PL/SQL

This section will provide a basic understanding of PL/SQL. This document will briefly cover the main concepts behind PL/SQL and provide brief examples illustrating the important facets of the language. Most of the information contained in this section is DIRECTLY extracted from ``PL/SQL User's Guide and Reference'' and all credit should be given to ORACLE. If you require more detailed information than provided in this section, consult the above stated manual. PL/SQL is Oracle's procedural language extension to SQL, the relational database language. PL/SQL fully integrates modern software engineering features such as data encapsulation, information hiding, overloading, and exception handling, and so brings state-of-the-art programming to the ORACLE Server and a variety of ORACLE tools.

Overview of PL/SQL
With PL/SQL, you can use SQL statements to manipulate ORACLE data and flow-of-control statements to process the data. Moreover, you can declare constants and variables, define subprograms (procedures and functions), and trap runtime errors. Thus, PL/SQL combines the data manipulating power of SQL with the data processing power of procedural languages. PL/SQL is a block-structured language. That is, the basic units (procedures, functions, and anonymous blocks) that make up a PL/SQL program are logical blocks, which can contain any number of nested sub-blocks. Typically, each logical block corresponds to a problem or subproblem to be solved. A block (or sub-block) lets you group logically related declarations and statements. That way you can place declarations close to where they are used. The declarations are local to the block and cease to exist when the block completes. [DECLARE -- declarations] BEGIN -- statements [EXCEPTION -- handlers] END;

FUNDAMENTALS of PL/SQL
Lexical Units PL/SQL is not case-sensitive, so lower-case letters are equivalent to corresponding upper-case letters except within string and character literals. A line of PL/SQL text contains groups of characters known as lexical units, which can be classified as follows:

•delimiters (simple and compound symbols) •identifiers, which include reserved words •literals •comments

1

A delimiter is a simple or compound symbol that has a special meaning to PL/SQL. For example, you use delimiters to represent arithmetic operations such as addition and subtraction. You use identifiers to name PL/SQL program objects and units, which include constants, variables, exceptions, cursors, subprograms, and packages. Some identifiers called RESERVED WORDS, have a special syntactic meaning to PL/SQL and so cannot be redefined. For flexibility, PL/SQL lets you enclose identifiers within double quotes. Quoted identifiers are seldom needed, but occasionally they can be useful. A literal is an explicit numeric, character, string, or Boolean value not represented by an identifier. •Two kinds of numeric literals can be used in arithmetic expressions: integers and reals. •String literal is a sequence of zero or more characters enclosed by single quotes. All string literals except the null string (`') belong to type CHAR. PL/SQL is case-sensitive within string literals. •Boolean literals are the predefined values TRUE and FALSE and the non-value NULL (which stands for a missing, unknown, or inapplicable value). Keep in mind that Boolean literals are not strings. The PL/SQL compiler ignores comments but you should not. Adding comments to your program promotes readability and aids understanding. PL/SQL supports two comment styles: single-line and multiline. Single-line comments begin with a double hyphen (--) anywhere on a line and extend to the end of the line. Multiline comments begin with a slashasterisk (/*), end with an asterisk-slash (*/), and can span multiple lines. You cannot nest comments.

Datatypes Every constant and variable has a datatype, which specifies a storage format, constraints, and valid range of values. PL/SQL provides a variety of predefined scalar and composite datatypes. A scalar type has no internal components. A composite type has internal components that can be manipulated individually. PL/SQL Datatypes are similar to SQL's Datatypes but some of the common datatypes are discussed again. For more information on the PL/SQL Datatypes see Chapter 2 of ``PL/SQL User's Guide and Reference.'' •(NUMBER) You use the NUMBER datatype to store fixed or floating point numbers of virtually any size. You can specify precision, which is the total number of digits, and scale, which determines where rounding occurs. NUMBER[(precision, scale)] You cannot use constants or variables to specify precision and scale; you must use an integer literals. •(CHAR) You use the CHAR datatype to store fixed-length character data. The CHAR datatype takes an optional parameter that lets you specify a maximum length up to 32767 bytes.

CHAR[(maximum_length)] You cannot use a constant or variable to specify the maximum length; you must use an integer literal. If you do not specify the maximum length, it defaults to 1. •(VARCHAR2) You use the VARCHAR2 datatype to store variable-length character data. The VARCHAR2 datatype takes a required parameter that lets you specify a maximum length up to 32767 bytes.

VARCHAR2(maximum_length) You cannot use a constant or variable to specify the maximum length; you must use an integer literal.

2

•(BOOLEAN) You use the BOOLEAN datatype to store the values TRUE and FALSE and the non-value NULL. Recall that NULL stands for a missing, unknown, or inapplicable value. The BOOLEAN datatype takes no parameters. •(DATE) You use the DATE datatype to store fixed-length date values. The DATE datatype takes no parameters. Valid dates for DATE variables include January 1, 4712 BC to December 31, 4712 AD. When stored in the database column, date values include the time of day in seconds since midnight. The date portion defaults to the first day of the current month; the time portion defaults to midnight.

Datatype Conversion Sometimes it is necessary to convert a value from one datatype to another. PL/SQL supports both explicit and implicit (automatic) datatype conversions. To specify conversions explicitly, you use built-in functions that convert values from one datatype to another. PL/SQL conversion functions are similar to those in SQL. For more information on conversion functions see Chapter 2 of ``PL/SQL User's Guide and Reference.'' When it makes sense, PL/SQL can convert the datatype of a value implicitly. This allows you to use literals, variables, and parameters of one type where another type is expected. If PL/SQL cannot determine which implicit conversion is needed, you get a compilation error. It is your responsibility to ensure that values are convertible. For instance, PL/SQL can convert the CHAR value '02-JUN-92' to a DATE value, but PL/SQL cannot convert the CHAR value 'YESTERDAY' to a DATE value.

Declarations Your program stores values in variables and constants. As the program executes, the values of variables can change, but the values of constants cannot. You can declare variables and constants in the declarative part of any PL/SQL block, subprogram, or package. Declarations allocate storage space for a value, specify its datatype, and name the storage location so that the value can be referenced. They can also assign an initial value and specify the NOT NULL constraint. birthdate DATE; emp_count SMALLINT := 0; acct_id VARCHAR2 (5) NOT NULL := 'AP001'; The first declaration names a variable of type DATE. The second declaration names a variable of type SMALLINT and uses the assignment operator (:=) to assign an initial value of zero to the variable. The third declaration names a variable of type VARCHAR2, specifies the NOT NULL constraint, and assigns an initial value of 'AP001' to the variable. In constant declarations, the reserved word CONSTANT must precede the type specifier. credit_limit CONSTANT REAL := 5000.00; •Using DEFAULT. If you prefer, you can use the reserved word DEFAULT instead of the assignment operator to initialize variables and constants. You can also use DEFAULT to initialize subprogram parameters, cursor parameters, and fields in a user-defined record. tax_year SMALLINT DEFAULT 92; valid BOOLEAN DEFAULT FALSE;

3

Using %TYPE to declare my_dname has two advantages. DECLARE dept_rec1 dept%ROWTYPE. . A %ROWTYPE declaration cannot include an initialization clause.•Using %TYPE. . or you can reference an owner.. For example in the declaration below. BEGIN SELECT * INTO emp_rec FROM emp WHERE .dept.2) variable. dept_rec2 dept%ROWTYPE. The %ROWTYPE attribute provides a record type that represents a row in a table (or view)... if the database definition of dname changes. dept_rec1 := dept_rec2. dname. the datatype of my_dname changes accordingly at run time. 4 .. END.. The record can store an entire row of data selected from the table or fetched by a cursor. Second.deptno = 20 THEN . credit REAL(7. or database column.. Columns in a row and corresponding fields in a record have the same names and datatypes. loc FROM dept. DECLARE emp_rec emp%ROWTYPE. there are two ways to assign values to all fields in a record at once. you use the dot notation.. dname. PL/SQL allows aggregate assignment between entire records if their declarations refer to the same table or cursor. First.. and column.ename := 'JOHNSON'. dept_rec3 c1%ROWTYPE. dept_rec4 c1%ROWTYPE. you need not know the exact datatype of dname. The %TYPE attribute provides the datatype of a variable. The %TYPE attribute is particularly useful when declaring variables that refer to database columns. table. BEGIN . CURSOR c1 is SELECT deptno. First.2).. However.. constant. The column values returned by the SELECT statement are stored in fields. Variables and constants declared using %TYPE are treated like those declared using a datatype name. you can assign the value of an expression to a specific field. loc FROM dept. IF emp_rec. In addition.dname%TYPE. To reference a field.. PL/SQL treats debit like a REAL(7. dept_rec4 := dept_rec3. my_dname scott. You can reference a table and column. CURSOR c1 IS SELECT deptno. emp_rec. •Using %ROWTYPE. debit credit%TYPE. dept_rec c1%ROWTYPE ..

