any SELECT statement without an ORDER BY, LIMIT,
820       FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE, or FOR KEY SHARE clause.
821
822       The EXCEPT operator computes the set of rows that are in the result of
823       the left SELECT statement but not in the result of the right one.
824
825       The result of EXCEPT does not contain any duplicate rows unless the ALL
826       option is specified. With ALL, a row that has m duplicates in the left
827       table and n duplicates in the right table will appear max(m-n,0) times
828       in the result set.  DISTINCT can be written to explicitly specify the
829       default behavior of eliminating duplicate rows.
830
831       Multiple EXCEPT operators in the same SELECT statement are evaluated
832       left to right, unless parentheses dictate otherwise.  EXCEPT binds at
833       the same level as UNION.
834
835       Currently, FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE and FOR KEY SHARE
836       cannot be specified either for an EXCEPT result or for any input of an
837       EXCEPT.
838
839   ORDER BY Clause
840       The optional ORDER BY clause has this general form:
841
842           ORDER BY expression [ ASC | DESC | USING operator ] [ NULLS { FIRST | LAST } ] [, ...]
843
844       The ORDER BY clause causes the result rows to be sorted according to
845       the specified expression(s). If two rows are equal according to the
846       leftmost expression, they are compared according to the next expression
847       and so on. If they are equal according to all specified expressions,
848       they are returned in an implementation-dependent order.
849
850       Each expression can be the name or ordinal number of an output column
851       (SELECT list item), or it can be an arbitrary expression formed from
852       input-column values.
853
854       The ordinal number refers to the ordinal (left-to-right) position of
855       the output column. This feature makes it possible to define an ordering
856       on the basis of a column that does not have a unique name. This is
857       never absolutely necessary because it is always possible to assign a
858       name to an output column using the AS clause.
859
860       It is also possible to use arbitrary expressions in the ORDER BY
861       clause, including columns that do not appear in the SELECT output list.
862       Thus the following statement is valid:
863
864           SELECT name FROM distributors ORDER BY code;
865
866       A limitation of this feature is that an ORDER BY clause applying to the
867       result of a UNION, INTERSECT, or EXCEPT clause can only specify an
868       output column name or number, not an expression.
869
870       If an ORDER BY expression is a simple name that matches both an output
871       column name and an input column name, ORDER BY will interpret it as the
872       output column name. This is the opposite of the choice that GROUP BY
873       will make in the same situation. This inconsistency is made to be
874       compatible with the SQL standard.
875
876       Optionally one can add the key word ASC (ascending) or DESC
877       (descending) after any expression in the ORDER BY clause. If not
878       specified, ASC is assumed by default. Alternatively, a specific
879       ordering operator name can be specified in the USING clause. An
880       ordering operator must be a less-than or greater-than member of some
881       B-tree operator family.  ASC is usually equivalent to USING < and DESC
882       is usually equivalent to USING >. (But the creator of a user-defined
883       data type can define exactly what the default sort ordering is, and it
884       might correspond to operators with other names.)
885
886       If NULLS LAST is specified, null values sort after all non-null values;
887       if NULLS FIRST is specified, null values sort before all non-null
888       values. If neither is specified, the default behavior is NULLS LAST
889       when ASC is specified or implied, and NULLS FIRST when DESC is
890       specified (thus, the default is to act as though nulls are larger than
891       non-nulls). When USING is specified, the default nulls ordering depends
892       on whether the operator is a less-than or greater-than operator.
893
894       Note that ordering options apply only to the expression they follow;
895       for example ORDER BY x, y DESC does not mean the same thing as ORDER BY
896       x DESC, y DESC.
897
898       Character-string data is sorted according to the collation that applies
899       to the column being sorted. That can be overridden at need by including
900       a COLLATE clause in the expression, for example ORDER BY mycolumn
901       COLLATE "en_US". For more information see Section 4.2.10 and
902       Section 24.2.
903
904   LIMIT Clause
905       The LIMIT clause consists of two independent sub-clauses:
906
907           LIMIT { count | ALL }
908           OFFSET start
909
910       The parameter count specifies the maximum number of rows to return,
911       while start specifies the number of rows to skip before starting to
912       return rows. When both are specified, start rows are skipped before
913       starting to count the count rows to be returned.
