PL/pgSQL is a loadable procedural language for the Postgres database system. This package was originally written by Jan Wieck. The design goals of PL/pgSQL were to create a loadable procedural language that can be used to create functions and trigger procedures, adds control structures to the SQL language, can perform complex computations, inherits all user defined types, functions and operators, can be defined to be trusted by the server, is easy to use. The PL/pgSQL call handler parses the functions source text and produces an internal binary instruction tree on the first time, the function is called by a backend. The produced bytecode is identified in the call handler by the object ID of the function. This ensures, that changing a function by a DROP/CREATE sequence will take effect without establishing a new database connection. For all expressions and SQL statements used in the function, the PL/pgSQL bytecode interpreter creates a prepared execution plan using the SPI managers SPI_prepare() and SPI_saveplan() functions. This is done the first time, the individual statement is processed in the PL/pgSQL function. Thus, a function with conditional code that contains many statements for which execution plans would be required, will only prepare and save those plans that are really used during the entire lifetime of the database connection. Except for input-/output-conversion and calculation functions for user defined types, anything that can be defined in C language functions can also be done with PL/pgSQL. It is possible to create complex conditional computation functions and later use them to define operators or use them in functional indices. The PL/pgSQL language is case insensitive. All keywords and identifiers can be used in mixed upper- and lowercase. PL/pgSQL is a block oriented language. A block is defined as [<<label>>] [DECLARE declarations] BEGIN statements END; There can be any number of subblocks in the statement section of a block. Subblocks can be used to hide variables from outside a block of statements. The variables declared in the declarations section preceding a block are initialized to their default values every time the block is entered, not only once per function call. It is important not to misunderstand the meaning of BEGIN/END for grouping statements in PL/pgSQL and the database commands for transaction control. Functions and trigger procedures cannot start or commit transactions and Postgres does not have nested transactions. There are two types of comments in PL/pgSQL. A double dash '--' starts a comment that extends to the end of the line. A '/*' starts a block comment that extends to the next occurence of '*/'. Block comments cannot be nested, but double dash comments can be enclosed into a block comment and a double dash can hide the block comment delimiters '/*' and '*/'. All variables, rows and records used in a block or it's subblocks must be declared in the declarations section of a block except for the loop variable of a FOR loop iterating over a range of integer values. Parameters given to a PL/pgSQL function are automatically declared with the usual identifiers $n. The declarations have the following syntax: name [ CONSTANT ] type [ NOT NULL ] [ DEFAULT | := value ]; Declares a variable of the specified base type. If the variable is declared as CONSTANT, the value cannot be changed. If NOT NULL is specified, an assignment of a NULL value results in a runtime error. Since the default value of all variables is the SQL NULL value, all variables declared as NOT NULL must also have a default value specified. The default value is evaluated ever time the function is called. So assigning 'now' to a variable of type datetime causes the variable to have the time of the actual function call, not when the function was precompiled into it's bytecode. name class%ROWTYPE; Declares a row with the structure of the given class. Class must be an existing table- or viewname of the database. The fields of the row are accessed in the dot notation. Parameters to a function can be composite types (complete table rows). In that case, the corresponding identifier $n will be a rowtype, but it must be aliased using the ALIAS command described below. Only the user attributes of a table row are accessible in the row, no Oid or other system attributes (hence the row could be from a view and view rows don't have useful system attributes). The fields of the rowtype inherit the tables fieldsizes or precision for char() etc. data types. name RECORD; Records are similar to rowtypes, but they have no predefined structure. They are used in selections and FOR loops to hold one actual database row from a SELECT operation. One and the same record can be used in different selections. Accessing a record or