1CREATE FUNCTION(7) PostgreSQL 10.7 Documentation CREATE FUNCTION(7)
2
6 CREATE_FUNCTION - define a new function
7
9 CREATE [ OR REPLACE ] FUNCTION
10 name ( [ [ argmode ] [ argname ] argtype [ { DEFAULT | = } default_expr ] [, ...] ] )
11 [ RETURNS rettype
12 | RETURNS TABLE ( column_name column_type [, ...] ) ]
13 { LANGUAGE lang_name
14 | TRANSFORM { FOR TYPE type_name } [, ... ]
15 | WINDOW
16 | IMMUTABLE | STABLE | VOLATILE | [ NOT ] LEAKPROOF
17 | CALLED ON NULL INPUT | RETURNS NULL ON NULL INPUT | STRICT
18 | [ EXTERNAL ] SECURITY INVOKER | [ EXTERNAL ] SECURITY DEFINER
19 | PARALLEL { UNSAFE | RESTRICTED | SAFE }
20 | COST execution_cost
21 | ROWS result_rows
22 | SET configuration_parameter { TO value | = value | FROM CURRENT }
23 | AS 'definition'
24 | AS 'obj_file', 'link_symbol'
25 } ...
26 [ WITH ( attribute [, ...] ) ]
27
29 CREATE FUNCTION defines a new function. CREATE OR REPLACE FUNCTION
30 will either create a new function, or replace an existing definition.
31 To be able to define a function, the user must have the USAGE privilege
32 on the language.
33 If a schema name is included, then the function is created in the
35 specified schema. Otherwise it is created in the current schema. The
36 name of the new function must not match any existing function with the
37 same input argument types in the same schema. However, functions of
38 different argument types can share a name (this is called overloading).
39 To replace the current definition of an existing function, use CREATE
41 OR REPLACE FUNCTION. It is not possible to change the name or argument
42 types of a function this way (if you tried, you would actually be
43 creating a new, distinct function). Also, CREATE OR REPLACE FUNCTION
44 will not let you change the return type of an existing function. To do
45 that, you must drop and recreate the function. (When using OUT
46 parameters, that means you cannot change the types of any OUT
47 parameters except by dropping the function.)
48 When CREATE OR REPLACE FUNCTION is used to replace an existing
50 function, the ownership and permissions of the function do not change.
51 All other function properties are assigned the values specified or
52 implied in the command. You must own the function to replace it (this
53 includes being a member of the owning role).
54 If you drop and then recreate a function, the new function is not the
56 same entity as the old; you will have to drop existing rules, views,
57 triggers, etc. that refer to the old function. Use CREATE OR REPLACE
58 FUNCTION to change a function definition without breaking objects that
59 refer to the function. Also, ALTER FUNCTION can be used to change most
60 of the auxiliary properties of an existing function.
61 The user that creates the function becomes the owner of the function.
63 To be able to create a function, you must have USAGE privilege on the
65 argument types and the return type.
66
68 name
69 The name (optionally schema-qualified) of the function to create.
70 argmode
72 The mode of an argument: IN, OUT, INOUT, or VARIADIC. If omitted,
73 the default is IN. Only OUT arguments can follow a VARIADIC one.
74 Also, OUT and INOUT arguments cannot be used together with the
75 RETURNS TABLE notation.
76 argname
78 The name of an argument. Some languages (including SQL and
79 PL/pgSQL) let you use the name in the function body. For other
80 languages the name of an input argument is just extra
81 documentation, so far as the function itself is concerned; but you
82 can use input argument names when calling a function to improve
83 readability (see Section 4.3). In any case, the name of an output
84 argument is significant, because it defines the column name in the
85 result row type. (If you omit the name for an output argument, the
86 system will choose a default column name.)
87 argtype
89 The data type(s) of the function's arguments (optionally
90 schema-qualified), if any. The argument types can be base,
91 composite, or domain types, or can reference the type of a table
92 column.
93 Depending on the implementation language it might also be allowed
95 to specify “pseudo-types” such as cstring. Pseudo-types indicate
96 that the actual argument type is either incompletely specified, or
97 outside the set of ordinary SQL data types.