914
915       If the count expression evaluates to NULL, it is treated as LIMIT ALL,
916       i.e., no limit. If start evaluates to NULL, it is treated the same as
917       OFFSET 0.
918
919       SQL:2008 introduced a different syntax to achieve the same result,
920       which PostgreSQL also supports. It is:
921
922           OFFSET start { ROW | ROWS }
923           FETCH { FIRST | NEXT } [ count ] { ROW | ROWS } { ONLY | WITH TIES }
924
925       In this syntax, the start or count value is required by the standard to
926       be a literal constant, a parameter, or a variable name; as a PostgreSQL
927       extension, other expressions are allowed, but will generally need to be
928       enclosed in parentheses to avoid ambiguity. If count is omitted in a
929       FETCH clause, it defaults to 1. The WITH TIES option is used to return
930       any additional rows that tie for the last place in the result set
931       according to the ORDER BY clause; ORDER BY is mandatory in this case,
932       and SKIP LOCKED is not allowed.  ROW and ROWS as well as FIRST and NEXT
933       are noise words that don't influence the effects of these clauses.
934       According to the standard, the OFFSET clause must come before the FETCH
935       clause if both are present; but PostgreSQL is laxer and allows either
936       order.
937
938       When using LIMIT, it is a good idea to use an ORDER BY clause that
939       constrains the result rows into a unique order. Otherwise you will get
940       an unpredictable subset of the query's rows — you might be asking for
941       the tenth through twentieth rows, but tenth through twentieth in what
942       ordering? You don't know what ordering unless you specify ORDER BY.
943
944       The query planner takes LIMIT into account when generating a query
945       plan, so you are very likely to get different plans (yielding different
946       row orders) depending on what you use for LIMIT and OFFSET. Thus, using
947       different LIMIT/OFFSET values to select different subsets of a query
948       result will give inconsistent results unless you enforce a predictable
949       result ordering with ORDER BY. This is not a bug; it is an inherent
950       consequence of the fact that SQL does not promise to deliver the
951       results of a query in any particular order unless ORDER BY is used to
952       constrain the order.
953
954       It is even possible for repeated executions of the same LIMIT query to
955       return different subsets of the rows of a table, if there is not an
956       ORDER BY to enforce selection of a deterministic subset. Again, this is
957       not a bug; determinism of the results is simply not guaranteed in such
958       a case.
959
960   The Locking Clause
961       FOR UPDATE, FOR NO KEY UPDATE, FOR SHARE and FOR KEY SHARE are locking
962       clauses; they affect how SELECT locks rows as they are obtained from
963       the table.
964
965       The locking clause has the general form
966
967           FOR lock_strength [ OF table_name [, ...] ] [ NOWAIT | SKIP LOCKED ]
968
969       where lock_strength can be one of
970
971           UPDATE
972           NO KEY UPDATE
973           SHARE
974           KEY SHARE
975
976       For more information on each row-level lock mode, refer to
977       Section 13.3.2.
978
979       To prevent the operation from waiting for other transactions to commit,
980       use either the NOWAIT or SKIP LOCKED option. With NOWAIT, the statement
981       reports an error, rather than waiting, if a selected row cannot be
982       locked immediately. With SKIP LOCKED, any selected rows that cannot be
983       immediately locked are skipped. Skipping locked rows provides an
984       inconsistent view of the data, so this is not suitable for general
985       purpose work, but can be used to avoid lock contention with multiple
986       consumers accessing a queue-like table. Note that NOWAIT and SKIP
987       LOCKED apply only to the row-level lock(s) — the required ROW SHARE
988       table-level lock is still taken in the ordinary way (see Chapter 13).
989       You can use LOCK with the NOWAIT option first, if you need to acquire
990       the table-level lock without waiting.
991
992       If specific tables are named in a locking clause, then only rows coming
993       from those tables are locked; any other tables used in the SELECT are
994       simply read as usual. A locking clause without a table list affects all
995       tables used in the statement. If a locking clause is applied to a view
996       or sub-query, it affects all tables used in the view or sub-query.