an attempt to assign a value to a record field when there is no actual row in it results in a runtime error. The NEW and OLD rows in a trigger are given to the procedure as records. This is necessary because in Postgres one and the same trigger procedure can handle trigger events for different tables. name ALIAS FOR $n; For better readability of the code it is possible to define an alias for a positional parameter to a function. This aliasing is required for composite types given as arguments to a function. The dot notation $1.salary as in SQL functions is not allowed in PL/pgSQL. RENAME oldname TO newname; Change the name of a variable, record or row. This is useful if NEW or OLD should be referenced by another name inside a trigger procedure. The type of a varible can be any of the existing basetypes of the database. type in the declarations section above is defined as: Postgres-basetype variable%TYPE class.field%TYPE variable is the name of a variable, previously declared in the same function, that is visible at this point. class is the name of an existing table or view where field is the name of an attribute. Using the class.field%TYPE causes PL/pgSQL to lookup the attributes definitions at the first call to the function during the lifetime of a backend. Have a table with a char(20) attribute and some PL/pgSQL functions that deal with it's content in local variables. Now someone decides that char(20) isn't enough, dumps the table, drops it, recreates it now with the attribute in question defined as char(40) and restores the data. Ha - he forgot about the funcitons. The computations inside them will truncate the values to 20 characters. But if they are defined using the class.field%TYPE declarations, they will automagically handle the size change or if the new table schema defines the attribute as text type. All expressions used in PL/pgSQL statements are processed using the backends executor. Expressions which appear to contain constants may in fact require run-time evaluation (e.g. 'now' for the datetime type) so it is impossible for the PL/pgSQL parser to identify real constant values other than the NULL keyword. All expressions are evaluated internally by executing a query SELECT expression using the SPI manager. In the expression, occurences of variable identifiers are substituted by parameters and the actual values from the variables are passed to the executor in the parameter array. All expressions used in a PL/pgSQL function are only prepared and saved once. The type checking done by the Postgres main parser has some side effects to the interpretation of constant values. In detail there is a difference between what the two functions CREATE FUNCTION logfunc1 (text) RETURNS datetime AS ' DECLARE logtxt ALIAS FOR $1; BEGIN INSERT INTO logtable VALUES (logtxt, ''now''); RETURN ''now''; END; ' LANGUAGE 'plpgsql'; and CREATE FUNCTION logfunc2 (text) RETURNS datetime AS ' DECLARE logtxt ALIAS FOR $1; curtime datetime; BEGIN curtime := ''now''; INSERT INTO logtable VALUES (logtxt, curtime); RETURN curtime; END; ' LANGUAGE 'plpgsql'; do. In the case of logfunc1(), the Postgres main parser knows when preparing the plan for the INSERT, that the string 'now' should be interpreted as datetime because the target field of logtable is of that type. Thus, it will make a constant from it at this time and this constant value is then used in all invocations of logfunc1() during the lifetime of the backend. Needless to say that this isn't what the programmer wanted. In the case of logfunc2(), the Postgres main parser does not know what type 'now' should become and therefor it returns a datatype of text containing the string 'now'. During the assignment to the local variable curtime, the PL/pgSQL interpreter casts this string to the datetime type by calling the text_out() and datetime_in() functions for the conversion. This type checking done by the Postgres main parser got implemented after PL/pgSQL was nearly done. It is a difference between 6.3 and 6.4 and affects all functions using the prepared plan feature of the SPI manager. Using a local variable in the above manner is currently the only way in PL/pgSQL to get those values interpreted correctly. If record fields are used in expressions or statements, the data types of fields should not change between calls of one and the same expression. Keep this in mind when writing trigger procedures that handle events for more than one table. Anything not understood by the PL/pgSQL parser as specified below will be put into a query and sent down to the database engine to execute. The resulting query should not return any data. Assignment An assignment of a value to a variable or row/record field is written as identifier := expression; If the expressions result data type