you cannot assign a list of column values to a record by using an assignment statement. -. if they are expressions. dname. -. BEGIN OPEN my_cursor. .0) wages.ename). However. including other declarative statements.END. You must declare a variable or constant before referencing it in other statements. ename FROM emp.illegal 5 . the following declaration is illegal: i. you can assign a list of column values to a record by using the SELECT and FETCH statement. -. But. must have aliases. DECLARE dept_rec dept%ROWTYPE. EXIT WHEN my_cursor%NOTFOUND. LOOP FETCH my_cursor INTO my_rec. Although you can retrieve entire records. my_rec. CLOSE my_cursor. you use an alias called wages: DECLARE CURSOR my_cursor IS SELECT sal + NVL(comm.. For example. Some languages allow you to declare a list of variables belonging to the same datatype. END. because dept_rec2 is based on a table and dept_rec3 is based on a cursor. For example. PL/SQL does not allow this. k SMALLINT. The column names must appear in the order in which they were defined by the CREATE TABLE or CREATE VIEW statement.. my_rec my_cursor%ROWTYPE. . In the following example. PL/SQL does allow the forward declaration of subprograms. as the example below shows. you cannot insert them. the following assignment is illegal: dept_rec2 := dept_rec3.. END. END IF. END LOOP.illegal Select-list items fetched by a cursor associated with %ROWTYPE must have simple names or.illegal Second. BEGIN SELECT deptno. the following statement is illegal: INSERT INTO dept VALUES (dept_rec). loc INTO dept_rec FROM dept WHERE deptno = 30. PL/SQL does not allow forward references.wages > 2000 THEN INSERT INTO temp VALUES (NULL. my_rec. j. IF my_rec. However..wages.

. to avoid ambiguity. BEGIN SELECT AVG(sal) INTO avg_sal FROM emp WHERE . In such cases. For example.. even if their datatypes differ.. BEGIN DELETE FROM emp WHERE ename = ename. you can prefix the table name with a username.. END. variables and parameters cannot share the same name.. use a block label to qualify references. the names of local variables and formal parameters take precedence over the names of database tables. ... procedures. In such cases. The names of database columns take precedence over the names of local variables and formal parameters. not just KING... For example. Or.. 6 . all declared identifiers must be unique.... as follows: <<main>> DECLARE ename CHAR(10) := 'KING'. the following DELETE statement removes all employees from the emp table. BEGIN DELETE FROM emp WHERE ename = main. BEGIN SELECT AVG(sal) INTO avg_sal FROM scott. exceptions. In potentially ambiguous SQL statements. as follows: PROECEDURE calc_bonus (emp NUMBER.. . .. . END. variables. . END. because ORACLE assumes that both enames in the WHERE clause refer to the database column: DECLARE ename CHAR(10) := 'KING'.emp WHERE . prefix the names of local variables and formal parameters with my_ as follows: DECLARE my_ename CHAR(10) := 'KING'. END.. the following SELECT statement fails because PL/SQL assumes that emp refers to the formal parameter: PROCEDURE calc_bonus (emp NUMBER. bonus OUT REAL) IS avg_sal REAL..Naming Conventions The same naming conventions apply to all PL/SQL program objects and units including constants. and packages.. So.. cursors. bonus OUT REAL) IS avg_sal REAL. functions. . Within the same scope.ename.

. Assignments Variables and constants are initialized every time a block or subprogram is entered. and any change in one does not affect the other. unless you expressly initialize a variable. By default. BEGIN SELECT AVG(sal) INTO avg_sal FROM emp WHERE job = calc_bonus. END IF.. both identifiers remain in scope. IF birthdate = outer. BEGIN . The two objects represented by the identifier are distinct. If you redeclare a identifier in a sub-block.. So. Although you cannot declare an identifier twice in the same block. job CHAR(15) := 'SALESMAN'. you cannot reference the global identifier unless you use a qualified name. or package) from which you can reference the identifier. BEGIN . DECLARE birthdate DATE. however. END.refers to parameter . only the local identifier is visible because you must use a qualified name to reference the global identifier. Within the sub-block. If a global identifier is redeclared in a sub-block.. An identifier is visible only in the regions from which you can reference the identifier using an unqualified name. Scope and Visibility References to an identifier are resolved according to its scope and visibility.. bonus OUT REAL) IS avg_sal REAL. subprogram.empno. END outer..refers to local variable SELECT ename INTO name FROM emp WHERE empno = calc_bonus.. -. as the following example shows: 7 .birthdate THEN . END. variables are initialized to NULL. Note that a block cannot reference identifiers declared in other blocks nested at the same level because those identifiers are neither local nor global to the block. its value is undefined. name CHAR(10)..The next example shows that you can use a subprogram name to qualify references to local variables and formal parameters: PROCEDURE calc_bonus (empno NUMBER. The qualifier can be the label of an enclosing block (or enclosing subprogram) as follows: <<outer>> DECLARE birthdate DATE. identifiers declared in a PL/SQL block are considered local to that block and global to all its sub-blocks. For example. The scope of an identifier is that region of a program unit (block.job. you can declare the same identifier in two different blocks. -.

the IF statement IF done = TRUE THEN . never reference a variable before you assign it a value. Only the values TRUE and FALSE and the nonvalue NULL can be assigned to a Boolean variable.. the following assignment is legal: DECLARE done BOOLEAN. Alternatively. BEGIN count := count + 1. you can use the SELECT or FETCH statement to have ORACLE assign values to a variable. END. Also. Therefore. So comparisons with the boolean values TRUE and FALSE are redundant. Expressions and Comparisons All expressions and comparisons are the same as those explained in the SQL Reference section. . Remember that a boolean variable is itself either true or false. there must be a corresponding variable in the INTO list. the relational operators return a Boolean value. For each item in the SELECT list. wages FROM emp WHERE empno = emp_id. if the condition evaluates to NULL. For example. Some guidelines follow to help you prevent falling into common traps.assigns a null to count END... BEGIN done := (count > 500). SELECT ename. sal + comm INTO last_name... When applied to PL/SQL expressions. assuming the variable done belongs to type BOOLEAN. 8 . can be simplified as follows: IF done THEN . do not compare real numbers for exact equality or inequality. You can avoid some common mistakes by keeping in mind the following rules: •comparisons involving nulls always yield NULL •applying the logical operator NOT to a null yields NULL •in conditional control statements. -. its associated sequence of statements is not executed Recall that applying the logical operator NULL to a null yields NULL.DELCARE count INTEGER. In general. each item must return a value that is implicitly convertible to the datatype of its corresponding variable. It is also a good idea to use parentheses when doing comparisons. So..

NVL returns the value of hire_date: start_date := NVL(hire_date. 'B'. you can use all the functions in procedural statements except the miscellaneous function DECODE. In the following example. In the next example. SYSDATE). You can use them wherever expressions of the same type are allowed. give you information about PL/SQL execution errors. NULL.If a null argument is passed to a function. Two functions. If a search is successful. after the assignment: new_string := REPLACE(old_string. •(SQLCODE) function SQLCODE return NUMBER Returns the number associated with the most recently raised exception. In addition. the corresponding result is returned. The NUMBER that SQLCODE returns is negative unless the ORACLE error is "no data found". The built-in functions fall into the following categories: • error-reporting functions • number functions • character functions • conversion functions • data functions • miscellaneous functions You can use all the built-in functions in SQL statements except the error-reporting functions SQLCODE and SQLERRM. 9 . Built-in Functions PL/SQL provides many powerful functions to help you manipulate data. •(NVL) If its first argument is null. if hire_date is null. Any search or result expression can be null. the function REPLACE returns the value of its first argument whether the optional third argument is present or not. SQLCODE returns the number of the associated ORACLE error. Most functions are the same as those discussed in SQL Reference section except the ones that are discussed below. my_string). For internal exceptions. SQLCODE always returns zero. you can nest them. the values of old_string and new_string are the same. which are paired with result expressions. in which case SQLCODE returns +100. SQLCODE and SQLERRM. •(REPLACE). the function NVL returns the value of its second argument. a null is returned except in the following cases: •(DECODE) The function DECODE compares its first argument to one or more search expressions. This function is meaningful only in an exception handler. For instance. Outside a handler. 'A'. otherwise. Furthermore. NULL. 4000). NVL returns the value of SYSDATE. 2000. if the value of rating is null. DECODE returns the value 1000: credit_limit := (rating. If its second argument is null. 1000.