997       However, these clauses do not apply to WITH queries referenced by the
998       primary query. If you want row locking to occur within a WITH query,
999       specify a locking clause within the WITH query.
1000
1001       Multiple locking clauses can be written if it is necessary to specify
1002       different locking behavior for different tables. If the same table is
1003       mentioned (or implicitly affected) by more than one locking clause,
1004       then it is processed as if it was only specified by the strongest one.
1005       Similarly, a table is processed as NOWAIT if that is specified in any
1006       of the clauses affecting it. Otherwise, it is processed as SKIP LOCKED
1007       if that is specified in any of the clauses affecting it.
1008
1009       The locking clauses cannot be used in contexts where returned rows
1010       cannot be clearly identified with individual table rows; for example
1011       they cannot be used with aggregation.
1012
1013       When a locking clause appears at the top level of a SELECT query, the
1014       rows that are locked are exactly those that are returned by the query;
1015       in the case of a join query, the rows locked are those that contribute
1016       to returned join rows. In addition, rows that satisfied the query
1017       conditions as of the query snapshot will be locked, although they will
1018       not be returned if they were updated after the snapshot and no longer
1019       satisfy the query conditions. If a LIMIT is used, locking stops once
1020       enough rows have been returned to satisfy the limit (but note that rows
1021       skipped over by OFFSET will get locked). Similarly, if a locking clause
1022       is used in a cursor's query, only rows actually fetched or stepped past
1023       by the cursor will be locked.
1024
1025       When a locking clause appears in a sub-SELECT, the rows locked are
1026       those returned to the outer query by the sub-query. This might involve
1027       fewer rows than inspection of the sub-query alone would suggest, since
1028       conditions from the outer query might be used to optimize execution of
1029       the sub-query. For example,
1030
1031           SELECT * FROM (SELECT * FROM mytable FOR UPDATE) ss WHERE col1 = 5;
1032
1033       will lock only rows having col1 = 5, even though that condition is not
1034       textually within the sub-query.
1035
1036       Previous releases failed to preserve a lock which is upgraded by a
1037       later savepoint. For example, this code:
1038
1039           BEGIN;
1040           SELECT * FROM mytable WHERE key = 1 FOR UPDATE;
1041           SAVEPOINT s;
1042           UPDATE mytable SET ... WHERE key = 1;
1043           ROLLBACK TO s;
1044
1045       would fail to preserve the FOR UPDATE lock after the ROLLBACK TO. This
1046       has been fixed in release 9.3.
1047
1048           Caution
1049           It is possible for a SELECT command running at the READ COMMITTED
1050           transaction isolation level and using ORDER BY and a locking clause
1051           to return rows out of order. This is because ORDER BY is applied
1052           first. The command sorts the result, but might then block trying to
1053           obtain a lock on one or more of the rows. Once the SELECT unblocks,
1054           some of the ordering column values might have been modified,
1055           leading to those rows appearing to be out of order (though they are
1056           in order in terms of the original column values). This can be
1057           worked around at need by placing the FOR UPDATE/SHARE clause in a
1058           sub-query, for example
1059
1060               SELECT * FROM (SELECT * FROM mytable FOR UPDATE) ss ORDER BY column1;
1061
1062           Note that this will result in locking all rows of mytable, whereas
1063           FOR UPDATE at the top level would lock only the actually returned
1064           rows. This can make for a significant performance difference,
1065           particularly if the ORDER BY is combined with LIMIT or other
1066           restrictions. So this technique is recommended only if concurrent
1067           updates of the ordering columns are expected and a strictly sorted
1068           result is required.
1069
1070           At the REPEATABLE READ or SERIALIZABLE transaction isolation level
1071           this would cause a serialization failure (with a SQLSTATE of
1072           '40001'), so there is no possibility of receiving rows out of order
1073           under these isolation levels.
