1 <!-- $PostgreSQL: pgsql/doc/src/sgml/plpgsql.sgml,v 1.149 2009/12/28 19:11:51 petere Exp $ -->
4 <title><application>PL/pgSQL</application> - <acronym>SQL</acronym> Procedural Language</title>
6 <indexterm zone="plpgsql">
7 <primary>PL/pgSQL</primary>
10 <sect1 id="plpgsql-overview">
11 <title>Overview</title>
14 <application>PL/pgSQL</application> is a loadable procedural
15 language for the <productname>PostgreSQL</productname> database
16 system. The design goals of <application>PL/pgSQL</> were to create
17 a loadable procedural language that
22 can be used to create functions and trigger procedures,
27 adds control structures to the <acronym>SQL</acronym> language,
32 can perform complex computations,
37 inherits all user-defined types, functions, and operators,
42 can be defined to be trusted by the server,
54 Functions created with <application>PL/pgSQL</application> can be
55 used anywhere that built-in functions could be used.
56 For example, it is possible to
57 create complex conditional computation functions and later use
58 them to define operators or use them in index expressions.
62 In <productname>PostgreSQL</> 8.5 and later,
63 <application>PL/pgSQL</application> is installed by default.
64 However it is still a loadable module, so especially security-conscious
65 administrators could choose to remove it.
68 <sect2 id="plpgsql-advantages">
69 <title>Advantages of Using <application>PL/pgSQL</application></title>
72 <acronym>SQL</acronym> is the language <productname>PostgreSQL</>
73 and most other relational databases use as query language. It's
74 portable and easy to learn. But every <acronym>SQL</acronym>
75 statement must be executed individually by the database server.
79 That means that your client application must send each query to
80 the database server, wait for it to be processed, receive and
81 process the results, do some computation, then send further
82 queries to the server. All this incurs interprocess
83 communication and will also incur network overhead if your client
84 is on a different machine than the database server.
88 With <application>PL/pgSQL</application> you can group a block of
89 computation and a series of queries <emphasis>inside</emphasis>
90 the database server, thus having the power of a procedural
91 language and the ease of use of SQL, but with considerable
92 savings of client/server communication overhead.
96 <listitem><para> Extra round trips between
97 client and server are eliminated </para></listitem>
99 <listitem><para> Intermediate results that the client does not
100 need do not have to be marshaled or transferred between server
101 and client </para></listitem>
103 <listitem><para> Multiple rounds of query
104 parsing can be avoided </para></listitem>
107 <para> This can result in a considerable performance increase as
108 compared to an application that does not use stored functions.
112 Also, with <application>PL/pgSQL</application> you can use all
113 the data types, operators and functions of SQL.
117 <sect2 id="plpgsql-args-results">
118 <title>Supported Argument and Result Data Types</title>
121 Functions written in <application>PL/pgSQL</application> can accept
122 as arguments any scalar or array data type supported by the server,
123 and they can return a result of any of these types. They can also
124 accept or return any composite type (row type) specified by name.
125 It is also possible to declare a <application>PL/pgSQL</application>
126 function as returning <type>record</>, which means that the result
127 is a row type whose columns are determined by specification in the
128 calling query, as discussed in <xref linkend="queries-tablefunctions">.
132 <application>PL/pgSQL</> functions can be declared to accept a variable
133 number of arguments by using the <literal>VARIADIC</> marker. This
134 works exactly the same way as for SQL functions, as discussed in
135 <xref linkend="xfunc-sql-variadic-functions">.
139 <application>PL/pgSQL</> functions can also be declared to accept
140 and return the polymorphic types
141 <type>anyelement</type>, <type>anyarray</type>, <type>anynonarray</type>,
142 and <type>anyenum</>. The actual
143 data types handled by a polymorphic function can vary from call to
144 call, as discussed in <xref linkend="extend-types-polymorphic">.
145 An example is shown in <xref linkend="plpgsql-declaration-parameters">.
149 <application>PL/pgSQL</> functions can also be declared to return
150 a <quote>set</> (or table) of any data type that can be returned as
151 a single instance. Such a function generates its output by executing
152 <command>RETURN NEXT</> for each desired element of the result
153 set, or by using <command>RETURN QUERY</> to output the result of
158 Finally, a <application>PL/pgSQL</> function can be declared to return
159 <type>void</> if it has no useful return value.
163 <application>PL/pgSQL</> functions can also be declared with output
164 parameters in place of an explicit specification of the return type.
165 This does not add any fundamental capability to the language, but
166 it is often convenient, especially for returning multiple values.
167 The <literal>RETURNS TABLE</> notation can also be used in place
168 of <literal>RETURNS SETOF</>.
172 Specific examples appear in
173 <xref linkend="plpgsql-declaration-parameters"> and
174 <xref linkend="plpgsql-statements-returning">.
179 <sect1 id="plpgsql-structure">
180 <title>Structure of <application>PL/pgSQL</application></title>
183 <application>PL/pgSQL</application> is a block-structured language.
184 The complete text of a function definition must be a
185 <firstterm>block</>. A block is defined as:
188 <optional> <<<replaceable>label</replaceable>>> </optional>
190 <replaceable>declarations</replaceable> </optional>
192 <replaceable>statements</replaceable>
193 END <optional> <replaceable>label</replaceable> </optional>;
198 Each declaration and each statement within a block is terminated
199 by a semicolon. A block that appears within another block must
200 have a semicolon after <literal>END</literal>, as shown above;
201 however the final <literal>END</literal> that
202 concludes a function body does not require a semicolon.
207 A common mistake is to write a semicolon immediately after
208 <literal>BEGIN</>. This is incorrect and will result in a syntax error.
213 A <replaceable>label</replaceable> is only needed if you want to
214 identify the block for use
215 in an <literal>EXIT</> statement, or to qualify the names of the
216 variables declared in the block. If a label is given after
217 <literal>END</>, it must match the label at the block's beginning.
221 All key words are case-insensitive.
222 Identifiers are implicitly converted to lowercase
223 unless double-quoted, just as they are in ordinary SQL commands.
227 Comments work the same way in <application>PL/pgSQL</> code as in
228 ordinary SQL. A double dash (<literal>--</literal>) starts a comment
229 that extends to the end of the line. A <literal>/*</literal> starts a
230 block comment that extends to the matching occurrence of
231 <literal>*/</literal>. Block comments nest.
235 Any statement in the statement section of a block
236 can be a <firstterm>subblock</>. Subblocks can be used for
237 logical grouping or to localize variables to a small group
238 of statements. Variables declared in a subblock mask any
239 similarly-named variables of outer blocks for the duration
240 of the subblock; but you can access the outer variables anyway
241 if you qualify their names with their block's label. For example:
243 CREATE FUNCTION somefunc() RETURNS integer AS $$
244 << outerblock >>
246 quantity integer := 30;
248 RAISE NOTICE 'Quantity here is %', quantity; -- Prints 30
254 quantity integer := 80;
256 RAISE NOTICE 'Quantity here is %', quantity; -- Prints 80
257 RAISE NOTICE 'Outer quantity here is %', outerblock.quantity; -- Prints 50
260 RAISE NOTICE 'Quantity here is %', quantity; -- Prints 50
270 There is actually a hidden <quote>outer block</> surrounding the body
271 of any <application>PL/pgSQL</> function. This block provides the
272 declarations of the function's parameters (if any), as well as some
273 special variables such as <literal>FOUND</literal> (see
274 <xref linkend="plpgsql-statements-diagnostics">). The outer block is
275 labeled with the function's name, meaning that parameters and special
276 variables can be qualified with the function's name.
281 It is important not to confuse the use of
282 <command>BEGIN</>/<command>END</> for grouping statements in
283 <application>PL/pgSQL</> with the similarly-named SQL commands
285 control. <application>PL/pgSQL</>'s <command>BEGIN</>/<command>END</>
286 are only for grouping; they do not start or end a transaction.
287 Functions and trigger procedures are always executed within a transaction
288 established by an outer query — they cannot start or commit that
289 transaction, since there would be no context for them to execute in.
290 However, a block containing an <literal>EXCEPTION</> clause effectively
291 forms a subtransaction that can be rolled back without affecting the
292 outer transaction. For more about that see <xref
293 linkend="plpgsql-error-trapping">.
297 <sect1 id="plpgsql-declarations">
298 <title>Declarations</title>
301 All variables used in a block must be declared in the
302 declarations section of the block.
303 (The only exceptions are that the loop variable of a <literal>FOR</> loop
304 iterating over a range of integer values is automatically declared as an
305 integer variable, and likewise the loop variable of a <literal>FOR</> loop
306 iterating over a cursor's result is automatically declared as a
311 <application>PL/pgSQL</> variables can have any SQL data type, such as
312 <type>integer</type>, <type>varchar</type>, and
317 Here are some examples of variable declarations:
322 myrow tablename%ROWTYPE;
323 myfield tablename.columnname%TYPE;
329 The general syntax of a variable declaration is:
331 <replaceable>name</replaceable> <optional> CONSTANT </optional> <replaceable>type</replaceable> <optional> NOT NULL </optional> <optional> { DEFAULT | := } <replaceable>expression</replaceable> </optional>;
333 The <literal>DEFAULT</> clause, if given, specifies the initial value assigned
334 to the variable when the block is entered. If the <literal>DEFAULT</> clause
335 is not given then the variable is initialized to the
336 <acronym>SQL</acronym> null value.
337 The <literal>CONSTANT</> option prevents the variable from being
338 assigned to, so that its value will remain constant for the duration of
340 If <literal>NOT NULL</>
341 is specified, an assignment of a null value results in a run-time
342 error. All variables declared as <literal>NOT NULL</>
343 must have a nonnull default value specified.
347 A variable's default value is evaluated and assigned to the variable
348 each time the block is entered (not just once per function call).
349 So, for example, assigning <literal>now()</literal> to a variable of type
350 <type>timestamp</type> causes the variable to have the
351 time of the current function call, not the time when the function was
358 quantity integer DEFAULT 32;
359 url varchar := 'http://mysite.com';
360 user_id CONSTANT integer := 10;
364 <sect2 id="plpgsql-declaration-parameters">
365 <title>Declaring Function Parameters</title>
368 Parameters passed to functions are named with the identifiers
369 <literal>$1</literal>, <literal>$2</literal>,
370 etc. Optionally, aliases can be declared for
371 <literal>$<replaceable>n</replaceable></literal>
372 parameter names for increased readability. Either the alias or the
373 numeric identifier can then be used to refer to the parameter value.
377 There are two ways to create an alias. The preferred way is to give a
378 name to the parameter in the <command>CREATE FUNCTION</command> command,
381 CREATE FUNCTION sales_tax(subtotal real) RETURNS real AS $$
383 RETURN subtotal * 0.06;
387 The other way, which was the only way available before
388 <productname>PostgreSQL</productname> 8.0, is to explicitly
389 declare an alias, using the declaration syntax
392 <replaceable>name</replaceable> ALIAS FOR $<replaceable>n</replaceable>;
395 The same example in this style looks like:
397 CREATE FUNCTION sales_tax(real) RETURNS real AS $$
399 subtotal ALIAS FOR $1;
401 RETURN subtotal * 0.06;
409 These two examples are not perfectly equivalent. In the first case,
410 <literal>subtotal</> could be referenced as
411 <literal>sales_tax.subtotal</>, but in the second case it could not.
412 (Had we attached a label to the inner block, <literal>subtotal</> could
413 be qualified with that label, instead.)
420 CREATE FUNCTION instr(varchar, integer) RETURNS integer AS $$
422 v_string ALIAS FOR $1;
425 -- some computations using v_string and index here
430 CREATE FUNCTION concat_selected_fields(in_t sometablename) RETURNS text AS $$
432 RETURN in_t.f1 || in_t.f3 || in_t.f5 || in_t.f7;
439 When a <application>PL/pgSQL</application> function is declared
440 with output parameters, the output parameters are given
441 <literal>$<replaceable>n</replaceable></literal> names and optional
442 aliases in just the same way as the normal input parameters. An
443 output parameter is effectively a variable that starts out NULL;
444 it should be assigned to during the execution of the function.
445 The final value of the parameter is what is returned. For instance,
446 the sales-tax example could also be done this way:
449 CREATE FUNCTION sales_tax(subtotal real, OUT tax real) AS $$
451 tax := subtotal * 0.06;
456 Notice that we omitted <literal>RETURNS real</> — we could have
457 included it, but it would be redundant.
461 Output parameters are most useful when returning multiple values.
462 A trivial example is:
465 CREATE FUNCTION sum_n_product(x int, y int, OUT sum int, OUT prod int) AS $$
473 As discussed in <xref linkend="xfunc-output-parameters">, this
474 effectively creates an anonymous record type for the function's
475 results. If a <literal>RETURNS</> clause is given, it must say
476 <literal>RETURNS record</>.
480 Another way to declare a <application>PL/pgSQL</application> function
481 is with <literal>RETURNS TABLE</>, for example:
484 CREATE FUNCTION extended_sales(p_itemno int) RETURNS TABLE(quantity int, total numeric) AS $$
486 RETURN QUERY SELECT quantity, quantity * price FROM sales WHERE itemno = p_itemno;
491 This is exactly equivalent to declaring one or more <literal>OUT</>
492 parameters and specifying <literal>RETURNS SETOF
493 <replaceable>sometype</></literal>.
497 When the return type of a <application>PL/pgSQL</application>
498 function is declared as a polymorphic type (<type>anyelement</type>,
499 <type>anyarray</type>, <type>anynonarray</type>, or <type>anyenum</>),
500 a special parameter <literal>$0</literal>
501 is created. Its data type is the actual return type of the function,
502 as deduced from the actual input types (see <xref
503 linkend="extend-types-polymorphic">).
504 This allows the function to access its actual return type
505 as shown in <xref linkend="plpgsql-declaration-type">.
506 <literal>$0</literal> is initialized to null and can be modified by
507 the function, so it can be used to hold the return value if desired,
508 though that is not required. <literal>$0</literal> can also be
509 given an alias. For example, this function works on any data type
510 that has a <literal>+</> operator:
513 CREATE FUNCTION add_three_values(v1 anyelement, v2 anyelement, v3 anyelement)
514 RETURNS anyelement AS $$
518 result := v1 + v2 + v3;
526 The same effect can be had by declaring one or more output parameters as
527 polymorphic types. In this case the
528 special <literal>$0</literal> parameter is not used; the output
529 parameters themselves serve the same purpose. For example:
532 CREATE FUNCTION add_three_values(v1 anyelement, v2 anyelement, v3 anyelement,
543 <sect2 id="plpgsql-declaration-alias">
544 <title><literal>ALIAS</></title>
547 <replaceable>newname</> ALIAS FOR <replaceable>oldname</>;
551 The <literal>ALIAS</> syntax is more general than is suggested in the
552 previous section: you can declare an alias for any variable, not just
553 function parameters. The main practical use for this is to assign
554 a different name for variables with predetermined names, such as
555 <varname>NEW</varname> or <varname>OLD</varname> within
564 updated ALIAS FOR new;
569 Since <literal>ALIAS</> creates two different ways to name the same
570 object, unrestricted use can be confusing. It's best to use it only
571 for the purpose of overriding predetermined names.
575 <sect2 id="plpgsql-declaration-type">
576 <title>Copying Types</title>
579 <replaceable>variable</replaceable>%TYPE
583 <literal>%TYPE</literal> provides the data type of a variable or
584 table column. You can use this to declare variables that will hold
585 database values. For example, let's say you have a column named
586 <literal>user_id</literal> in your <literal>users</literal>
587 table. To declare a variable with the same data type as
588 <literal>users.user_id</> you write:
590 user_id users.user_id%TYPE;
595 By using <literal>%TYPE</literal> you don't need to know the data
596 type of the structure you are referencing, and most importantly,
597 if the data type of the referenced item changes in the future (for
598 instance: you change the type of <literal>user_id</>
599 from <type>integer</type> to <type>real</type>), you might not need
600 to change your function definition.
604 <literal>%TYPE</literal> is particularly valuable in polymorphic
605 functions, since the data types needed for internal variables can
606 change from one call to the next. Appropriate variables can be
607 created by applying <literal>%TYPE</literal> to the function's
608 arguments or result placeholders.
613 <sect2 id="plpgsql-declaration-rowtypes">
614 <title>Row Types</title>
617 <replaceable>name</replaceable> <replaceable>table_name</replaceable><literal>%ROWTYPE</literal>;
618 <replaceable>name</replaceable> <replaceable>composite_type_name</replaceable>;
622 A variable of a composite type is called a <firstterm>row</>
623 variable (or <firstterm>row-type</> variable). Such a variable
624 can hold a whole row of a <command>SELECT</> or <command>FOR</>
625 query result, so long as that query's column set matches the
626 declared type of the variable.
627 The individual fields of the row value
628 are accessed using the usual dot notation, for example
629 <literal>rowvar.field</literal>.
633 A row variable can be declared to have the same type as the rows of
634 an existing table or view, by using the
635 <replaceable>table_name</replaceable><literal>%ROWTYPE</literal>
636 notation; or it can be declared by giving a composite type's name.
637 (Since every table has an associated composite type of the same name,
638 it actually does not matter in <productname>PostgreSQL</> whether you
639 write <literal>%ROWTYPE</literal> or not. But the form with
640 <literal>%ROWTYPE</literal> is more portable.)
644 Parameters to a function can be
645 composite types (complete table rows). In that case, the
646 corresponding identifier <literal>$<replaceable>n</replaceable></> will be a row variable, and fields can
647 be selected from it, for example <literal>$1.user_id</literal>.
651 Only the user-defined columns of a table row are accessible in a
652 row-type variable, not the OID or other system columns (because the
653 row could be from a view). The fields of the row type inherit the
654 table's field size or precision for data types such as
655 <type>char(<replaceable>n</>)</type>.