98 The type of a column is referenced by writing
100 table_name.column_name%TYPE. Using this feature can sometimes help
101 make a function independent of changes to the definition of a
102 table.
103 default_expr
105 An expression to be used as default value if the parameter is not
106 specified. The expression has to be coercible to the argument type
107 of the parameter. Only input (including INOUT) parameters can have
108 a default value. All input parameters following a parameter with a
109 default value must have default values as well.
110 rettype
112 The return data type (optionally schema-qualified). The return type
113 can be a base, composite, or domain type, or can reference the type
114 of a table column. Depending on the implementation language it
115 might also be allowed to specify “pseudo-types” such as cstring. If
116 the function is not supposed to return a value, specify void as the
117 return type.
118 When there are OUT or INOUT parameters, the RETURNS clause can be
120 omitted. If present, it must agree with the result type implied by
121 the output parameters: RECORD if there are multiple output
122 parameters, or the same type as the single output parameter.
123 The SETOF modifier indicates that the function will return a set of
125 items, rather than a single item.
126 The type of a column is referenced by writing
128 table_name.column_name%TYPE.
129 column_name
131 The name of an output column in the RETURNS TABLE syntax. This is
132 effectively another way of declaring a named OUT parameter, except
133 that RETURNS TABLE also implies RETURNS SETOF.
134 column_type
136 The data type of an output column in the RETURNS TABLE syntax.
137 lang_name
139 The name of the language that the function is implemented in. It
140 can be sql, c, internal, or the name of a user-defined procedural
141 language, e.g. plpgsql. Enclosing the name in single quotes is
142 deprecated and requires matching case.
143 TRANSFORM { FOR TYPE type_name } [, ... ] }
145 Lists which transforms a call to the function should apply.
146 Transforms convert between SQL types and language-specific data
147 types; see CREATE TRANSFORM (CREATE_TRANSFORM(7)). Procedural
148 language implementations usually have hardcoded knowledge of the
149 built-in types, so those don't need to be listed here. If a
150 procedural language implementation does not know how to handle a
151 type and no transform is supplied, it will fall back to a default
152 behavior for converting data types, but this depends on the
153 implementation.
154 WINDOW
156 WINDOW indicates that the function is a window function rather than
157 a plain function. This is currently only useful for functions
158 written in C. The WINDOW attribute cannot be changed when replacing
159 an existing function definition.
160 IMMUTABLE
162 STABLE
163 VOLATILE
164 These attributes inform the query optimizer about the behavior of
165 the function. At most one choice can be specified. If none of these
166 appear, VOLATILE is the default assumption.
167 IMMUTABLE indicates that the function cannot modify the database
169 and always returns the same result when given the same argument
170 values; that is, it does not do database lookups or otherwise use
171 information not directly present in its argument list. If this
172 option is given, any call of the function with all-constant
173 arguments can be immediately replaced with the function value.
174 STABLE indicates that the function cannot modify the database, and
176 that within a single table scan it will consistently return the
177 same result for the same argument values, but that its result could
178 change across SQL statements. This is the appropriate selection for
179 functions whose results depend on database lookups, parameter
180 variables (such as the current time zone), etc. (It is
181 inappropriate for AFTER triggers that wish to query rows modified
182 by the current command.) Also note that the current_timestamp
183 family of functions qualify as stable, since their values do not
184 change within a transaction.
185 VOLATILE indicates that the function value can change even within a
187 single table scan, so no optimizations can be made. Relatively few
188 database functions are volatile in this sense; some examples are
189 random(), currval(), timeofday(). But note that any function that
190 has side-effects must be classified volatile, even if its result is
191 quite predictable, to prevent calls from being optimized away; an
192 example is setval().
193 For additional details see Section 37.6.
195 LEAKPROOF
197 LEAKPROOF indicates that the function has no side effects. It
198 reveals no information about its arguments other than by its return
199 value. For example, a function which throws an error message for
200 some argument values but not others, or which includes the argument
201 values in any error message, is not leakproof. This affects how the
202 system executes queries against views created with the
203 security_barrier option or tables with row level security enabled.
