1CREATE FUNCTION(7) PostgreSQL 16.1 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 }
17 | [ NOT ] LEAKPROOF
18 | { CALLED ON NULL INPUT | RETURNS NULL ON NULL INPUT | STRICT }
19 | { [ EXTERNAL ] SECURITY INVOKER | [ EXTERNAL ] SECURITY DEFINER }
20 | PARALLEL { UNSAFE | RESTRICTED | SAFE }
21 | COST execution_cost
22 | ROWS result_rows
23 | SUPPORT support_function
24 | SET configuration_parameter { TO value | = value | FROM CURRENT }
25 | AS 'definition'
26 | AS 'obj_file', 'link_symbol'
27 | sql_body
28 } ...
29
31 CREATE FUNCTION defines a new function. CREATE OR REPLACE FUNCTION
32 will either create a new function, or replace an existing definition.
33 To be able to define a function, the user must have the USAGE privilege
34 on the language.
35 If a schema name is included, then the function is created in the
37 specified schema. Otherwise it is created in the current schema. The
38 name of the new function must not match any existing function or
39 procedure with the same input argument types in the same schema.
40 However, functions and procedures of different argument types can share
41 a name (this is called overloading).
42 To replace the current definition of an existing function, use CREATE
44 OR REPLACE FUNCTION. It is not possible to change the name or argument
45 types of a function this way (if you tried, you would actually be
46 creating a new, distinct function). Also, CREATE OR REPLACE FUNCTION
47 will not let you change the return type of an existing function. To do
48 that, you must drop and recreate the function. (When using OUT
49 parameters, that means you cannot change the types of any OUT
50 parameters except by dropping the function.)
51 When CREATE OR REPLACE FUNCTION is used to replace an existing
53 function, the ownership and permissions of the function do not change.
54 All other function properties are assigned the values specified or
55 implied in the command. You must own the function to replace it (this
56 includes being a member of the owning role).
57 If you drop and then recreate a function, the new function is not the
59 same entity as the old; you will have to drop existing rules, views,
60 triggers, etc. that refer to the old function. Use CREATE OR REPLACE
61 FUNCTION to change a function definition without breaking objects that
62 refer to the function. Also, ALTER FUNCTION can be used to change most
63 of the auxiliary properties of an existing function.
64 The user that creates the function becomes the owner of the function.
66 To be able to create a function, you must have USAGE privilege on the
68 argument types and the return type.
69 Refer to Section 38.3 for further information on writing functions.
71
73 name
74 The name (optionally schema-qualified) of the function to create.
75 argmode
77 The mode of an argument: IN, OUT, INOUT, or VARIADIC. If omitted,
78 the default is IN. Only OUT arguments can follow a VARIADIC one.
79 Also, OUT and INOUT arguments cannot be used together with the
80 RETURNS TABLE notation.
81 argname
83 The name of an argument. Some languages (including SQL and
84 PL/pgSQL) let you use the name in the function body. For other
85 languages the name of an input argument is just extra
86 documentation, so far as the function itself is concerned; but you
87 can use input argument names when calling a function to improve
88 readability (see Section 4.3). In any case, the name of an output
89 argument is significant, because it defines the column name in the
90 result row type. (If you omit the name for an output argument, the
91 system will choose a default column name.)
92 argtype
94 The data type(s) of the function's arguments (optionally
95 schema-qualified), if any. The argument types can be base,
96 composite, or domain types, or can reference the type of a table
97 column.
98 Depending on the implementation language it might also be allowed
100 to specify “pseudo-types” such as cstring. Pseudo-types indicate
101 that the actual argument type is either incompletely specified, or
102 outside the set of ordinary SQL data types.
103 The type of a column is referenced by writing
105 table_name.column_name%TYPE. Using this feature can sometimes help
106 make a function independent of changes to the definition of a
107 table.
108 default_expr
110 An expression to be used as default value if the parameter is not
111 specified. The expression has to be coercible to the argument type
112 of the parameter. Only input (including INOUT) parameters can have
113 a default value. All input parameters following a parameter with a
114 default value must have default values as well.
115 rettype
117 The return data type (optionally schema-qualified). The return type
118 can be a base, composite, or domain type, or can reference the type
119 of a table column. Depending on the implementation language it
120 might also be allowed to specify “pseudo-types” such as cstring. If
121 the function is not supposed to return a value, specify void as the
122 return type.