10 . You can use the information to write an application audit trail table or to determine the language and character set are in use. •(UID) function UID return NUMBER Returns the unique identification number assigned to the current ORACLE user. the number of rows in a PL/SQL table can increase dynamically. The message begins with the ORACLE error code. in which case SQLERRM returns the message associated with error_number. SQLERRM is meaningful only in an exception handler.'' For internal exceptions.\\ The string str can have any of the following values: •'ENTRYID' returns an auditing entry identifier •'LANGUAGE' returns the language. which are modeled as (but not the same as) database tables. For user-defined exceptions. in which case SQLERRM returns the corresponding error message. •(USERENV) function USERENV (str VARCHAR2) return VARCHAR2 Returns information about the current session. Like the size of a database table. SQLERRM returns the message associated with the ORACLE error that occurred. in which case SQLCODE returns that error number. SQLCODE returns +1 unless you used the pragma EXCEPTION_INIT to associate the exception with an ORACLE error number. The following miscellaneous functions may be of use to you in PL/SQL coding.For user-defined exceptions. USER takes no arguments. You can pass the argument error_number to SQLERRM. territory. the size of a PL/SQL table is unconstrained. The PL/SQL table grows as new rows are added. SQLERRM returns the message ``User-Defined Exception'' unless you used the pragma EXCEPTION_INIT to associate the exception with an ORACLE error number. PL/SQL tables use a primary key to give you array-like access to rows. SQLERRM with no argument always returns the message ``ORA-0000:normal. successful completion. and database character set in use •'SESSIONID' returns the auditing session identifier •'TERMINAL' returns the operating system identifier for the session terminal You cannot specify the 'ENTRYID' or 'SESSIONID' option in SQL statements that access a remote database. PL/SQL Tables PL/SQL provides two composite datatypes: TABLE and RECORD. Objects of type TABLE are called PL/SQL tables. •(USER) function USER return VARCHAR2 Returns the username of the current ORACLE user. That is. Outside a handler. UID takes no arguments. •(SQLERRM) function SQLERRM [(error_number NUMBER)] return CHAR Returns the error message associated with the current value of SQLCODE.

you specify a primary key value using the array-like syntax plsql_table_name(primary_key_value) where primary_key_value belongs to type BINARY_INTEGER. the following declaration is illegal: ename_tab EnameTabTyp := ('CASEY'. neither of which can be named. As a result. You can use the %TYPE attribute to specify a column datatype. You can declare TABLE types in the declarative part of any block.ename%TYPE INDEX BY BINARY_INTEGER. For example. you cannot initialize a PL/SQL table in its declaration.PL/SQL tables can have one column and a primary key. 'STUART'.. ename_tab EnameTabTyp. For example. where type_name is a type specifier used in subsequent declarations of PL/SQL tables and column_type is any scalar (not composite) datatype such as CHAR. repeatedly fetches rows of values from the cursor into fields in the record. In the example below. PL/SQL tables must be declared in two steps. 'CHU'). The magnitude range of a BINARY_INTEGER value is -2**31-1 .. A cursor FOR loop implicitly declares its loop index as a record.. You can assign the value of a PL/SQL expression to a specific row using the following syntax: plsql_table_name(primary_key_value) := plsql_expression. as follows: ename_tab EnameTabTyp The identifier ename_tab represents an entire PL/SQL table. 2**31-1. First. you reference the third row in PL/SQL table ename_tab as follows: ename_tab(3) . or package using the syntax: TYPE type_name IS TABLE OF { column_type | variable%TYPE | table. To reference rows in a PL/SQL table. then declare PL/SQL tables of that type. DATE.ename%TYPE INDEX BY BINARY_INTEGER. you can declare PL/SQL tables of that type. subprogram.. The column can belong to any scalar type.sal%TYPE INDEX BY BINARY_INTEGER. you define a TABLE type.. then closes the cursor.. TYPE SalTabTyp IS TABLE OF emp. you declare a TABLE type called EnameTabTyp: DECLARE TYPE EnameTabTyp IS TABLE OF emp. sal_tab SalTabTyp. Once you define type EnameTabTyp. PL/SQL table is unconstrained because its primary key can assume any value in the range of values defined for BINARY_INTEGER.column%TYPE } [NOT NULL] INDEX BY BINARY_INTEGER. In the following example. you use a cursor FOR loop to load two PL/SQL tables. DECLARE TYPE EnameTabTyp IS TABLE OF emp. opens the cursor associated with a given query. but the primary key must belong to type BINARY_INTEGER. . 11 . or NUMBER.

--process the tables process_sals(ename_tab. hire date. subprogram. sal_tab). which can belong to different datatypes. you can use a simple workaround to delete entire PL/SQL tables. or package using the syntax 7 TYPE type_name IS RECORD (field_name1 {field_type | variable%TYPE | table. and so on. You can declare RECORD types in the declarative part of any block. records must be declared in two steps. Likewise. Later.. There is no straightforward way to delete rows from a PL/SQL table because the DELETE statement cannot specify PL/SQL tables. you define a RECORD type. when you want to delete the original PL/SQL tables. Until a row is assigned a value. (field_name2 {field_type | variable%TYPE | table. sal_tab(i) := emprec. When PL/SQL runs out of memory it raises the predefined exception STORAGE_ERROR. BEGIN -. However. sal FROM emp) LOOP i := i + 1. salary. If you try to reference an uninitialized row.load employee names and salaries into PL/SQL tables For emprec IN (SELECT ename.. Therefore.. This data is dissimilar in type but logically related.sal.column%TYPE | table%ROWTYPE} [NOT NULL]. A record that contains such fields as the name. Unlike PL/SQL tables. Remember that the size of PL/SQL table is unconstrained so. The composite datatype RECORD lifts those restrictions. For example. First.i BINARY_INTEGER := 0. that if you want to maintain a row count.. constrained only by available memory. simply assign the empty table to them.. . Setting a row to NULL does not work because the row remains and does not raise the exception NO_DATA_FOUND when referenced. You must use a loop to INSERT values from a PL/SQL table into a database column. .ename. . objects of type RECORD are called records. salary. you cannot specify the datatypes of fields in the record or define fields of your own. END LOOP. As you might expect. you must use a loop to FETCH values from a database column into a PL/SQL table. you cannot reference PL/SQL tables in the INTO clause of a SELECT statement.. 12 . suppose you have different kinds of data about an employee such as name. Although you cannot delete individual rows from a PL/SQL table. A PL/SQL table can grow large. END.column%TYPE | table%ROWTYPE} [NOT NULL]. ename_tab(i) := emprec. it does not exist. User-defined Records You can use the %ROWTYPE attribute to declare a record that represents a row in a table or a row fetched by a cursor. ). you must declare a variable for that purpose. and hire date of an employee would let you treat the data as a logical unit. records have uniquely named fields. declare another PL/SQL table of the same type and leave it empty. then declare userdefined records of that type. First. Like PL/SQL tables. PL/SQL raises the predefined exception NO_DATA_FOUND.

Instead of assigning values separately to each field in a record... First. You cannot assign a list of values to a record by using an assignment statement.dname%TYPE). DECLARE TYPE DeptRecTyp IS RECORD (deptno NUMBER(2) NOT NULL := 20. BEGIN SELECT deptno. For instance. you declare a RECORD type named DeptRecTyp: DECLARE TYPE DeptRecTyp is RECORD (deptno NUMBER(2) NOT NULL := 20. you can declare records of that type. Second. loc dept..illegal Also. You can use the %TYPE or %ROWTYPE attribute to specify a field datatype. . END. you use the dot notation and the following syntax: record_name. Just make sure the column names appear in the same order as the fields in your record. you can assign a list of column values to a record by using the SELECT or FETCH statement. you can assign one record to another if they belong to the same datatype. the following IF condition is illegal: IF dept_rec1 = dept_rec2 THEN -. This can be done in two ways. value3. a userdefined record and a %ROWTYPE record always belong to different types.dname%TYPE.. 13 .field_name You can assign the value of a PL/SQL expression to a specific field by using the following syntax: record_name.. Even if their fields match exactly. dname..field_name := plsql_expression. records of different types cannot be assigned to each other. . records cannot be tested for equality or inequality. . you can assign values to all fields at once. Furthermore. value2. -. To reference individual fields in a record.. dname dept. So. dname dept. The identifier dept_rec represents an entire record. dept_rec DeptRecTyp.).illegal . loc INTO dept_rec FROM dept WHERE deptno = 30. as the example below shows. the following syntax is illegal: record_name := (value1.where type_name is a type specifier used in subsequent declarations of records and field_type is any datatype including RECORD and TABLE. END IF.dname%TYPE. loc dept. Once you define type DeptRecTyp.. In the following example. .dname%TYPE). as follows: dept_rec DeptRecTyp...