659 Here is an example of using composite types. <structname>table1</>
660 and <structname>table2</> are existing tables having at least the
664 CREATE FUNCTION merge_fields(t_row table1) RETURNS text AS $$
666 t2_row table2%ROWTYPE;
668 SELECT * INTO t2_row FROM table2 WHERE ... ;
669 RETURN t_row.f1 || t2_row.f3 || t_row.f5 || t2_row.f7;
673 SELECT merge_fields(t.*) FROM table1 t WHERE ... ;
678 <sect2 id="plpgsql-declaration-records">
679 <title>Record Types</title>
682 <replaceable>name</replaceable> RECORD;
686 Record variables are similar to row-type variables, but they have no
687 predefined structure. They take on the actual row structure of the
688 row they are assigned during a <command>SELECT</> or <command>FOR</> command. The substructure
689 of a record variable can change each time it is assigned to.
690 A consequence of this is that until a record variable is first assigned
691 to, it has no substructure, and any attempt to access a
692 field in it will draw a run-time error.
696 Note that <literal>RECORD</> is not a true data type, only a placeholder.
697 One should also realize that when a <application>PL/pgSQL</application>
698 function is declared to return type <type>record</>, this is not quite the
699 same concept as a record variable, even though such a function might
700 use a record variable to hold its result. In both cases the actual row
701 structure is unknown when the function is written, but for a function
702 returning <type>record</> the actual structure is determined when the
703 calling query is parsed, whereas a record variable can change its row
704 structure on-the-fly.
709 <sect1 id="plpgsql-expressions">
710 <title>Expressions</title>
713 All expressions used in <application>PL/pgSQL</application>
714 statements are processed using the server's main
715 <acronym>SQL</acronym> executor. For example, when you write
716 a <application>PL/pgSQL</application> statement like
718 IF <replaceable>expression</replaceable> THEN ...
720 <application>PL/pgSQL</application> will evaluate the expression by
723 SELECT <replaceable>expression</replaceable>
725 to the main SQL engine. While forming the <command>SELECT</> command,
726 any occurrences of <application>PL/pgSQL</application> variable names
727 are replaced by parameters, as discussed in detail in
728 <xref linkend="plpgsql-var-subst">.
729 This allows the query plan for the <command>SELECT</command> to
730 be prepared just once and then reused for subsequent
731 evaluations with different values of the variables. Thus, what
732 really happens on first use of an expression is essentially a
733 <command>PREPARE</> command. For example, if we have declared
734 two integer variables <literal>x</> and <literal>y</>, and we write
738 what happens behind the scenes is equivalent to
740 PREPARE <replaceable>statement_name</>(integer, integer) AS SELECT $1 < $2;
742 and then this prepared statement is <command>EXECUTE</>d for each
743 execution of the <command>IF</> statement, with the current values
744 of the <application>PL/pgSQL</application> variables supplied as
746 The query plan prepared in this way is saved for the life of the database
747 connection, as described in
748 <xref linkend="plpgsql-plan-caching">. Normally these details are
749 not important to a <application>PL/pgSQL</application> user, but
750 they are useful to know when trying to diagnose a problem.
754 <sect1 id="plpgsql-statements">
755 <title>Basic Statements</title>
758 In this section and the following ones, we describe all the statement
759 types that are explicitly understood by
760 <application>PL/pgSQL</application>.
761 Anything not recognized as one of these statement types is presumed
762 to be an SQL command and is sent to the main database engine to execute,
763 as described in <xref linkend="plpgsql-statements-sql-noresult">
764 and <xref linkend="plpgsql-statements-sql-onerow">.
767 <sect2 id="plpgsql-statements-assignment">
768 <title>Assignment</title>
771 An assignment of a value to a <application>PL/pgSQL</application>
772 variable is written as:
774 <replaceable>variable</replaceable> := <replaceable>expression</replaceable>;
776 As explained previously, the expression in such a statement is evaluated
777 by means of an SQL <command>SELECT</> command sent to the main
778 database engine. The expression must yield a single value (possibly
779 a row value, if the variable is a row or record variable). The target
780 variable can be a simple variable (optionally qualified with a block
781 name), a field of a row or record variable, or an element of an array
782 that is a simple variable or field.
786 If the expression's result data type doesn't match the variable's
787 data type, or the variable has a specific size/precision
788 (like <type>char(20)</type>), the result value will be implicitly
789 converted by the <application>PL/pgSQL</application> interpreter using
790 the result type's output-function and
791 the variable type's input-function. Note that this could potentially
792 result in run-time errors generated by the input function, if the
793 string form of the result value is not acceptable to the input function.
799 tax := subtotal * 0.06;
800 my_record.user_id := 20;
805 <sect2 id="plpgsql-statements-sql-noresult">
806 <title>Executing a Command With No Result</title>
809 For any SQL command that does not return rows, for example
810 <command>INSERT</> without a <literal>RETURNING</> clause, you can
811 execute the command within a <application>PL/pgSQL</application> function
812 just by writing the command.
816 Any <application>PL/pgSQL</application> variable name appearing
817 in the command text is treated as a parameter, and then the
818 current value of the variable is provided as the parameter value
819 at run time. This is exactly like the processing described earlier
820 for expressions; for details see <xref linkend="plpgsql-var-subst">.
824 When executing a SQL command in this way,
825 <application>PL/pgSQL</application> plans the command just once
826 and re-uses the plan on subsequent executions, for the life of
827 the database connection. The implications of this are discussed
828 in detail in <xref linkend="plpgsql-plan-caching">.
832 Sometimes it is useful to evaluate an expression or <command>SELECT</>
833 query but discard the result, for example when calling a function
834 that has side-effects but no useful result value. To do
835 this in <application>PL/pgSQL</application>, use the
836 <command>PERFORM</command> statement:
839 PERFORM <replaceable>query</replaceable>;
842 This executes <replaceable>query</replaceable> and discards the
843 result. Write the <replaceable>query</replaceable> the same
844 way you would write an SQL <command>SELECT</> command, but replace the
845 initial keyword <command>SELECT</> with <command>PERFORM</command>.
846 <application>PL/pgSQL</application> variables will be
847 substituted into the query just as for commands that return no result,
848 and the plan is cached in the same way. Also, the special variable
849 <literal>FOUND</literal> is set to true if the query produced at
850 least one row, or false if it produced no rows (see
851 <xref linkend="plpgsql-statements-diagnostics">).
856 One might expect that writing <command>SELECT</command> directly
857 would accomplish this result, but at
858 present the only accepted way to do it is
859 <command>PERFORM</command>. A SQL command that can return rows,
860 such as <command>SELECT</command>, will be rejected as an error
861 unless it has an <literal>INTO</> clause as discussed in the
869 PERFORM create_mv('cs_session_page_requests_mv', my_query);
874 <sect2 id="plpgsql-statements-sql-onerow">
875 <title>Executing a Query with a Single-Row Result</title>
877 <indexterm zone="plpgsql-statements-sql-onerow">
878 <primary>SELECT INTO</primary>
879 <secondary>in PL/pgSQL</secondary>
882 <indexterm zone="plpgsql-statements-sql-onerow">
883 <primary>RETURNING INTO</primary>
884 <secondary>in PL/pgSQL</secondary>
888 The result of a SQL command yielding a single row (possibly of multiple
889 columns) can be assigned to a record variable, row-type variable, or list
890 of scalar variables. This is done by writing the base SQL command and
891 adding an <literal>INTO</> clause. For example,
894 SELECT <replaceable>select_expressions</replaceable> INTO <optional>STRICT</optional> <replaceable>target</replaceable> FROM ...;
895 INSERT ... RETURNING <replaceable>expressions</replaceable> INTO <optional>STRICT</optional> <replaceable>target</replaceable>;
896 UPDATE ... RETURNING <replaceable>expressions</replaceable> INTO <optional>STRICT</optional> <replaceable>target</replaceable>;
897 DELETE ... RETURNING <replaceable>expressions</replaceable> INTO <optional>STRICT</optional> <replaceable>target</replaceable>;
900 where <replaceable>target</replaceable> can be a record variable, a row
901 variable, or a comma-separated list of simple variables and
903 <application>PL/pgSQL</application> variables will be
904 substituted into the rest of the query, and the plan is cached,
905 just as described above for commands that do not return rows.
906 This works for <command>SELECT</>,
907 <command>INSERT</>/<command>UPDATE</>/<command>DELETE</> with
908 <literal>RETURNING</>, and utility commands that return row-set
909 results (such as <command>EXPLAIN</>).
910 Except for the <literal>INTO</> clause, the SQL command is the same
911 as it would be written outside <application>PL/pgSQL</application>.
916 Note that this interpretation of <command>SELECT</> with <literal>INTO</>
917 is quite different from <productname>PostgreSQL</>'s regular
918 <command>SELECT INTO</command> command, wherein the <literal>INTO</>
919 target is a newly created table. If you want to create a table from a
920 <command>SELECT</> result inside a
921 <application>PL/pgSQL</application> function, use the syntax
922 <command>CREATE TABLE ... AS SELECT</command>.
927 If a row or a variable list is used as target, the query's result columns
928 must exactly match the structure of the target as to number and data
929 types, or else a run-time error
930 occurs. When a record variable is the target, it automatically
931 configures itself to the row type of the query result columns.
935 The <literal>INTO</> clause can appear almost anywhere in the SQL
936 command. Customarily it is written either just before or just after
937 the list of <replaceable>select_expressions</replaceable> in a
938 <command>SELECT</> command, or at the end of the command for other
939 command types. It is recommended that you follow this convention
940 in case the <application>PL/pgSQL</application> parser becomes
941 stricter in future versions.
945 If <literal>STRICT</literal> is not specified in the <literal>INTO</>
946 clause, then <replaceable>target</replaceable> will be set to the first
947 row returned by the query, or to nulls if the query returned no rows.
948 (Note that <quote>the first row</> is not
949 well-defined unless you've used <literal>ORDER BY</>.) Any result rows
950 after the first row are discarded.
951 You can check the special <literal>FOUND</literal> variable (see
952 <xref linkend="plpgsql-statements-diagnostics">) to
953 determine whether a row was returned:
956 SELECT * INTO myrec FROM emp WHERE empname = myname;
958 RAISE EXCEPTION 'employee % not found', myname;
962 If the <literal>STRICT</literal> option is specified, the query must
963 return exactly one row or a run-time error will be reported, either
964 <literal>NO_DATA_FOUND</> (no rows) or <literal>TOO_MANY_ROWS</>
965 (more than one row). You can use an exception block if you wish
966 to catch the error, for example:
970 SELECT * INTO STRICT myrec FROM emp WHERE empname = myname;
972 WHEN NO_DATA_FOUND THEN
973 RAISE EXCEPTION 'employee % not found', myname;
974 WHEN TOO_MANY_ROWS THEN
975 RAISE EXCEPTION 'employee % not unique', myname;
978 Successful execution of a command with <literal>STRICT</>
979 always sets <literal>FOUND</literal> to true.
983 For <command>INSERT</>/<command>UPDATE</>/<command>DELETE</> with
984 <literal>RETURNING</>, <application>PL/pgSQL</application> reports
985 an error for more than one returned row, even when
986 <literal>STRICT</literal> is not specified. This is because there
987 is no option such as <literal>ORDER BY</> with which to determine
988 which affected row should be returned.
993 The <literal>STRICT</> option matches the behavior of
994 Oracle PL/SQL's <command>SELECT INTO</command> and related statements.
999 To handle cases where you need to process multiple result rows
1000 from a SQL query, see <xref linkend="plpgsql-records-iterating">.
1005 <sect2 id="plpgsql-statements-executing-dyn">
1006 <title>Executing Dynamic Commands</title>
1009 Oftentimes you will want to generate dynamic commands inside your
1010 <application>PL/pgSQL</application> functions, that is, commands
1011 that will involve different tables or different data types each
1012 time they are executed. <application>PL/pgSQL</application>'s
1013 normal attempts to cache plans for commands (as discussed in
1014 <xref linkend="plpgsql-plan-caching">) will not work in such
1015 scenarios. To handle this sort of problem, the
1016 <command>EXECUTE</command> statement is provided:
1019 EXECUTE <replaceable class="command">command-string</replaceable> <optional> INTO <optional>STRICT</optional> <replaceable>target</replaceable> </optional> <optional> USING <replaceable>expression</replaceable> <optional>, ...</optional> </optional>;
1022 where <replaceable>command-string</replaceable> is an expression
1023 yielding a string (of type <type>text</type>) containing the
1024 command to be executed. The optional <replaceable>target</replaceable>
1025 is a record variable, a row variable, or a comma-separated list of
1026 simple variables and record/row fields, into which the results of
1027 the command will be stored. The optional <literal>USING</> expressions
1028 supply values to be inserted into the command.
1032 No substitution of <application>PL/pgSQL</> variables is done on the
1033 computed command string. Any required variable values must be inserted
1034 in the command string as it is constructed; or you can use parameters
1039 Also, there is no plan caching for commands executed via
1040 <command>EXECUTE</command>. Instead, the
1041 command is prepared each time the statement is run. Thus the command
1042 string can be dynamically created within the function to perform
1043 actions on different tables and columns.
1047 The <literal>INTO</literal> clause specifies where the results of
1048 a SQL command returning rows should be assigned. If a row
1049 or variable list is provided, it must exactly match the structure
1050 of the query's results (when a
1051 record variable is used, it will configure itself to match the
1052 result structure automatically). If multiple rows are returned,
1053 only the first will be assigned to the <literal>INTO</literal>
1054 variable. If no rows are returned, NULL is assigned to the
1055 <literal>INTO</literal> variable(s). If no <literal>INTO</literal>
1056 clause is specified, the query results are discarded.
1060 If the <literal>STRICT</> option is given, an error is reported
1061 unless the query produces exactly one row.
1065 The command string can use parameter values, which are referenced
1066 in the command as <literal>$1</>, <literal>$2</>, etc.
1067 These symbols refer to values supplied in the <literal>USING</>
1068 clause. This method is often preferable to inserting data values
1069 into the command string as text: it avoids run-time overhead of
1070 converting the values to text and back, and it is much less prone
1071 to SQL-injection attacks since there is no need for quoting or escaping.
1074 EXECUTE 'SELECT count(*) FROM mytable WHERE inserted_by = $1 AND inserted <= $2'
1076 USING checked_user, checked_date;
1081 Note that parameter symbols can only be used for data values
1082 — if you want to use dynamically determined table or column
1083 names, you must insert them into the command string textually.
1084 For example, if the preceding query needed to be done against a
1085 dynamically selected table, you could do this:
1087 EXECUTE 'SELECT count(*) FROM '
1088 || tabname::regclass
1089 || ' WHERE inserted_by = $1 AND inserted <= $2'
1091 USING checked_user, checked_date;
1093 Another restriction on parameter symbols is that they only work in
1094 <command>SELECT</>, <command>INSERT</>, <command>UPDATE</>, and
1095 <command>DELETE</> commands. In other statement
1096 types (generically called utility statements), you must insert
1097 values textually even if they are just data values.
1101 An <command>EXECUTE</> with a simple constant command string and some
1102 <literal>USING</> parameters, as in the first example above, is
1103 functionally equivalent to just writing the command directly in
1104 <application>PL/pgSQL</application> and allowing replacement of
1105 <application>PL/pgSQL</application> variables to happen automatically.
1106 The important difference is that <command>EXECUTE</> will re-plan
1107 the command on each execution, generating a plan that is specific
1108 to the current parameter values; whereas
1109 <application>PL/pgSQL</application> normally creates a generic plan
1110 and caches it for re-use. In situations where the best plan depends
1111 strongly on the parameter values, <command>EXECUTE</> can be
1112 significantly faster; while when the plan is not sensitive to parameter
1113 values, re-planning will be a waste.
1117 <command>SELECT INTO</command> is not currently supported within
1118 <command>EXECUTE</command>; instead, execute a plain <command>SELECT</>
1119 command and specify <literal>INTO</> as part of the <command>EXECUTE</>
1125 The <application>PL/pgSQL</application>
1126 <command>EXECUTE</command> statement is not related to the
1127 <xref linkend="sql-execute" endterm="sql-execute-title"> SQL
1128 statement supported by the
1129 <productname>PostgreSQL</productname> server. The server's
1130 <command>EXECUTE</command> statement cannot be used directly within
1131 <application>PL/pgSQL</> functions (and is not needed).
1135 <example id="plpgsql-quote-literal-example">
1136 <title>Quoting values in dynamic queries</title>
1139 <primary>quote_ident</primary>
1140 <secondary>use in PL/PgSQL</secondary>
1144 <primary>quote_literal</primary>
1145 <secondary>use in PL/PgSQL</secondary>
1149 <primary>quote_nullable</primary>
1150 <secondary>use in PL/PgSQL</secondary>
1154 When working with dynamic commands you will often have to handle escaping
1155 of single quotes. The recommended method for quoting fixed text in your
1156 function body is dollar quoting. (If you have legacy code that does
1157 not use dollar quoting, please refer to the
1158 overview in <xref linkend="plpgsql-quote-tips">, which can save you
1159 some effort when translating said code to a more reasonable scheme.)
1163 Dynamic values that are to be inserted into the constructed
1164 query require careful handling since they might themselves contain
1166 An example (this assumes that you are using dollar quoting for the
1167 function as a whole, so the quote marks need not be doubled):
1169 EXECUTE 'UPDATE tbl SET '
1170 || quote_ident(colname)
1172 || quote_literal(newvalue)
1174 || quote_literal(keyvalue);
1179 This example demonstrates the use of the
1180 <function>quote_ident</function> and
1181 <function>quote_literal</function> functions (see <xref
1182 linkend="functions-string">). For safety, expressions containing column
1183 or table identifiers should be passed through
1184 <function>quote_ident</function> before insertion in a dynamic query.