123 When there are OUT or INOUT parameters, the RETURNS clause can be
125 omitted. If present, it must agree with the result type implied by
126 the output parameters: RECORD if there are multiple output
127 parameters, or the same type as the single output parameter.
128 The SETOF modifier indicates that the function will return a set of
130 items, rather than a single item.
131 The type of a column is referenced by writing
133 table_name.column_name%TYPE.
134 column_name
136 The name of an output column in the RETURNS TABLE syntax. This is
137 effectively another way of declaring a named OUT parameter, except
138 that RETURNS TABLE also implies RETURNS SETOF.
139 column_type
141 The data type of an output column in the RETURNS TABLE syntax.
142 lang_name
144 The name of the language that the function is implemented in. It
145 can be sql, c, internal, or the name of a user-defined procedural
146 language, e.g., plpgsql. The default is sql if sql_body is
147 specified. Enclosing the name in single quotes is deprecated and
148 requires matching case.
149 TRANSFORM { FOR TYPE type_name } [, ... ] }
151 Lists which transforms a call to the function should apply.
152 Transforms convert between SQL types and language-specific data
153 types; see CREATE TRANSFORM (CREATE_TRANSFORM(7)). Procedural
154 language implementations usually have hardcoded knowledge of the
155 built-in types, so those don't need to be listed here. If a
156 procedural language implementation does not know how to handle a
157 type and no transform is supplied, it will fall back to a default
158 behavior for converting data types, but this depends on the
159 implementation.
160 WINDOW
162 WINDOW indicates that the function is a window function rather than
163 a plain function. This is currently only useful for functions
164 written in C. The WINDOW attribute cannot be changed when replacing
165 an existing function definition.
166 IMMUTABLE
168 STABLE
169 VOLATILE
170 These attributes inform the query optimizer about the behavior of
171 the function. At most one choice can be specified. If none of these
172 appear, VOLATILE is the default assumption.
173 IMMUTABLE indicates that the function cannot modify the database
175 and always returns the same result when given the same argument
176 values; that is, it does not do database lookups or otherwise use
177 information not directly present in its argument list. If this
178 option is given, any call of the function with all-constant
179 arguments can be immediately replaced with the function value.
180 STABLE indicates that the function cannot modify the database, and
182 that within a single table scan it will consistently return the
183 same result for the same argument values, but that its result could
184 change across SQL statements. This is the appropriate selection for
185 functions whose results depend on database lookups, parameter
186 variables (such as the current time zone), etc. (It is
187 inappropriate for AFTER triggers that wish to query rows modified
188 by the current command.) Also note that the current_timestamp
189 family of functions qualify as stable, since their values do not
190 change within a transaction.
191 VOLATILE indicates that the function value can change even within a
193 single table scan, so no optimizations can be made. Relatively few
194 database functions are volatile in this sense; some examples are
195 random(), currval(), timeofday(). But note that any function that
196 has side-effects must be classified volatile, even if its result is
197 quite predictable, to prevent calls from being optimized away; an
198 example is setval().
199 For additional details see Section 38.7.
201 LEAKPROOF
203 LEAKPROOF indicates that the function has no side effects. It
204 reveals no information about its arguments other than by its return
205 value. For example, a function which throws an error message for
206 some argument values but not others, or which includes the argument
207 values in any error message, is not leakproof. This affects how the
208 system executes queries against views created with the
209 security_barrier option or tables with row level security enabled.
210 The system will enforce conditions from security policies and
211 security barrier views before any user-supplied conditions from the
212 query itself that contain non-leakproof functions, in order to
213 prevent the inadvertent exposure of data. Functions and operators
214 marked as leakproof are assumed to be trustworthy, and may be
215 executed before conditions from security policies and security
216 barrier views. In addition, functions which do not take arguments
217 or which are not passed any arguments from the security barrier
218 view or table do not have to be marked as leakproof to be executed
219 before security conditions. See CREATE VIEW (CREATE_VIEW(7)) and
220 Section 41.5. This option can only be set by the superuser.
221 CALLED ON NULL INPUT
223 RETURNS NULL ON NULL INPUT
224 STRICT
225 CALLED ON NULL INPUT (the default) indicates that the function will
226 be called normally when some of its arguments are null. It is then
227 the function author's responsibility to check for null values if
228 necessary and respond appropriately.