time TimeTyp. as follows: IF condition1 THEN sequence_of_statements1.time := meeting. party.. .time. . The iteration structure executes a sequence of statements repeatedly as long as a condition holds true. CONTROL STRUCTURES According to the structure theorem. You can assign one nested record to another if they belong to the same datatype. meeting. TYPE MeetingTyp IS RECORD (day DATE. Conditional Control: IF Statements Often. TYPE PartyTyp IS RECORD (date DATE. A condition is any variable or expression that returns a Boolean value (TRUE. a record can be the component of another record. meeting.hour := 10. That is. That is. -. or NULL). -. purpose CHAR(50)). then executes one sequence of statements instead of another. Such assignments are allowed even if the containing records belong to different datatypes.minute := 45.nested record place CHAR(20). depending on whether the condition is true or false. 14 .. BEGIN meeting.PL/SQL lets you declare and reference nested records.time.. and IF-THEN-ELSIF. as follows: DECLARE TYPE TimeTyp IS RECORD (minute SMALLINT. it is necessary to take alternative actions depending on circumstances. meeting MeetingTyp. There are three forms of IF statements: IF-THEN. whether the sequence is executed or not depends on the value of a condition. ELSE sequence_of_statements3. seminar. The sequence structure simply executes a sequence of statements in the order in which they occur..time. FALSE. The IF statement lets you execute a sequence of statements conditionally. party PartyTyp. seminar MeetingTyp. END. IF-THEN-ELSE. The selection structure tests a condition. hour SMALLINT). ELSIF condition2 THEN sequence_of_statements2. any computer program can be written using the basic control structures which can be combined in any way necessary to deal with a given problem. time TimeTyp.time := meeting.nested record loc CHAR(15)).time.day := '26-Jun-91'. The third form of IF statement uses the keyword ELSIF (NOT ELSEIF) to introduce additional conditions.

..END IF. the loop completes and control passes to the next statement after the loop. as follows: LOOP sequence_of_statements3. as follows: <<label_name>> LOOP sequence_of_statements. the loop cannot complete. LOOP FETCH c1 INTO . If further processing is undesirable or impossible.. which encloses a sequence of statements between the keywords LOOP and END LOOP. The EXIT statement forces a loop to complete unconditionally. When the EXIT statement is encountered. must appear at the beginning of the LOOP statement. EXIT WHEN c1%NOTFOUND. -. statements within the loop must change the value of the condition. then control resumes at the top of the loop. The label. 15 .. END LOOP. .. -. an undeclared identifier enclosed by double angle brackets.. You can place one or more EXIT statements anywhere inside a loop. . END LOOP. With each iteration of the loop. When an EXIT statement is encountered. -.control resumes here The EXIT-WHEN statement allows a loop to complete conditionally.. IF . Iterative Control: LOOP and EXIT Statements LOOP statements let you execute a sequence of statements multiple times. So. and FOR-LOOP. LOOP The simplest form of LOOP statement is the basic (or infinite) loop. There are two forms of EXIT statements: EXIT and EXIT-WHEN. CLOSE c1. loops can be labeled... END LOOP... the sequence of statements is executed..exit loop immediately END IF.exit loop if condition is true . WHILE-LOOP. EXIT. If the condition evaluates to TRUE. but nowhere outside a loop. Like PL/SQL blocks. Until the condition evaluates to TRUE. the condition in the WHEN clause is evaluated.. THEN . you can use the EXIT statement to complete the loop. LOOP . the loop completes immediately and control passes to the next statement.. There are three forms of LOOP statements: LOOP.

. However.. LOOP . END LOOP outer. as follows: WHILE condition LOOP sequence_of_statements. Optionally. FOR i IN 1. With either form of EXIT statement. the condition is evaluated. but any enclosing loop. which is enclosed by the keywords FOR and LOOP. If the condition evaluates to FALSE or NULL. .. -. END LOOP.upper_bound LOOP sequence_of_statements.. The range is part of an iteration scheme. . FOR loops iterate over a specified range of integers... Simply label the enclosing loop that you want to complete. the number of iterations through a FOR loop is known before the loop is entered..emp_count LOOP .. END LOOP.. .exit both loops END LOOP.. If the condition evaluates to TRUE. as the following example shows: SELECT COUNT(empno) INTO emp_count FROM emp. the sequence might execute zero times. FOR-LOOP Whereas the number of iteration through a WHILE loop is unknown until the loop completes.. you can complete not only the current loop. Since the condition is tested at the top of the loop. FOR counter IN [REVERSE] lower_bound. Before each iteration of the loop. then use the label in an EXIT statement.. END LOOP. the label name can also appear at the end of the LOOP statement.. WHILE-LOOP The WHILE-LOOP statement associates a condition with a sequence of statements enclosed by the keywords LOOP and END LOOP... <<outer>> LOOP . The lower bound need not be 1. PL/SQL lets you determine the loop range dynamically at run time. EXIT outer WHEN .. the loop counter increment (or decrement) must be 1. 16 . the sequence of statements is executed.END LOOP [label_name]. the loop is bypassed and control passes to the next statement. then control resumes at the top of the loop.

two UPDATE statements might credit one bank account and debit another. At the same instant. The NULL statement can make the meaning and action of conditional statements clear and so improve readability. but any enclosing loop. You need not explicitly declare the loop counter because it is implicitly declared as a local variable of type INTEGER. it can simplify logic enough to warrant its use. The EXIT statement allows a FOR loop to complete prematurely. or the system control command ALTER SYSTEM. ORACLE uses transactions to ensure data integrity. You can complete not only the current loop. A GOTO statement cannot branch out of a subprogram. Data Manipulation To manipulate ORACLE data. Finally. the NULL statement is a handy way to create stubs when designing applications from the top down.. For example. However. ORACLE makes permanent or undoes all database changes made by a transaction. functions. A GOTO statement cannot branch from one IF statement clause to another. Hence. Occasionally. session control commands such as SET ROLES... ORACLE detects the error and rolls back the transaction. A GOTO statement cannot branch into an IF statement. a GOTO statement cannot branch from an exception handler into the current block. If your program fails in the middle of a transaction. It can. Transaction Control ORACLE is transaction oriented.. and LOCK TABLE commands. or sub-block. END. You can manipulate ORACLE data flexibly and safely because PL/SQL supports all SQL data manipulation commands (except EXPLAIN PLAN).The loop counter is defined only within the loop. The NULL statement explicitly specifies inaction. Sequential Control: GOTO and NULL statements Unlike the IF and LOOP statements. that is.. The structure of PL/SQL is such that the GOTO statement is seldom needed. Interaction With ORACLE SQL Support By extending SQL. the database is restored to its former state automatically. SELECT.. . LOOP statement. it does nothing other than pass control to the next statement. transaction control commands. improve readability. you use the INSERT. the GOTO and NULL statements are not crucial to PL/SQL programming. and operators. Also. You cannot reference it outside the loop. however. PL/SQL offers a unique combination of power and ease of use. UPDATE. GOTO insert_row. pseudocolumns. 17 . DELETE. A transaction is a series of SQL data manipulation statements that does a logical unit of work. <<insert_row>> INSERT INTO emp VALUES . PL/SQL does not support data definition commands such as CREATE. BEGIN .

CURRVAL returns the current value in a specified sequence. you can reference pseudocolumns in SQL statements. The root is level 1. For example. and SET TRANSACTION commands to control transactions. If a SELECT statement includes an ORDER BY clause. LEVEL. then the following statement inserts a new employee number into the emp table: INSERT INTO emp VALUES (empno_seq.NEXTVAL. However. ROLLBACK ends the current transaction and undoes any changes made since the transaction began. For instance. When you create a sequence. COMMIT makes permanent any database changes made during the current transaction. A sequence is a database object that generates sequential numbers. Furthermore. ROWNUM returns a number indicating the order in which a row was selected from a table. LEVEL is used with the SELECT CONNECT BY statement to organize rows from a database table into a tree structure. Assume that you have declared empno_seq as a database sequence. PL/SQL declares a cursor implicitly for all SQL data manipulation statements. Cursor Management PL/SQL uses two types of cursors: implicit and explicit. SAVEPOINT. you can select values from a pseudocolumn. SAVEPOINT marks the current point in the processing of a transaction. ROLLBACK. you can specify its initial value and an increment. Before you can reference CURRVAL in a session. NEXTVAL returns the next value in a database sequence. ROWID. for queries that return more than one row. In the START WITH clause. LEVEL returns the level number of a node in a tree structure.You use the COMMIT.. ROWID returns the rowid (binary address) of a row in a database table. However. Operators PL/SQL lets you use all the SQL comparison. SQL Functions PL/SQL lets you use all the SQL functions including group functions. Used with ROLLBACK. including queries that return only one row. NEXTVAL. and ROWNUM. set. you cannot insert values into. and row operators in SQL statements. undoes part of a transaction. you must use NEXTVAL to generate a number. other users cannot see them. They are called pseudocolumns because they are not actual columns in a table but behave like columns. children of the root are level 2 and so on. update values in. Until you commit your changes. SQL Pseudocolumns PL/SQL recognizes the following SQL pseudocolumns. ROWNUMs are assigned to the retrieved rows before the sort is done. you must declare an explicit cursor or use a cursor FOR loop Explicit Cursors 18 . .. which summarize entire columns of ORACLE data. you specify a condition that identifies the root of the tree. which return specific data items: CURRVAL. new_ename. You specify the direction in which the query walks the tree (down from the root or up from the branches) with the PRIOR operator. or delete values from a pseudocolumn. SET TRANSACTION establishes a read-only transaction.).