1185 Expressions containing values that should be literal strings in the
1186 constructed command should be passed through <function>quote_literal</>.
1187 These functions take the appropriate steps to return the input text
1188 enclosed in double or single quotes respectively, with any embedded
1189 special characters properly escaped.
1193 Because <function>quote_literal</function> is labelled
1194 <literal>STRICT</literal>, it will always return null when called with a
1195 null argument. In the above example, if <literal>newvalue</> or
1196 <literal>keyvalue</> were null, the entire dynamic query string would
1197 become null, leading to an error from <command>EXECUTE</command>.
1198 You can avoid this problem by using the <function>quote_nullable</>
1199 function, which works the same as <function>quote_literal</> except that
1200 when called with a null argument it returns the string <literal>NULL</>.
1203 EXECUTE 'UPDATE tbl SET '
1204 || quote_ident(colname)
1206 || quote_nullable(newvalue)
1208 || quote_nullable(keyvalue);
1210 If you are dealing with values that might be null, you should usually
1211 use <function>quote_nullable</> in place of <function>quote_literal</>.
1215 As always, care must be taken to ensure that null values in a query do
1216 not deliver unintended results. For example the <literal>WHERE</> clause
1218 'WHERE key = ' || quote_nullable(keyvalue)
1220 will never succeed if <literal>keyvalue</> is null, because the
1221 result of using the equality operator <literal>=</> with a null operand
1222 is always null. If you wish null to work like an ordinary key value,
1223 you would need to rewrite the above as
1225 'WHERE key IS NOT DISTINCT FROM ' || quote_nullable(keyvalue)
1227 (At present, <literal>IS NOT DISTINCT FROM</> is handled much less
1228 efficiently than <literal>=</>, so don't do this unless you must.
1229 See <xref linkend="functions-comparison"> for
1230 more information on nulls and <literal>IS DISTINCT</>.)
1234 Note that dollar quoting is only useful for quoting fixed text.
1235 It would be a very bad idea to try to write this example as:
1237 EXECUTE 'UPDATE tbl SET '
1238 || quote_ident(colname)
1241 || '$$ WHERE key = '
1242 || quote_literal(keyvalue);
1244 because it would break if the contents of <literal>newvalue</>
1245 happened to contain <literal>$$</>. The same objection would
1246 apply to any other dollar-quoting delimiter you might pick.
1247 So, to safely quote text that is not known in advance, you
1248 <emphasis>must</> use <function>quote_literal</>,
1249 <function>quote_nullable</>, or <function>quote_ident</>, as appropriate.
1254 A much larger example of a dynamic command and
1255 <command>EXECUTE</command> can be seen in <xref
1256 linkend="plpgsql-porting-ex2">, which builds and executes a
1257 <command>CREATE FUNCTION</> command to define a new function.
1261 <sect2 id="plpgsql-statements-diagnostics">
1262 <title>Obtaining the Result Status</title>
1265 There are several ways to determine the effect of a command. The
1266 first method is to use the <command>GET DIAGNOSTICS</command>
1267 command, which has the form:
1270 GET DIAGNOSTICS <replaceable>variable</replaceable> = <replaceable>item</replaceable> <optional> , ... </optional>;
1273 This command allows retrieval of system status indicators. Each
1274 <replaceable>item</replaceable> is a key word identifying a state
1275 value to be assigned to the specified variable (which should be
1276 of the right data type to receive it). The currently available
1277 status items are <varname>ROW_COUNT</>, the number of rows
1278 processed by the last <acronym>SQL</acronym> command sent to
1279 the <acronym>SQL</acronym> engine, and <varname>RESULT_OID</>,
1280 the OID of the last row inserted by the most recent
1281 <acronym>SQL</acronym> command. Note that <varname>RESULT_OID</>
1282 is only useful after an <command>INSERT</command> command into a
1283 table containing OIDs.
1289 GET DIAGNOSTICS integer_var = ROW_COUNT;
1294 The second method to determine the effects of a command is to check the
1295 special variable named <literal>FOUND</literal>, which is of
1296 type <type>boolean</type>. <literal>FOUND</literal> starts out
1297 false within each <application>PL/pgSQL</application> function call.
1298 It is set by each of the following types of statements:
1303 A <command>SELECT INTO</command> statement sets
1304 <literal>FOUND</literal> true if a row is assigned, false if no
1310 A <command>PERFORM</> statement sets <literal>FOUND</literal>
1311 true if it produces (and discards) one or more rows, false if
1317 <command>UPDATE</>, <command>INSERT</>, and <command>DELETE</>
1318 statements set <literal>FOUND</literal> true if at least one
1319 row is affected, false if no row is affected.
1324 A <command>FETCH</> statement sets <literal>FOUND</literal>
1325 true if it returns a row, false if no row is returned.
1330 A <command>MOVE</> statement sets <literal>FOUND</literal>
1331 true if it successfully repositions the cursor, false otherwise.
1337 A <command>FOR</> statement sets <literal>FOUND</literal> true
1338 if it iterates one or more times, else false. This applies to
1339 all four variants of the <command>FOR</> statement (integer
1340 <command>FOR</> loops, record-set <command>FOR</> loops,
1341 dynamic record-set <command>FOR</> loops, and cursor
1342 <command>FOR</> loops).
1343 <literal>FOUND</literal> is set this way when the
1344 <command>FOR</> loop exits; inside the execution of the loop,
1345 <literal>FOUND</literal> is not modified by the
1346 <command>FOR</> statement, although it might be changed by the
1347 execution of other statements within the loop body.
1352 A <command>RETURN QUERY</command> and <command>RETURN QUERY
1353 EXECUTE</command> statements set <literal>FOUND</literal>
1354 true if the query returns at least one row, false if no row
1360 Other <application>PL/pgSQL</application> statements do not change
1361 the state of <literal>FOUND</literal>.
1362 Note in particular that <command>EXECUTE</command>
1363 changes the output of <command>GET DIAGNOSTICS</command>, but
1364 does not change <literal>FOUND</literal>.
1368 <literal>FOUND</literal> is a local variable within each
1369 <application>PL/pgSQL</application> function; any changes to it
1370 affect only the current function.
1375 <sect2 id="plpgsql-statements-null">
1376 <title>Doing Nothing At All</title>
1379 Sometimes a placeholder statement that does nothing is useful.
1380 For example, it can indicate that one arm of an if/then/else
1381 chain is deliberately empty. For this purpose, use the
1382 <command>NULL</command> statement:
1390 For example, the following two fragments of code are equivalent:
1395 WHEN division_by_zero THEN
1396 NULL; -- ignore the error
1404 WHEN division_by_zero THEN -- ignore the error
1407 Which is preferable is a matter of taste.
1412 In Oracle's PL/SQL, empty statement lists are not allowed, and so
1413 <command>NULL</> statements are <emphasis>required</> for situations
1414 such as this. <application>PL/pgSQL</application> allows you to
1415 just write nothing, instead.
1422 <sect1 id="plpgsql-control-structures">
1423 <title>Control Structures</title>
1426 Control structures are probably the most useful (and
1427 important) part of <application>PL/pgSQL</>. With
1428 <application>PL/pgSQL</>'s control structures,
1429 you can manipulate <productname>PostgreSQL</> data in a very
1430 flexible and powerful way.
1433 <sect2 id="plpgsql-statements-returning">
1434 <title>Returning From a Function</title>
1437 There are two commands available that allow you to return data
1438 from a function: <command>RETURN</command> and <command>RETURN
1443 <title><command>RETURN</></title>
1446 RETURN <replaceable>expression</replaceable>;
1450 <command>RETURN</command> with an expression terminates the
1451 function and returns the value of
1452 <replaceable>expression</replaceable> to the caller. This form
1453 is used for <application>PL/pgSQL</> functions that do
1458 When returning a scalar type, any expression can be used. The
1459 expression's result will be automatically cast into the
1460 function's return type as described for assignments. To return a
1461 composite (row) value, you must write a record or row variable
1462 as the <replaceable>expression</replaceable>.
1466 If you declared the function with output parameters, write just
1467 <command>RETURN</command> with no expression. The current values
1468 of the output parameter variables will be returned.
1472 If you declared the function to return <type>void</type>, a
1473 <command>RETURN</command> statement can be used to exit the function
1474 early; but do not write an expression following
1475 <command>RETURN</command>.
1479 The return value of a function cannot be left undefined. If
1480 control reaches the end of the top-level block of the function
1481 without hitting a <command>RETURN</command> statement, a run-time
1482 error will occur. This restriction does not apply to functions
1483 with output parameters and functions returning <type>void</type>,
1484 however. In those cases a <command>RETURN</command> statement is
1485 automatically executed if the top-level block finishes.
1490 <title><command>RETURN NEXT</> and <command>RETURN QUERY</command></title>
1492 <primary>RETURN NEXT</primary>
1493 <secondary>in PL/PgSQL</secondary>
1496 <primary>RETURN QUERY</primary>
1497 <secondary>in PL/PgSQL</secondary>
1501 RETURN NEXT <replaceable>expression</replaceable>;
1502 RETURN QUERY <replaceable>query</replaceable>;
1503 RETURN QUERY EXECUTE <replaceable class="command">command-string</replaceable> <optional> USING <replaceable>expression</replaceable> <optional>, ...</optional> </optional>;
1507 When a <application>PL/pgSQL</> function is declared to return
1508 <literal>SETOF <replaceable>sometype</></literal>, the procedure
1509 to follow is slightly different. In that case, the individual
1510 items to return are specified by a sequence of <command>RETURN
1511 NEXT</command> or <command>RETURN QUERY</command> commands, and
1512 then a final <command>RETURN</command> command with no argument
1513 is used to indicate that the function has finished executing.
1514 <command>RETURN NEXT</command> can be used with both scalar and
1515 composite data types; with a composite result type, an entire
1516 <quote>table</quote> of results will be returned.
1517 <command>RETURN QUERY</command> appends the results of executing
1518 a query to the function's result set. <command>RETURN
1519 NEXT</command> and <command>RETURN QUERY</command> can be freely
1520 intermixed in a single set-returning function, in which case
1521 their results will be concatenated.
1525 <command>RETURN NEXT</command> and <command>RETURN
1526 QUERY</command> do not actually return from the function —
1527 they simply append zero or more rows to the function's result
1528 set. Execution then continues with the next statement in the
1529 <application>PL/pgSQL</> function. As successive
1530 <command>RETURN NEXT</command> or <command>RETURN
1531 QUERY</command> commands are executed, the result set is built
1532 up. A final <command>RETURN</command>, which should have no
1533 argument, causes control to exit the function (or you can just
1534 let control reach the end of the function).
1538 <command>RETURN QUERY</command> has a variant
1539 <command>RETURN QUERY EXECUTE</command>, which specifies the
1540 query to be executed dynamically. Parameter expressions can
1541 be inserted into the computed query string via <literal>USING</>,
1542 in just the same way as in the <command>EXECUTE</> command.
1546 If you declared the function with output parameters, write just
1547 <command>RETURN NEXT</command> with no expression. On each
1548 execution, the current values of the output parameter
1549 variable(s) will be saved for eventual return as a row of the
1550 result. Note that you must declare the function as returning
1551 <literal>SETOF record</literal> when there are multiple output
1552 parameters, or <literal>SETOF <replaceable>sometype</></literal>
1553 when there is just one output parameter of type
1554 <replaceable>sometype</>, in order to create a set-returning
1555 function with output parameters.
1559 Here is an example of a function using <command>RETURN
1563 CREATE TABLE foo (fooid INT, foosubid INT, fooname TEXT);
1564 INSERT INTO foo VALUES (1, 2, 'three');
1565 INSERT INTO foo VALUES (4, 5, 'six');
1567 CREATE OR REPLACE FUNCTION getAllFoo() RETURNS SETOF foo AS
1572 FOR r IN SELECT * FROM foo
1575 -- can do some processing here
1576 RETURN NEXT r; -- return current row of SELECT
1581 LANGUAGE 'plpgsql' ;
1583 SELECT * FROM getallfoo();
1589 The current implementation of <command>RETURN NEXT</command>
1590 and <command>RETURN QUERY</command> stores the entire result set
1591 before returning from the function, as discussed above. That
1592 means that if a <application>PL/pgSQL</> function produces a
1593 very large result set, performance might be poor: data will be
1594 written to disk to avoid memory exhaustion, but the function
1595 itself will not return until the entire result set has been
1596 generated. A future version of <application>PL/pgSQL</> might
1597 allow users to define set-returning functions
1598 that do not have this limitation. Currently, the point at
1599 which data begins being written to disk is controlled by the
1600 <xref linkend="guc-work-mem">
1601 configuration variable. Administrators who have sufficient
1602 memory to store larger result sets in memory should consider
1603 increasing this parameter.
1609 <sect2 id="plpgsql-conditionals">
1610 <title>Conditionals</title>
1613 <command>IF</> and <command>CASE</> statements let you execute
1614 alternative commands based on certain conditions.
1615 <application>PL/pgSQL</> has three forms of <command>IF</>:
1618 <para><literal>IF ... THEN</></>
1621 <para><literal>IF ... THEN ... ELSE</></>
1624 <para><literal>IF ... THEN ... ELSIF ... THEN ... ELSE</></>
1628 and two forms of <command>CASE</>:
1631 <para><literal>CASE ... WHEN ... THEN ... ELSE ... END CASE</></>
1634 <para><literal>CASE WHEN ... THEN ... ELSE ... END CASE</></>
1640 <title><literal>IF-THEN</></title>
1643 IF <replaceable>boolean-expression</replaceable> THEN
1644 <replaceable>statements</replaceable>
1649 <literal>IF-THEN</literal> statements are the simplest form of
1650 <literal>IF</literal>. The statements between
1651 <literal>THEN</literal> and <literal>END IF</literal> will be
1652 executed if the condition is true. Otherwise, they are
1659 IF v_user_id <> 0 THEN
1660 UPDATE users SET email = v_email WHERE user_id = v_user_id;
1667 <title><literal>IF-THEN-ELSE</></title>
1670 IF <replaceable>boolean-expression</replaceable> THEN
1671 <replaceable>statements</replaceable>
1673 <replaceable>statements</replaceable>
1678 <literal>IF-THEN-ELSE</literal> statements add to
1679 <literal>IF-THEN</literal> by letting you specify an
1680 alternative set of statements that should be executed if the
1681 condition is not true. (Note this includes the case where the
1682 condition evaluates to NULL.)
1688 IF parentid IS NULL OR parentid = ''
1692 RETURN hp_true_filename(parentid) || '/' || fullname;
1697 IF v_count > 0 THEN
1698 INSERT INTO users_count (count) VALUES (v_count);
1708 <title><literal>IF-THEN-ELSIF</></title>
1711 IF <replaceable>boolean-expression</replaceable> THEN
1712 <replaceable>statements</replaceable>
1713 <optional> ELSIF <replaceable>boolean-expression</replaceable> THEN
1714 <replaceable>statements</replaceable>
1715 <optional> ELSIF <replaceable>boolean-expression</replaceable> THEN
1716 <replaceable>statements</replaceable>
1721 <replaceable>statements</replaceable> </optional>
1726 Sometimes there are more than just two alternatives.
1727 <literal>IF-THEN-ELSIF</> provides a convenient
1728 method of checking several alternatives in turn.
1729 The <literal>IF</> conditions are tested successively
1730 until the first one that is true is found. Then the
1731 associated statement(s) are executed, after which control
1732 passes to the next statement after <literal>END IF</>.
1733 (Any subsequent <literal>IF</> conditions are <emphasis>not</>
1734 tested.) If none of the <literal>IF</> conditions is true,
1735 then the <literal>ELSE</> block (if any) is executed.
1744 ELSIF number > 0 THEN
1745 result := 'positive';
1746 ELSIF number < 0 THEN
1747 result := 'negative';
1749 -- hmm, the only other possibility is that number is null
1756 The key word <literal>ELSIF</> can also be spelled
1761 An alternative way of accomplishing the same task is to nest
1762 <literal>IF-THEN-ELSE</literal> statements, as in the
1766 IF demo_row.sex = 'm' THEN
1767 pretty_sex := 'man';
1769 IF demo_row.sex = 'f' THEN
1770 pretty_sex := 'woman';
1777 However, this method requires writing a matching <literal>END IF</>
1778 for each <literal>IF</>, so it is much more cumbersome than
1779 using <literal>ELSIF</> when there are many alternatives.
1784 <title>Simple <literal>CASE</></title>
1787 CASE <replaceable>search-expression</replaceable>
1788 WHEN <replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> <optional> ... </optional></optional> THEN
1789 <replaceable>statements</replaceable>
1790 <optional> WHEN <replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> <optional> ... </optional></optional> THEN
1791 <replaceable>statements</replaceable>
1794 <replaceable>statements</replaceable> </optional>
1799 The simple form of <command>CASE</> provides conditional execution
1800 based on equality of operands. The <replaceable>search-expression</>
1801 is evaluated (once) and successively compared to each
1802 <replaceable>expression</> in the <literal>WHEN</> clauses.
1803 If a match is found, then the corresponding
1804 <replaceable>statements</replaceable> are executed, and then control
1805 passes to the next statement after <literal>END CASE</>. (Subsequent
1806 <literal>WHEN</> expressions are not evaluated.) If no match is
1807 found, the <literal>ELSE</> <replaceable>statements</replaceable> are
1808 executed; but if <literal>ELSE</> is not present, then a
1809 <literal>CASE_NOT_FOUND</literal> exception is raised.