1074
1075   TABLE Command
1076       The command
1077
1078           TABLE name
1079
1080       is equivalent to
1081
1082           SELECT * FROM name
1083
1084       It can be used as a top-level command or as a space-saving syntax
1085       variant in parts of complex queries. Only the WITH, UNION, INTERSECT,
1086       EXCEPT, ORDER BY, LIMIT, OFFSET, FETCH and FOR locking clauses can be
1087       used with TABLE; the WHERE clause and any form of aggregation cannot be
1088       used.
1089

EXAMPLES

1091       To join the table films with the table distributors:
1092
1093           SELECT f.title, f.did, d.name, f.date_prod, f.kind
1094               FROM distributors d, films f
1095               WHERE f.did = d.did
1096
1097                  title       | did |     name     | date_prod  |   kind
1098           -------------------+-----+--------------+------------+----------
1099            The Third Man     | 101 | British Lion | 1949-12-23 | Drama
1100            The African Queen | 101 | British Lion | 1951-08-11 | Romantic
1101            ...
1102
1103       To sum the column len of all films and group the results by kind:
1104
1105           SELECT kind, sum(len) AS total FROM films GROUP BY kind;
1106
1107              kind   | total
1108           ----------+-------
1109            Action   | 07:34
1110            Comedy   | 02:58
1111            Drama    | 14:28
1112            Musical  | 06:42
1113            Romantic | 04:38
1114
1115       To sum the column len of all films, group the results by kind and show
1116       those group totals that are less than 5 hours:
1117
1118           SELECT kind, sum(len) AS total
1119               FROM films
1120               GROUP BY kind
1121               HAVING sum(len) < interval '5 hours';
1122
1123              kind   | total
1124           ----------+-------
1125            Comedy   | 02:58
1126            Romantic | 04:38
1127
1128       The following two examples are identical ways of sorting the individual
1129       results according to the contents of the second column (name):
1130
1131           SELECT * FROM distributors ORDER BY name;
1132           SELECT * FROM distributors ORDER BY 2;
1133
1134            did |       name
1135           -----+------------------
1136            109 | 20th Century Fox
1137            110 | Bavaria Atelier
1138            101 | British Lion
1139            107 | Columbia
1140            102 | Jean Luc Godard
1141            113 | Luso films
1142            104 | Mosfilm
1143            103 | Paramount
1144            106 | Toho
1145            105 | United Artists
1146            111 | Walt Disney
1147            112 | Warner Bros.
1148            108 | Westward
1149
1150       The next example shows how to obtain the union of the tables
1151       distributors and actors, restricting the results to those that begin
1152       with the letter W in each table. Only distinct rows are wanted, so the
1153       key word ALL is omitted.
1154
1155           distributors:               actors:
1156            did |     name              id |     name
1157           -----+--------------        ----+----------------
1158            108 | Westward               1 | Woody Allen
1159            111 | Walt Disney            2 | Warren Beatty
1160            112 | Warner Bros.           3 | Walter Matthau
1161            ...                         ...