Forward references are not allowed in PL/SQL. The formal parameters of a cursor must be IN parameter. factor := factor + 1. CLOSE c1. . For cursors declared using the FOR UPDATE clause. one. you must declare a cursor before referencing it in other statements.sal%TYPE. you can explicitly define a cursor to process the rows. Each time FETCH is executed. ename FROM emp WHERE sal > median. you use three commands to control the cursor: OPEN. subprogram. my_job emp.here factor equals 2 LOOP FETCH c1 INTO my_sal. parameter. The cursor name is an undeclared identifier. factor INTEGER := 2. You define a cursor in a declarative part of a PL/SQL block. their datatypes must be compatible. the cursor advances to the next row in the active set. which consists of all rows that meet the query search criteria. So.does not affect FETCH END LOOP. and CLOSE..job%TYPE. the query can reference PL/SQL variables within its scope: DECLARE my_sal emp. BEGIN .The set of rows returned by a query can consist of zero. Also. When a query returns multiple rows. Then. -. Any variables in the WHERE clause of the query associated with the cursor are evaluated only when the cursor is OPENed. For each column value returned by the query associated with the cursor. CURSOR c1 (median IN NUMBER) IS SELECT job.]) ] IS where parameter stands for the following syntax: variable_name [IN] datatype [{:= | DEFAULT} value] OPENing the cursor executes the query and identifies the active set.. you name it and associate it with a specific query. as the example below shows. .. or multiple rows. it is used only to reference a query. When you declare a cursor. depending on how many rows meet your search criteria. As the following example shows. 19 .. Rows in the active set are not retrieved when the OPEN statement is executed. A cursor parameter can appear in a query wherever a constant can appear. To declare formal cursor parameters. not a PL/SQL variable. Cursors can take parameters.. EXIT WHEN c1%NOTFOUND.. the FETCH statement retrieves the rows. you use the syntax: CURSOR name [ (parameter [. there must be a corresponding variable in the INTO list. the OPEN statement also locks those rows. or package by naming it and specifying a query. OPEN c1. Rather. FETCH. -. END. The FETCH statement retrieves the rows in the active set one at a time. cursor c1 IS SELECT factor*sal FROM emp WHERE job = my_job.

When you open its cursor. you find out if that exception was raised by checking %NOTFOUND. so when that happens. Thereafter.Explicit Cursor Attributes Each cursor that you explicitly define has four attributes: %NOTFOUND. The features of %NOTFOUND are similar to those of the explicit cursor attributes but you must bear in mind the following: if a SELECT INTO fails to return a row. •Using %NOTFOUND. Implicit Cursor Attributes The SQL cursor has four attributes: %NOTFOUND. PL/SQL lets you refer to the most recent implicit cursor as the ``SQL'' cursor. Before the first fetch. •Using %NOTFOUND. although you cannot use the OPEN. %ISOPEN evaluates to FALSE. the loop is never exited unless your EXIT WHEN statement is as follows: EXIT WHEN c1%NOTFOUND OR c1%NOTFOUND IS NULL. and CLOSE statements to control an implicit cursor. the predefined exception NO_DATA_FOUND is raised whether you check %NOTFOUND on the next line or not. Implicit Cursors ORACLE implicitly opens a cursor to process each SQL statement not associated with an explicitly declared cursor. %ROWCOUNT. it evaluates to TRUE if the last fetch returned a row or to FALSE if no row was returned. •Using %ISOPEN. and SELECT INTO statements. %ROWCOUNT is zeroed. So. •Using %FOUND. Before the first fetch. FETCH is expected to fail eventually. if FETCH never executes successfully. 20 . %FOUND. %FOUND evaluates to NULL. If the last fetch failed to return a row (because the active set was empty). and %ISOPEN. these attributes let you access information about the execution of INSERT. So. %NOTFOUND evaluates to TRUE. otherwise. %NOTFOUND evaluates to FALSE. The check for %NOTFOUND on the next line would be useless because when NO_DATA_FOUND is raised. it returns the number of rows fetched so far. When appended to the cursor name (SQL). You can open multiple cursors. and %ISOPEN. you can still use cursor attributes to access information about the most recently executed SQL statement. %NOTFOUND evaluates to NULL. these attributes let you access useful information about the execution of a multirow query. %ROWCOUNT. DELETE. UPDATE. my_empno NUMBER(4). Rows are FETCHed from the active set one at a time. When appended to the cursor name. %FOUND is the logical opposite of %NOTFOUND. When a cursor is OPENed. If the last fetch returned a row. then use %NOTFOUND to tell which cursors have rows left to fetch. the rows that satisfy the associated query are identified and form the active set. Thereafter. Instead of coding a NO_DATA_FOUND handler. After an explicit cursor is open but before the first fetch. In this situation %NOTFOUND is useful in the OTHERS exception handler.2). You can use implicit cursor attributes in procedural statements but not in SQL statements. normal execution stops and control transfers to the exception-handling part of the block. %ISOPEN evaluates to TRUE if its cursor is open. %ROWCOUNT returns a zero. no exception is raised. You can use explicit cursor attributes in procedural statements but not in SQL statements. %FOUND. DECLARE my_sal NUMBER(7. FETCH. •Using %ROWCOUNT. The number is incremented if the latest fetch returned a row.

BEGIN . you compute the total wages paid to employees in that department..check for 'no data found' .. . In the following example. -.. That is because group functions such as AVG and SUM always return a value or a null. Then. then closes the cursor when all rows have been processed or when you exit the loop. Packaged Cursors You can separate a cursor specification from its body for placement in a package by using the RETURN clause: CREATE PACKAGE emp_actions AS /* Declare cursor specification */ CURSOR c1 RETURN emp%ROWTYPE .. you pass a department number. This way. SELECT sal INTO my_sal FROM emp WHERE empno = my_empno. A cursor specification has no SELECT statement because the RETURN clause defines the datatype of the result value. END emp_action. Also. %ROWCOUNT and %ISOPEN. repeatedly fetches rows of values from the active set into fields in the record. A cursor body must have a SELECT statement and the same RETURN clause as its corresponding cursor specification. Furthermore.. you determine how many employees have salaries higher than $2000 and how many have commissions larger than their salaries.. These attributes are similar in use to those of explicit cursor attributes. END..might raise NO_DATA_FOUND EXCEPTION WHEN OTHERS THEN IF SQL%NOTFOUND THEN -. CREATE PACKAGE BODY emp_actions AS /* Define cursor body */ CURSOR c1 RETURN emp%ROWTYPE SELECT * FROM emp WHERE sal > 3000. END emp_actions. 21 . you can change the cursor body without changing the cursor specification. the number and datatypes of select-list items in the SELECT statement must match the RETURN clause. a SELECT INTO that calls a SQL group function never raises the exception NO_DATA_FOUND.. opens a cursor.. Cursor FOR Loops You can use a cursor FOR loop to simplify coding. However. . A cursor FOR loop implicitly declares its loop index as a record of type %ROWTYPE. END IF. •Using %FOUND.. You can pass parameters to a cursor used in a cursor FOR loop.

