1CREATE FUNCTION(7) SQL Commands CREATE FUNCTION(7)
2
6 CREATE FUNCTION - define a new function
7
10 CREATE [ OR REPLACE ] FUNCTION
11 name ( [ [ argmode ] [ argname ] argtype [ { DEFAULT | = } defexpr ] [, ...] ] )
12 [ RETURNS rettype
13 | RETURNS TABLE ( colname coltype [, ...] ) ]
14 { LANGUAGE langname
15 | WINDOW
16 | IMMUTABLE | STABLE | VOLATILE
17 | CALLED ON NULL INPUT | RETURNS NULL ON NULL INPUT | STRICT
18 | [ EXTERNAL ] SECURITY INVOKER | [ EXTERNAL ] SECURITY DEFINER
19 | COST execution_cost
20 | ROWS result_rows
21 | SET configuration_parameter { TO value | = value | FROM CURRENT }
22 | AS 'definition'
23 | AS 'obj_file', 'link_symbol'
24 } ...
25 [ WITH ( attribute [, ...] ) ]
26
29 CREATE FUNCTION defines a new function. CREATE OR REPLACE FUNCTION
30 will either create a new function, or replace an existing definition.
31 If a schema name is included, then the function is created in the spec‐
33 ified schema. Otherwise it is created in the current schema. The name
34 of the new function must not match any existing function with the same
35 input argument types in the same schema. However, functions of differ‐
36 ent argument types can share a name (this is called overloading).
37 To replace the current definition of an existing function, use CREATE
39 OR REPLACE FUNCTION. It is not possible to change the name or argument
40 types of a function this way (if you tried, you would actually be cre‐
41 ating a new, distinct function). Also, CREATE OR REPLACE FUNCTION will
42 not let you change the return type of an existing function. To do that,
43 you must drop and recreate the function. (When using OUT parameters,
44 that means you cannot change the names or types of any OUT parameters
45 except by dropping the function.)
46 If you drop and then recreate a function, the new function is not the
48 same entity as the old; you will have to drop existing rules, views,
49 triggers, etc. that refer to the old function. Use CREATE OR REPLACE
50 FUNCTION to change a function definition without breaking objects that
51 refer to the function. Also, ALTER FUNCTION can be used to change most
52 of the auxiliary properties of an existing function.
53 The user that creates the function becomes the owner of the function.
55
57 name The name (optionally schema-qualified) of the function to cre‐
58 ate.
59 argmode
61 The mode of an argument: IN, OUT, INOUT, or VARIADIC. If omit‐
62 ted, the default is IN. Only OUT arguments can follow a VARI‐
63 ADIC one. Also, OUT and INOUT arguments cannot be used together
64 with the RETURNS TABLE notation.
65 argname
67 The name of an argument. Some languages (including PL/pgSQL, but
68 currently not SQL) let you use the name in the function body.
69 For other languages the name of an input argument is just extra
70 documentation. But the name of an output argument is signifi‐
71 cant, since it defines the column name in the result row type.
72 (If you omit the name for an output argument, the system will
73 choose a default column name.)
74 argtype
76 The data type(s) of the function's arguments (optionally schema-
77 qualified), if any. The argument types can be base, composite,
78 or domain types, or can reference the type of a table column.
79 Depending on the implementation language it might also be
81 allowed to specify ``pseudotypes'' such as cstring. Pseudotypes
82 indicate that the actual argument type is either incompletely
83 specified, or outside the set of ordinary SQL data types.
84 The type of a column is referenced by writing tablename.column‐
86 name%TYPE. Using this feature can sometimes help make a func‐
87 tion independent of changes to the definition of a table.
88 defexpr
90 An expression to be used as default value if the parameter is
91 not specified. The expression has to be coercible to the argu‐
92 ment type of the parameter. Only input (including INOUT) param‐
93 eters can have a default value. All input parameters following a
94 parameter with a default value must have default values as well.
95 rettype
97 The return data type (optionally schema-qualified). The return
98 type can be a base, composite, or domain type, or can reference
99 the type of a table column. Depending on the implementation
100 language it might also be allowed to specify ``pseudotypes''
101 such as cstring. If the function is not supposed to return a
102 value, specify void as the return type.