204 The system will enforce conditions from security policies and
205 security barrier views before any user-supplied conditions from the
206 query itself that contain non-leakproof functions, in order to
207 prevent the inadvertent exposure of data. Functions and operators
208 marked as leakproof are assumed to be trustworthy, and may be
209 executed before conditions from security policies and security
210 barrier views. In addition, functions which do not take arguments
211 or which are not passed any arguments from the security barrier
212 view or table do not have to be marked as leakproof to be executed
213 before security conditions. See CREATE VIEW (CREATE_VIEW(7)) and
214 Section 40.5. This option can only be set by the superuser.
215 CALLED ON NULL INPUT
217 RETURNS NULL ON NULL INPUT
218 STRICT
219 CALLED ON NULL INPUT (the default) indicates that the function will
220 be called normally when some of its arguments are null. It is then
221 the function author's responsibility to check for null values if
222 necessary and respond appropriately.
223 RETURNS NULL ON NULL INPUT or STRICT indicates that the function
225 always returns null whenever any of its arguments are null. If this
226 parameter is specified, the function is not executed when there are
227 null arguments; instead a null result is assumed automatically.
228 [EXTERNAL] SECURITY INVOKER
230 [EXTERNAL] SECURITY DEFINER
231 SECURITY INVOKER indicates that the function is to be executed with
232 the privileges of the user that calls it. That is the default.
233 SECURITY DEFINER specifies that the function is to be executed with
234 the privileges of the user that owns it.
235 The key word EXTERNAL is allowed for SQL conformance, but it is
237 optional since, unlike in SQL, this feature applies to all
238 functions not only external ones.
239 PARALLEL
241 PARALLEL UNSAFE indicates that the function can't be executed in
242 parallel mode and the presence of such a function in an SQL
243 statement forces a serial execution plan. This is the default.
244 PARALLEL RESTRICTED indicates that the function can be executed in
245 parallel mode, but the execution is restricted to parallel group
246 leader. PARALLEL SAFE indicates that the function is safe to run
247 in parallel mode without restriction.
248 Functions should be labeled parallel unsafe if they modify any
250 database state, or if they make changes to the transaction such as
251 using sub-transactions, or if they access sequences or attempt to
252 make persistent changes to settings (e.g. setval). They should be
253 labeled as parallel restricted if they access temporary tables,
254 client connection state, cursors, prepared statements, or
255 miscellaneous backend-local state which the system cannot
256 synchronize in parallel mode (e.g. setseed cannot be executed
257 other than by the group leader because a change made by another
258 process would not be reflected in the leader). In general, if a
259 function is labeled as being safe when it is restricted or unsafe,
260 or if it is labeled as being restricted when it is in fact unsafe,
261 it may throw errors or produce wrong answers when used in a
262 parallel query. C-language functions could in theory exhibit
263 totally undefined behavior if mislabeled, since there is no way for
264 the system to protect itself against arbitrary C code, but in most
265 likely cases the result will be no worse than for any other
266 function. If in doubt, functions should be labeled as UNSAFE, which
267 is the default.
268 execution_cost
270 A positive number giving the estimated execution cost for the
271 function, in units of cpu_operator_cost. If the function returns a
272 set, this is the cost per returned row. If the cost is not
273 specified, 1 unit is assumed for C-language and internal functions,
274 and 100 units for functions in all other languages. Larger values
275 cause the planner to try to avoid evaluating the function more
276 often than necessary.
277 result_rows
279 A positive number giving the estimated number of rows that the
280 planner should expect the function to return. This is only allowed
281 when the function is declared to return a set. The default
282 assumption is 1000 rows.
283 configuration_parameter
285 value
286 The SET clause causes the specified configuration parameter to be
287 set to the specified value when the function is entered, and then
288 restored to its prior value when the function exits. SET FROM
289 CURRENT saves the value of the parameter that is current when
290 CREATE FUNCTION is executed as the value to be applied when the
291 function is entered.
292 If a SET clause is attached to a function, then the effects of a
294 SET LOCAL command executed inside the function for the same
295 variable are restricted to the function: the configuration
296 parameter's prior value is still restored at function exit.