the FOR UPDATE clause must appear at the end of the cursor declaration.comm. dept WHERE emp. dname FROM emp. However.DECLARE CURSOR emp_cursor(dnum NUMBER) IS SELECT sal. •Using a LOCK TABLE statement lets you lock entire database tables in a specified lock mode so that you can share or deny access to tables while maintaining their integrity. comm FROM emp WHERE deptno = dnum. BEGIN /* the number of iterations will equal the number of rows * * returned by emp_cursor */ FOR emp_record IN emp_cursor(20) LOOP emp_record. END IF. DECLARE CURSOR c1 IS SELECT empno. you can request specific data locks on rows or tables when it is to your advantage to override default locking. Table locks are released when your transaction issues a COMMIT or ROLLBACK. When querying multiple tables. COMMIT.deptno = dept. 'Total Wages: ' || TO_CHAR(total_wages)).comm.sal + emp_record. LOCK TABLE emp IN ROW SHARE MODE NOWAIT. The rows are unlocked when you COMMIT the transaction.2) := 0. DECLARE CURSOR c1 IS SELECT ename. When declaring a cursor that will be referenced in the WHERE CURRENT OF clause of an UPDATE or DELETE statement. as the following example shows.comm := NVL(emp_record.deptno AND job = 'MANAGER' FOR UPDATE OF sal. you cannot FETCH from a FOR UPDATE cursor after a COMMIT. If present. END. total_wages := total_wages + emp_record. All rows in the active set are locked when you OPEN the cursor. The FOR UPDATE clause indicates that rows will be updated or deleted and locks all rows in the active set. •Using FOR UPDATE. total_wages NUMBER(11. sal FROM emp WHERE job = 'SALESMAN' AND comm > sal FOR UPDATE. So. Overriding Default Locking By default ORACLE locks data structures for you automatically. higher_comm NUMBER(4) := 0. 22 . END LOOP. you can use the FOR UPDATE OF clause to confine row locking to particular tables.0).sal > 2000 THEN high_paid := high_paid + 1. you must use the FOR UPDATE clause to acquire exclusive row locks. higher_comm. IF emp_record. high_paid NUMBER(4) := 0. INSERT INTO temp VALUES (high_paid.

you can use database triggers to •audit data modification •log events transparently •enforce complex business rules •derive column values automatically •implement complex security authorizations •maintain replicate tables You can associate up to 12 database triggers with a give table. For the trigger to fire. :new. A database trigger can also have DECLARE and EXCEPTION sections.END block. including subprogram calls. unlike subprograms. however.reorderable = 'T') BEGIN /* trigger action */ IF :new. The database trigger named reorder ensures that a part is reordered when its quantity on hand drops below the reorder point. ORACLE executes (fires) the database trigger automatically whenever a given SQL operation affects the table. and a trigger action.table FOR EACH ROW /* trigger constraint */ WHEN (new. you can reference :new and :old values of changing rows. Within a database trigger.reorder_point THEN INSERT INTO pending_orders VALUES (:new. not the current user. If the trigger statement fails.qty_on_hand < :new. CREATE TRIGGER reorder /* triggering event */ AFTER UPDATE OF qty_on_hand ON inventory -. any SQL or procedural statement. Among other things. it is rolled back. 23 . Except for transaction control statements such as COMMIT and ROLLBACK.part_no.reorder_qty. The triggering event specifies the SQL data manipulation statement that affects the table. database triggers are invoked implicitly. the database trigger fires and an anonymous PL/SQL block performs the action. a database trigger fires once per table. can appear in the BEGIN.. You can use the REFERENCING clause (not shown) to replace :new and :old with other correlation names. The FOR EACH ROW option specifies that the trigger fires once per row. In this case.. Database triggers fire with the privileges of the owner. END. the Boolean expression in the WHEN clause must evaluate to TRUE. END IF. SYSDATE). the owner must have appropriate access to all objects referenced by the trigger action. So. Notice that the colon is not used in the WHEN clause. A database trigger has three parts: a triggering event. The prefix :new is a correlation name that refers to the newly updated column value. which must be invoked explicitly. When the event occurs. The keyword AFTER specifies that the database trigger fires after the update is done. an optional trigger constraint. The example below illustrates transparent event logging. By default.Database Triggers A database trigger is a stored PL/SQL program unit associated with a specific database table. So. The name in the ON clause identifies the database table associated with the database trigger. the statement is UPDATE.

Exceptions can be internally defined (by the runtime system) or user-defined. ELSIF (:new. * * or increase exceeds 10%.sal < minsal OR :new.1 * :old. you write separate routines called exception handlers. The other internal exceptions can be given names. */ IF (:new. or package. user-defined exceptions must be given names. normal execution stops and control transfers to the exception handling part of your PL/SQL block or subprogram. Some common internal exceptions have predefined names. User-defined exceptions must be raised explicitly by RAISE statements. For a full discussion of database triggers. maxsal FROM sals WHERE job = :new. */ SELECT losal. you might define an exception named insufficient_funds to flag an overdrawn bank accounts. 'Negative increase'). Error Handling Overview In PL/SQL a warning or error condition is called an exception.sal > 1. To handle raised exceptions.sal > maxsal) THEN raise_application_error(-20225. hisal INTO minsal. More information on built-in procedures is provided later in this chapter. That is. BEGIN /* Get salary range for a given job from table sals. After an exception handler runs. END IF: END. control returns to the host environment. Examples of internally defined exceptions include division by zero and out of memory. such as ZERO_DIVIDE and STORAGE_ERROR. Unlike internal exceptions. maxsal NUMBER.job. ELSIF (:new. an exception is raised. 'Increase exceeds 10%'). which can also raise predefined exceptions. You can define exceptions of your own in the declarative part of any PL/SQL block. Internal exceptions are raised implicitly (automatically) by the runtime system. When an error occurs. the current block stops executing and the enclosing block resumes with the next statement. /* If salary is out of range.sal < :old. 'Salary out of range').sal) THEN raise_application_error(-20320. which lets you issue user-defined error messages: CREATE TRIGGER check_salary BEFORE INSERT OR UPDATE OF sal. increase is negative.job != 'PRESIDENT') DECLARE minsal NUMBER. see ``ORACLE7 Server Application Developer's Guide''. For example.The next example shows that the trigger action can include calls to the built-in ORACLE procedure raise_application_error. If there is no enclosing block. raise an exception. subprogram. 24 .sal) THEN raise_application_error(-20325. job ON emp FOR EACH ROW WHEN (new.

•INVALID_NUMBER is raised in a SQL statement if the conversion of a character string to a number fails. Just add an exception handler to your PL/SQL block. every time you issue a command. For example. You can write handlers for predefined exceptions using the names shown below: Exception Name ORACLE Error SQLCODE Value CURSOR_ALREADY OPEN ORA-06511 -6511 DUP_VAL_ON_INDEX ORA-00001 -1 INVALID_CURSOR ORA-01001 -1001 INVALID_NUMBER ORA-01722 -1722 LOGIN_DENIED ORA-01017 -1017 NO_DATA_FOUND ORA-01403 +100 NOT_LOGGED_ON ORA-01012 -1012 PROGRAM_ERROR ORA-06501 -6501 STORAGE_ERROR ORA-06500 -6500 TIMEOUT_ON_RESOURCE ORA-00051 -51 TOO_MANY_ROWS ORA-01422 -1422 VALUE_ERROR ORA-06502 -6502 ZERO_DIVIDE ORA-01476 -1476 •CURSOR_ALREADY_OPEN is raised if you try to OPEN an already open cursor. •STORAGE_ERROR is raised if PL/SQL runs out of memory or if memory is corrupted. •DUP_VAL_ON_INDEX is raised if you try to store duplicate values in a database column that is constrained by a unique index. Exceptions also improve reliability. but exceptions must be handled by name. Every ORACLE error has a number. For example. Without exception handling. •NO_DATA_FOUND is raised if a SELECT INTO statement returns no rows or if you reference an uninitialized row in a PL/SQL table. So. •INVALID_CURSOR is raised if you try an illegal cursor operation. PL/SQL predefines some common ORACLE errors as exceptions. •NOT_LOGGED_ON is raised if your PL/SQL program issues a database call without being logged on to ORACLE. •PROGRAM_ERROR is raised if PL/SQL has an internal problem. Predefined Exceptions An internal exception is raised explicitly whenever your PL/SQL program violates an ORACLE rule or exceeds a system-dependent limit. If the exception is ever raised in that block (or any sub-block). you must check for execution errors. which defines the PL/SQL environment. •LOGIN_DENIED is raised if you try logging on to ORACLE with an invalid username/password. You need not worry about checking for an error at every point it might occur. So. if you try to CLOSE an unopened cursor.Advantages of Exceptions Using exceptions for error handling has several advantages. PL/SQL declares predefined exceptions globally in package STANDARD. •TIMEOUT_ON_RESOURCE is raised if a timeout occurs while ORACLE is waiting for a resource. the predefined exception NO_DATA_FOUND is raised if a SELECT INTO statement returns no rows. you need not declare them yourself. 25 . you can be sure it will be handled.