1813 Here is a simple example:
1818 msg := 'one or two';
1820 msg := 'other value than one or two';
1827 <title>Searched <literal>CASE</></title>
1831 WHEN <replaceable>boolean-expression</replaceable> THEN
1832 <replaceable>statements</replaceable>
1833 <optional> WHEN <replaceable>boolean-expression</replaceable> THEN
1834 <replaceable>statements</replaceable>
1837 <replaceable>statements</replaceable> </optional>
1842 The searched form of <command>CASE</> provides conditional execution
1843 based on truth of boolean expressions. Each <literal>WHEN</> clause's
1844 <replaceable>boolean-expression</replaceable> is evaluated in turn,
1845 until one is found that yields <literal>true</>. Then the
1846 corresponding <replaceable>statements</replaceable> are executed, and
1847 then control passes to the next statement after <literal>END CASE</>.
1848 (Subsequent <literal>WHEN</> expressions are not evaluated.)
1849 If no true result is found, the <literal>ELSE</>
1850 <replaceable>statements</replaceable> are executed;
1851 but if <literal>ELSE</> is not present, then a
1852 <literal>CASE_NOT_FOUND</literal> exception is raised.
1860 WHEN x BETWEEN 0 AND 10 THEN
1861 msg := 'value is between zero and ten';
1862 WHEN x BETWEEN 11 AND 20 THEN
1863 msg := 'value is between eleven and twenty';
1869 This form of <command>CASE</> is entirely equivalent to
1870 <literal>IF-THEN-ELSIF</>, except for the rule that reaching
1871 an omitted <literal>ELSE</> clause results in an error rather
1878 <sect2 id="plpgsql-control-structures-loops">
1879 <title>Simple Loops</title>
1881 <indexterm zone="plpgsql-control-structures-loops">
1882 <primary>loop</primary>
1883 <secondary>in PL/pgSQL</secondary>
1887 With the <literal>LOOP</>, <literal>EXIT</>,
1888 <literal>CONTINUE</>, <literal>WHILE</>, and <literal>FOR</>
1889 statements, you can arrange for your <application>PL/pgSQL</>
1890 function to repeat a series of commands.
1894 <title><literal>LOOP</></title>
1897 <optional> <<<replaceable>label</replaceable>>> </optional>
1899 <replaceable>statements</replaceable>
1900 END LOOP <optional> <replaceable>label</replaceable> </optional>;
1904 <literal>LOOP</> defines an unconditional loop that is repeated
1905 indefinitely until terminated by an <literal>EXIT</> or
1906 <command>RETURN</command> statement. The optional
1907 <replaceable>label</replaceable> can be used by <literal>EXIT</>
1908 and <literal>CONTINUE</literal> statements within nested loops to
1909 specify which loop those statements refer to.
1914 <title><literal>EXIT</></title>
1917 <primary>EXIT</primary>
1918 <secondary>in PL/pgSQL</secondary>
1922 EXIT <optional> <replaceable>label</replaceable> </optional> <optional> WHEN <replaceable>boolean-expression</replaceable> </optional>;
1926 If no <replaceable>label</replaceable> is given, the innermost
1927 loop is terminated and the statement following <literal>END
1928 LOOP</> is executed next. If <replaceable>label</replaceable>
1929 is given, it must be the label of the current or some outer
1930 level of nested loop or block. Then the named loop or block is
1931 terminated and control continues with the statement after the
1932 loop's/block's corresponding <literal>END</>.
1936 If <literal>WHEN</> is specified, the loop exit occurs only if
1937 <replaceable>boolean-expression</> is true. Otherwise, control passes
1938 to the statement after <literal>EXIT</>.
1942 <literal>EXIT</> can be used with all types of loops; it is
1943 not limited to use with unconditional loops.
1948 <literal>BEGIN</literal> block, <literal>EXIT</literal> passes
1949 control to the next statement after the end of the block.
1950 Note that a label must be used for this purpose; an unlabelled
1951 <literal>EXIT</literal> is never considered to match a
1952 <literal>BEGIN</literal> block. (This is a change from
1953 pre-8.4 releases of <productname>PostgreSQL</productname>, which
1954 would allow an unlabelled <literal>EXIT</literal> to match
1955 a <literal>BEGIN</literal> block.)
1962 -- some computations
1963 IF count > 0 THEN
1969 -- some computations
1970 EXIT WHEN count > 0; -- same result as previous example
1973 <<ablock>>
1975 -- some computations
1976 IF stocks > 100000 THEN
1977 EXIT ablock; -- causes exit from the BEGIN block
1979 -- computations here will be skipped when stocks > 100000
1986 <title><literal>CONTINUE</></title>
1989 <primary>CONTINUE</primary>
1990 <secondary>in PL/pgSQL</secondary>
1994 CONTINUE <optional> <replaceable>label</replaceable> </optional> <optional> WHEN <replaceable>boolean-expression</replaceable> </optional>;
1998 If no <replaceable>label</> is given, the next iteration of
1999 the innermost loop is begun. That is, all statements remaining
2000 in the loop body are skipped, and control returns
2001 to the loop control expression (if any) to determine whether
2002 another loop iteration is needed.
2003 If <replaceable>label</> is present, it
2004 specifies the label of the loop whose execution will be
2009 If <literal>WHEN</> is specified, the next iteration of the
2010 loop is begun only if <replaceable>boolean-expression</> is
2011 true. Otherwise, control passes to the statement after
2012 <literal>CONTINUE</>.
2016 <literal>CONTINUE</> can be used with all types of loops; it
2017 is not limited to use with unconditional loops.
2024 -- some computations
2025 EXIT WHEN count > 100;
2026 CONTINUE WHEN count < 50;
2027 -- some computations for count IN [50 .. 100]
2035 <title><literal>WHILE</></title>
2038 <primary>WHILE</primary>
2039 <secondary>in PL/pgSQL</secondary>
2043 <optional> <<<replaceable>label</replaceable>>> </optional>
2044 WHILE <replaceable>boolean-expression</replaceable> LOOP
2045 <replaceable>statements</replaceable>
2046 END LOOP <optional> <replaceable>label</replaceable> </optional>;
2050 The <literal>WHILE</> statement repeats a
2051 sequence of statements so long as the
2052 <replaceable>boolean-expression</replaceable>
2053 evaluates to true. The expression is checked just before
2054 each entry to the loop body.
2060 WHILE amount_owed > 0 AND gift_certificate_balance > 0 LOOP
2061 -- some computations here
2065 -- some computations here
2071 <sect3 id="plpgsql-integer-for">
2072 <title><literal>FOR</> (integer variant)</title>
2075 <optional> <<<replaceable>label</replaceable>>> </optional>
2076 FOR <replaceable>name</replaceable> IN <optional> REVERSE </optional> <replaceable>expression</replaceable> .. <replaceable>expression</replaceable> <optional> BY <replaceable>expression</replaceable> </optional> LOOP
2077 <replaceable>statements</replaceable>
2078 END LOOP <optional> <replaceable>label</replaceable> </optional>;
2082 This form of <literal>FOR</> creates a loop that iterates over a range
2083 of integer values. The variable
2084 <replaceable>name</replaceable> is automatically defined as type
2085 <type>integer</> and exists only inside the loop (any existing
2086 definition of the variable name is ignored within the loop).
2087 The two expressions giving
2088 the lower and upper bound of the range are evaluated once when entering
2089 the loop. If the <literal>BY</> clause isn't specified the iteration
2090 step is 1, otherwise it's the value specified in the <literal>BY</>
2091 clause, which again is evaluated once on loop entry.
2092 If <literal>REVERSE</> is specified then the step value is
2093 subtracted, rather than added, after each iteration.
2097 Some examples of integer <literal>FOR</> loops:
2100 -- i will take on the values 1,2,3,4,5,6,7,8,9,10 within the loop
2103 FOR i IN REVERSE 10..1 LOOP
2104 -- i will take on the values 10,9,8,7,6,5,4,3,2,1 within the loop
2107 FOR i IN REVERSE 10..1 BY 2 LOOP
2108 -- i will take on the values 10,8,6,4,2 within the loop
2114 If the lower bound is greater than the upper bound (or less than,
2115 in the <literal>REVERSE</> case), the loop body is not
2116 executed at all. No error is raised.
2120 If a <replaceable>label</replaceable> is attached to the
2121 <literal>FOR</> loop then the integer loop variable can be
2122 referenced with a qualified name, using that
2123 <replaceable>label</replaceable>.
2128 <sect2 id="plpgsql-records-iterating">
2129 <title>Looping Through Query Results</title>
2132 Using a different type of <literal>FOR</> loop, you can iterate through
2133 the results of a query and manipulate that data
2134 accordingly. The syntax is:
2136 <optional> <<<replaceable>label</replaceable>>> </optional>
2137 FOR <replaceable>target</replaceable> IN <replaceable>query</replaceable> LOOP
2138 <replaceable>statements</replaceable>
2139 END LOOP <optional> <replaceable>label</replaceable> </optional>;
2141 The <replaceable>target</replaceable> is a record variable, row variable,
2142 or comma-separated list of scalar variables.
2143 The <replaceable>target</replaceable> is successively assigned each row
2144 resulting from the <replaceable>query</replaceable> and the loop body is
2145 executed for each row. Here is an example:
2147 CREATE FUNCTION cs_refresh_mviews() RETURNS integer AS $$
2151 PERFORM cs_log('Refreshing materialized views...');
2153 FOR mviews IN SELECT * FROM cs_materialized_views ORDER BY sort_key LOOP
2155 -- Now "mviews" has one record from cs_materialized_views
2157 PERFORM cs_log('Refreshing materialized view ' || quote_ident(mviews.mv_name) || ' ...');
2158 EXECUTE 'TRUNCATE TABLE ' || quote_ident(mviews.mv_name);
2159 EXECUTE 'INSERT INTO ' || quote_ident(mviews.mv_name) || ' ' || mviews.mv_query;
2162 PERFORM cs_log('Done refreshing materialized views.');
2165 $$ LANGUAGE plpgsql;
2168 If the loop is terminated by an <literal>EXIT</> statement, the last
2169 assigned row value is still accessible after the loop.
2173 The <replaceable>query</replaceable> used in this type of <literal>FOR</>
2174 statement can be any SQL command that returns rows to the caller:
2175 <command>SELECT</> is the most common case,
2176 but you can also use <command>INSERT</>, <command>UPDATE</>, or
2177 <command>DELETE</> with a <literal>RETURNING</> clause. Some utility
2178 commands such as <command>EXPLAIN</> will work too.
2182 <application>PL/pgSQL</> variables are substituted into the query text,
2183 and the query plan is cached for possible re-use, as discussed in
2184 detail in <xref linkend="plpgsql-var-subst"> and
2185 <xref linkend="plpgsql-plan-caching">.
2189 The <literal>FOR-IN-EXECUTE</> statement is another way to iterate over
2192 <optional> <<<replaceable>label</replaceable>>> </optional>
2193 FOR <replaceable>target</replaceable> IN EXECUTE <replaceable>text_expression</replaceable> <optional> USING <replaceable>expression</replaceable> <optional>, ...</optional> </optional> LOOP
2194 <replaceable>statements</replaceable>
2195 END LOOP <optional> <replaceable>label</replaceable> </optional>;
2197 This is like the previous form, except that the source query
2198 is specified as a string expression, which is evaluated and replanned
2199 on each entry to the <literal>FOR</> loop. This allows the programmer to
2200 choose the speed of a preplanned query or the flexibility of a dynamic
2201 query, just as with a plain <command>EXECUTE</command> statement.
2202 As with <command>EXECUTE</command>, parameter values can be inserted
2203 into the dynamic command via <literal>USING</>.
2207 Another way to specify the query whose results should be iterated
2208 through is to declare it as a cursor. This is described in
2209 <xref linkend="plpgsql-cursor-for-loop">.
2213 <sect2 id="plpgsql-error-trapping">
2214 <title>Trapping Errors</title>
2217 <primary>exceptions</primary>
2218 <secondary>in PL/PgSQL</secondary>
2222 By default, any error occurring in a <application>PL/pgSQL</>
2223 function aborts execution of the function, and indeed of the
2224 surrounding transaction as well. You can trap errors and recover
2225 from them by using a <command>BEGIN</> block with an
2226 <literal>EXCEPTION</> clause. The syntax is an extension of the
2227 normal syntax for a <command>BEGIN</> block:
2230 <optional> <<<replaceable>label</replaceable>>> </optional>
2232 <replaceable>declarations</replaceable> </optional>
2234 <replaceable>statements</replaceable>
2236 WHEN <replaceable>condition</replaceable> <optional> OR <replaceable>condition</replaceable> ... </optional> THEN
2237 <replaceable>handler_statements</replaceable>
2238 <optional> WHEN <replaceable>condition</replaceable> <optional> OR <replaceable>condition</replaceable> ... </optional> THEN
2239 <replaceable>handler_statements</replaceable>
2246 If no error occurs, this form of block simply executes all the
2247 <replaceable>statements</replaceable>, and then control passes
2248 to the next statement after <literal>END</>. But if an error
2249 occurs within the <replaceable>statements</replaceable>, further
2250 processing of the <replaceable>statements</replaceable> is
2251 abandoned, and control passes to the <literal>EXCEPTION</> list.
2252 The list is searched for the first <replaceable>condition</replaceable>
2253 matching the error that occurred. If a match is found, the
2254 corresponding <replaceable>handler_statements</replaceable> are
2255 executed, and then control passes to the next statement after
2256 <literal>END</>. If no match is found, the error propagates out
2257 as though the <literal>EXCEPTION</> clause were not there at all:
2258 the error can be caught by an enclosing block with
2259 <literal>EXCEPTION</>, or if there is none it aborts processing
2264 The <replaceable>condition</replaceable> names can be any of
2265 those shown in <xref linkend="errcodes-appendix">. A category
2266 name matches any error within its category. The special
2267 condition name <literal>OTHERS</> matches every error type except
2268 <literal>QUERY_CANCELED</>. (It is possible, but often unwise,
2269 to trap <literal>QUERY_CANCELED</> by name.) Condition names are
2270 not case-sensitive. Also, an error condition can be specified
2271 by <literal>SQLSTATE</> code; for example these are equivalent:
2273 WHEN division_by_zero THEN ...
2274 WHEN SQLSTATE '22012' THEN ...
2279 If a new error occurs within the selected
2280 <replaceable>handler_statements</replaceable>, it cannot be caught
2281 by this <literal>EXCEPTION</> clause, but is propagated out.
2282 A surrounding <literal>EXCEPTION</> clause could catch it.
2286 When an error is caught by an <literal>EXCEPTION</> clause,
2287 the local variables of the <application>PL/pgSQL</> function
2288 remain as they were when the error occurred, but all changes
2289 to persistent database state within the block are rolled back.
2290 As an example, consider this fragment:
2293 INSERT INTO mytab(firstname, lastname) VALUES('Tom', 'Jones');
2295 UPDATE mytab SET firstname = 'Joe' WHERE lastname = 'Jones';
2299 WHEN division_by_zero THEN
2300 RAISE NOTICE 'caught division_by_zero';
2305 When control reaches the assignment to <literal>y</>, it will
2306 fail with a <literal>division_by_zero</> error. This will be caught by
2307 the <literal>EXCEPTION</> clause. The value returned in the
2308 <command>RETURN</> statement will be the incremented value of
2309 <literal>x</>, but the effects of the <command>UPDATE</> command will
2310 have been rolled back. The <command>INSERT</> command preceding the
2311 block is not rolled back, however, so the end result is that the database
2312 contains <literal>Tom Jones</> not <literal>Joe Jones</>.
2317 A block containing an <literal>EXCEPTION</> clause is significantly
2318 more expensive to enter and exit than a block without one. Therefore,
2319 don't use <literal>EXCEPTION</> without need.
2324 Within an exception handler, the <varname>SQLSTATE</varname>
2325 variable contains the error code that corresponds to the
2326 exception that was raised (refer to <xref
2327 linkend="errcodes-table"> for a list of possible error
2328 codes). The <varname>SQLERRM</varname> variable contains the
2329 error message associated with the exception. These variables are
2330 undefined outside exception handlers.
2333 <example id="plpgsql-upsert-example">
2334 <title>Exceptions with <command>UPDATE</>/<command>INSERT</></title>
2337 This example uses exception handling to perform either
2338 <command>UPDATE</> or <command>INSERT</>, as appropriate:
2341 CREATE TABLE db (a INT PRIMARY KEY, b TEXT);
2343 CREATE FUNCTION merge_db(key INT, data TEXT) RETURNS VOID AS
2347 -- first try to update the key
2348 UPDATE db SET b = data WHERE a = key;
2352 -- not there, so try to insert the key
2353 -- if someone else inserts the same key concurrently,
2354 -- we could get a unique-key failure
2356 INSERT INTO db(a,b) VALUES (key, data);
2358 EXCEPTION WHEN unique_violation THEN
2359 -- do nothing, and loop to try the UPDATE again
2366 SELECT merge_db(1, 'david');
2367 SELECT merge_db(1, 'dennis');
2375 <sect1 id="plpgsql-cursors">
2376 <title>Cursors</title>
2378 <indexterm zone="plpgsql-cursors">
2379 <primary>cursor</primary>
2380 <secondary>in PL/pgSQL</secondary>
2384 Rather than executing a whole query at once, it is possible to set
2385 up a <firstterm>cursor</> that encapsulates the query, and then read
2386 the query result a few rows at a time. One reason for doing this is
2387 to avoid memory overrun when the result contains a large number of
2388 rows. (However, <application>PL/pgSQL</> users do not normally need
2389 to worry about that, since <literal>FOR</> loops automatically use a cursor
2390 internally to avoid memory problems.) A more interesting usage is to
2391 return a reference to a cursor that a function has created, allowing the
2392 caller to read the rows. This provides an efficient way to return
2393 large row sets from functions.
2396 <sect2 id="plpgsql-cursor-declarations">
2397 <title>Declaring Cursor Variables</title>
2400 All access to cursors in <application>PL/pgSQL</> goes through
2401 cursor variables, which are always of the special data type
2402 <type>refcursor</>. One way to create a cursor variable
2403 is just to declare it as a variable of type <type>refcursor</>.