103 When there are OUT or INOUT parameters, the RETURNS clause can
105 be omitted. If present, it must agree with the result type
106 implied by the output parameters: RECORD if there are multiple
107 output parameters, or the same type as the single output parame‐
108 ter.
109 The SETOF modifier indicates that the function will return a set
111 of items, rather than a single item.
112 The type of a column is referenced by writing tablename.column‐
114 name%TYPE.
115 colname
117 The name of an output column in the RETURNS TABLE syntax. This
118 is effectively another way of declaring a named OUT parameter,
119 except that RETURNS TABLE also implies RETURNS SETOF.
120 coltype
122 The data type of an output column in the RETURNS TABLE syntax.
123 langname
125 The name of the language that the function is implemented in.
126 Can be SQL, C, internal, or the name of a user-defined procedur‐
127 al language. For backward compatibility, the name can be
128 enclosed by single quotes.
129 WINDOW WINDOW indicates that the function is a window function rather
131 than a plain function. This is currently only useful for func‐
132 tions written in C. The WINDOW attribute cannot be changed when
133 replacing an existing function definition.
134 IMMUTABLE
136 STABLE
138 VOLATILE
140 These attributes inform the query optimizer about the behavior
141 of the function. At most one choice can be specified. If none of
142 these appear, VOLATILE is the default assumption.
143 IMMUTABLE indicates that the function cannot modify the database
145 and always returns the same result when given the same argument
146 values; that is, it does not do database lookups or otherwise
147 use information not directly present in its argument list. If
148 this option is given, any call of the function with all-constant
149 arguments can be immediately replaced with the function value.
150 STABLE indicates that the function cannot modify the database,
152 and that within a single table scan it will consistently return
153 the same result for the same argument values, but that its
154 result could change across SQL statements. This is the appropri‐
155 ate selection for functions whose results depend on database
156 lookups, parameter variables (such as the current time zone),
157 etc. Also note that the current_timestamp family of functions
158 qualify as stable, since their values do not change within a
159 transaction.
160 VOLATILE indicates that the function value can change even
162 within a single table scan, so no optimizations can be made.
163 Relatively few database functions are volatile in this sense;
164 some examples are random(), currval(), timeofday(). But note
165 that any function that has side-effects must be classified
166 volatile, even if its result is quite predictable, to prevent
167 calls from being optimized away; an example is setval().
168 For additional details see in the documentation.
170 CALLED ON NULL INPUT
172 RETURNS NULL ON NULL INPUT
174 STRICT CALLED ON NULL INPUT (the default) indicates that the function
176 will be called normally when some of its arguments are null. It
177 is then the function author's responsibility to check for null
178 values if necessary and respond appropriately.
179 RETURNS NULL ON NULL INPUT or STRICT indicates that the function
181 always returns null whenever any of its arguments are null. If
182 this parameter is specified, the function is not executed when
183 there are null arguments; instead a null result is assumed auto‐
184 matically.
185 [EXTERNAL] SECURITY INVOKER
187 [EXTERNAL] SECURITY DEFINER
189 SECURITY INVOKER indicates that the function is to be executed
190 with the privileges of the user that calls it. That is the
191 default. SECURITY DEFINER specifies that the function is to be
192 executed with the privileges of the user that created it.
193 The key word EXTERNAL is allowed for SQL conformance, but it is
195 optional since, unlike in SQL, this feature applies to all func‐
196 tions not only external ones.
197 execution_cost
199 A positive number giving the estimated execution cost for the
200 function, in units of cpu_operator_cost. If the function returns
201 a set, this is the cost per returned row. If the cost is not
202 specified, 1 unit is assumed for C-language and internal func‐
203 tions, and 100 units for functions in all other languages.
204 Larger values cause the planner to try to avoid evaluating the
205 function more often than necessary.
206 result_rows
208 A positive number giving the estimated number of rows that the
209 planner should expect the function to return. This is only
210 allowed when the function is declared to return a set. The
211 default assumption is 1000 rows.
212 configuration_parameter
214 value The SET clause causes the specified configuration parameter to
216 be set to the specified value when the function is entered, and
217 then restored to its prior value when the function exits. SET
218 FROM CURRENT saves the session's current value of the parameter
219 as the value to be applied when the function is entered.