26 .handle the error . acct_num NUMBER(5). DECLARE insufficient_privileges EXCEPTION. Unlike predefined exceptions. or constraint error occurs.SELECT privileges ----------------------------------------------------BEGIN . PRAGMA EXCEPTION_INIT(insufficient_privileges.ORACLE returns error number -1031 if.•TOO_MANY_ROWS is raised if a SELECT INTO statement returns more than one row. User-defined Exceptions PL/SQL lets you define exceptions of your own. the same scope rules apply to variables and exceptions. But remember. A pragma is a compiler directive. That allows you to refer to any internal exception by name and to write a specific handler for it. However. Exceptions can be declared only in the declarative part of a PL/SQL block.. You code the pragma EXCEPTION_INIT in the declarative part of a PL/SQL block. subprogram. an exception is an error condition... •VALUE_ERROR is raised if an arithmetic. or package. or package using the syntax PRAGMA EXCEPTION_INIT(exception_name. where exception_name is the name of a previously declared exception. which can be thought of as a parenthetical remark to the compiler. EXCEPTION WHEN insufficient_privileges THEN -. user-defined exceptions must be declared and must be raised explicitly by RAISE statements. To handle unnamed internal exceptions. In PL/SQL. ORACLE_error_number). subprogram. -1031). You declare an exception by introducing its name. END. not an object. truncation. followed by the keyword EXCEPTION. the predefined pragma EXCEPTION_INIT tells the compiler to associate an exception name with an ORACLE error number. DECLARE past_due EXCEPTION. you must use the OTHERS handler or the pragma EXCEPTION_INIT.. BEGIN Exceptions and variable declarations are similar. for example -. •ZERO_DIVIDE is raised if you try to divide a number by zero.you try to UPDATE a table for which you have only -. Unlike variables. •Using EXCEPTION_INIT. conversion. -----------------------------------------------------. exceptions cannot appear in assignment statements or SQL statements.

PRAGMA EXCEPTION_INIT(null_salary. However.. which it can process using the error-reporting functions SQLCODE and SQLERRM in an OTHERS handler. IF number_on_hand < 1 THEN RAISE out_of_stock. other user-defined exceptions must be raised explicitly by RAISE statements. An application can call raise_application_error only from an executing stored subprogram. You can code a RAISE statement for a given exception anywhere within the scope of that exception. 27 . •Using RAISE statement.•Using raise_application_error. -20101). PROCEDURE raise_salary (emp_id NUMBER. When called. number_on_hand NUMBER(4). 'Salary is missing'). increase NUMBER) IS current_salary NUMBER..-20999 and error_message is a character string up to 512 bytes in length. DECLARE . rolls back any database changes it made. END IF.. Furthermore. ELSE UPDATE emp SET sal = current_salary + increase WHERE empno = emp_id. This package includes a procedure named raise_application_error. it can use EXCEPTION_INIT to map specific error numbers returned by raise_application_error to exceptions of its own. PL/SQL blocks and subprograms should RAISE an exception only when an error makes it undesirable or impossible to finish processing.. BEGIN . A package named DBMS_STANDARD (part of the Procedural Database Extention) provides language facilities that help your application interact with ORACLE.. which lets you issue user-defined error messages from a stored subprogram or database trigger. IF current_salary is NULL THEN raise_application_error(-20101. raise_application_error ends a subprogram.. END raise_salary. and returns a user-defined error message to the application.. BEGIN SELECT sal INTO current_salary FROM emp WHERE empno = emp_id. The calling application gets a PL/SQL exception. The calling syntax is raise_application_error(error_number. as are user-defined exceptions that you have associated with an ORACLE error number using EXCEPTION_INIT. DECLARE out_of_stock EXCEPTION. error_message). . How Exceptions Are Raised Internal exceptions are raised implicitly by the runtime system. where error_number is a negative integer in the range -20000. null_salary EXCEPTION.

reraises the current exception.end of sub-block ----------------------------EXCEPTION WHEN out_of_balance THEN .reraise the current exception END... which is allowed only in an exception handler. 28 . BEGIN . it must appear by itself. you can use an exception handler written for the predefined exception to process other errors. Sometimes.raise the exception END IF. then pass it to an enclosing block. .. separating them by the keyword OR. THEN RAISE out_of_balance. which is always the last handler in a block or subprogram. IF .handle the error differently . acts as the handler for all exceptions not named specifically. handle it locally.. as shown below DECLARE out_of_balance EXCEPTION.handle the error RAISE... The optional OTHERS exception handler..beginning of sub-block ----------------------BEGIN . The keyword OTHERS cannot appear in the list of exception names. Omitting the exception name in a RAISE statement.. If you want two or more exceptions to execute the same sequence of statements. Handling Raised Exceptions When an exception is raised. END.END IF. simply place a RAISE statement in the local handler. -.handle the error END... normal execution of your PL/SQL block or subprogram stops and control transfers to its exception-handling part and control does NOT return to where the exception was raised. you want to reraise an exception. that is. In other words. You can also raise a predefined exception explicitly: RAISE INVALID_NUMBER. . EXCEPTION WHEN out_of_balance THEN -. ------------.. Use of the OTHERS handler guarantees that no exception will go unhandled.. ------------. -. EXCEPTION WHEN out_of_stock THEN -. That way. list the exception names in the WHEN clause. To reraise an exception. you cannot resume processing where you left off.

The string function SUBSTR ensures that a VALUE_ERROR exception (for truncation) is not raised when you assign the value of SQLERRM to err_msg. EXCEPTION WHEN . Generally. You cannot use SQLCODE and SQLERRM directly in a SQL statement... subprograms have a declarative part. Procedures A procedure is a subprogram that performs a specific action. PL/SQL has two types of subprograms called procedures and functions. you use a procedure to perform an action and a function to compute a value... THEN ..handle the error differently WHEN . where parameter stands for the following syntax var_name [IN | OUT | IN OUT] datatype [{:= | DEFAULT} value] 29 . [. You write procedures using the syntax PROCEDURE name [ (parameter. err_msg CHAR(100).handle the error differently . parameter. WHEN OTHERS THEN err_num := SQLCODE. then use the variables in the SQL statement. INSERT INTO errors VALUES (err_num.... 100). SQLCODE and SQLERRM are especially useful in the OTHERS exception handler because they tell you which internal exception was raised. 1.. THEN . err_msg). • Using SQLCODE and SQLERRM. an executable part... WHEN OTHERS THEN .. Instead. you must assign their values to local variables.]) ] IS [local declarations] BEGIN executable statements [EXCEPTION] exception-handlers] END [name]. END. Subprograms Subprograms are named PL/SQL blocks that can take parameters and be invoked. DECLARE err_num NUMBER.handle the error differently END.. and an optional exception-handling part.. OR . .. Like unnamed or anonymous PL/SQL blocks. BEGIN . err_msg := SUBSTR(SQLERRM.

. ESLE UPDATE emp SET sal = sal + increase WHERE empno = emp_id.Unlike the datatype specifier in a variable declaration. the datatype specifier in a parameter declaration must be unconstrained. where parameter stands for the following syntax var_name [IN | OUT | IN OUT] datatype [{:= | DEFAULT} value] The function body begins with the keyword IS and ends with the keyword RETURN clause. [EXCEPTION WHEN NO_DATA_FOUND THEN INSERT INTO emp_audit VALUES (emp_id. IF current_salary IS NULL THEN RAISE salary_missing. parameter. You write functions using the syntax FUNCTION name [ (parameter.. The procedure body begins with the keyword IS and ends with the keyword END followed by an optional procedure name. Functions A function is a subprogram that computes a value. except that functions have a RETURN clause. (name CHAR(20) ) IS -. PROCEDURE raise_salary (emp_id INTEGER. which specifies the datatype of the result value. 'Salary is null'). [.. 'No such number'). but not in SQL statements.. should be CHAR The procedure specification begins with the keyword PROCEDURE and ends with the procedure name or a parameter list.]) ] RETURN datatype IS [local declarations] BEGIN executable statements [EXCEPTION exception-handlers] END [name]. WHEN salary_missing THEN INSERT INTO emp_audit VALUES (emp_id.illegal. PROCEDURE . END IF. salary_missing EXCEPTION. END raise_salary. BEGIN SELECT sal INTO current_salary FROM emp WHERE empno = emp_id. Calls to user-defined functions can appear in procedural statements.. increase REAL) IS current_salary REAL. 30 . Functions and procedure are structured alike.

Otherwise. In procedures. hisal INTO min_sal. END. max_sal REAL.. The statement simply returns control to the caller before the normal end of the procedure is reached. DECLARE PROCEDURE calc_rating (.. Forward Declarations PL/SQL requires that you declare an identifier before using it.forward declaration /* Define subprogram in alphabetical order */ PROCEDURE award_bonus (.FUNCTION sal_ok (salary REAL... A function must contain at least one RETURN statement. You can place the subprogram body anywhere after the forward declaration. END sal_ok.. which is evaluated when the RETURN statement is reached.. in functions. or package. but they must appear in the same block.. 31 . A subprogram can contain several RETURN statements. Execution then resumes with the statement following the subprogram call. subprogram. END. Therefore.) IS BEGIN calc_rating(. none of which need be the last lexical statement. title REAL) RETURN BOOLEAN IS min_sal REAL.. .). RETURN Statement The RETURN statement immediatedly completes the execution of a subprogram and returns control to the caller. RETURN (salary >= min_sal) AND (salary <= max_sal).. PROCEDURE calc_rating (. . max_sal FROM sals WHERE job = title. BEGIN SELECT losal. A forward declaration consists of a subprogram specification terminated by a semicolon. PL/SQL solves the problem of subprograms used before they are declared by providing a special subprogram declaration called forward declaration. Although the formal parameter list appears in the forward declaration. it must also appear in the subprogram body.. -. you must declare a subprogram before calling it.).) IS BEGIN . a RETURN statement must contain an expression... PL/SQL raises the predefined exception PROGRAM_ERROR at run time. However. a RETURN statement cannot contain an expression.