2404 Another way is to use the cursor declaration syntax,
2405 which in general is:
2407 <replaceable>name</replaceable> <optional> <optional> NO </optional> SCROLL </optional> CURSOR <optional> ( <replaceable>arguments</replaceable> ) </optional> FOR <replaceable>query</replaceable>;
2409 (<literal>FOR</> can be replaced by <literal>IS</> for
2410 <productname>Oracle</productname> compatibility.)
2411 If <literal>SCROLL</> is specified, the cursor will be capable of
2412 scrolling backward; if <literal>NO SCROLL</> is specified, backward
2413 fetches will be rejected; if neither specification appears, it is
2414 query-dependent whether backward fetches will be allowed.
2415 <replaceable>arguments</replaceable>, if specified, is a
2416 comma-separated list of pairs <literal><replaceable>name</replaceable>
2417 <replaceable>datatype</replaceable></literal> that define names to be
2418 replaced by parameter values in the given query. The actual
2419 values to substitute for these names will be specified later,
2420 when the cursor is opened.
2427 curs2 CURSOR FOR SELECT * FROM tenk1;
2428 curs3 CURSOR (key integer) IS SELECT * FROM tenk1 WHERE unique1 = key;
2430 All three of these variables have the data type <type>refcursor</>,
2431 but the first can be used with any query, while the second has
2432 a fully specified query already <firstterm>bound</> to it, and the last
2433 has a parameterized query bound to it. (<literal>key</> will be
2434 replaced by an integer parameter value when the cursor is opened.)
2435 The variable <literal>curs1</>
2436 is said to be <firstterm>unbound</> since it is not bound to
2437 any particular query.
2441 <sect2 id="plpgsql-cursor-opening">
2442 <title>Opening Cursors</title>
2445 Before a cursor can be used to retrieve rows, it must be
2446 <firstterm>opened</>. (This is the equivalent action to the SQL
2447 command <command>DECLARE CURSOR</>.) <application>PL/pgSQL</> has
2448 three forms of the <command>OPEN</> statement, two of which use unbound
2449 cursor variables while the third uses a bound cursor variable.
2454 Bound cursor variables can also be used without explicitly opening the cursor,
2455 via the <command>FOR</> statement described in
2456 <xref linkend="plpgsql-cursor-for-loop">.
2461 <title><command>OPEN FOR</command> <replaceable>query</replaceable></title>
2464 OPEN <replaceable>unbound_cursorvar</replaceable> <optional> <optional> NO </optional> SCROLL </optional> FOR <replaceable>query</replaceable>;
2468 The cursor variable is opened and given the specified query to
2469 execute. The cursor cannot be open already, and it must have been
2470 declared as an unbound cursor variable (that is, as a simple
2471 <type>refcursor</> variable). The query must be a
2472 <command>SELECT</command>, or something else that returns rows
2473 (such as <command>EXPLAIN</>). The query
2474 is treated in the same way as other SQL commands in
2475 <application>PL/pgSQL</>: <application>PL/pgSQL</>
2476 variable names are substituted, and the query plan is cached for
2477 possible reuse. When a <application>PL/pgSQL</>
2478 variable is substituted into the cursor query, the value that is
2479 substituted is the one it has at the time of the <command>OPEN</>;
2480 subsequent changes to the variable will not affect the cursor's
2482 The <literal>SCROLL</> and <literal>NO SCROLL</>
2483 options have the same meanings as for a bound cursor.
2489 OPEN curs1 FOR SELECT * FROM foo WHERE key = mykey;
2495 <title><command>OPEN FOR EXECUTE</command></title>
2498 OPEN <replaceable>unbound_cursorvar</replaceable> <optional> <optional> NO </optional> SCROLL </optional> FOR EXECUTE <replaceable class="command">query_string</replaceable>;
2502 The cursor variable is opened and given the specified query to
2503 execute. The cursor cannot be open already, and it must have been
2504 declared as an unbound cursor variable (that is, as a simple
2505 <type>refcursor</> variable). The query is specified as a string
2506 expression, in the same way as in the <command>EXECUTE</command>
2507 command. As usual, this gives flexibility so the query plan can vary
2508 from one run to the next (see <xref linkend="plpgsql-plan-caching">),
2509 and it also means that variable substitution is not done on the
2511 The <literal>SCROLL</> and
2512 <literal>NO SCROLL</> options have the same meanings as for a bound
2519 OPEN curs1 FOR EXECUTE 'SELECT * FROM ' || quote_ident($1);
2525 <title>Opening a Bound Cursor</title>
2528 OPEN <replaceable>bound_cursorvar</replaceable> <optional> ( <replaceable>argument_values</replaceable> ) </optional>;
2532 This form of <command>OPEN</command> is used to open a cursor
2533 variable whose query was bound to it when it was declared. The
2534 cursor cannot be open already. A list of actual argument value
2535 expressions must appear if and only if the cursor was declared to
2536 take arguments. These values will be substituted in the query.
2537 The query plan for a bound cursor is always considered cacheable;
2538 there is no equivalent of <command>EXECUTE</command> in this case.
2539 Notice that <literal>SCROLL</> and
2540 <literal>NO SCROLL</> cannot be specified, as the cursor's scrolling
2541 behavior was already determined.
2545 Note that because variable substitution is done on the bound
2546 cursor's query, there are two ways to pass values into the cursor:
2547 either with an explicit argument to <command>OPEN</>, or
2548 implicitly by referencing a <application>PL/pgSQL</> variable
2549 in the query. However, only variables declared before the bound
2550 cursor was declared will be substituted into it. In either case
2551 the value to be passed is determined at the time of the
2565 <sect2 id="plpgsql-cursor-using">
2566 <title>Using Cursors</title>
2569 Once a cursor has been opened, it can be manipulated with the
2570 statements described here.
2574 These manipulations need not occur in the same function that
2575 opened the cursor to begin with. You can return a <type>refcursor</>
2576 value out of a function and let the caller operate on the cursor.
2577 (Internally, a <type>refcursor</> value is simply the string name
2578 of a so-called portal containing the active query for the cursor. This name
2579 can be passed around, assigned to other <type>refcursor</> variables,
2580 and so on, without disturbing the portal.)
2584 All portals are implicitly closed at transaction end. Therefore
2585 a <type>refcursor</> value is usable to reference an open cursor
2586 only until the end of the transaction.
2590 <title><literal>FETCH</></title>
2593 FETCH <optional> <replaceable>direction</replaceable> { FROM | IN } </optional> <replaceable>cursor</replaceable> INTO <replaceable>target</replaceable>;
2597 <command>FETCH</command> retrieves the next row from the
2598 cursor into a target, which might be a row variable, a record
2599 variable, or a comma-separated list of simple variables, just like
2600 <command>SELECT INTO</command>. If there is no next row, the
2601 target is set to NULL(s). As with <command>SELECT
2602 INTO</command>, the special variable <literal>FOUND</literal> can
2603 be checked to see whether a row was obtained or not.
2607 The <replaceable>direction</replaceable> clause can be any of the
2608 variants allowed in the SQL <xref linkend="sql-fetch"
2609 endterm="sql-fetch-title"> command except the ones that can fetch
2610 more than one row; namely, it can be
2615 <literal>ABSOLUTE</> <replaceable>count</replaceable>,
2616 <literal>RELATIVE</> <replaceable>count</replaceable>,
2617 <literal>FORWARD</>, or
2618 <literal>BACKWARD</>.
2619 Omitting <replaceable>direction</replaceable> is the same
2620 as specifying <literal>NEXT</>.
2621 <replaceable>direction</replaceable> values that require moving
2622 backward are likely to fail unless the cursor was declared or opened
2623 with the <literal>SCROLL</> option.
2627 <replaceable>cursor</replaceable> must be the name of a <type>refcursor</>
2628 variable that references an open cursor portal.
2634 FETCH curs1 INTO rowvar;
2635 FETCH curs2 INTO foo, bar, baz;
2636 FETCH LAST FROM curs3 INTO x, y;
2637 FETCH RELATIVE -2 FROM curs4 INTO x;
2643 <title><literal>MOVE</></title>
2646 MOVE <optional> <replaceable>direction</replaceable> { FROM | IN } </optional> <replaceable>cursor</replaceable>;
2650 <command>MOVE</command> repositions a cursor without retrieving
2651 any data. <command>MOVE</command> works exactly like the
2652 <command>FETCH</command> command, except it only repositions the
2653 cursor and does not return the row moved to. As with <command>SELECT
2654 INTO</command>, the special variable <literal>FOUND</literal> can
2655 be checked to see whether there was a next row to move to.
2659 The <replaceable>direction</replaceable> clause can be any of the
2660 variants allowed in the SQL <xref linkend="sql-fetch"
2661 endterm="sql-fetch-title"> command, namely
2666 <literal>ABSOLUTE</> <replaceable>count</replaceable>,
2667 <literal>RELATIVE</> <replaceable>count</replaceable>,
2669 <literal>FORWARD</> <optional> <replaceable>count</replaceable> | <literal>ALL</> </optional>, or
2670 <literal>BACKWARD</> <optional> <replaceable>count</replaceable> | <literal>ALL</> </optional>.
2671 Omitting <replaceable>direction</replaceable> is the same
2672 as specifying <literal>NEXT</>.
2673 <replaceable>direction</replaceable> values that require moving
2674 backward are likely to fail unless the cursor was declared or opened
2675 with the <literal>SCROLL</> option.
2682 MOVE LAST FROM curs3;
2683 MOVE RELATIVE -2 FROM curs4;
2684 MOVE FORWARD 2 FROM curs4;
2690 <title><literal>UPDATE/DELETE WHERE CURRENT OF</></title>
2693 UPDATE <replaceable>table</replaceable> SET ... WHERE CURRENT OF <replaceable>cursor</replaceable>;
2694 DELETE FROM <replaceable>table</replaceable> WHERE CURRENT OF <replaceable>cursor</replaceable>;
2698 When a cursor is positioned on a table row, that row can be updated
2699 or deleted using the cursor to identify the row. There are
2700 restrictions on what the cursor's query can be (in particular,
2701 no grouping) and it's best to use <literal>FOR UPDATE</> in the
2702 cursor. For more information see the
2703 <xref linkend="sql-declare" endterm="sql-declare-title">
2710 UPDATE foo SET dataval = myval WHERE CURRENT OF curs1;
2716 <title><literal>CLOSE</></title>
2719 CLOSE <replaceable>cursor</replaceable>;
2723 <command>CLOSE</command> closes the portal underlying an open
2724 cursor. This can be used to release resources earlier than end of
2725 transaction, or to free up the cursor variable to be opened again.
2737 <title>Returning Cursors</title>
2740 <application>PL/pgSQL</> functions can return cursors to the
2741 caller. This is useful to return multiple rows or columns,
2742 especially with very large result sets. To do this, the function
2743 opens the cursor and returns the cursor name to the caller (or simply
2744 opens the cursor using a portal name specified by or otherwise known
2745 to the caller). The caller can then fetch rows from the cursor. The
2746 cursor can be closed by the caller, or it will be closed automatically
2747 when the transaction closes.
2751 The portal name used for a cursor can be specified by the
2752 programmer or automatically generated. To specify a portal name,
2753 simply assign a string to the <type>refcursor</> variable before
2754 opening it. The string value of the <type>refcursor</> variable
2755 will be used by <command>OPEN</> as the name of the underlying portal.
2756 However, if the <type>refcursor</> variable is null,
2757 <command>OPEN</> automatically generates a name that does not
2758 conflict with any existing portal, and assigns it to the
2759 <type>refcursor</> variable.
2764 A bound cursor variable is initialized to the string value
2765 representing its name, so that the portal name is the same as
2766 the cursor variable name, unless the programmer overrides it
2767 by assignment before opening the cursor. But an unbound cursor
2768 variable defaults to the null value initially, so it will receive
2769 an automatically-generated unique name, unless overridden.
2774 The following example shows one way a cursor name can be supplied by
2778 CREATE TABLE test (col text);
2779 INSERT INTO test VALUES ('123');
2781 CREATE FUNCTION reffunc(refcursor) RETURNS refcursor AS '
2783 OPEN $1 FOR SELECT col FROM test;
2789 SELECT reffunc('funccursor');
2790 FETCH ALL IN funccursor;
2796 The following example uses automatic cursor name generation:
2799 CREATE FUNCTION reffunc2() RETURNS refcursor AS '
2803 OPEN ref FOR SELECT col FROM test;
2812 --------------------
2813 <unnamed cursor 1>
2816 FETCH ALL IN "<unnamed cursor 1>";
2822 The following example shows one way to return multiple cursors
2823 from a single function:
2826 CREATE FUNCTION myfunc(refcursor, refcursor) RETURNS SETOF refcursor AS $$
2828 OPEN $1 FOR SELECT * FROM table_1;
2830 OPEN $2 FOR SELECT * FROM table_2;
2833 $$ LANGUAGE plpgsql;
2835 -- need to be in a transaction to use cursors.
2838 SELECT * FROM myfunc('a', 'b');
2848 <sect2 id="plpgsql-cursor-for-loop">
2849 <title>Looping Through a Cursor's Result</title>
2852 There is a variant of the <command>FOR</> statement that allows
2853 iterating through the rows returned by a cursor. The syntax is:
2856 <optional> <<<replaceable>label</replaceable>>> </optional>
2857 FOR <replaceable>recordvar</replaceable> IN <replaceable>bound_cursorvar</replaceable> <optional> ( <replaceable>argument_values</replaceable> ) </optional> LOOP
2858 <replaceable>statements</replaceable>
2859 END LOOP <optional> <replaceable>label</replaceable> </optional>;
2862 The cursor variable must have been bound to some query when it was
2863 declared, and it <emphasis>cannot</> be open already. The
2864 <command>FOR</> statement automatically opens the cursor, and it closes
2865 the cursor again when the loop exits. A list of actual argument value
2866 expressions must appear if and only if the cursor was declared to take
2867 arguments. These values will be substituted in the query, in just
2868 the same way as during an <command>OPEN</>.
2869 The variable <replaceable>recordvar</replaceable> is automatically
2870 defined as type <type>record</> and exists only inside the loop (any
2871 existing definition of the variable name is ignored within the loop).
2872 Each row returned by the cursor is successively assigned to this
2873 record variable and the loop body is executed.
2879 <sect1 id="plpgsql-errors-and-messages">
2880 <title>Errors and Messages</title>
2883 <primary>RAISE</primary>
2887 <primary>reporting errors</primary>
2888 <secondary>in PL/PgSQL</secondary>
2892 Use the <command>RAISE</command> statement to report messages and
2896 RAISE <optional> <replaceable class="parameter">level</replaceable> </optional> '<replaceable class="parameter">format</replaceable>' <optional>, <replaceable class="parameter">expression</replaceable> <optional>, ...</optional></optional> <optional> USING <replaceable class="parameter">option</replaceable> = <replaceable class="parameter">expression</replaceable> <optional>, ... </optional> </optional>;
2897 RAISE <optional> <replaceable class="parameter">level</replaceable> </optional> <replaceable class="parameter">condition_name</> <optional> USING <replaceable class="parameter">option</replaceable> = <replaceable class="parameter">expression</replaceable> <optional>, ... </optional> </optional>;
2898 RAISE <optional> <replaceable class="parameter">level</replaceable> </optional> SQLSTATE '<replaceable class="parameter">sqlstate</>' <optional> USING <replaceable class="parameter">option</replaceable> = <replaceable class="parameter">expression</replaceable> <optional>, ... </optional> </optional>;
2899 RAISE <optional> <replaceable class="parameter">level</replaceable> </optional> USING <replaceable class="parameter">option</replaceable> = <replaceable class="parameter">expression</replaceable> <optional>, ... </optional>;
2903 The <replaceable class="parameter">level</replaceable> option specifies
2904 the error severity. Allowed levels are <literal>DEBUG</literal>,
2905 <literal>LOG</literal>, <literal>INFO</literal>,
2906 <literal>NOTICE</literal>, <literal>WARNING</literal>,
2907 and <literal>EXCEPTION</literal>, with <literal>EXCEPTION</literal>
2909 <literal>EXCEPTION</literal> raises an error (which normally aborts the
2910 current transaction); the other levels only generate messages of different
2912 Whether messages of a particular priority are reported to the client,
2913 written to the server log, or both is controlled by the
2914 <xref linkend="guc-log-min-messages"> and
2915 <xref linkend="guc-client-min-messages"> configuration
2916 variables. See <xref linkend="runtime-config"> for more
2921 After <replaceable class="parameter">level</replaceable> if any,
2922 you can write a <replaceable class="parameter">format</replaceable>
2923 (which must be a simple string literal, not an expression). The
2924 format string specifies the error message text to be reported.
2925 The format string can be followed
2926 by optional argument expressions to be inserted into the message.
2927 Inside the format string, <literal>%</literal> is replaced by the
2928 string representation of the next optional argument's value. Write
2929 <literal>%%</literal> to emit a literal <literal>%</literal>.
2933 In this example, the value of <literal>v_job_id</> will replace the
2934 <literal>%</literal> in the string:
2936 RAISE NOTICE 'Calling cs_create_job(%)', v_job_id;
2941 You can attach additional information to the error report by writing
2942 <literal>USING</> followed by <replaceable
2943 class="parameter">option</replaceable> = <replaceable
2944 class="parameter">expression</replaceable> items. The allowed
2945 <replaceable class="parameter">option</replaceable> keywords are
2946 <literal>MESSAGE</>, <literal>DETAIL</>, <literal>HINT</>, and
2947 <literal>ERRCODE</>, while each <replaceable
2948 class="parameter">expression</replaceable> can be any string-valued
2950 <literal>MESSAGE</> sets the error message text (this option can't
2951 be used in the form of <command>RAISE</> that includes a format
2952 string before <literal>USING</>).