1162
1163           SELECT distributors.name
1164               FROM distributors
1165               WHERE distributors.name LIKE 'W%'
1166           UNION
1167           SELECT actors.name
1168               FROM actors
1169               WHERE actors.name LIKE 'W%';
1170
1171                 name
1172           ----------------
1173            Walt Disney
1174            Walter Matthau
1175            Warner Bros.
1176            Warren Beatty
1177            Westward
1178            Woody Allen
1179
1180       This example shows how to use a function in the FROM clause, both with
1181       and without a column definition list:
1182
1183           CREATE FUNCTION distributors(int) RETURNS SETOF distributors AS $$
1184               SELECT * FROM distributors WHERE did = $1;
1185           $$ LANGUAGE SQL;
1186
1187           SELECT * FROM distributors(111);
1188            did |    name
1189           -----+-------------
1190            111 | Walt Disney
1191
1192           CREATE FUNCTION distributors_2(int) RETURNS SETOF record AS $$
1193               SELECT * FROM distributors WHERE did = $1;
1194           $$ LANGUAGE SQL;
1195
1196           SELECT * FROM distributors_2(111) AS (f1 int, f2 text);
1197            f1  |     f2
1198           -----+-------------
1199            111 | Walt Disney
1200
1201       Here is an example of a function with an ordinality column added:
1202
1203           SELECT * FROM unnest(ARRAY['a','b','c','d','e','f']) WITH ORDINALITY;
1204            unnest | ordinality
1205           --------+----------
1206            a      |        1
1207            b      |        2
1208            c      |        3
1209            d      |        4
1210            e      |        5
1211            f      |        6
1212           (6 rows)
1213
1214       This example shows how to use a simple WITH clause:
1215
1216           WITH t AS (
1217               SELECT random() as x FROM generate_series(1, 3)
1218             )
1219           SELECT * FROM t
1220           UNION ALL
1221           SELECT * FROM t
1222
1223                    x
1224           --------------------
1225             0.534150459803641
1226             0.520092216785997
1227            0.0735620250925422
1228             0.534150459803641
1229             0.520092216785997
1230            0.0735620250925422
1231
1232       Notice that the WITH query was evaluated only once, so that we got two
1233       sets of the same three random values.
1234
1235       This example uses WITH RECURSIVE to find all subordinates (direct or
1236       indirect) of the employee Mary, and their level of indirectness, from a
1237       table that shows only direct subordinates:
1238
1239           WITH RECURSIVE employee_recursive(distance, employee_name, manager_name) AS (
1240               SELECT 1, employee_name, manager_name
1241               FROM employee
1242               WHERE manager_name = 'Mary'
1243             UNION ALL
1244               SELECT er.distance + 1, e.employee_name, e.manager_name
1245               FROM employee_recursive er, employee e
1246               WHERE er.employee_name = e.manager_name
1247             )
1248           SELECT distance, employee_name FROM employee_recursive;
1249
1250       Notice the typical form of recursive queries: an initial condition,
1251       followed by UNION, followed by the recursive part of the query. Be sure
1252       that the recursive part of the query will eventually return no tuples,
1253       or else the query will loop indefinitely. (See Section 7.8 for more
1254       examples.)
1255
1256       This example uses LATERAL to apply a set-returning function
1257       get_product_names() for each row of the manufacturers table:
1258
1259           SELECT m.name AS mname, pname
1260           FROM manufacturers m, LATERAL get_product_names(m.id) pname;
1261
1262       Manufacturers not currently having any products would not appear in the
1263       result, since it is an inner join. If we wished to include the names of
1264       such manufacturers in the result, we could do:
1265
1266           SELECT m.name AS mname, pname
1267           FROM manufacturers m LEFT JOIN LATERAL get_product_names(m.id) pname ON true;
1268

COMPATIBILITY

1270       Of course, the SELECT statement is compatible with the SQL standard.
1271       But there are some extensions and some missing features.
1272
1273   Omitted FROM Clauses
1274       PostgreSQL allows one to omit the FROM clause. It has a straightforward
1275       use to compute the results of simple expressions:
1276
1277           SELECT 2+2;
1278
1279            ?column?
1280           ----------
1281                   4
1282
1283       Some other SQL databases cannot do this except by introducing a dummy
1284       one-row table from which to do the SELECT.
1285
1286   Empty SELECT Lists
1287       The list of output expressions after SELECT can be empty, producing a
1288       zero-column result table. This is not valid syntax according to the SQL
1289       standard.  PostgreSQL allows it to be consistent with allowing
1290       zero-column tables. However, an empty list is not allowed when DISTINCT
1291       is used.
1292
1293   Omitting the AS Key Word
1294       In the SQL standard, the optional key word AS can be omitted before an
1295       output column name whenever the new column name is a valid column name
1296       (that is, not the same as any reserved keyword).  PostgreSQL is
1297       slightly more restrictive: AS is required if the new column name
1298       matches any keyword at all, reserved or not. Recommended practice is to
1299       use AS or double-quote output column names, to prevent any possible
1300       conflict against future keyword additions.
1301
1302       In FROM items, both the standard and PostgreSQL allow AS to be omitted
1303       before an alias that is an unreserved keyword. But this is impractical
1304       for output column names, because of syntactic ambiguities.