297 However, an ordinary SET command (without LOCAL) overrides the SET
298 clause, much as it would do for a previous SET LOCAL command: the
299 effects of such a command will persist after function exit, unless
300 the current transaction is rolled back.
301 See SET(7) and Chapter 19 for more information about allowed
303 parameter names and values.
304 definition
306 A string constant defining the function; the meaning depends on the
307 language. It can be an internal function name, the path to an
308 object file, an SQL command, or text in a procedural language.
309 It is often helpful to use dollar quoting (see Section 4.1.2.4) to
311 write the function definition string, rather than the normal single
312 quote syntax. Without dollar quoting, any single quotes or
313 backslashes in the function definition must be escaped by doubling
314 them.
315 obj_file, link_symbol
317 This form of the AS clause is used for dynamically loadable C
318 language functions when the function name in the C language source
319 code is not the same as the name of the SQL function. The string
320 obj_file is the name of the shared library file containing the
321 compiled C function, and is interpreted as for the LOAD(7) command.
322 The string link_symbol is the function's link symbol, that is, the
323 name of the function in the C language source code. If the link
324 symbol is omitted, it is assumed to be the same as the name of the
325 SQL function being defined. The C names of all functions must be
326 different, so you must give overloaded C functions different C
327 names (for example, use the argument types as part of the C names).
328 When repeated CREATE FUNCTION calls refer to the same object file,
330 the file is only loaded once per session. To unload and reload the
331 file (perhaps during development), start a new session.
332 attribute
334 The historical way to specify optional pieces of information about
335 the function. The following attributes can appear here:
336 isStrict
338 Equivalent to STRICT or RETURNS NULL ON NULL INPUT.
339 isCachable
341 isCachable is an obsolete equivalent of IMMUTABLE; it's still
342 accepted for backwards-compatibility reasons.
343 Attribute names are not case-sensitive.
345 Refer to Section 37.3 for further information on writing functions.
347
349 PostgreSQL allows function overloading; that is, the same name can be
350 used for several different functions so long as they have distinct
351 input argument types. Whether or not you use it, this capability
352 entails security precautions when calling functions in databases where
353 some users mistrust other users; see Section 10.3.
354 Two functions are considered the same if they have the same names and
356 input argument types, ignoring any OUT parameters. Thus for example
357 these declarations conflict:
358 CREATE FUNCTION foo(int) ...
360 CREATE FUNCTION foo(int, out text) ...
361 Functions that have different argument type lists will not be
363 considered to conflict at creation time, but if defaults are provided
364 they might conflict in use. For example, consider
365 CREATE FUNCTION foo(int) ...
367 CREATE FUNCTION foo(int, int default 42) ...
368 A call foo(10) will fail due to the ambiguity about which function
370 should be called.
371
373 The full SQL type syntax is allowed for declaring a function's
374 arguments and return value. However, parenthesized type modifiers
375 (e.g., the precision field for type numeric) are discarded by CREATE
376 FUNCTION. Thus for example CREATE FUNCTION foo (varchar(10)) ... is
377 exactly the same as CREATE FUNCTION foo (varchar) ....
378 When replacing an existing function with CREATE OR REPLACE FUNCTION,
380 there are restrictions on changing parameter names. You cannot change
381 the name already assigned to any input parameter (although you can add
382 names to parameters that had none before). If there is more than one
383 output parameter, you cannot change the names of the output parameters,
384 because that would change the column names of the anonymous composite
385 type that describes the function's result. These restrictions are made
386 to ensure that existing calls of the function do not stop working when
387 it is replaced.
388 If a function is declared STRICT with a VARIADIC argument, the
390 strictness check tests that the variadic array as a whole is non-null.
391 The function will still be called if the array has null elements.
392
394 Here are some trivial examples to help you get started. For more
395 information and examples, see Section 37.3.
