embedded sql john ortiz. lecture 15embedded sql2 why isn’t interactive sql enough? how to do this...

Embedded SQL

John Ortiz

Lecture 15 Embedded SQL 2

Why Isn’t Interactive SQL Enough? How to do this using interactive SQL?

Print a well-formatted transcript of a student with either a name or an id

What does it take for someone to use SQL? To know schemas, syntax, semantics,

mathematics, logic, … Solution?

Write application programs to help naïve users to manipulate the data.

How to access database from within a program?


Idea of Embedded SQL Combine the power of both SQL & a

general purpose programming language. Use (embedded) SQL to perform data

retrieval and updates. Use the general purpose programming

language (host PL) to perform more complex data processing and to provide a friendly user interface.


Oracle API Support embedded SQL through five host

PLs (pro*languages)C/C++, Cobol, PL/I, Ada, Pascal

Oracle8i supports Java/JDBC and SQLJ SQL stmts are placed in host PL programs Data flow from database to program

variables and vice versa Two step compilation:

Precompilation: prog.pc prog.cc Compilation: prog.cc prog.o


A Sample Pro*C/C++ Program The program is sample1.pc Common tasks:

Declare variables interfacing SQL & host PL

Prepare for any SQL error Include sqlca (communication area) Use whenever sqlerror, … Provide for error processing

Connect to database Issue SQL statements Disconnect the database


PL & SQL Variables Every variable used to get data from & pass

data to database must be declared in a declare section

exec sql begin declare section; varchar user_name[20], passwd[10]; … exec sql end declare section; Must use SQL data types Can appear in both SQL (preceded by a

colon :user_name) & PL statements (no colon)


PL & SQL Variables (cont.) Oracle Pro*C pre-processor will convert varchar user_name[20]; to: struct { unsigned short len; unsigned char arr[20]; } user_name; SQL string is not ‘\0’-ended, but C string

is Variables must have comparable types to

get data from table columns


Handle SQL Errors SQLCA (p.121): SQL Communication area is

a predefined data structure used to pass control information from Oracle to application program Error code: sqlca.sqlcode=0 successful; > 0 value not found; < 0 error

Can also use predefined error types in whenever: sqlerror, not found, sqlwarning, …

Common error handling actions: goto Label, continue, stop, do func, do break, do return, …


Embedded SQL Statements Every SQL statement is preceded by exec

sql Can use all SQL statements plus special

ones. Connect, disconnect Whenever Select … into … from … Rollback Commit Statements declare and use cursors Statements define and execute dynamic

queries


Transaction Control Statements A transaction is modeled by a sequence of

database update operations Operations of a transaction should either all

be successfully executed or none is executed

When a failure occurs, all updates done so far need to be undone

Explicit control: Start: set transaction [read only, read

write] Undo: rollback End: commit


Sample Program Using A Cursor How does a program handle query result

containing more than one tuple? Use a cursor. See sample2.pc A cursor is a “window” through which one

tuple can be accessed. A cursor must be declared with a query Open cursor executes the query Fetch cursor moves to the next tuple A cursor can be closed and re-opened


Dynamic SQL Create SQL statements at run time and then

execute the newly created statements. General framework:

Declare a host string variable. Place an SQL statement in the variable at

run-time. Let the DBMS parse & execute the SQL

statement in the host variable.


Why Dynamic SQL Consider relation: Projects(Pno, Name, Budget, Duration) How to write a program to delete projects

that satisfy some yet unknown conditions? delete Projects where <unknown> Some possible types of conditions:

(1) Budget > 2000000(2) Budget > 2000000 and Duration > 24(3)Name = 'Manned Spacecraft to Mars‘

Why isn’t static SQL a viable solution?


Methods of Dynamic SQL Method 1: Non-query statement w/o host

var. delete from emp where deptno = 20 Method 2: Non-query w/ known # of input

var. delete from emp where empno = :eid Method 3: Query w/ known select-list &

input host variables. select ename from emp where deptno

= :dno Method 4: Query w/ unknown select-list &

var. select <unknown> from emp


Execute-Immediate Statement A sample program execute immediate causes the dynamic SQL

statement to be compiled and executed immediately.

General execute immediate statement: exec sql execute immediate :host_variable; Implement Method 1.


Prepare-Execute Statement A sample program prepare stmt_name from :host_variable

compiles the SQL query in host_variable and then saves the prepared result in stmt_name. stmt_name is an SQL identifier and needs

not be explicitly declared. Implement Method 2: The host_variable can

contain any SQL statement other than a query and can have a known number of placeholders.

Prepare once, run many times (with different values in input host variables).


Pro*C/C++ and PL/SQL On Oracle, Pro*C/C++ programs may

contain any SQL statement and PL/SQL blocks.

See this sample program Precompiled and stored PL/SQL

procedures and functions can be used directly in embedded SQL statement

exec sql execute p(…)


Oracle JDBC JDBC (Java Database Connectivity): API

that allows Java applications and applets to connect to Oracle databases, send SQL statements, and receive data from databases.

Need to set up CLASSPATH environment variable.

See QueryUnivDB.java


JDBC: Program Tasks Import Java SQL API library Load JDBC driver: Class.forName(…) Create a DB connection DriveManager.getConnection(<connect

string>) Create a query statement object Create a result object by executing the

query Process the result Close the query Close the connection


Oracle SQLJ Java with embedded SQL (prec. w/ #sql) Part of the new SQL 1999 Standard. Much easier to use:

Connect to database Create iterators for query results

Oracle8i only supports JDK1.1 Need two compilation steps:

sqlj file[.sqlj] javac file.java

See QueryUnivDB.sqlj


Embedded SQL Summary Many additional statements handling

communication between database and PL system

Special handling of multiple tuple result (cursor)

Must handle errors and exceptions generated by DBMS

Must pay special attention to program structure (goto, whenever not found,… )

JDBC & SQLJ have much better DB-PL interface


Look Ahead Next topic: Storage System Read textbook:

Chapter 5

embedded sql john ortiz. lecture 15embedded sql2 why isn’t interactive sql enough? how to do this...

Documents