220 See SET [set(7)] and in the documentation for more information
222 about allowed parameter names and values.
223 definition
225 A string constant defining the function; the meaning depends on
226 the language. It can be an internal function name, the path to
227 an object file, an SQL command, or text in a procedural lan‐
228 guage.
229 obj_file, link_symbol
231 This form of the AS clause is used for dynamically loadable C
232 language functions when the function name in the C language
233 source code is not the same as the name of the SQL function. The
234 string obj_file is the name of the file containing the dynami‐
235 cally loadable object, and link_symbol is the function's link
236 symbol, that is, the name of the function in the C language
237 source code. If the link symbol is omitted, it is assumed to be
238 the same as the name of the SQL function being defined.
239 attribute
241 The historical way to specify optional pieces of information
242 about the function. The following attributes can appear here:
243 isStrict
245 Equivalent to STRICT or RETURNS NULL ON NULL INPUT.
246 isCachable
248 isCachable is an obsolete equivalent of IMMUTABLE; it's
249 still accepted for backwards-compatibility reasons.
250 Attribute names are not case-sensitive.
252
254 Refer to in the documentation for further information on writing func‐
255 tions.
256 The full SQL type syntax is allowed for declaring a function's argu‐
258 ments and return value. However, parenthesized type modifiers (e.g.,
259 the precision field for type numeric) are discarded by CREATE FUNCTION.
260 Thus for example CREATE FUNCTION foo (varchar(10)) ... is exactly the
261 same as CREATE FUNCTION foo (varchar) ....
262 PostgreSQL allows function overloading; that is, the same name can be
264 used for several different functions so long as they have distinct
265 input argument types. However, the C names of all functions must be
266 different, so you must give overloaded C functions different C names
267 (for example, use the argument types as part of the C names).
268 Two functions are considered the same if they have the same names and
270 input argument types, ignoring any OUT parameters. Thus for example
271 these declarations conflict:
272 CREATE FUNCTION foo(int) ...
274 CREATE FUNCTION foo(int, out text) ...
275 Functions that have different argument type lists will not be consid‐
278 ered to conflict at creation time, but if defaults are provided they
279 might conflict in use. For example, consider
280 CREATE FUNCTION foo(int) ...
282 CREATE FUNCTION foo(int, int default 42) ...
283 A call foo(10) will fail due to the ambiguity about which function
285 should be called.
286 When repeated CREATE FUNCTION calls refer to the same object file, the
288 file is only loaded once per session. To unload and reload the file
289 (perhaps during development), start a new session.
290 Use DROP FUNCTION [drop_function(7)] to remove user-defined functions.
292 It is often helpful to use dollar quoting (see in the documentation) to
294 write the function definition string, rather than the normal single
295 quote syntax. Without dollar quoting, any single quotes or backslashes
296 in the function definition must be escaped by doubling them.
297 If a SET clause is attached to a function, then the effects of a SET
299 LOCAL command executed inside the function for the same variable are
300 restricted to the function: the configuration parameter's prior value
301 is still restored at function exit. However, an ordinary SET command
302 (without LOCAL) overrides the SET clause, much as it would do for a
303 previous SET LOCAL command: the effects of such a command will persist
304 after function exit, unless the current transaction is rolled back.
305 To be able to define a function, the user must have the USAGE privilege
307 on the language.
308 When CREATE OR REPLACE FUNCTION is used to replace an existing func‐
310 tion, the ownership and permissions of the function do not change. All
311 other function properties are assigned the values specified or implied
312 in the command. You must own the function to replace it (this includes
313 being a member of the owning role).
314 If a function is declared STRICT with a VARIADIC argument, the strict‐
316 ness check tests that the variadic array as a whole is non-null. The
317 function will still be called if the array has null elements.
318
320 Here are some trivial examples to help you get started. For more infor‐
321 mation and examples, see in the documentation.
