Postgresql 中文操作指南

9.21. Aggregate Functions #

Aggregate functions 从一组输入值中计算一个单一结果。内置通用聚合函数列在 Table 9.59 中，而统计聚合函数列在 Table 9.60 中。内置组内有序集聚合函数列在 Table 9.61 中，而内置组内假设集函数列在 Table 9.62 中。分组操作与聚合函数密切相关，列在 Table 9.63 中。聚合函数的特殊语法注意事项在 Section 4.2.7 中进行了解释。请查阅 Section 2.7 以获取其他介绍性信息。

Aggregate functions compute a single result from a set of input values. The built-in general-purpose aggregate functions are listed in Table 9.59 while statistical aggregates are in Table 9.60. The built-in within-group ordered-set aggregate functions are listed in Table 9.61 while the built-in within-group hypothetical-set ones are in Table 9.62. Grouping operations, which are closely related to aggregate functions, are listed in Table 9.63. The special syntax considerations for aggregate functions are explained in Section 4.2.7. Consult Section 2.7 for additional introductory information.

支持 Partial Mode 的聚合函数有资格参与各种优化，例如并行聚合。

Aggregate functions that support Partial Mode are eligible to participate in various optimizations, such as parallel aggregation.

Table 9.59. General-Purpose Aggregate Functions

Function

Description

Partial Mode

any_value ( anyelement ) → same as input type

Returns an arbitrary value from the non-null input values.

Yes

array_agg ( anynonarray ) → anyarray

Collects all the input values, including nulls, into an array.

Yes

array_agg ( anyarray ) → anyarray

Concatenates all the input arrays into an array of one higher dimension. (The inputs must all have the same dimensionality, and cannot be empty or null.)

Yes

avg ( smallint ) → numeric

avg ( integer ) → numeric

avg ( bigint ) → numeric

avg ( numeric ) → numeric

avg ( real ) → double precision

avg ( double precision ) → double precision

avg ( interval ) → interval

Computes the average (arithmetic mean) of all the non-null input values.

Yes

bit_and ( smallint ) → smallint

bit_and ( integer ) → integer

bit_and ( bigint ) → bigint

bit_and ( bit ) → bit

Computes the bitwise AND of all non-null input values.

Yes

bit_or ( smallint ) → smallint

bit_or ( integer ) → integer

bit_or ( bigint ) → bigint

bit_or ( bit ) → bit

Computes the bitwise OR of all non-null input values.

Yes

bit_xor ( smallint ) → smallint

bit_xor ( integer ) → integer

bit_xor ( bigint ) → bigint

bit_xor ( bit ) → bit

Computes the bitwise exclusive OR of all non-null input values. Can be useful as a checksum for an unordered set of values.

Yes

bool_and ( boolean ) → boolean

Returns true if all non-null input values are true, otherwise false.

Yes

bool_or ( boolean ) → boolean

Returns true if any non-null input value is true, otherwise false.

Yes

count ( * ) → bigint

Computes the number of input rows.

Yes

count ( "any" ) → bigint

Computes the number of input rows in which the input value is not null.

Yes

every ( boolean ) → boolean

This is the SQL standard’s equivalent to bool_and.

Yes

json_agg ( anyelement ) → json

jsonb_agg ( anyelement ) → jsonb

Collects all the input values, including nulls, into a JSON array. Values are converted to JSON as per to_json or to_jsonb.

json_objectagg ( [ { key_expression { VALUE

':' } value_expression } ] [ { NULL

ABSENT } ON NULL ] [ { WITH

WITHOUT } UNIQUE [ KEYS ] ] [ RETURNING data_type [ FORMAT JSON [ ENCODING UTF8 ] ] ])

Behaves like json_object, but as an aggregate function, so it only takes one key_expression and one value_expression parameter.