1305
1306   ONLY and Inheritance
1307       The SQL standard requires parentheses around the table name when
1308       writing ONLY, for example SELECT * FROM ONLY (tab1), ONLY (tab2) WHERE
1309       ....  PostgreSQL considers these parentheses to be optional.
1310
1311       PostgreSQL allows a trailing * to be written to explicitly specify the
1312       non-ONLY behavior of including child tables. The standard does not
1313       allow this.
1314
1315       (These points apply equally to all SQL commands supporting the ONLY
1316       option.)
1317
1318   TABLESAMPLE Clause Restrictions
1319       The TABLESAMPLE clause is currently accepted only on regular tables and
1320       materialized views. According to the SQL standard it should be possible
1321       to apply it to any FROM item.
1322
1323   Function Calls in FROM
1324       PostgreSQL allows a function call to be written directly as a member of
1325       the FROM list. In the SQL standard it would be necessary to wrap such a
1326       function call in a sub-SELECT; that is, the syntax FROM func(...) alias
1327       is approximately equivalent to FROM LATERAL (SELECT func(...)) alias.
1328       Note that LATERAL is considered to be implicit; this is because the
1329       standard requires LATERAL semantics for an UNNEST() item in FROM.
1330       PostgreSQL treats UNNEST() the same as other set-returning functions.
1331
1332   Namespace Available to GROUP BY and ORDER BY
1333       In the SQL-92 standard, an ORDER BY clause can only use output column
1334       names or numbers, while a GROUP BY clause can only use expressions
1335       based on input column names.  PostgreSQL extends each of these clauses
1336       to allow the other choice as well (but it uses the standard's
1337       interpretation if there is ambiguity).  PostgreSQL also allows both
1338       clauses to specify arbitrary expressions. Note that names appearing in
1339       an expression will always be taken as input-column names, not as
1340       output-column names.
1341
1342       SQL:1999 and later use a slightly different definition which is not
1343       entirely upward compatible with SQL-92. In most cases, however,
1344       PostgreSQL will interpret an ORDER BY or GROUP BY expression the same
1345       way SQL:1999 does.
1346
1347   Functional Dependencies
1348       PostgreSQL recognizes functional dependency (allowing columns to be
1349       omitted from GROUP BY) only when a table's primary key is included in
1350       the GROUP BY list. The SQL standard specifies additional conditions
1351       that should be recognized.
1352
1353   LIMIT and OFFSET
1354       The clauses LIMIT and OFFSET are PostgreSQL-specific syntax, also used
1355       by MySQL. The SQL:2008 standard has introduced the clauses OFFSET ...
1356       FETCH {FIRST|NEXT} ...  for the same functionality, as shown above in
1357       LIMIT Clause. This syntax is also used by IBM DB2. (Applications
1358       written for Oracle frequently use a workaround involving the
1359       automatically generated rownum column, which is not available in
1360       PostgreSQL, to implement the effects of these clauses.)
1361
1362   FOR NO KEY UPDATE, FOR UPDATE, FOR SHARE, FOR KEY SHARE
1363       Although FOR UPDATE appears in the SQL standard, the standard allows it
1364       only as an option of DECLARE CURSOR.  PostgreSQL allows it in any
1365       SELECT query as well as in sub-SELECTs, but this is an extension. The
1366       FOR NO KEY UPDATE, FOR SHARE and FOR KEY SHARE variants, as well as the
1367       NOWAIT and SKIP LOCKED options, do not appear in the standard.
1368
1369   Data-Modifying Statements in WITH
1370       PostgreSQL allows INSERT, UPDATE, and DELETE to be used as WITH
1371       queries. This is not found in the SQL standard.
1372
1373   Nonstandard Clauses
1374       DISTINCT ON ( ... ) is an extension of the SQL standard.
1375
1376       ROWS FROM( ... ) is an extension of the SQL standard.
1377
1378       The MATERIALIZED and NOT MATERIALIZED options of WITH are extensions of
1379       the SQL standard.
1380
1381
1382
1383PostgreSQL 14.3                      2022                            SELECT(7)
Impressum