2953 <literal>DETAIL</> supplies an error detail message, while
2954 <literal>HINT</> supplies a hint message.
2955 <literal>ERRCODE</> specifies the error code (SQLSTATE) to report,
2956 either by condition name as shown in <xref linkend="errcodes-appendix">,
2957 or directly as a five-character SQLSTATE code.
2961 This example will abort the transaction with the given error message
2964 RAISE EXCEPTION 'Nonexistent ID --> %', user_id USING HINT = 'Please check your user id';
2969 These two examples show equivalent ways of setting the SQLSTATE:
2971 RAISE 'Duplicate user ID: %', user_id USING ERRCODE = 'unique_violation';
2972 RAISE 'Duplicate user ID: %', user_id USING ERRCODE = '23505';
2977 There is a second <command>RAISE</> syntax in which the main argument
2978 is the condition name or SQLSTATE to be reported, for example:
2980 RAISE division_by_zero;
2981 RAISE SQLSTATE '22012';
2983 In this syntax, <literal>USING</> can be used to supply a custom
2984 error message, detail, or hint. Another way to do the earlier
2987 RAISE unique_violation USING MESSAGE = 'Duplicate user ID: ' || user_id;
2992 Still another variant is to write <literal>RAISE USING</> or <literal>RAISE
2993 <replaceable class="parameter">level</replaceable> USING</> and put
2994 everything else into the <literal>USING</> list.
2998 The last variant of <command>RAISE</> has no parameters at all.
2999 This form can only be used inside a <literal>BEGIN</> block's
3000 <literal>EXCEPTION</> clause;
3001 it causes the error currently being handled to be re-thrown to the
3002 next enclosing block.
3006 If no condition name nor SQLSTATE is specified in a
3007 <command>RAISE EXCEPTION</command> command, the default is to use
3008 <literal>RAISE_EXCEPTION</> (<literal>P0001</>). If no message
3009 text is specified, the default is to use the condition name or
3010 SQLSTATE as message text.
3015 When specifying an error code by SQLSTATE code, you are not
3016 limited to the predefined error codes, but can select any
3017 error code consisting of five digits and/or upper-case ASCII
3018 letters, other than <literal>00000</>. It is recommended that
3019 you avoid throwing error codes that end in three zeroes, because
3020 these are category codes and can only be trapped by trapping
3027 <sect1 id="plpgsql-trigger">
3028 <title>Trigger Procedures</title>
3030 <indexterm zone="plpgsql-trigger">
3031 <primary>trigger</primary>
3032 <secondary>in PL/pgSQL</secondary>
3036 <application>PL/pgSQL</application> can be used to define trigger
3037 procedures. A trigger procedure is created with the
3038 <command>CREATE FUNCTION</> command, declaring it as a function with
3039 no arguments and a return type of <type>trigger</type>. Note that
3040 the function must be declared with no arguments even if it expects
3041 to receive arguments specified in <command>CREATE TRIGGER</> —
3042 trigger arguments are passed via <varname>TG_ARGV</>, as described
3047 When a <application>PL/pgSQL</application> function is called as a
3048 trigger, several special variables are created automatically in the
3049 top-level block. They are:
3053 <term><varname>NEW</varname></term>
3056 Data type <type>RECORD</type>; variable holding the new
3057 database row for <command>INSERT</>/<command>UPDATE</> operations in row-level
3058 triggers. This variable is <symbol>NULL</symbol> in statement-level triggers
3059 and for <command>DELETE</command> operations.
3065 <term><varname>OLD</varname></term>
3068 Data type <type>RECORD</type>; variable holding the old
3069 database row for <command>UPDATE</>/<command>DELETE</> operations in row-level
3070 triggers. This variable is <symbol>NULL</symbol> in statement-level triggers
3071 and for <command>INSERT</command> operations.
3077 <term><varname>TG_NAME</varname></term>
3080 Data type <type>name</type>; variable that contains the name of the trigger actually
3087 <term><varname>TG_WHEN</varname></term>
3090 Data type <type>text</type>; a string of either
3091 <literal>BEFORE</literal> or <literal>AFTER</literal>
3092 depending on the trigger's definition.
3098 <term><varname>TG_LEVEL</varname></term>
3101 Data type <type>text</type>; a string of either
3102 <literal>ROW</literal> or <literal>STATEMENT</literal>
3103 depending on the trigger's definition.
3109 <term><varname>TG_OP</varname></term>
3112 Data type <type>text</type>; a string of
3113 <literal>INSERT</literal>, <literal>UPDATE</literal>,
3114 <literal>DELETE</literal>, or <literal>TRUNCATE</>
3115 telling for which operation the trigger was fired.
3121 <term><varname>TG_RELID</varname></term>
3124 Data type <type>oid</type>; the object ID of the table that caused the
3131 <term><varname>TG_RELNAME</varname></term>
3134 Data type <type>name</type>; the name of the table that caused the trigger
3135 invocation. This is now deprecated, and could disappear in a future
3136 release. Use <literal>TG_TABLE_NAME</> instead.
3142 <term><varname>TG_TABLE_NAME</varname></term>
3145 Data type <type>name</type>; the name of the table that
3146 caused the trigger invocation.
3152 <term><varname>TG_TABLE_SCHEMA</varname></term>
3155 Data type <type>name</type>; the name of the schema of the
3156 table that caused the trigger invocation.
3162 <term><varname>TG_NARGS</varname></term>
3165 Data type <type>integer</type>; the number of arguments given to the trigger
3166 procedure in the <command>CREATE TRIGGER</command> statement.
3172 <term><varname>TG_ARGV[]</varname></term>
3175 Data type array of <type>text</type>; the arguments from
3176 the <command>CREATE TRIGGER</command> statement.
3177 The index counts from 0. Invalid
3178 indexes (less than 0 or greater than or equal to <varname>tg_nargs</>)
3179 result in a null value.
3187 A trigger function must return either <symbol>NULL</symbol> or a
3188 record/row value having exactly the structure of the table the
3189 trigger was fired for.
3193 Row-level triggers fired <literal>BEFORE</> can return null to signal the
3194 trigger manager to skip the rest of the operation for this row
3195 (i.e., subsequent triggers are not fired, and the
3196 <command>INSERT</>/<command>UPDATE</>/<command>DELETE</> does not occur
3197 for this row). If a nonnull
3198 value is returned then the operation proceeds with that row value.
3199 Returning a row value different from the original value
3200 of <varname>NEW</> alters the row that will be inserted or
3201 updated. Thus, if the trigger function wants the triggering
3202 action to succeed normally without altering the row
3203 value, <varname>NEW</varname> (or a value equal thereto) has to be
3204 returned. To alter the row to be stored, it is possible to
3205 replace single values directly in <varname>NEW</> and return the
3206 modified <varname>NEW</>, or to build a complete new record/row to
3207 return. In the case of a before-trigger
3208 on <command>DELETE</command>, the returned value has no direct
3209 effect, but it has to be nonnull to allow the trigger action to
3210 proceed. Note that <varname>NEW</varname> is null
3211 in <command>DELETE</command> triggers, so returning that is
3212 usually not sensible. A useful idiom in <command>DELETE</command>
3213 triggers might be to return <varname>OLD</varname>.
3217 The return value of a row-level trigger
3218 fired <literal>AFTER</literal> or a statement-level trigger
3219 fired <literal>BEFORE</> or <literal>AFTER</> is
3220 always ignored; it might as well be null. However, any of these types of
3221 triggers might still abort the entire operation by raising an error.
3225 <xref linkend="plpgsql-trigger-example"> shows an example of a
3226 trigger procedure in <application>PL/pgSQL</application>.
3229 <example id="plpgsql-trigger-example">
3230 <title>A <application>PL/pgSQL</application> Trigger Procedure</title>
3233 This example trigger ensures that any time a row is inserted or updated
3234 in the table, the current user name and time are stamped into the
3235 row. And it checks that an employee's name is given and that the
3236 salary is a positive value.
3243 last_date timestamp,
3247 CREATE FUNCTION emp_stamp() RETURNS trigger AS $emp_stamp$
3249 -- Check that empname and salary are given
3250 IF NEW.empname IS NULL THEN
3251 RAISE EXCEPTION 'empname cannot be null';
3253 IF NEW.salary IS NULL THEN
3254 RAISE EXCEPTION '% cannot have null salary', NEW.empname;
3257 -- Who works for us when she must pay for it?
3258 IF NEW.salary < 0 THEN
3259 RAISE EXCEPTION '% cannot have a negative salary', NEW.empname;
3262 -- Remember who changed the payroll when
3263 NEW.last_date := current_timestamp;
3264 NEW.last_user := current_user;
3267 $emp_stamp$ LANGUAGE plpgsql;
3269 CREATE TRIGGER emp_stamp BEFORE INSERT OR UPDATE ON emp
3270 FOR EACH ROW EXECUTE PROCEDURE emp_stamp();
3275 Another way to log changes to a table involves creating a new table that
3276 holds a row for each insert, update, or delete that occurs. This approach
3277 can be thought of as auditing changes to a table.
3278 <xref linkend="plpgsql-trigger-audit-example"> shows an example of an
3279 audit trigger procedure in <application>PL/pgSQL</application>.
3282 <example id="plpgsql-trigger-audit-example">
3283 <title>A <application>PL/pgSQL</application> Trigger Procedure For Auditing</title>
3286 This example trigger ensures that any insert, update or delete of a row
3287 in the <literal>emp</literal> table is recorded (i.e., audited) in the <literal>emp_audit</literal> table.
3288 The current time and user name are stamped into the row, together with
3289 the type of operation performed on it.
3294 empname text NOT NULL,
3298 CREATE TABLE emp_audit(
3299 operation char(1) NOT NULL,
3300 stamp timestamp NOT NULL,
3301 userid text NOT NULL,
3302 empname text NOT NULL,
3306 CREATE OR REPLACE FUNCTION process_emp_audit() RETURNS TRIGGER AS $emp_audit$
3309 -- Create a row in emp_audit to reflect the operation performed on emp,
3310 -- make use of the special variable TG_OP to work out the operation.
3312 IF (TG_OP = 'DELETE') THEN
3313 INSERT INTO emp_audit SELECT 'D', now(), user, OLD.*;
3315 ELSIF (TG_OP = 'UPDATE') THEN
3316 INSERT INTO emp_audit SELECT 'U', now(), user, NEW.*;
3318 ELSIF (TG_OP = 'INSERT') THEN
3319 INSERT INTO emp_audit SELECT 'I', now(), user, NEW.*;
3322 RETURN NULL; -- result is ignored since this is an AFTER trigger
3324 $emp_audit$ LANGUAGE plpgsql;
3326 CREATE TRIGGER emp_audit
3327 AFTER INSERT OR UPDATE OR DELETE ON emp
3328 FOR EACH ROW EXECUTE PROCEDURE process_emp_audit();
3333 One use of triggers is to maintain a summary table
3334 of another table. The resulting summary can be used in place of the
3335 original table for certain queries — often with vastly reduced run
3337 This technique is commonly used in Data Warehousing, where the tables
3338 of measured or observed data (called fact tables) might be extremely large.
3339 <xref linkend="plpgsql-trigger-summary-example"> shows an example of a
3340 trigger procedure in <application>PL/pgSQL</application> that maintains
3341 a summary table for a fact table in a data warehouse.
3345 <example id="plpgsql-trigger-summary-example">
3346 <title>A <application>PL/pgSQL</application> Trigger Procedure For Maintaining A Summary Table</title>
3349 The schema detailed here is partly based on the <emphasis>Grocery Store
3350 </emphasis> example from <emphasis>The Data Warehouse Toolkit</emphasis>
3356 -- Main tables - time dimension and sales fact.
3358 CREATE TABLE time_dimension (
3359 time_key integer NOT NULL,
3360 day_of_week integer NOT NULL,
3361 day_of_month integer NOT NULL,
3362 month integer NOT NULL,
3363 quarter integer NOT NULL,
3364 year integer NOT NULL
3366 CREATE UNIQUE INDEX time_dimension_key ON time_dimension(time_key);
3368 CREATE TABLE sales_fact (
3369 time_key integer NOT NULL,
3370 product_key integer NOT NULL,
3371 store_key integer NOT NULL,
3372 amount_sold numeric(12,2) NOT NULL,
3373 units_sold integer NOT NULL,
3374 amount_cost numeric(12,2) NOT NULL
3376 CREATE INDEX sales_fact_time ON sales_fact(time_key);
3379 -- Summary table - sales by time.
3381 CREATE TABLE sales_summary_bytime (
3382 time_key integer NOT NULL,
3383 amount_sold numeric(15,2) NOT NULL,
3384 units_sold numeric(12) NOT NULL,
3385 amount_cost numeric(15,2) NOT NULL
3387 CREATE UNIQUE INDEX sales_summary_bytime_key ON sales_summary_bytime(time_key);
3390 -- Function and trigger to amend summarized column(s) on UPDATE, INSERT, DELETE.
3392 CREATE OR REPLACE FUNCTION maint_sales_summary_bytime() RETURNS TRIGGER AS $maint_sales_summary_bytime$
3394 delta_time_key integer;
3395 delta_amount_sold numeric(15,2);
3396 delta_units_sold numeric(12);
3397 delta_amount_cost numeric(15,2);
3400 -- Work out the increment/decrement amount(s).
3401 IF (TG_OP = 'DELETE') THEN
3403 delta_time_key = OLD.time_key;
3404 delta_amount_sold = -1 * OLD.amount_sold;
3405 delta_units_sold = -1 * OLD.units_sold;
3406 delta_amount_cost = -1 * OLD.amount_cost;
3408 ELSIF (TG_OP = 'UPDATE') THEN
3410 -- forbid updates that change the time_key -
3411 -- (probably not too onerous, as DELETE + INSERT is how most
3412 -- changes will be made).
3413 IF ( OLD.time_key != NEW.time_key) THEN
3414 RAISE EXCEPTION 'Update of time_key : % -> % not allowed', OLD.time_key, NEW.time_key;
3417 delta_time_key = OLD.time_key;
3418 delta_amount_sold = NEW.amount_sold - OLD.amount_sold;
3419 delta_units_sold = NEW.units_sold - OLD.units_sold;
3420 delta_amount_cost = NEW.amount_cost - OLD.amount_cost;
3422 ELSIF (TG_OP = 'INSERT') THEN
3424 delta_time_key = NEW.time_key;
3425 delta_amount_sold = NEW.amount_sold;
3426 delta_units_sold = NEW.units_sold;
3427 delta_amount_cost = NEW.amount_cost;
3432 -- Insert or update the summary row with the new values.
3433 <<insert_update>>
3435 UPDATE sales_summary_bytime
3436 SET amount_sold = amount_sold + delta_amount_sold,
3437 units_sold = units_sold + delta_units_sold,
3438 amount_cost = amount_cost + delta_amount_cost
3439 WHERE time_key = delta_time_key;
3441 EXIT insert_update WHEN found;
3444 INSERT INTO sales_summary_bytime (
3459 WHEN UNIQUE_VIOLATION THEN
3462 END LOOP insert_update;
3467 $maint_sales_summary_bytime$ LANGUAGE plpgsql;
3469 CREATE TRIGGER maint_sales_summary_bytime
3470 AFTER INSERT OR UPDATE OR DELETE ON sales_fact
3471 FOR EACH ROW EXECUTE PROCEDURE maint_sales_summary_bytime();
3473 INSERT INTO sales_fact VALUES(1,1,1,10,3,15);
3474 INSERT INTO sales_fact VALUES(1,2,1,20,5,35);
3475 INSERT INTO sales_fact VALUES(2,2,1,40,15,135);
3476 INSERT INTO sales_fact VALUES(2,3,1,10,1,13);
3477 SELECT * FROM sales_summary_bytime;
3478 DELETE FROM sales_fact WHERE product_key = 1;
3479 SELECT * FROM sales_summary_bytime;
3480 UPDATE sales_fact SET units_sold = units_sold * 2;
3481 SELECT * FROM sales_summary_bytime;
3487 <sect1 id="plpgsql-implementation">
3488 <title><application>PL/pgSQL</> Under the Hood</title>
3491 This section discusses some implementation details that are
3492 frequently important for <application>PL/pgSQL</> users to know.
3495 <sect2 id="plpgsql-var-subst">
3496 <title>Variable Substitution</title>
3499 SQL statements and expressions within a <application>PL/pgSQL</> function
3500 can refer to variables and parameters of the function. Behind the scenes,
3501 <application>PL/pgSQL</> substitutes query parameters for such references.
3502 Parameters will only be substituted in places where a parameter or
3503 column reference is syntactically allowed. As an extreme case, consider
3504 this example of poor programming style:
3506 INSERT INTO foo (foo) VALUES (foo);
3508 The first occurrence of <literal>foo</> must syntactically be a table
3509 name, so it will not be substituted, even if the function has a variable
3510 named <literal>foo</>. The second occurrence must be the name of a
3511 column of the table, so it will not be substituted either. Only the
3512 third occurrence is a candidate to be a reference to the function's
3518 <productname>PostgreSQL</productname> versions before 8.5 would try
3519 to substitute the variable in all three cases, leading to syntax errors.
3524 Since the names of variables are syntactically no different from the names
3525 of table columns, there can be ambiguity in statements that also refer to
3526 tables: is a given name meant to refer to a table column, or a variable?
3527 Let's change the previous example to
3529 INSERT INTO dest (col) SELECT foo + bar FROM src;
3531 Here, <literal>dest</> and <literal>src</> must be table names, and
3532 <literal>col</> must be a column of <literal>dest</>, but <literal>foo</>
3533 and <literal>bar</> might reasonably be either variables of the function
3534 or columns of <literal>src</>.