SELECT json_objectagg(k:v) FROM (VALUES ('a'::text,current_date),('b',current_date + 1)) AS t(k,v) → { "a" : "2022-05-10", "b" : "2022-05-11" }

json_object_agg ( key "any", value "any" ) → json

jsonb_object_agg ( key "any", value "any" ) → jsonb

Collects all the key/value pairs into a JSON object. Key arguments are coerced to text; value arguments are converted as per to_json or to_jsonb. Values can be null, but keys cannot.

json_object_agg_strict ( key "any", value "any" ) → json

jsonb_object_agg_strict ( key "any", value "any" ) → jsonb

Collects all the key/value pairs into a JSON object. Key arguments are coerced to text; value arguments are converted as per to_json or to_jsonb. The key can not be null. If the value is null then the entry is skipped,

json_object_agg_unique ( key "any", value "any" ) → json

jsonb_object_agg_unique ( key "any", value "any" ) → jsonb

Collects all the key/value pairs into a JSON object. Key arguments are coerced to text; value arguments are converted as per to_json or to_jsonb. Values can be null, but keys cannot. If there is a duplicate key an error is thrown.

json_arrayagg ( [ value_expression ] [ ORDER BY sort_expression ] [ { NULL

ABSENT } ON NULL ] [ RETURNING data_type [ FORMAT JSON [ ENCODING UTF8 ] ] ])

Behaves in the same way as json_array but as an aggregate function so it only takes one value_expression parameter. If ABSENT ON NULL is specified, any NULL values are omitted. If ORDER BY is specified, the elements will appear in the array in that order rather than in the input order.

SELECT json_arrayagg(v) FROM (VALUES(2),(1)) t(v) → [2, 1]

json_object_agg_unique_strict ( key "any", value "any" ) → json

jsonb_object_agg_unique_strict ( key "any", value "any" ) → jsonb

max ( see text ) → same as input type

Computes the maximum of the non-null input values. Available for any numeric, string, date/time, or enum type, as well as inet, interval, money, oid, pg_lsn, tid, xid8, and arrays of any of these types.

Yes

min ( see text ) → same as input type

Computes the minimum of the non-null input values. Available for any numeric, string, date/time, or enum type, as well as inet, interval, money, oid, pg_lsn, tid, xid8, and arrays of any of these types.

Yes

range_agg ( value anyrange ) → anymultirange

range_agg ( value anymultirange ) → anymultirange

Computes the union of the non-null input values.

range_intersect_agg ( value anyrange ) → anyrange

range_intersect_agg ( value anymultirange ) → anymultirange

Computes the intersection of the non-null input values.

json_agg_strict ( anyelement ) → json

jsonb_agg_strict ( anyelement ) → jsonb

Collects all the input values, skipping nulls, into a JSON array. Values are converted to JSON as per to_json or to_jsonb.

string_agg ( value text, delimiter text ) → text

string_agg ( value bytea, delimiter bytea ) → bytea

Concatenates the non-null input values into a string. Each value after the first is preceded by the corresponding delimiter (if it’s not null).

Yes

sum ( smallint ) → bigint

sum ( integer ) → bigint

sum ( bigint ) → numeric

sum ( numeric ) → numeric

sum ( real ) → real

sum ( double precision ) → double precision

sum ( interval ) → interval

sum ( money ) → money

Computes the sum of the non-null input values.

Yes

xmlagg ( xml ) → xml

Concatenates the non-null XML input values (see Section 9.15.1.7).

应该注意，除了 count 外，这些函数在不选择任何行时会返回 null 值。尤其是 sum，没有行会返回 null，而不是人们期望的零；当没有任何输入行时，array_agg 会返回 null，而不是空数组。coalesce 函数可用于在必要时用零或空数组替换 null。

It should be noted that except for count, these functions return a null value when no rows are selected. In particular, sum of no rows returns null, not zero as one might expect, and array_agg returns null rather than an empty array when there are no input rows. The coalesce function can be used to substitute zero or an empty array for null when necessary.

聚合函数 array_agg, json_agg, jsonb_agg, json_agg_strict, jsonb_agg_strict, json_object_agg, jsonb_object_agg, json_object_agg_strict, jsonb_object_agg_strict, json_object_agg_unique, jsonb_object_agg_unique, json_object_agg_unique_strict, jsonb_object_agg_unique_strict, string_agg_和 _xmlagg，类似的用户定义聚合函数产生意义显著的不同结果值，具体取决于输入值顺序。此顺序默认未指定，但可以通过在聚合调用中编写 _ORDER BY_语句进行控制，如 Section 4.2.7所示。或者，通常会提供来自排序子查询的输入值。例如：