229 RETURNS NULL ON NULL INPUT or STRICT indicates that the function
231 always returns null whenever any of its arguments are null. If this
232 parameter is specified, the function is not executed when there are
233 null arguments; instead a null result is assumed automatically.
234 [EXTERNAL] SECURITY INVOKER
236 [EXTERNAL] SECURITY DEFINER
237 SECURITY INVOKER indicates that the function is to be executed with
238 the privileges of the user that calls it. That is the default.
239 SECURITY DEFINER specifies that the function is to be executed with
240 the privileges of the user that owns it. For information on how to
241 write SECURITY DEFINER functions safely, see below.
242 The key word EXTERNAL is allowed for SQL conformance, but it is
244 optional since, unlike in SQL, this feature applies to all
245 functions not only external ones.
246 PARALLEL
248 PARALLEL UNSAFE indicates that the function can't be executed in
249 parallel mode and the presence of such a function in an SQL
250 statement forces a serial execution plan. This is the default.
251 PARALLEL RESTRICTED indicates that the function can be executed in
252 parallel mode, but the execution is restricted to parallel group
253 leader. PARALLEL SAFE indicates that the function is safe to run
254 in parallel mode without restriction.
255 Functions should be labeled parallel unsafe if they modify any
257 database state, or if they make changes to the transaction such as
258 using sub-transactions, or if they access sequences or attempt to
259 make persistent changes to settings (e.g., setval). They should be
260 labeled as parallel restricted if they access temporary tables,
261 client connection state, cursors, prepared statements, or
262 miscellaneous backend-local state which the system cannot
263 synchronize in parallel mode (e.g., setseed cannot be executed
264 other than by the group leader because a change made by another
265 process would not be reflected in the leader). In general, if a
266 function is labeled as being safe when it is restricted or unsafe,
267 or if it is labeled as being restricted when it is in fact unsafe,
268 it may throw errors or produce wrong answers when used in a
269 parallel query. C-language functions could in theory exhibit
270 totally undefined behavior if mislabeled, since there is no way for
271 the system to protect itself against arbitrary C code, but in most
272 likely cases the result will be no worse than for any other
273 function. If in doubt, functions should be labeled as UNSAFE, which
274 is the default.
275 COST execution_cost
277 A positive number giving the estimated execution cost for the
278 function, in units of cpu_operator_cost. If the function returns a
279 set, this is the cost per returned row. If the cost is not
280 specified, 1 unit is assumed for C-language and internal functions,
281 and 100 units for functions in all other languages. Larger values
282 cause the planner to try to avoid evaluating the function more
283 often than necessary.
284 ROWS result_rows
286 A positive number giving the estimated number of rows that the
287 planner should expect the function to return. This is only allowed
288 when the function is declared to return a set. The default
289 assumption is 1000 rows.
290 SUPPORT support_function
292 The name (optionally schema-qualified) of a planner support
293 function to use for this function. See Section 38.11 for details.
294 You must be superuser to use this option.
295 configuration_parameter
297 value
298 The SET clause causes the specified configuration parameter to be
299 set to the specified value when the function is entered, and then
300 restored to its prior value when the function exits. SET FROM
301 CURRENT saves the value of the parameter that is current when
302 CREATE FUNCTION is executed as the value to be applied when the
303 function is entered.
304 If a SET clause is attached to a function, then the effects of a
306 SET LOCAL command executed inside the function for the same
307 variable are restricted to the function: the configuration
308 parameter's prior value is still restored at function exit.
309 However, an ordinary SET command (without LOCAL) overrides the SET
310 clause, much as it would do for a previous SET LOCAL command: the
311 effects of such a command will persist after function exit, unless
312 the current transaction is rolled back.
313 See SET(7) and Chapter 20 for more information about allowed
315 parameter names and values.
316 definition
318 A string constant defining the function; the meaning depends on the
319 language. It can be an internal function name, the path to an
320 object file, an SQL command, or text in a procedural language.
321 It is often helpful to use dollar quoting (see Section 4.1.2.4) to
323 write the function definition string, rather than the normal single
324 quote syntax. Without dollar quoting, any single quotes or
325 backslashes in the function definition must be escaped by doubling
326 them.
327 obj_file, link_symbol
329 This form of the AS clause is used for dynamically loadable C
330 language functions when the function name in the C language source
331 code is not the same as the name of the SQL function. The string
332 obj_file is the name of the shared library file containing the
333 compiled C function, and is interpreted as for the LOAD command.
