Archive for the ‘Oracle XE’ Category
SQL Logic Overkill, again …
It’s interesting to watch people try to solve problems. For example, the student is required to use a scalar subquery in a SQL lab exercise that I wrote. It should be a simple fix. The problem is structured with an incorrect foreign key value in an external CSV file and the restriction that you can not replace the value in the external CSV file. I hoped that students would see the easiest option was to write a scalar subquery in the SELECT clause to replace the value found in the external file. There’s even a hint about how to use a scalar subquery.
Students who are new to SQL can take very interesting approaches to solve problems. The flexibility of SQL can lead them to solve problems in interesting ways. While the following solution worked to solve the problem, it’s wrong on two levels:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 | INSERT INTO TRANSACTION (SELECT transaction_s1.NEXTVAL , tr.transaction_account , CASE WHEN NOT tr.transaction_type = (SELECT common_lookup_id FROM common_lookup WHERE common_lookup_table = 'TRANSACTION' AND common_lookup_column = 'TRANSACTION_TYPE' AND common_lookup_type = 'CREDIT') THEN cl.common_lookup_id END AS transaction_type , tr.transaction_date , (tr.transaction_amount / 1.06) AS transaction_amount , tr.rental_id , tr.payment_method_type , tr.payment_account_number , tr.created_by , tr.creation_date , tr.last_updated_by , tr.last_update_date FROM transaction_reversal tr CROSS JOIN common_lookup cl WHERE cl.common_lookup_table = 'TRANSACTION' AND cl.common_lookup_column = 'TRANSACTION_TYPE' AND cl.common_lookup_type = 'CREDIT'); |
The CASE statement on lines 4 through 12 substitutes a value only when the source value is not a match. That means if the source file is ever correct a null value would become the transaction_type column value, which would make the statement fail because the transaction_type column is NOT NULL constrained in the target transaction table. Therefore, the logic of the student’s approach requires adding an ELSE clause to the CASE statement for the event that the source file is ever corrected. The modified CASE statement would be =the following:
4 5 6 7 8 9 10 11 12 13 14 | , CASE WHEN NOT tr.transaction_type = (SELECT common_lookup_id FROM common_lookup WHERE common_lookup_table = 'TRANSACTION' AND common_lookup_column = 'TRANSACTION_TYPE' AND common_lookup_type = 'CREDIT') THEN cl.common_lookup_id ELSE tr.transaction_type END AS transaction_type |
The second element of student thought at issue is the CROSS JOIN to the in-line view. It does one thing right and another wrong. It uses the unique key to identify a single row, which effectively adds all the columns for that one row to all rows returned from the external transaction_reversal table. The CROSS JOIN is a correct approach to adding values for computations to a query when you need those columns for computations. The problem with this CROSS JOIN logic may not be immediately obvious when you write it in ANSI SQL 1992 syntax, but it should become obvious when you replace the inline view with a Common Table Expression (CTE) in ANSI SQL 1999 syntax, like:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 | INSERT INTO TRANSACTION (WITH cte AS (SELECT * FROM common_lookup WHERE common_lookup_table = 'TRANSACTION' AND common_lookup_column = 'TRANSACTION_TYPE' AND common_lookup_type = 'CREDIT') SELECT transaction_s1.NEXTVAL , tr.transaction_account , cte.common_lookup_id AS transaction_type , tr.transaction_date , (tr.transaction_amount / 1.06) AS transaction_amount , tr.rental_id , tr.payment_method_type , tr.payment_account_number , tr.created_by , tr.creation_date , tr.last_updated_by , tr.last_update_date FROM transaction_reversal tr CROSS JOIN cte); |
Unfortunately, you would discover that Oracle Database 11g does not support the use of an ANSI SQL 1999 WITH clause inside as the source for an INSERT statement. Oracle Database 12c does support the use of the ANSI SQL 1999 WITH clause inside a subquery of an INSERT statement. That’s an “Oops!” for Oracle 11g because that means the Oracle database fails to meet the ANSI SQL 1999 compliance test. 😉 Great that they fixed it in Oracle 12c. While the nested query would work in Oracle as an ordinary query (outside of an INSERT statement). It raises the following error when you embed it in an INSERT statement:
ERROR AT line 20: ORA-32034: unsupported USE OF WITH clause |
The WITH clause does highlight a key problem with the idea of a CROSS JOIN in this situation. You don’t need all the columns from the common_lookup table. You only need the common_lookup_id column. That make the CROSS JOIN approach suboptimal if it worked.