The aggregate functions array_agg, json_agg, jsonb_agg, json_agg_strict, jsonb_agg_strict, json_object_agg, jsonb_object_agg, json_object_agg_strict, jsonb_object_agg_strict, json_object_agg_unique, jsonb_object_agg_unique, json_object_agg_unique_strict, jsonb_object_agg_unique_strict, string_agg, and xmlagg, as well as similar user-defined aggregate functions, produce meaningfully different result values depending on the order of the input values. This ordering is unspecified by default, but can be controlled by writing an ORDER BY clause within the aggregate call, as shown in Section 4.2.7. Alternatively, supplying the input values from a sorted subquery will usually work. For example:

SELECT xmlagg(x) FROM (SELECT x FROM test ORDER BY y DESC) AS tab;

当心如果外部查询级别包含其他处理，如连接，此方法可能会失败，因为这可能导致在计算聚合之前重新对子查询的输出进行排序。

Beware that this approach can fail if the outer query level contains additional processing, such as a join, because that might cause the subquery’s output to be reordered before the aggregate is computed.

Note

布尔聚合 bool_and 和 bool_or 对应于标准 SQL 聚合 every 和 any 或 some。PostgreSQL 支持 every，但不支持 any 或 some，因为标准语法中存在歧义：

The boolean aggregates bool_and and bool_or correspond to the standard SQL aggregates every and any or some. PostgreSQL supports every, but not any or some, because there is an ambiguity built into the standard syntax:

SELECT b1 = ANY((SELECT b2 FROM t2 ...)) FROM t1 ...;

在此，ANY 可被认为要么是引入一个子查询，要么是作为聚合函数，如果子查询返回一行具有布尔值的查询。因此，无法向这些聚合赋予标准名称。

Here ANY can be considered either as introducing a subquery, or as being an aggregate function, if the subquery returns one row with a Boolean value. Thus the standard name cannot be given to these aggregates.

Note

习惯于使用其他 SQL 数据库管理系统的用户可能会对将 count 聚合应用于整个表时的性能感到失望。例如：

Users accustomed to working with other SQL database management systems might be disappointed by the performance of the count aggregate when it is applied to the entire table. A query like:

SELECT count(*) FROM sometable;

需要与表大小成比例的 esforço：PostgreSQL 将需要扫描整个表或某个索引的全部，其中包括表中的所有行。

will require effort proportional to the size of the table: PostgreSQL will need to scan either the entire table or the entirety of an index that includes all rows in the table.

Table 9.60列出了统计分析中通常使用的聚合函数。（这些函数只是为了避免常见聚合函数的排列混乱而单独列出。）显示为接受 numeric_type_的函数适用于所有类型 _smallint, integer, bigint, numeric, real_和 _double precision。说明中提及 _N_处，指所有输入表达式均非 Null 的输入行数。在所有情况下，如果计算没有意义（例如当 _N_为零时），将返回 null。

Table 9.60 shows aggregate functions typically used in statistical analysis. (These are separated out merely to avoid cluttering the listing of more-commonly-used aggregates.) Functions shown as accepting numeric_type are available for all the types smallint, integer, bigint, numeric, real, and double precision. Where the description mentions N, it means the number of input rows for which all the input expressions are non-null. In all cases, null is returned if the computation is meaningless, for example when N is zero.

Table 9.60. Aggregate Functions for Statistics

Function

Description

Partial Mode

corr ( Y double precision, X double precision ) → double precision

Computes the correlation coefficient.

Yes

covar_pop ( Y double precision, X double precision ) → double precision

Computes the population covariance.

Yes

covar_samp ( Y double precision, X double precision ) → double precision

Computes the sample covariance.

Yes

regr_avgx ( Y double precision, X double precision ) → double precision

Computes the average of the independent variable, sum(_X)/N_.

Yes

regr_avgy ( Y double precision, X double precision ) → double precision

Computes the average of the dependent variable, sum(_Y)/N_.

Yes

regr_count ( Y double precision, X double precision ) → bigint

Computes the number of rows in which both inputs are non-null.

Yes

regr_intercept ( Y double precision, X double precision ) → double precision

Computes the y-intercept of the least-squares-fit linear equation determined by the (X, Y) pairs.