Actual versus Formal Parameters Subprograms pass information using parameters. it cannot be a constant or expression. The actual parameter that corresponds to an OUT formal parameter must be a variable. it cannot be assigned a value. Unlike OUT and IN OUT parameters. 32 . the value is lost when you call the subprogram. an OUT parameter acts like an uninitialized variable. •an IN parameter lets you pass values to the subprogram being called. where they are invisible to applications. When calling a subprogram. you can write the actual parameters using either positional or named notation. Therefore. the call to the procedure raise_salary can be made as follows: raise_salary(emp. Inside the subprogram. it cannot be a constant or expression.Packaged Subprograms Forward declarations also let you group logically related subprograms in a package. Inside the subprogram. and the subprogram bodies go in the package body. packages allow you to hide implementation details. The actual parameter that corresponds to an IN OUT formal parameter must be a variable. An OUT actual parameter can (but need not) have a value before the subprogram is called. the second uses named notation. an IN parameter can be initialized to default values. IN (the default). OUT. Therefore. Before exiting a subprogram. The variables declared in a subprogram specification and referenced in the subprogram body are formal parameters. an IN parameter acts like a constant. The variables or expressions referenced in the parameter list of a subprogram call are actual parameters. inc). If you exit successfully. Thus. However. an IN OUT parameter acts like an initialized variable. The subprogram specifications go in the package specification. •an IN OUT parameter lets you pass initial values to the subprogram being called and return updated values to the caller. However. increase => inc) The first procedure call uses positional notation. For example. PL/SQL assigns values to the actual parameters. Otherwise. •an OUT parameter lets you return values to the caller of a subprogram. avoid using the OUT and IN OUT modes with functions. if you exit with an unhandled exception. and the third uses mixed notation. Inside the subprogram. PL/SQL does not assign values to the actual parameters. can be used with any subprogram. The actual parameter and its corresponding formal parameter must belong to compatible datatypes. The three parameter modes. emp_id => emp) raise_salary(emp. However. and IN OUT. its value cannot be assigned to another variable or reassigned to itself. the values of corresponding actual parameters are indeterminate. Parameter Modes You use parameter modes to define the behavior of formal parameters. explicitly assign values to all OUT formal parameters. raise_salary(increase => inc.

although sometimes the body is unnecessary. Stored Subprograms Subprograms can be compiled separately and stored permanently in an ORACLE database. PL/SQL determines which of the two procedures is being called by checking their formal parameters. hiredate_tab DateTabTyp. you cannot overload two functions that differ only in return type even if the types are in different families. an ORACLE Precompiler application. you can use the same name for several different subprograms as long as their formal parameters differ in number.0. memory savings. objects. That is. END initialize. END initialize.. another stored subprogram. or datatype family. You can call stored subprograms from a database trigger. n INTEGER) IS BEGIN FOR i IN 1. n INTEGER) IS BEGIN FOR i IN 1.. better performance. You can place the two overloaded initialize procedures in the same block. END LOOP. Because the processing in these two procedures is the same. You cannot overload the names of stand-alone subprograms. subprogram. and tighter security. a specification and a body. PACKAGES A package is a database object that groups logically related PL/SQL types. For example you can grant users EXECUTE access to a stored procedure that updates the emp table. ready to be executed. You cannot overload two subprograms if their formal parameters differ only in datatype and the different datatypes are in the same family (REAL and INTEGER). That way. but cannot arbitrarily manipulate table data.n LOOP tab(i) := 0. You can restrict users to specific database operations by granting access only through subprograms. TYPE RealTabTyp IS TABLE OF REAL INDEX BY BINARY_INTEGER.. PROCEDURE initialize (tab OUT RealTabTyp. Stored subprograms offer higher productivity. and subprograms. or package. . You might write the following procedures to initialize the PL/SQL tables named initialize for hiredate_tab and sal_tab. order. Stored subprograms can help enforce data security. PROCEDURE initialize (tab OUT DateTabTyp. users can call the procedure. application integrity. but not grant them access to the table itself.n LOOP tab(i) := SYSDATE.. Packages usually have two parts. Finally. or an ORACLE tool such as SQL*Plus.Overloading PL/SQL lets you overload subprogram names. The specification is the 33 . END LOOP. DECLARE TYPE DateTabTyp IS TABLE OF DATE INDEX BY BINARY_INTEGER. it is logical to give them the same name. You cannot overload two subprograms if their formal parameters differ only in name or parameter mode. an OCI application. sal_tab RealTabTyp.

body CURSOR desc_salary (emp_id NUMBER) RETURN EmpRecTyp IS SELECT empno. PROCEDURE hire_employee (ename CHAR. exceptions. PROCEDURE fire_employee (emp_id NUMBER) IS BEGIN DELETE FROM emp WHERE empno = emp_id.initialization statements] END [name]. and subprograms available for use. BEGIN INSERT INTO emp VALUES (empno_seq. sal NUMBER. the format of a package is similar to that of a subprogram: PACKAGE name IS -. PROCEDURE hire_employee (ename CHAR. sal. SYSDATE.subprogram bodies [BEGIN -. and so implements the specification.private type and object declarations -. ename. job. comm. job CHAR. a cursor.body (hidden part) -. Packages offer several advantages: modularity.specification (visible part) -. or nested. job CHAR. variables.subprogram specifications END [name}. deptno ). In the following example. PROCEDURE fire_employee (emp_id NUMBER). salary REAL). comm NUMBER. 34 . CURSOR desc_salary (emp_id NUMBER) RETURN EmpRecTyp. passed parameters. END emp_actions. END hire_employee. CREATE PACKAGE BODY emp_actions AS -. comm NUMBER. constants. Unlike subprograms.interface to your application. packages cannot be called. a record type. The body fully defines cursors and subprograms. information hiding. mgr.specification TYPE EmpRecTyp is RECORD (emp_id INTEGER. sal NUMBER. easier application design. added functionality. Packages are created interactively with SQL*Plus using the CREATE PACKAGE and CREATE PACKAGE BODY commands. END fire_employee. END emp_actions. it declares the types. sal FROM emp ORDER BY sal DESC.NEXTVAL. deptno NUMBER ). deptno NUMBER ). PACKAGE BODY name IS -. mgr NUMBER. cursors. Still. mgr NUMBER.public type and object declarations -. and better performance. and two employment procedures are packaged: CREATE PACKAGE emp_actions AS -.

the declarative part of a package body can contain subprogram bodies. So. you use dot notation as follows: package_name. the declared types and objects are inaccessible except from within the package body. the package body contains the definition of every cursor and subprogram declared in the package specification. objects. objects. So. The specification lists the package resources available to applications. Changes to a package body do not require ORACLE to recompile dependent procedures. and subprograms that must be visible to users of the package. Place in a specification only the types.object_name package_name. All information your application needs to use the resources is in the specification. which typically holds statements that initialize some of the variables previously declared in the package. As a result. Only subprograms and cursors have an underlying implementation or definition. However. That is. which define types and objects necessary for the internal workings of the package. the initialization part of a package is run only once. Unlike a package specification. Keep in mind that subprograms defined in a package body are accessible outside the package only if their specification also appear in the package specification. if a specification declares only types. Avoid writing packages that duplicate some feature already provided by ORACLE. The scope of these declarations is local to your database schema and global to the package. To reference the types. Package specifications reflect the design of your application. The package body can also contain private declarations. a package cannot be called or passed parameters.type_name package_name. Following the declarative part of a package body is the optional initialization part. The initialization part of a package plays a minor role because.The Package Specification The package specification contains public declarations.subprogram_name The Package Body The package body implements the package specification. Guidelines When writing packages. unlike subprograms. the package body is unnecessary. place as few items as possible in a package specification. and exceptions. The scope of these declarations is local to the package body. keep them as general as possible so they can be reused in future applications. constants. and subprograms declared within a package specification. To reduce the need for recompiling when code is changed. variables. 35 . changes to a package specification require ORACLE to recompile every stored subprogram that references the package. the first time you reference the package. define them before the package bodies. Therefore.

Sign up to vote on this title
UsefulNot useful