3538 By default, <application>PL/pgSQL</> will report an error if a name
3539 in a SQL statement could refer to either a variable or a table column.
3540 You can fix such a problem by renaming the variable or column,
3541 or by qualifying the ambiguous reference, or by telling
3542 <application>PL/pgSQL</> which interpretation to prefer.
3546 The simplest solution is to rename the variable or column.
3547 A common coding rule is to use a
3548 different naming convention for <application>PL/pgSQL</application>
3549 variables than you use for column names. For example,
3550 if you consistently name function variables
3551 <literal>v_<replaceable>something</></literal> while none of your
3552 column names start with <literal>v_</>, no conflicts will occur.
3556 Alternatively you can qualify ambiguous references to make them clear.
3557 In the above example, <literal>src.foo</> would be an unambiguous reference
3558 to the table column. To create an unambiguous reference to a variable,
3559 declare it in a labeled block and use the block's label
3560 (see <xref linkend="plpgsql-structure">). For example,
3562 <<block>>
3567 INSERT INTO dest (col) SELECT block.foo + bar FROM src;
3569 Here <literal>block.foo</> means the variable even if there is a column
3570 <literal>foo</> in <literal>src</>. Function parameters, as well as
3571 special variables such as <literal>FOUND</>, can be qualified by the
3572 function's name, because they are implicitly declared in an outer block
3573 labeled with the function's name.
3577 Sometimes it is impractical to fix all the ambiguous references in a
3578 large body of <application>PL/pgSQL</> code. In such cases you can
3579 specify that <application>PL/pgSQL</> should resolve ambiguous references
3580 as the variable (which is compatible with <application>PL/pgSQL</>'s
3581 behavior before <productname>PostgreSQL</productname> 8.5), or as the
3582 table column (which is compatible with some other systems such as
3583 <productname>Oracle</productname>).
3587 <primary><varname>plpgsql.variable_conflict</> configuration parameter</primary>
3591 To change this behavior on a system-wide basis, set the configuration
3592 parameter <literal>plpgsql.variable_conflict</> to one of
3593 <literal>error</>, <literal>use_variable</>, or
3594 <literal>use_column</> (where <literal>error</> is the factory default).
3595 This parameter affects subsequent compilations
3596 of statements in <application>PL/pgSQL</> functions, but not statements
3597 already compiled in the current session. To set the parameter before
3598 <application>PL/pgSQL</> has been loaded, it is necessary to have added
3599 <quote><literal>plpgsql</></> to the <xref
3600 linkend="guc-custom-variable-classes"> list in
3601 <filename>postgresql.conf</filename>. Because changing this setting
3602 can cause unexpected changes in the behavior of <application>PL/pgSQL</>
3603 functions, it can only be changed by a superuser.
3607 You can also set the behavior on a function-by-function basis, by
3608 inserting one of these special commands at the start of the function
3611 #variable_conflict error
3612 #variable_conflict use_variable
3613 #variable_conflict use_column
3615 These commands affect only the function they are written in, and override
3616 the setting of <literal>plpgsql.variable_conflict</>. An example is
3618 CREATE FUNCTION stamp_user(id int, comment text) RETURNS void AS $$
3619 #variable_conflict use_variable
3621 curtime timestamp := now();
3623 UPDATE users SET last_modified = curtime, comment = comment
3624 WHERE users.id = id;
3626 $$ LANGUAGE plpgsql;
3628 In the <literal>UPDATE</> command, <literal>curtime</>, <literal>comment</>,
3629 and <literal>id</> will refer to the function's variable and parameters
3630 whether or not <literal>users</> has columns of those names. Notice
3631 that we had to qualify the reference to <literal>users.id</> in the
3632 <literal>WHERE</> clause to make it refer to the table column.
3633 But we did not have to qualify the reference to <literal>comment</>
3634 as a target in the <literal>UPDATE</> list, because syntactically
3635 that must be a column of <literal>users</>. We could write the same
3636 function without depending on the <literal>variable_conflict</> setting
3639 CREATE FUNCTION stamp_user(id int, comment text) RETURNS void AS $$
3642 curtime timestamp := now();
3644 UPDATE users SET last_modified = fn.curtime, comment = stamp_user.comment
3645 WHERE users.id = stamp_user.id;
3647 $$ LANGUAGE plpgsql;
3652 Variable substitution does not happen in the command string given
3653 to <command>EXECUTE</> or one of its variants. If you need to
3654 insert a varying value into such a command, do so as part of
3655 constructing the string value, or use <literal>USING</>, as illustrated in
3656 <xref linkend="plpgsql-statements-executing-dyn">.
3660 Variable substitution currently works only in <command>SELECT</>,
3661 <command>INSERT</>, <command>UPDATE</>, and <command>DELETE</> commands,
3662 because the main SQL engine allows query parameters only in these
3663 commands. To use a non-constant name or value in other statement
3664 types (generically called utility statements), you must construct
3665 the utility statement as a string and <command>EXECUTE</> it.
3670 <sect2 id="plpgsql-plan-caching">
3671 <title>Plan Caching</title>
3674 The <application>PL/pgSQL</> interpreter parses the function's source
3675 text and produces an internal binary instruction tree the first time the
3676 function is called (within each session). The instruction tree
3677 fully translates the
3678 <application>PL/pgSQL</> statement structure, but individual
3679 <acronym>SQL</acronym> expressions and <acronym>SQL</acronym> commands
3680 used in the function are not translated immediately.
3684 As each expression and <acronym>SQL</acronym> command is first
3685 executed in the function, the <application>PL/pgSQL</> interpreter
3686 creates a prepared execution plan (using the
3687 <acronym>SPI</acronym> manager's <function>SPI_prepare</function>
3688 and <function>SPI_saveplan</function>
3689 functions).<indexterm><primary>preparing a query</><secondary>in
3690 PL/pgSQL</></> Subsequent visits to that expression or command
3691 reuse the prepared plan. Thus, a function with conditional code
3692 that contains many statements for which execution plans might be
3693 required will only prepare and save those plans that are really
3694 used during the lifetime of the database connection. This can
3695 substantially reduce the total amount of time required to parse
3696 and generate execution plans for the statements in a
3697 <application>PL/pgSQL</> function. A disadvantage is that errors
3698 in a specific expression or command cannot be detected until that
3699 part of the function is reached in execution. (Trivial syntax
3700 errors will be detected during the initial parsing pass, but
3701 anything deeper will not be detected until execution.)
3705 A saved plan will be re-planned automatically if there is any schema
3706 change to any table used in the query, or if any user-defined function
3707 used in the query is redefined. This makes the re-use of prepared plans
3708 transparent in most cases, but there are corner cases where a stale plan
3709 might be re-used. An example is that dropping and re-creating a
3710 user-defined operator won't affect already-cached plans; they'll continue
3711 to call the original operator's underlying function, if that has not been
3712 changed. When necessary, the cache can be flushed by starting a fresh
3717 Because <application>PL/pgSQL</application> saves execution plans
3718 in this way, SQL commands that appear directly in a
3719 <application>PL/pgSQL</application> function must refer to the
3720 same tables and columns on every execution; that is, you cannot use
3721 a parameter as the name of a table or column in an SQL command. To get
3722 around this restriction, you can construct dynamic commands using
3723 the <application>PL/pgSQL</application> <command>EXECUTE</command>
3724 statement — at the price of constructing a new execution plan on
3729 Another important point is that the prepared plans are parameterized
3730 to allow the values of <application>PL/pgSQL</application> variables
3731 to change from one use to the next, as discussed in detail above.
3732 Sometimes this means that a plan is less efficient than it would be
3733 if generated for a specific variable value. As an example, consider
3735 SELECT * INTO myrec FROM dictionary WHERE word LIKE search_term;
3737 where <literal>search_term</> is a <application>PL/pgSQL</application>
3738 variable. The cached plan for this query will never use an index on
3739 <structfield>word</>, since the planner cannot assume that the
3740 <literal>LIKE</> pattern will be left-anchored at run time. To use
3741 an index the query must be planned with a specific constant
3742 <literal>LIKE</> pattern provided. This is another situation where
3743 <command>EXECUTE</command> can be used to force a new plan to be
3744 generated for each execution.
3748 The mutable nature of record variables presents another problem in this
3749 connection. When fields of a record variable are used in
3750 expressions or statements, the data types of the fields must not
3751 change from one call of the function to the next, since each
3752 expression will be planned using the data type that is present
3753 when the expression is first reached. <command>EXECUTE</command> can be
3754 used to get around this problem when necessary.
3758 If the same function is used as a trigger for more than one table,
3759 <application>PL/pgSQL</application> prepares and caches plans
3760 independently for each such table — that is, there is a cache
3761 for each trigger function and table combination, not just for each
3762 function. This alleviates some of the problems with varying
3763 data types; for instance, a trigger function will be able to work
3764 successfully with a column named <literal>key</> even if it happens
3765 to have different types in different tables.
3769 Likewise, functions having polymorphic argument types have a separate
3770 plan cache for each combination of actual argument types they have been
3771 invoked for, so that data type differences do not cause unexpected
3776 Plan caching can sometimes have surprising effects on the interpretation
3777 of time-sensitive values. For example there
3778 is a difference between what these two functions do:
3781 CREATE FUNCTION logfunc1(logtxt text) RETURNS void AS $$
3783 INSERT INTO logtable VALUES (logtxt, 'now');
3785 $$ LANGUAGE plpgsql;
3791 CREATE FUNCTION logfunc2(logtxt text) RETURNS void AS $$
3796 INSERT INTO logtable VALUES (logtxt, curtime);
3798 $$ LANGUAGE plpgsql;
3803 In the case of <function>logfunc1</function>, the
3804 <productname>PostgreSQL</productname> main parser knows when
3805 preparing the plan for the <command>INSERT</command> that the
3806 string <literal>'now'</literal> should be interpreted as
3807 <type>timestamp</type>, because the target column of
3808 <classname>logtable</classname> is of that type. Thus,
3809 <literal>'now'</literal> will be converted to a constant when the
3810 <command>INSERT</command> is planned, and then used in all
3811 invocations of <function>logfunc1</function> during the lifetime
3812 of the session. Needless to say, this isn't what the programmer
3817 In the case of <function>logfunc2</function>, the
3818 <productname>PostgreSQL</productname> main parser does not know
3819 what type <literal>'now'</literal> should become and therefore
3820 it returns a data value of type <type>text</type> containing the string
3821 <literal>now</literal>. During the ensuing assignment
3822 to the local variable <varname>curtime</varname>, the
3823 <application>PL/pgSQL</application> interpreter casts this
3824 string to the <type>timestamp</type> type by calling the
3825 <function>text_out</function> and <function>timestamp_in</function>
3826 functions for the conversion. So, the computed time stamp is updated
3827 on each execution as the programmer expects.
3834 <sect1 id="plpgsql-development-tips">
3835 <title>Tips for Developing in <application>PL/pgSQL</application></title>
3838 One good way to develop in
3839 <application>PL/pgSQL</> is to use the text editor of your
3840 choice to create your functions, and in another window, use
3841 <application>psql</application> to load and test those functions.
3842 If you are doing it this way, it
3843 is a good idea to write the function using <command>CREATE OR
3844 REPLACE FUNCTION</>. That way you can just reload the file to update
3845 the function definition. For example:
3847 CREATE OR REPLACE FUNCTION testfunc(integer) RETURNS integer AS $$
3849 $$ LANGUAGE plpgsql;
3854 While running <application>psql</application>, you can load or reload such
3855 a function definition file with:
3859 and then immediately issue SQL commands to test the function.
3863 Another good way to develop in <application>PL/pgSQL</> is with a
3864 GUI database access tool that facilitates development in a
3865 procedural language. One example of such as a tool is
3866 <application>pgAdmin</>, although others exist. These tools often
3867 provide convenient features such as escaping single quotes and
3868 making it easier to recreate and debug functions.
3871 <sect2 id="plpgsql-quote-tips">
3872 <title>Handling of Quotation Marks</title>
3875 The code of a <application>PL/pgSQL</> function is specified in
3876 <command>CREATE FUNCTION</command> as a string literal. If you
3877 write the string literal in the ordinary way with surrounding
3878 single quotes, then any single quotes inside the function body
3879 must be doubled; likewise any backslashes must be doubled (assuming
3880 escape string syntax is used).
3881 Doubling quotes is at best tedious, and in more complicated cases
3882 the code can become downright incomprehensible, because you can
3883 easily find yourself needing half a dozen or more adjacent quote marks.
3884 It's recommended that you instead write the function body as a
3885 <quote>dollar-quoted</> string literal (see <xref
3886 linkend="sql-syntax-dollar-quoting">). In the dollar-quoting
3887 approach, you never double any quote marks, but instead take care to
3888 choose a different dollar-quoting delimiter for each level of
3889 nesting you need. For example, you might write the <command>CREATE
3890 FUNCTION</command> command as:
3892 CREATE OR REPLACE FUNCTION testfunc(integer) RETURNS integer AS $PROC$
3894 $PROC$ LANGUAGE plpgsql;
3896 Within this, you might use quote marks for simple literal strings in
3897 SQL commands and <literal>$$</> to delimit fragments of SQL commands
3898 that you are assembling as strings. If you need to quote text that
3899 includes <literal>$$</>, you could use <literal>$Q$</>, and so on.
3903 The following chart shows what you have to do when writing quote
3904 marks without dollar quoting. It might be useful when translating
3905 pre-dollar quoting code into something more comprehensible.
3910 <term>1 quotation mark</term>
3913 To begin and end the function body, for example:
3915 CREATE FUNCTION foo() RETURNS integer AS '
3919 Anywhere within a single-quoted function body, quote marks
3920 <emphasis>must</> appear in pairs.
3926 <term>2 quotation marks</term>
3929 For string literals inside the function body, for example:
3931 a_output := ''Blah'';
3932 SELECT * FROM users WHERE f_name=''foobar'';
3934 In the dollar-quoting approach, you'd just write:
3937 SELECT * FROM users WHERE f_name='foobar';
3939 which is exactly what the <application>PL/pgSQL</> parser would see
3946 <term>4 quotation marks</term>
3949 When you need a single quotation mark in a string constant inside the
3950 function body, for example:
3952 a_output := a_output || '' AND name LIKE ''''foobar'''' AND xyz''
3954 The value actually appended to <literal>a_output</literal> would be:
3955 <literal> AND name LIKE 'foobar' AND xyz</literal>.
3958 In the dollar-quoting approach, you'd write:
3960 a_output := a_output || $$ AND name LIKE 'foobar' AND xyz$$
3962 being careful that any dollar-quote delimiters around this are not
3963 just <literal>$$</>.
3969 <term>6 quotation marks</term>
3972 When a single quotation mark in a string inside the function body is
3973 adjacent to the end of that string constant, for example:
3975 a_output := a_output || '' AND name LIKE ''''foobar''''''
3977 The value appended to <literal>a_output</literal> would then be:
3978 <literal> AND name LIKE 'foobar'</literal>.
3981 In the dollar-quoting approach, this becomes:
3983 a_output := a_output || $$ AND name LIKE 'foobar'$$
3990 <term>10 quotation marks</term>
3993 When you want two single quotation marks in a string constant (which
3994 accounts for 8 quotation marks) and this is adjacent to the end of that
3995 string constant (2 more). You will probably only need that if
3996 you are writing a function that generates other functions, as in
3997 <xref linkend="plpgsql-porting-ex2">.
4000 a_output := a_output || '' if v_'' ||
4001 referrer_keys.kind || '' like ''''''''''
4002 || referrer_keys.key_string || ''''''''''
4003 then return '''''' || referrer_keys.referrer_type
4004 || ''''''; end if;'';
4006 The value of <literal>a_output</literal> would then be:
4008 if v_... like ''...'' then return ''...''; end if;
4012 In the dollar-quoting approach, this becomes:
4014 a_output := a_output || $$ if v_$$ || referrer_keys.kind || $$ like '$$
4015 || referrer_keys.key_string || $$'
4016 then return '$$ || referrer_keys.referrer_type
4019 where we assume we only need to put single quote marks into
4020 <literal>a_output</literal>, because it will be re-quoted before use.
4029 <!-- **** Porting from Oracle PL/SQL **** -->
4031 <sect1 id="plpgsql-porting">
4032 <title>Porting from <productname>Oracle</productname> PL/SQL</title>
4034 <indexterm zone="plpgsql-porting">
4035 <primary>Oracle</primary>
4036 <secondary>porting from PL/SQL to PL/pgSQL</secondary>
4039 <indexterm zone="plpgsql-porting">
4040 <primary>PL/SQL (Oracle)</primary>
4041 <secondary>porting to PL/pgSQL</secondary>
4045 This section explains differences between
4046 <productname>PostgreSQL</>'s <application>PL/pgSQL</application>
4047 language and Oracle's <application>PL/SQL</application> language,
4048 to help developers who port applications from
4049 <trademark class="registered">Oracle</> to <productname>PostgreSQL</>.
4053 <application>PL/pgSQL</application> is similar to PL/SQL in many
4054 aspects. It is a block-structured, imperative language, and all
4055 variables have to be declared. Assignments, loops, conditionals
4056 are similar. The main differences you should keep in mind when
4057 porting from <application>PL/SQL</> to
4058 <application>PL/pgSQL</application> are:
4063 If a name used in a SQL command could be either a column name of a
4064 table or a reference to a variable of the function,
4065 <application>PL/SQL</> treats it as a column name. This corresponds
4066 to <application>PL/pgSQL</>'s
4067 <literal>plpgsql.variable_conflict</> = <literal>use_column</>
4068 behavior, which is not the default,
4069 as explained in <xref linkend="plpgsql-var-subst">.