doesn't match the variables data type, or the variable has a size/precision that is known (as for char(20)), the result value will be implicitly casted by the PL/pgSQL bytecode interpreter using the result types output- and the variables type input-functions. Note that this could potentially result in runtime errors generated by the types input functions. An assignment of a complete selection into a record or row can be done by SELECT INTO target expressions FROM ...; target can be a record, a row variable or a comma separated list of variables and record-/row-fields. if a row or a variable list is used as target, the selected values must exactly match the structure of the target(s) or a runtime error occurs. The FROM keyword can be followed by any valid qualification, grouping, sorting etc. that can be given for a SELECT statement. There is a special variable named FOUND of type bool that can be used immediately after a SELECT INTO to check if an assignment had success. SELECT INTO myrec * FROM EMP WHERE empname = myname; IF NOT FOUND THEN RAISE EXCEPTION ''employee % not found'', myname; END IF; If the selection returns multiple rows, only the first is moved into the target fields. All others are silently discarded. Calling another function All functions defined in a Prostgres database return a value. Thus, the normal way to call a function is to execute a SELECT query or doing an assignment (resulting in a PL/pgSQL internal SELECT). But there are cases where someone isn't interested int the functions result. PERFORM query executes a 'SELECT query' over the SPI manager and discards the result. Identifiers like local variables are still substituted into parameters. Returning from the function RETURN expression The function terminates and the value of expression will be returned to the upper executor. The return value of a function cannot be undefined. If control reaches the end of the toplevel block of the function without hitting a RETURN statement, a runtime error will occur. The expressions result will be automatically casted into the functions return type as described for assignments. Aborting and messages As indicated in the above examples there is a RAISE statement that can throw messages into the Postgres elog mechanism. RAISE level format'' [, identifier [...]]; Inside the format, "%" is used as a placeholder for the subsequent comma-separated identifiers. Possible levels are DEBUG (silently suppressed in production running databases), NOTICE (written into the database log and forwarded to the client application) and EXCEPTION (written into the database log and aborting the transaction). Conditionals IF expression THEN statements [ELSE statements] END IF; The expression must return a value that at least can be casted into a boolean type. Loops There are multiple types of loops. [<<label>>] LOOP statements END LOOP; An unconditional loop that must be terminated explicitly by an EXIT statement. The optional label can be used by EXIT statements of nested loops to specify which level of nesting should be terminated. [<<label>>] WHILE expression LOOP statements END LOOP; A conditional loop that is executed as long as the evaluation of expression is true. [<<label>>] FOR name IN [ REVERSE ] expression .. expression LOOP statements END LOOP; A loop that iterates over a range of integer values. The variable name is automatically created as type integer and exists only inside the loop. The two expressions giving the lower and upper bound of the range are evaluated only when entering the loop. The iteration step is always 1. [<<label>>] FOR record | row IN select_clause LOOP statements END LOOP; The record or row is assigned all the rows resulting from the select clause and the statements executed for each. If the loop is terminated with an EXIT statement, the last assigned row is still accessible after the loop. EXIT [ label ] [ WHEN expression ]; If no label given, the innermost loop is terminated and the statement following END LOOP is executed next. If label is given, it must be the label of the current or an upper level of nested loop blocks. Then the named loop or block is terminated and control continues with the statement after the loops/blocks corresponding END. PL/pgSQL can be used to define trigger procedures. They are created with the usual CREATE FUNCTION command as a function with no arguments and a return type of OPAQUE. There are some Postgres specific details in functions used as trigger procedures. First they have some special variables created automatically in the toplevel blocks declaration section. They are NEW Datatype RECORD; variable holding the new database row on INSERT/UPDATE operations on ROW level triggers. OLD Datatype RECORD; variable holding the old database row on UPDATE/DELETE