396 CREATE FUNCTION add(integer, integer) RETURNS integer
398 AS 'select $1 + $2;'
399 LANGUAGE SQL
400 IMMUTABLE
401 RETURNS NULL ON NULL INPUT;
402 Increment an integer, making use of an argument name, in PL/pgSQL:
404 CREATE OR REPLACE FUNCTION increment(i integer) RETURNS integer AS $$
406 BEGIN
407 RETURN i + 1;
408 END;
409 $$ LANGUAGE plpgsql;
410 Return a record containing multiple output parameters:
412 CREATE FUNCTION dup(in int, out f1 int, out f2 text)
414 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
415 LANGUAGE SQL;
416 SELECT * FROM dup(42);
418 You can do the same thing more verbosely with an explicitly named
420 composite type:
421 CREATE TYPE dup_result AS (f1 int, f2 text);
423 CREATE FUNCTION dup(int) RETURNS dup_result
425 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
426 LANGUAGE SQL;
427 SELECT * FROM dup(42);
429 Another way to return multiple columns is to use a TABLE function:
431 CREATE FUNCTION dup(int) RETURNS TABLE(f1 int, f2 text)
433 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
434 LANGUAGE SQL;
435 SELECT * FROM dup(42);
437 However, a TABLE function is different from the preceding examples,
439 because it actually returns a set of records, not just one record.
440
442 Because a SECURITY DEFINER function is executed with the privileges of
443 the user that owns it, care is needed to ensure that the function
444 cannot be misused. For security, search_path should be set to exclude
445 any schemas writable by untrusted users. This prevents malicious users
446 from creating objects (e.g., tables, functions, and operators) that
447 mask objects intended to be used by the function. Particularly
448 important in this regard is the temporary-table schema, which is
449 searched first by default, and is normally writable by anyone. A secure
450 arrangement can be obtained by forcing the temporary schema to be
451 searched last. To do this, write pg_temp as the last entry in
452 search_path. This function illustrates safe usage:
453 CREATE FUNCTION check_password(uname TEXT, pass TEXT)
455 RETURNS BOOLEAN AS $$
456 DECLARE passed BOOLEAN;
457 BEGIN
458 SELECT (pwd = $2) INTO passed
459 FROM pwds
460 WHERE username = $1;
461 RETURN passed;
463 END;
464 $$ LANGUAGE plpgsql
465 SECURITY DEFINER
466 -- Set a secure search_path: trusted schema(s), then 'pg_temp'.
467 SET search_path = admin, pg_temp;
468 This function's intention is to access a table admin.pwds. But without
470 the SET clause, or with a SET clause mentioning only admin, the
471 function could be subverted by creating a temporary table named pwds.
472 Before PostgreSQL version 8.3, the SET clause was not available, and so
474 older functions may contain rather complicated logic to save, set, and
475 restore search_path. The SET clause is far easier to use for this
476 purpose.
477 Another point to keep in mind is that by default, execute privilege is
479 granted to PUBLIC for newly created functions (see GRANT(7) for more
480 information). Frequently you will wish to restrict use of a security
481 definer function to only some users. To do that, you must revoke the
482 default PUBLIC privileges and then grant execute privilege selectively.
483 To avoid having a window where the new function is accessible to all,
484 create it and set the privileges within a single transaction. For
485 example:
486 BEGIN;
488 CREATE FUNCTION check_password(uname TEXT, pass TEXT) ... SECURITY DEFINER;
489 REVOKE ALL ON FUNCTION check_password(uname TEXT, pass TEXT) FROM PUBLIC;
490 GRANT EXECUTE ON FUNCTION check_password(uname TEXT, pass TEXT) TO admins;
491 COMMIT;
492
494 A CREATE FUNCTION command is defined in SQL:1999 and later. The
495 PostgreSQL version is similar but not fully compatible. The attributes
496 are not portable, neither are the different available languages.
497 For compatibility with some other database systems, argmode can be
499 written either before or after argname. But only the first way is
500 standard-compliant.
501 For parameter defaults, the SQL standard specifies only the syntax with
503 the DEFAULT key word. The syntax with = is used in T-SQL and Firebird.
504
506 ALTER FUNCTION (ALTER_FUNCTION(7)), DROP FUNCTION (DROP_FUNCTION(7)),
507 GRANT(7), LOAD(7), REVOKE(7)
508PostgreSQL 10.7 2019 CREATE FUNCTION(7)