The complex logic in the original approach is wasted. That’s true because the common_lookup_id value can be supplied to each row as the value from a scalar subquery. The scalar query runs once and the result is placed in the return set for each row. You implement the scalar subquery in the SELECT clause, like:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | INSERT INTO TRANSACTION (SELECT transaction_s1.NEXTVAL , tr.transaction_account , (SELECT common_lookup_id FROM common_lookup WHERE common_lookup_table = 'TRANSACTION' AND common_lookup_column = 'TRANSACTION_TYPE' AND common_lookup_type = 'CREDIT') AS transaction_type , tr.transaction_date , (tr.transaction_amount / 1.06) AS transaction_amount , tr.rental_id , tr.payment_method_type , tr.payment_account_number , tr.created_by , tr.creation_date , tr.last_updated_by , tr.last_update_date FROM transaction_reversal tr); |
There really was no intent or logical outcome where the value from the original CASE statement would be different than the subquery’s common_lookup_id value. That fact makes adding an ELSE clause useless, and the solution viable though inefficient. Also, there was no need for the additional columns from the common_lookup table because they are unused. The subquery on lines 4 through 8 provides the optimal solution and improved efficiency.
Developers should ask themselves two questions when they write SQL:
- If my logic is so elegant why do I need it to be so elegant?
- Is there a simpler solution to provide the desired result set?
If there aren’t good answers to both questions, they should re-write it. I hope the examples answer questions and help folks solve problems.
Debug PL/SQL Web Pages
What happens when you can’t get a PL/SQL Web Toolkit to work because it only prints to a web page? That’s more tedious because any dbms_output.put_line command you embed only prints to a SQL*Plus session. The answer is quite simple, you create a test case and test it inside a SQL*Plus environment.
Here’s a sample web page that fails to run successfully …
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 | CREATE OR REPLACE PROCEDURE html_table_values ( name_array OWA_UTIL.VC_ARR , value_array OWA_UTIL.VC_ARR ) IS BEGIN /* Print debug to SQL*Plus session. */ FOR i IN 1..name_array.COUNT LOOP DBMS_OUTPUT.put_line('Value ['||name_array(i)||'='||value_array(i)||']'); END LOOP; /* Open HTML page with the PL/SQL toolkit. */ htp.print('<!DOCTYPE html>'); htp.htmlopen; htp.headopen; htp.htitle('Test'); htp.headclose; htp.bodyopen; htp.line; htp.print('Test'); htp.line; htp.bodyclose; htp.htmlclose; END; / |
You can test the program with the following anonymous block as the SYSTEM user, which is equivalent to the following URL:
http://localhost:8080/db/html_table_values?begin=1004&end=1012 |
The following test program lets you work:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | DECLARE x OWA_UTIL.VC_ARR; y OWA_UTIL.VC_ARR; BEGIN /* Insert first row element. */ x(1) := 'begin'; y(1) := '1004'; /* Insert second row element. */ x(2) := 'end'; y(2) := '1012'; /* Call the anonymous schema's web page. */ anonymous.html_table_values(x,y); END; / |
It should print:
Value [begin=1004] Value [end=1012] |
I hope this helps those looking for a solution.
SQL*Plus Tricks
Have you ever wondered how to leverage substitution variables in anonymous block programs? There are several tricks that you can use beyond passing numeric and string values to local variable. The generic default appears to take a number unless you cast it as a string but that’s not really the whole story. The first two are standard examples of how to use numeric and string substitution values.
The following accept a numeric substitution value:
1 2 3 4 5 6 7 8 9 10 | DECLARE lv_input NUMBER; BEGIN /* Assign substitution value to local variable. */ lv_input := &input; /* Print the local variable. */ dbms_output.put_line('['||lv_input||']'); END; / |
The following accept a string substitution value, casts the input as a string, assigns the string value to a 4,000 character length local variable, checks whether the 4,000 character length is greater than 10, and assigns the first 10 characters to the lv_parse_input variable:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | DECLARE lv_unparsed_input VARCHAR2(4000); lv_parsed_input VARCHAR2(10); BEGIN /* Assign substitution value to local variable. */ lv_unparsed_input := '&input'; /* Check size of input value. */ IF LENGTH(lv_unparsed_input) > 10 THEN lv_parsed_input := SUBSTR(lv_unparsed_input,1,10); END IF; /* Print the local variable. */ dbms_output.put_line('Print {lv_parsed_input}: ['||lv_parsed_input||']'); END; / |
Next, let’s examine two tricks. The first passes a case insensitive variable name and the second passes a case sensitive variable name as a parameter to an anonymous block program.
This declares an anonymous block program that uses a substitution value as a variable name:
1 2 3 4 5 6 7 | DECLARE mine VARCHAR2(10) := 'Default'; BEGIN /* Print the local variable's value. */ dbms_output.put_line('Print {mine} variable value: ['||&input||']'); END; / |
When you run the anonymous block, you’re prompted for an input variable. You provide a case insensitive variable name as the input value:
Enter value for input: MINE
old 5: dbms_output.put_line('['||&input||']');
new 5: dbms_output.put_line('['||MINE||']');
Print {mine} variable value: [Default] |
The downside of this approach, yields an ORA-06550 and PLS-00201 exception. Neither of these can be caught because Oracle raises the errors during parsing when the variable name isn’t a 100% case insensitive match. The same type of problem occurs in the next example when the input variable isn’t a 100% case sensitive match.