Yes

regr_r2 ( Y double precision, X double precision ) → double precision

Computes the square of the correlation coefficient.

Yes

regr_slope ( Y double precision, X double precision ) → double precision

Computes the slope of the least-squares-fit linear equation determined by the (X, Y) pairs.

Yes

regr_sxx ( Y double precision, X double precision ) → double precision

Computes the “sum of squares” of the independent variable, sum(_X^2) - sum(X)^2/N_.

Yes

regr_sxy ( Y double precision, X double precision ) → double precision

Computes the “sum of products” of independent times dependent variables, sum(_X*Y) - sum(X) * sum(Y)/N_.

Yes

regr_syy ( Y double precision, X double precision ) → double precision

Computes the “sum of squares” of the dependent variable, sum(_Y^2) - sum(Y)^2/N_.

Yes

stddev ( numeric_type ) → double precision for real or double precision, otherwise numeric

This is a historical alias for stddev_samp.

Yes

stddev_pop ( numeric_type ) → double precision for real or double precision, otherwise numeric

Computes the population standard deviation of the input values.

Yes

stddev_samp ( numeric_type ) → double precision for real or double precision, otherwise numeric

Computes the sample standard deviation of the input values.

Yes

variance ( numeric_type ) → double precision for real or double precision, otherwise numeric

This is a historical alias for var_samp.

Yes

var_pop ( numeric_type ) → double precision for real or double precision, otherwise numeric

Computes the population variance of the input values (square of the population standard deviation).

Yes

var_samp ( numeric_type ) → double precision for real or double precision, otherwise numeric

Computes the sample variance of the input values (square of the sample standard deviation).

Yes

Table 9.61列出了一些使用 ordered-set aggregate_语法的聚合函数。这些函数有时称为“逆分布”函数。其聚合输入由 _ORDER BY_引入，这些函数还可以使用未被聚合但仅计算一次的 _direct argument。所有这些函数都忽略其聚合输入中的 null 值。对于那些接受 _fraction_参数的函数，分数值必须介于 0 到 1 之间；否则将引发错误。但是，null _fraction_值只会产生 null 结果。

Table 9.61 shows some aggregate functions that use the ordered-set aggregate syntax. These functions are sometimes referred to as “inverse distribution” functions. Their aggregated input is introduced by ORDER BY, and they may also take a direct argument that is not aggregated, but is computed only once. All these functions ignore null values in their aggregated input. For those that take a fraction parameter, the fraction value must be between 0 and 1; an error is thrown if not. However, a null fraction value simply produces a null result.

Table 9.61. Ordered-Set Aggregate Functions

Function

Description

Partial Mode

mode () WITHIN GROUP ( ORDER BY anyelement ) → anyelement

Computes the mode, the most frequent value of the aggregated argument (arbitrarily choosing the first one if there are multiple equally-frequent values). The aggregated argument must be of a sortable type.

percentile_cont ( fraction double precision ) WITHIN GROUP ( ORDER BY double precision ) → double precision

percentile_cont ( fraction double precision ) WITHIN GROUP ( ORDER BY interval ) → interval

Computes the continuous percentile, a value corresponding to the specified fraction within the ordered set of aggregated argument values. This will interpolate between adjacent input items if needed.

percentile_cont ( fractions double precision[] ) WITHIN GROUP ( ORDER BY double precision ) → double precision[]

percentile_cont ( fractions double precision[] ) WITHIN GROUP ( ORDER BY interval ) → interval[]

Computes multiple continuous percentiles. The result is an array of the same dimensions as the fractions parameter, with each non-null element replaced by the (possibly interpolated) value corresponding to that percentile.

percentile_disc ( fraction double precision ) WITHIN GROUP ( ORDER BY anyelement ) → anyelement

Computes the discrete percentile, the first value within the ordered set of aggregated argument values whose position in the ordering equals or exceeds the specified fraction. The aggregated argument must be of a sortable type.

percentile_disc ( fractions double precision[] ) WITHIN GROUP ( ORDER BY anyelement ) → anyarray

Computes multiple discrete percentiles. The result is an array of the same dimensions as the fractions parameter, with each non-null element replaced by the input value corresponding to that percentile. The aggregated argument must be of a sortable type.