4070 It's often best to avoid such ambiguities in the first place,
4071 but if you have to port a large amount of code that depends on
4072 this behavior, setting <literal>variable_conflict</> may be the
4079 In <productname>PostgreSQL</> the function body must be written as
4080 a string literal. Therefore you need to use dollar quoting or escape
4081 single quotes in the function body. (See <xref
4082 linkend="plpgsql-quote-tips">.)
4088 Instead of packages, use schemas to organize your functions
4095 Since there are no packages, there are no package-level variables
4096 either. This is somewhat annoying. You can keep per-session state
4097 in temporary tables instead.
4103 Integer <command>FOR</> loops with <literal>REVERSE</> work
4104 differently: <application>PL/SQL</> counts down from the second
4105 number to the first, while <application>PL/pgSQL</> counts down
4106 from the first number to the second, requiring the loop bounds
4107 to be swapped when porting. This incompatibility is unfortunate
4108 but is unlikely to be changed. (See <xref
4109 linkend="plpgsql-integer-for">.)
4115 <command>FOR</> loops over queries (other than cursors) also work
4116 differently: the target variable(s) must have been declared,
4117 whereas <application>PL/SQL</> always declares them implicitly.
4118 An advantage of this is that the variable values are still accessible
4119 after the loop exits.
4125 There are various notational differences for the use of cursor
4134 <title>Porting Examples</title>
4137 <xref linkend="pgsql-porting-ex1"> shows how to port a simple
4138 function from <application>PL/SQL</> to <application>PL/pgSQL</>.
4141 <example id="pgsql-porting-ex1">
4142 <title>Porting a Simple Function from <application>PL/SQL</> to <application>PL/pgSQL</></title>
4145 Here is an <productname>Oracle</productname> <application>PL/SQL</> function:
4147 CREATE OR REPLACE FUNCTION cs_fmt_browser_version(v_name varchar,
4151 IF v_version IS NULL THEN
4154 RETURN v_name || '/' || v_version;
4162 Let's go through this function and see the differences compared to
4163 <application>PL/pgSQL</>:
4168 The <literal>RETURN</literal> key word in the function
4169 prototype (not the function body) becomes
4170 <literal>RETURNS</literal> in
4171 <productname>PostgreSQL</productname>.
4172 Also, <literal>IS</> becomes <literal>AS</>, and you need to
4173 add a <literal>LANGUAGE</> clause because <application>PL/pgSQL</>
4174 is not the only possible function language.
4180 In <productname>PostgreSQL</>, the function body is considered
4181 to be a string literal, so you need to use quote marks or dollar
4182 quotes around it. This substitutes for the terminating <literal>/</>
4183 in the Oracle approach.
4189 The <literal>show errors</literal> command does not exist in
4190 <productname>PostgreSQL</>, and is not needed since errors are
4191 reported automatically.
4198 This is how this function would look when ported to
4199 <productname>PostgreSQL</>:
4202 CREATE OR REPLACE FUNCTION cs_fmt_browser_version(v_name varchar,
4204 RETURNS varchar AS $$
4206 IF v_version IS NULL THEN
4209 RETURN v_name || '/' || v_version;
4211 $$ LANGUAGE plpgsql;
4217 <xref linkend="plpgsql-porting-ex2"> shows how to port a
4218 function that creates another function and how to handle the
4219 ensuing quoting problems.
4222 <example id="plpgsql-porting-ex2">
4223 <title>Porting a Function that Creates Another Function from <application>PL/SQL</> to <application>PL/pgSQL</></title>
4226 The following procedure grabs rows from a
4227 <command>SELECT</command> statement and builds a large function
4228 with the results in <literal>IF</literal> statements, for the
4233 This is the Oracle version:
4235 CREATE OR REPLACE PROCEDURE cs_update_referrer_type_proc IS
4236 CURSOR referrer_keys IS
4237 SELECT * FROM cs_referrer_keys
4239 func_cmd VARCHAR(4000);
4241 func_cmd := 'CREATE OR REPLACE FUNCTION cs_find_referrer_type(v_host IN VARCHAR,
4242 v_domain IN VARCHAR, v_url IN VARCHAR) RETURN VARCHAR IS BEGIN';
4244 FOR referrer_key IN referrer_keys LOOP
4245 func_cmd := func_cmd ||
4246 ' IF v_' || referrer_key.kind
4247 || ' LIKE ''' || referrer_key.key_string
4248 || ''' THEN RETURN ''' || referrer_key.referrer_type
4252 func_cmd := func_cmd || ' RETURN NULL; END;';
4254 EXECUTE IMMEDIATE func_cmd;
4262 Here is how this function would end up in <productname>PostgreSQL</>:
4264 CREATE OR REPLACE FUNCTION cs_update_referrer_type_proc() RETURNS void AS $func$
4266 CURSOR referrer_keys IS
4267 SELECT * FROM cs_referrer_keys
4272 func_body := 'BEGIN';
4274 FOR referrer_key IN referrer_keys LOOP
4275 func_body := func_body ||
4276 ' IF v_' || referrer_key.kind
4277 || ' LIKE ' || quote_literal(referrer_key.key_string)
4278 || ' THEN RETURN ' || quote_literal(referrer_key.referrer_type)
4282 func_body := func_body || ' RETURN NULL; END;';
4285 'CREATE OR REPLACE FUNCTION cs_find_referrer_type(v_host varchar,
4288 RETURNS varchar AS '
4289 || quote_literal(func_body)
4290 || ' LANGUAGE plpgsql;' ;
4294 $func$ LANGUAGE plpgsql;
4296 Notice how the body of the function is built separately and passed
4297 through <literal>quote_literal</> to double any quote marks in it. This
4298 technique is needed because we cannot safely use dollar quoting for
4299 defining the new function: we do not know for sure what strings will
4300 be interpolated from the <structfield>referrer_key.key_string</> field.
4301 (We are assuming here that <structfield>referrer_key.kind</> can be
4302 trusted to always be <literal>host</>, <literal>domain</>, or
4303 <literal>url</>, but <structfield>referrer_key.key_string</> might be
4304 anything, in particular it might contain dollar signs.) This function
4305 is actually an improvement on the Oracle original, because it will
4306 not generate broken code when <structfield>referrer_key.key_string</> or
4307 <structfield>referrer_key.referrer_type</> contain quote marks.
4312 <xref linkend="plpgsql-porting-ex3"> shows how to port a function
4313 with <literal>OUT</> parameters and string manipulation.
4314 <productname>PostgreSQL</> does not have a built-in
4315 <function>instr</function> function, but you can create one
4316 using a combination of other
4317 functions.<indexterm><primary>instr</></indexterm> In <xref
4318 linkend="plpgsql-porting-appendix"> there is a
4319 <application>PL/pgSQL</application> implementation of
4320 <function>instr</function> that you can use to make your porting
4324 <example id="plpgsql-porting-ex3">
4325 <title>Porting a Procedure With String Manipulation and
4326 <literal>OUT</> Parameters from <application>PL/SQL</> to
4327 <application>PL/pgSQL</></title>
4330 The following <productname>Oracle</productname> PL/SQL procedure is used
4331 to parse a URL and return several elements (host, path, and query).
4335 This is the Oracle version:
4337 CREATE OR REPLACE PROCEDURE cs_parse_url(
4339 v_host OUT VARCHAR, -- This will be passed back
4340 v_path OUT VARCHAR, -- This one too
4341 v_query OUT VARCHAR) -- And this one
4349 a_pos1 := instr(v_url, '//');
4354 a_pos2 := instr(v_url, '/', a_pos1 + 2);
4356 v_host := substr(v_url, a_pos1 + 2);
4361 v_host := substr(v_url, a_pos1 + 2, a_pos2 - a_pos1 - 2);
4362 a_pos1 := instr(v_url, '?', a_pos2 + 1);
4365 v_path := substr(v_url, a_pos2);
4369 v_path := substr(v_url, a_pos2, a_pos1 - a_pos2);
4370 v_query := substr(v_url, a_pos1 + 1);
4378 Here is a possible translation into <application>PL/pgSQL</>:
4380 CREATE OR REPLACE FUNCTION cs_parse_url(
4382 v_host OUT VARCHAR, -- This will be passed back
4383 v_path OUT VARCHAR, -- This one too
4384 v_query OUT VARCHAR) -- And this one
4393 a_pos1 := instr(v_url, '//');
4398 a_pos2 := instr(v_url, '/', a_pos1 + 2);
4400 v_host := substr(v_url, a_pos1 + 2);
4405 v_host := substr(v_url, a_pos1 + 2, a_pos2 - a_pos1 - 2);
4406 a_pos1 := instr(v_url, '?', a_pos2 + 1);
4409 v_path := substr(v_url, a_pos2);
4413 v_path := substr(v_url, a_pos2, a_pos1 - a_pos2);
4414 v_query := substr(v_url, a_pos1 + 1);
4416 $$ LANGUAGE plpgsql;
4419 This function could be used like this:
4421 SELECT * FROM cs_parse_url('http://foobar.com/query.cgi?baz');
4427 <xref linkend="plpgsql-porting-ex4"> shows how to port a procedure
4428 that uses numerous features that are specific to Oracle.
4431 <example id="plpgsql-porting-ex4">
4432 <title>Porting a Procedure from <application>PL/SQL</> to <application>PL/pgSQL</></title>
4438 CREATE OR REPLACE PROCEDURE cs_create_job(v_job_id IN INTEGER) IS
4439 a_running_job_count INTEGER;
4440 PRAGMA AUTONOMOUS_TRANSACTION;<co id="co.plpgsql-porting-pragma">
4442 LOCK TABLE cs_jobs IN EXCLUSIVE MODE;<co id="co.plpgsql-porting-locktable">
4444 SELECT count(*) INTO a_running_job_count FROM cs_jobs WHERE end_stamp IS NULL;
4446 IF a_running_job_count > 0 THEN
4447 COMMIT; -- free lock<co id="co.plpgsql-porting-commit">
4448 raise_application_error(-20000, 'Unable to create a new job: a job is currently running.');
4451 DELETE FROM cs_active_job;
4452 INSERT INTO cs_active_job(job_id) VALUES (v_job_id);
4455 INSERT INTO cs_jobs (job_id, start_stamp) VALUES (v_job_id, sysdate);
4457 WHEN dup_val_on_index THEN NULL; -- don't worry if it already exists
4467 Procedures like this can easily be converted into <productname>PostgreSQL</>
4468 functions returning <type>void</type>. This procedure in
4469 particular is interesting because it can teach us some things:
4472 <callout arearefs="co.plpgsql-porting-pragma">
4474 There is no <literal>PRAGMA</literal> statement in <productname>PostgreSQL</>.
4478 <callout arearefs="co.plpgsql-porting-locktable">
4480 If you do a <command>LOCK TABLE</command> in <application>PL/pgSQL</>,
4481 the lock will not be released until the calling transaction is
4486 <callout arearefs="co.plpgsql-porting-commit">
4488 You cannot issue <command>COMMIT</> in a
4489 <application>PL/pgSQL</application> function. The function is
4490 running within some outer transaction and so <command>COMMIT</>
4491 would imply terminating the function's execution. However, in
4492 this particular case it is not necessary anyway, because the lock
4493 obtained by the <command>LOCK TABLE</command> will be released when
4501 This is how we could port this procedure to <application>PL/pgSQL</>:
4504 CREATE OR REPLACE FUNCTION cs_create_job(v_job_id integer) RETURNS void AS $$
4506 a_running_job_count integer;
4508 LOCK TABLE cs_jobs IN EXCLUSIVE MODE;
4510 SELECT count(*) INTO a_running_job_count FROM cs_jobs WHERE end_stamp IS NULL;
4512 IF a_running_job_count > 0 THEN
4513 RAISE EXCEPTION 'Unable to create a new job: a job is currently running';<co id="co.plpgsql-porting-raise">
4516 DELETE FROM cs_active_job;
4517 INSERT INTO cs_active_job(job_id) VALUES (v_job_id);
4520 INSERT INTO cs_jobs (job_id, start_stamp) VALUES (v_job_id, now());
4522 WHEN unique_violation THEN <co id="co.plpgsql-porting-exception">
4523 -- don't worry if it already exists
4526 $$ LANGUAGE plpgsql;
4530 <callout arearefs="co.plpgsql-porting-raise">
4532 The syntax of <literal>RAISE</> is considerably different from
4533 Oracle's statement, although the basic case <literal>RAISE</>
4534 <replaceable class="parameter">exception_name</replaceable> works
4538 <callout arearefs="co.plpgsql-porting-exception">
4540 The exception names supported by <application>PL/pgSQL</> are
4541 different from Oracle's. The set of built-in exception names
4542 is much larger (see <xref linkend="errcodes-appendix">). There
4543 is not currently a way to declare user-defined exception names,
4544 although you can throw user-chosen SQLSTATE values instead.
4549 The main functional difference between this procedure and the
4550 Oracle equivalent is that the exclusive lock on the <literal>cs_jobs</>
4551 table will be held until the calling transaction completes. Also, if
4552 the caller later aborts (for example due to an error), the effects of
4553 this procedure will be rolled back.
4558 <sect2 id="plpgsql-porting-other">
4559 <title>Other Things to Watch For</title>
4562 This section explains a few other things to watch for when porting
4563 Oracle <application>PL/SQL</> functions to
4564 <productname>PostgreSQL</productname>.
4567 <sect3 id="plpgsql-porting-exceptions">
4568 <title>Implicit Rollback after Exceptions</title>
4571 In <application>PL/pgSQL</>, when an exception is caught by an
4572 <literal>EXCEPTION</> clause, all database changes since the block's
4573 <literal>BEGIN</> are automatically rolled back. That is, the behavior
4574 is equivalent to what you'd get in Oracle with:
4590 If you are translating an Oracle procedure that uses
4591 <command>SAVEPOINT</> and <command>ROLLBACK TO</> in this style,
4592 your task is easy: just omit the <command>SAVEPOINT</> and
4593 <command>ROLLBACK TO</>. If you have a procedure that uses
4594 <command>SAVEPOINT</> and <command>ROLLBACK TO</> in a different way
4595 then some actual thought will be required.
4600 <title><command>EXECUTE</command></title>
4603 The <application>PL/pgSQL</> version of
4604 <command>EXECUTE</command> works similarly to the
4605 <application>PL/SQL</> version, but you have to remember to use
4606 <function>quote_literal</function> and
4607 <function>quote_ident</function> as described in <xref
4608 linkend="plpgsql-statements-executing-dyn">. Constructs of the
4609 type <literal>EXECUTE 'SELECT * FROM $1';</literal> will not work
4610 reliably unless you use these functions.
4614 <sect3 id="plpgsql-porting-optimization">
4615 <title>Optimizing <application>PL/pgSQL</application> Functions</title>
4618 <productname>PostgreSQL</> gives you two function creation
4619 modifiers to optimize execution: <quote>volatility</> (whether
4620 the function always returns the same result when given the same
4621 arguments) and <quote>strictness</quote> (whether the function
4622 returns null if any argument is null). Consult the <xref
4623 linkend="sql-createfunction" endterm="sql-createfunction-title">
4624 reference page for details.
4628 When making use of these optimization attributes, your
4629 <command>CREATE FUNCTION</command> statement might look something
4633 CREATE FUNCTION foo(...) RETURNS integer AS $$
4635 $$ LANGUAGE plpgsql STRICT IMMUTABLE;
4641 <sect2 id="plpgsql-porting-appendix">
4642 <title>Appendix</title>
4645 This section contains the code for a set of Oracle-compatible
4646 <function>instr</function> functions that you can use to simplify
4647 your porting efforts.
4652 -- instr functions that mimic Oracle's counterpart
4653 -- Syntax: instr(string1, string2, [n], [m]) where [] denotes optional parameters.
4655 -- Searches string1 beginning at the nth character for the mth occurrence
4656 -- of string2. If n is negative, search backwards. If m is not passed,
4657 -- assume 1 (search starts at first character).
4660 CREATE FUNCTION instr(varchar, varchar) RETURNS integer AS $$
4664 pos:= instr($1, $2, 1);
4667 $$ LANGUAGE plpgsql STRICT IMMUTABLE;
4670 CREATE FUNCTION instr(string varchar, string_to_search varchar, beg_index integer)
4671 RETURNS integer AS $$
4673 pos integer NOT NULL DEFAULT 0;
4679 IF beg_index > 0 THEN
4680 temp_str := substring(string FROM beg_index);
4681 pos := position(string_to_search IN temp_str);
4686 RETURN pos + beg_index - 1;
4689 ss_length := char_length(string_to_search);
4690 length := char_length(string);
4691 beg := length + beg_index - ss_length + 2;
4693 WHILE beg > 0 LOOP
4694 temp_str := substring(string FROM beg FOR ss_length);
4695 pos := position(string_to_search IN temp_str);
4707 $$ LANGUAGE plpgsql STRICT IMMUTABLE;
4710 CREATE FUNCTION instr(string varchar, string_to_search varchar,
4711 beg_index integer, occur_index integer)
4712 RETURNS integer AS $$
4714 pos integer NOT NULL DEFAULT 0;
4715 occur_number integer NOT NULL DEFAULT 0;
4722 IF beg_index > 0 THEN
4724 temp_str := substring(string FROM beg_index);
4726 FOR i IN 1..occur_index LOOP
4727 pos := position(string_to_search IN temp_str);
4730 beg := beg + pos - 1;
4735 temp_str := substring(string FROM beg + 1);
4744 ss_length := char_length(string_to_search);
4745 length := char_length(string);
4746 beg := length + beg_index - ss_length + 2;
4748 WHILE beg > 0 LOOP
4749 temp_str := substring(string FROM beg FOR ss_length);
4750 pos := position(string_to_search IN temp_str);
4753 occur_number := occur_number + 1;
4755 IF occur_number = occur_index THEN
4766 $$ LANGUAGE plpgsql STRICT IMMUTABLE;
projects / postgresql / blob