334 The string link_symbol is the function's link symbol, that is, the
335 name of the function in the C language source code. If the link
336 symbol is omitted, it is assumed to be the same as the name of the
337 SQL function being defined. The C names of all functions must be
338 different, so you must give overloaded C functions different C
339 names (for example, use the argument types as part of the C names).
340 When repeated CREATE FUNCTION calls refer to the same object file,
342 the file is only loaded once per session. To unload and reload the
343 file (perhaps during development), start a new session.
344 sql_body
346 The body of a LANGUAGE SQL function. This can either be a single
347 statement
348 RETURN expression
350 or a block
352 BEGIN ATOMIC
354 statement;
355 statement;
356 ...
357 statement;
358 END
359 This is similar to writing the text of the function body as a
361 string constant (see definition above), but there are some
362 differences: This form only works for LANGUAGE SQL, the string
363 constant form works for all languages. This form is parsed at
364 function definition time, the string constant form is parsed at
365 execution time; therefore this form cannot support polymorphic
366 argument types and other constructs that are not resolvable at
367 function definition time. This form tracks dependencies between the
368 function and objects used in the function body, so DROP ... CASCADE
369 will work correctly, whereas the form using string literals may
370 leave dangling functions. Finally, this form is more compatible
371 with the SQL standard and other SQL implementations.
372
374 PostgreSQL allows function overloading; that is, the same name can be
375 used for several different functions so long as they have distinct
376 input argument types. Whether or not you use it, this capability
377 entails security precautions when calling functions in databases where
378 some users mistrust other users; see Section 10.3.
379 Two functions are considered the same if they have the same names and
381 input argument types, ignoring any OUT parameters. Thus for example
382 these declarations conflict:
383 CREATE FUNCTION foo(int) ...
385 CREATE FUNCTION foo(int, out text) ...
386 Functions that have different argument type lists will not be
388 considered to conflict at creation time, but if defaults are provided
389 they might conflict in use. For example, consider
390 CREATE FUNCTION foo(int) ...
392 CREATE FUNCTION foo(int, int default 42) ...
393 A call foo(10) will fail due to the ambiguity about which function
395 should be called.
396
398 The full SQL type syntax is allowed for declaring a function's
399 arguments and return value. However, parenthesized type modifiers
400 (e.g., the precision field for type numeric) are discarded by CREATE
401 FUNCTION. Thus for example CREATE FUNCTION foo (varchar(10)) ... is
402 exactly the same as CREATE FUNCTION foo (varchar) ....
403 When replacing an existing function with CREATE OR REPLACE FUNCTION,
405 there are restrictions on changing parameter names. You cannot change
406 the name already assigned to any input parameter (although you can add
407 names to parameters that had none before). If there is more than one
408 output parameter, you cannot change the names of the output parameters,
409 because that would change the column names of the anonymous composite
410 type that describes the function's result. These restrictions are made
411 to ensure that existing calls of the function do not stop working when
412 it is replaced.
413 If a function is declared STRICT with a VARIADIC argument, the
415 strictness check tests that the variadic array as a whole is non-null.
416 The function will still be called if the array has null elements.
417
419 Add two integers using an SQL function:
420 CREATE FUNCTION add(integer, integer) RETURNS integer
422 AS 'select $1 + $2;'
423 LANGUAGE SQL
424 IMMUTABLE
425 RETURNS NULL ON NULL INPUT;
426 The same function written in a more SQL-conforming style, using
428 argument names and an unquoted body:
429 CREATE FUNCTION add(a integer, b integer) RETURNS integer
431 LANGUAGE SQL
432 IMMUTABLE
433 RETURNS NULL ON NULL INPUT
434 RETURN a + b;
435 Increment an integer, making use of an argument name, in PL/pgSQL:
437 CREATE OR REPLACE FUNCTION increment(i integer) RETURNS integer AS $$
439 BEGIN
440 RETURN i + 1;
441 END;
442 $$ LANGUAGE plpgsql;
443 Return a record containing multiple output parameters:
445 CREATE FUNCTION dup(in int, out f1 int, out f2 text)
447 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
448 LANGUAGE SQL;
449 SELECT * FROM dup(42);
451 You can do the same thing more verbosely with an explicitly named
453 composite type:
454 CREATE TYPE dup_result AS (f1 int, f2 text);
456 CREATE FUNCTION dup(int) RETURNS dup_result
458 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
459 LANGUAGE SQL;
460 SELECT * FROM dup(42);
462 Another way to return multiple columns is to use a TABLE function:
464 CREATE FUNCTION dup(int) RETURNS TABLE(f1 int, f2 text)
466 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
467 LANGUAGE SQL;
468 SELECT * FROM dup(42);
470 However, a TABLE function is different from the preceding examples,
472 because it actually returns a set of records, not just one record.