Table 9.62中列出的每个“假设集”聚合都与 Section 9.22中定义的同名窗口函数相关联。在每种情况下，聚合结果都是相关窗口函数针对由 _args_构建的“假设”行将返回的值（如果已将这样的行添加到由 _sorted_args_表示的排序行组中）。对于这些函数中的每一个，_args_中提供的直接参数列表必须与 _sorted_args_中提供的聚合参数的数量和类型匹配。与大多数内置聚合不同，这些聚合并非严格的，也就是说，它们不会删除包含 Null 值的输入行。Null 值按照 _ORDER BY_语句中指定的规则进行排序。

Each of the “hypothetical-set” aggregates listed in Table 9.62 is associated with a window function of the same name defined in Section 9.22. In each case, the aggregate’s result is the value that the associated window function would have returned for the “hypothetical” row constructed from args, if such a row had been added to the sorted group of rows represented by the sorted_args. For each of these functions, the list of direct arguments given in args must match the number and types of the aggregated arguments given in sorted_args. Unlike most built-in aggregates, these aggregates are not strict, that is they do not drop input rows containing nulls. Null values sort according to the rule specified in the ORDER BY clause.

Table 9.62. Hypothetical-Set Aggregate Functions

Function

Description

Partial Mode

rank ( args ) WITHIN GROUP ( ORDER BY sorted_args ) → bigint

Computes the rank of the hypothetical row, with gaps; that is, the row number of the first row in its peer group.

dense_rank ( args ) WITHIN GROUP ( ORDER BY sorted_args ) → bigint

Computes the rank of the hypothetical row, without gaps; this function effectively counts peer groups.

percent_rank ( args ) WITHIN GROUP ( ORDER BY sorted_args ) → double precision

Computes the relative rank of the hypothetical row, that is (rank - 1) / (total rows - 1). The value thus ranges from 0 to 1 inclusive.

cume_dist ( args ) WITHIN GROUP ( ORDER BY sorted_args ) → double precision

Computes the cumulative distribution, that is (number of rows preceding or peers with hypothetical row) / (total rows). The value thus ranges from 1/N to 1.

Table 9.63. Grouping Operations

Function

Description

GROUPING ( group_by_expression(s) ) → integer

Returns a bit mask indicating which GROUP BY expressions are not included in the current grouping set. Bits are assigned with the rightmost argument corresponding to the least-significant bit; each bit is 0 if the corresponding expression is included in the grouping criteria of the grouping set generating the current result row, and 1 if it is not included.

Table 9.63中显示的分组操作与分组集（请参见 Section 7.2.4）结合使用，以区分结果行。_GROUPING_函数的参数并未实际求值，但它们必须与相关查询级别的 _GROUP BY_语句中提供的表达式完全匹配。例如：

The grouping operations shown in Table 9.63 are used in conjunction with grouping sets (see Section 7.2.4) to distinguish result rows. The arguments to the GROUPING function are not actually evaluated, but they must exactly match expressions given in the GROUP BY clause of the associated query level. For example:

=> SELECT * FROM items_sold;
 make  | model | sales
-------+-------+-------
 Foo   | GT    |  10
 Foo   | Tour  |  20
 Bar   | City  |  15
 Bar   | Sport |  5
(4 rows)

=> SELECT make, model, GROUPING(make,model), sum(sales) FROM items_sold GROUP BY ROLLUP(make,model);
 make  | model | grouping | sum
-------+-------+----------+-----
 Foo   | GT    |        0 | 10
 Foo   | Tour  |        0 | 20
 Bar   | City  |        0 | 15
 Bar   | Sport |        0 | 5
 Foo   |       |        1 | 30
 Bar   |       |        1 | 20
       |       |        3 | 50
(7 rows)

这里，前四行中的 grouping 值 0 表明它们已按两个分组列正常分组。值 1 表明 model 未在倒数第二行中按它分组，而值 3 表明 make 和 model 均未在最后一行中按它们分组（因此，它是对所有输入行的聚合）。

Here, the grouping value 0 in the first four rows shows that those have been grouped normally, over both the grouping columns. The value 1 indicates that model was not grouped by in the next-to-last two rows, and the value 3 indicates that neither make nor model was grouped by in the last row (which therefore is an aggregate over all the input rows).