322 CREATE FUNCTION add(integer, integer) RETURNS integer
324 AS 'select $1 + $2;'
325 LANGUAGE SQL
326 IMMUTABLE
327 RETURNS NULL ON NULL INPUT;
328 Increment an integer, making use of an argument name, in PL/pgSQL:
331 CREATE OR REPLACE FUNCTION increment(i integer) RETURNS integer AS $$
333 BEGIN
334 RETURN i + 1;
335 END;
336 $$ LANGUAGE plpgsql;
337 Return a record containing multiple output parameters:
340 CREATE FUNCTION dup(in int, out f1 int, out f2 text)
342 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
343 LANGUAGE SQL;
344 SELECT * FROM dup(42);
346 You can do the same thing more verbosely with an explicitly named com‐
348 posite type:
349 CREATE TYPE dup_result AS (f1 int, f2 text);
351 CREATE FUNCTION dup(int) RETURNS dup_result
353 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
354 LANGUAGE SQL;
355 SELECT * FROM dup(42);
357 Another way to return multiple columns is to use a TABLE function:
359 CREATE FUNCTION dup(int) RETURNS TABLE(f1 int, f2 text)
361 AS $$ SELECT $1, CAST($1 AS text) || ' is text' $$
362 LANGUAGE SQL;
363 SELECT * FROM dup(42);
365 However, a TABLE function is different from the preceding examples,
367 because it actually returns a set of records, not just one record.
368
370 Because a SECURITY DEFINER function is executed with the privileges of
371 the user that created it, care is needed to ensure that the function
372 cannot be misused. For security, search_path should be set to exclude
373 any schemas writable by untrusted users. This prevents malicious users
374 from creating objects that mask objects used by the function. Particu‐
375 larly important in this regard is the temporary-table schema, which is
376 searched first by default, and is normally writable by anyone. A secure
377 arrangement can be had by forcing the temporary schema to be searched
378 last. To do this, write pg_temp as the last entry in search_path. This
379 function illustrates safe usage:
380 CREATE FUNCTION check_password(uname TEXT, pass TEXT)
382 RETURNS BOOLEAN AS $$
383 DECLARE passed BOOLEAN;
384 BEGIN
385 SELECT (pwd = $2) INTO passed
386 FROM pwds
387 WHERE username = $1;
388 RETURN passed;
390 END;
391 $$ LANGUAGE plpgsql
392 SECURITY DEFINER
393 -- Set a secure search_path: trusted schema(s), then 'pg_temp'.
394 SET search_path = admin, pg_temp;
395 Before PostgreSQL version 8.3, the SET option was not available, and so
398 older functions may contain rather complicated logic to save, set, and
399 restore search_path. The SET option is far easier to use for this pur‐
400 pose.
401 Another point to keep in mind is that by default, execute privilege is
403 granted to PUBLIC for newly created functions (see GRANT [grant(7)] for
404 more information). Frequently you will wish to restrict use of a secu‐
405 rity definer function to only some users. To do that, you must revoke
406 the default PUBLIC privileges and then grant execute privilege selec‐
407 tively. To avoid having a window where the new function is accessible
408 to all, create it and set the privileges within a single transaction.
409 For example:
410 BEGIN;
412 CREATE FUNCTION check_password(uname TEXT, pass TEXT) ... SECURITY DEFINER;
413 REVOKE ALL ON FUNCTION check_password(uname TEXT, pass TEXT) FROM PUBLIC;
414 GRANT EXECUTE ON FUNCTION check_password(uname TEXT, pass TEXT) TO admins;
415 COMMIT;
416
419 A CREATE FUNCTION command is defined in SQL:1999 and later. The Post‐
420 greSQL version is similar but not fully compatible. The attributes are
421 not portable, neither are the different available languages.
422 For compatibility with some other database systems, argmode can be
424 written either before or after argname. But only the first way is
425 standard-compliant.
426 The SQL standard does not specify parameter defaults. The syntax with
428 the DEFAULT key word is from Oracle, and it is somewhat in the spirit
429 of the standard: SQL/PSM uses it for variable default values. The syn‐
430 tax with = is used in T-SQL and Firebird.
431
433 ALTER FUNCTION [alter_function(7)], DROP FUNCTION [drop_function(7)],
434 GRANT [grant(7)], LOAD [load(7)], REVOKE [revoke(7)], createlang [cre‐
435 atelang(1)]
436SQL - Language Statements 2014-02-17 CREATE FUNCTION(7)