operations on ROW level triggers. TG_NAME Datatype name; variable that contains the name of the trigger actually fired. TG_WHEN Datatype text; a string of either 'BEFORE' or 'AFTER' depending on the triggers definition. TG_LEVEL Datatype text; a string of either 'ROW' or 'STATEMENT' depending on the triggers definition. TG_OP Datatype text; a string of 'INSERT', 'UPDATE' or 'DELETE' telling for which operation the trigger is actually fired. TG_RELID Datatype oid; the object ID of the table that caused the trigger invocation. TG_RELNAME Datatype name; the name of the table that caused the trigger invocation. TG_NARGS Datatype integer; the number of arguments given to the trigger procedure in the CREATE TRIGGER statement. TG_ARGV[] Datatype array of text; the arguments from the CREATE TRIGGER statement. The index counts from 0 and can be given as an expression. Invalid indices (< 0 or >= tg_nargs) result in a NULL value. Second they must return either NULL or a record/row containing exactly the structure of the table the trigger was fired for. Triggers fired AFTER might always return a NULL value with no effect. Triggers fired BEFORE signal the trigger manager to skip the operation for this actual row when returning NULL. Otherwise, the returned record/row replaces the inserted/updated row in the operation. It is possible to replace single values directly in NEW and return that or to build a complete new record/row to return. Postgres does not have a very smart exception handling model. Whenever the parser, planner/optimizer or executor decide that a statement cannot be processed any longer, the whole transaction gets aborted and the system jumps back into the mainloop to get the next query from the client application. It is possible to hook into the error mechanism to notice that this happens. But currently it's impossible to tell what really caused the abort (input/output conversion error, floating point error, parse error). And it is possible that the database backend is in an inconsistent state at this point so returning to the upper executor or issuing more commands might corrupt the whole database. And even if, at this point the information, that the transaction is aborted, is already sent to the client application, so resuming operation does not make any sense. Thus, the only thing PL/pgSQL currently does when it encounters an abort during execution of a function or trigger procedure is to write some additional DEBUG level log messages telling in which function and where (line number and type of statement) this happened. Here are only a few functions to demonstrate how easy PL/pgSQL functions can be written. For more complex examples the programmer might look at the regression test for PL/pgSQL. One painful detail of writing functions in PL/pgSQL is the handling of single quotes. The functions source text on CREATE FUNCTION must be a literal string. Single quotes inside of literal strings must be either doubled or quoted with a backslash. We are still looking for an elegant alternative. In the meantime, doubling the single qoutes as in the examples below should be used. Any solution for this in future versions of Postgres will be upward compatible. The following two PL/pgSQL functions are identical to their counterparts from the C language function discussion. CREATE FUNCTION add_one (int4) RETURNS int4 AS ' BEGIN RETURN $1 + 1; END; ' LANGUAGE 'plpgsql'; CREATE FUNCTION concat_text (text, text) RETURNS text AS ' BEGIN RETURN $1 || $2; END; ' LANGUAGE 'plpgsql'; Again it is the PL/pgSQL equivalent to the example from The C functions. CREATE FUNCTION c_overpaid (EMP, int4) RETURNS bool AS ' DECLARE emprec ALIAS FOR $1; sallim ALIAS FOR $2; BEGIN IF emprec.salary ISNULL THEN RETURN ''f''; END IF; RETURN emprec.salary > sallim; END; ' LANGUAGE 'plpgsql'; This trigger ensures, that any time a row is inserted or updated in the table, the current username and time are stamped into the row. And it ensures that an employees name is given and that the salary is a positive value. CREATE TABLE emp ( empname text, salary int4, last_date datetime, last_user name); CREATE FUNCTION emp_stamp () RETURNS OPAQUE AS ' BEGIN -- Check that empname and salary are given IF NEW.empname ISNULL THEN RAISE EXCEPTION ''empname cannot be NULL value''; END IF; IF NEW.salary ISNULL THEN RAISE EXCEPTION ''% cannot have NULL salary'', NEW.empname; END IF; -- Who works for us when she must pay for? IF NEW.salary < 0 THEN RAISE EXCEPTION ''% cannot have a negative salary'', NEW.empname; END IF; -- Remember who changed the payroll when NEW.last_date := ''now''; NEW.last_user := getpgusername(); RETURN NEW; END; ' LANGUAGE 'plpgsql'; CREATE TRIGGER emp_stamp BEFORE INSERT OR UPDATE ON emp FOR EACH ROW EXECUTE PROCEDURE emp_stamp();