473
475 Because a SECURITY DEFINER function is executed with the privileges of
476 the user that owns it, care is needed to ensure that the function
477 cannot be misused. For security, search_path should be set to exclude
478 any schemas writable by untrusted users. This prevents malicious users
479 from creating objects (e.g., tables, functions, and operators) that
480 mask objects intended to be used by the function. Particularly
481 important in this regard is the temporary-table schema, which is
482 searched first by default, and is normally writable by anyone. A secure
483 arrangement can be obtained by forcing the temporary schema to be
484 searched last. To do this, write pg_temp as the last entry in
485 search_path. This function illustrates safe usage:
486 CREATE FUNCTION check_password(uname TEXT, pass TEXT)
488 RETURNS BOOLEAN AS $$
489 DECLARE passed BOOLEAN;
490 BEGIN
491 SELECT (pwd = $2) INTO passed
492 FROM pwds
493 WHERE username = $1;
494 RETURN passed;
496 END;
497 $$ LANGUAGE plpgsql
498 SECURITY DEFINER
499 -- Set a secure search_path: trusted schema(s), then 'pg_temp'.
500 SET search_path = admin, pg_temp;
501 This function's intention is to access a table admin.pwds. But without
503 the SET clause, or with a SET clause mentioning only admin, the
504 function could be subverted by creating a temporary table named pwds.
505 If the security definer function intends to create roles, and if it is
507 running as a non-superuser, createrole_self_grant should also be set to
508 a known value using the SET clause.
509 Another point to keep in mind is that by default, execute privilege is
511 granted to PUBLIC for newly created functions (see Section 5.7 for more
512 information). Frequently you will wish to restrict use of a security
513 definer function to only some users. To do that, you must revoke the
514 default PUBLIC privileges and then grant execute privilege selectively.
515 To avoid having a window where the new function is accessible to all,
516 create it and set the privileges within a single transaction. For
517 example:
518 BEGIN;
520 CREATE FUNCTION check_password(uname TEXT, pass TEXT) ... SECURITY DEFINER;
521 REVOKE ALL ON FUNCTION check_password(uname TEXT, pass TEXT) FROM PUBLIC;
522 GRANT EXECUTE ON FUNCTION check_password(uname TEXT, pass TEXT) TO admins;
523 COMMIT;
524
526 A CREATE FUNCTION command is defined in the SQL standard. The
527 PostgreSQL implementation can be used in a compatible way but has many
528 extensions. Conversely, the SQL standard specifies a number of optional
529 features that are not implemented in PostgreSQL.
530 The following are important compatibility issues:
532 • OR REPLACE is a PostgreSQL extension.
534 • For compatibility with some other database systems, argmode can be
536 written either before or after argname. But only the first way is
537 standard-compliant.
538 • For parameter defaults, the SQL standard specifies only the syntax
540 with the DEFAULT key word. The syntax with = is used in T-SQL and
541 Firebird.
542 • The SETOF modifier is a PostgreSQL extension.
544 • Only SQL is standardized as a language.
546 • All other attributes except CALLED ON NULL INPUT and RETURNS NULL
548 ON NULL INPUT are not standardized.
549 • For the body of LANGUAGE SQL functions, the SQL standard only
551 specifies the sql_body form.
552 Simple LANGUAGE SQL functions can be written in a way that is both
554 standard-conforming and portable to other implementations. More complex
555 functions using advanced features, optimization attributes, or other
556 languages will necessarily be specific to PostgreSQL in a significant
557 way.
558
560 ALTER FUNCTION (ALTER_FUNCTION(7)), DROP FUNCTION (DROP_FUNCTION(7)),
561 GRANT(7), LOAD(7), REVOKE(7)
562PostgreSQL 16.1 2023 CREATE FUNCTION(7)