You can rewrite the program to handle case insensitive variables like this:
1 2 3 4 5 6 7 | DECLARE "Mine" VARCHAR2(10) := 'Default'; BEGIN /* Print the local variable's value. */ dbms_output.put_line('Print {mine} variable value: ['||"&input"||']'); END; / |
When you run the anonymous block, you’re prompted for an input variable. You provide a case sensitive variable name as the input value:
Enter value for input: Mine
old 5: dbms_output.put_line('['||&input||']');
new 5: dbms_output.put_line('['||"Mine"||']');
Print {Mine} variable value: [Default] |
Hope this helps those looking for a solution.
Use an object in a query?
Using an Oracle object type’s instance in a query is a powerful capability. Unfortunately, Oracle’s SQL syntax doesn’t make it immediately obvious how to do it. Most get far enough to put it in a runtime view (a subquery in the FROM clause), but then they get errors like this:
SELECT instance.get_type() * ERROR AT line 4: ORA-00904: "INSTANCE"."GET_TYPE": invalid identifier |
The problem is how Oracle treats runtime views, which appears to me as a casting error. Somewhat like the ORDER BY clause irregularity that I noted in July, the trick is complete versus incomplete syntax. The following query fails and generates the foregoing error:
1 2 3 4 | SELECT instance.get_type() AS object_type , instance.to_string() AS object_content FROM (SELECT dependent()AS instance FROM dual); |
If you add a table alias, or name, to the runtime view on line 4, it works fine:
1 2 3 4 | SELECT cte.instance.get_type() AS object_type , cte.instance.to_string() AS object_content FROM (SELECT dependent() AS instance FROM dual) cte; |
That is the trick. You use an alias for the query, which assigns the alias like a table reference. The reference lets you access instance methods in the scope of a query. Different columns in the query’s SELECT-list may return different results from different methods from the same instance of the object type.
You can also raise an exception if you forget the open and close parentheses for a method call to a UDT, which differs from how Oracle treats no argument functions and procedures. That type of error would look like this:
SELECT cte.instance.get_type AS object_type * ERROR AT line 1: ORA-00904: : invalid identifier |
It is an invalid identifier because there’s no public variable get_type, and a method is only found by using the parenthesis and a list of parameters where they’re required.
The object source code is visible by clicking on the expandable label below.
Setup Object Types ↓
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 | DROP TYPE dependent_t; DROP TYPE base_t; SET SERVEROUTPUT ON SIZE UNLIMITED -- Create an object. CREATE OR REPLACE TYPE base_t IS OBJECT ( TYPE VARCHAR2(20) , CONSTRUCTOR FUNCTION base_t RETURN SELF AS RESULT , MEMBER FUNCTION get_type RETURN VARCHAR2 , MEMBER FUNCTION to_string RETURN VARCHAR2) INSTANTIABLE NOT FINAL; / -- Create an object body. CREATE OR REPLACE TYPE BODY base_t IS CONSTRUCTOR FUNCTION base_t RETURN SELF AS RESULT IS BEGIN RETURN; END base_t; MEMBER FUNCTION get_type RETURN VARCHAR2 IS BEGIN RETURN self.TYPE; END; MEMBER FUNCTION to_string RETURN VARCHAR2 IS BEGIN RETURN self.TYPE; END to_string; END; / -- Create a subtype. CREATE OR REPLACE TYPE dependent UNDER base_t ( child VARCHAR2(40) , CONSTRUCTOR FUNCTION dependent RETURN SELF AS RESULT , CONSTRUCTOR FUNCTION dependent ( child VARCHAR2 ) RETURN SELF AS RESULT , OVERRIDING MEMBER FUNCTION get_type RETURN VARCHAR2 , OVERRIDING MEMBER FUNCTION to_string RETURN VARCHAR2) INSTANTIABLE NOT FINAL; / -- Create a subtype body. CREATE OR REPLACE TYPE BODY dependent IS CONSTRUCTOR FUNCTION dependent RETURN SELF AS RESULT IS BEGIN /* Assign subtype name to type. */ self.TYPE := 'DEPENDENT'; RETURN; END dependent; CONSTRUCTOR FUNCTION dependent ( child VARCHAR2 ) RETURN SELF AS RESULT IS lv_dependent DEPENDENT := dependent(); BEGIN /* Assign default constructor to self instance. */ self := lv_dependent; /* Assign parameters to object instance. */ self.TYPE := 'DEPENDENT'; self.child := child; RETURN; END dependent; OVERRIDING MEMBER FUNCTION get_type RETURN VARCHAR2 IS BEGIN RETURN self.TYPE; END; OVERRIDING MEMBER FUNCTION to_string RETURN VARCHAR2 IS BEGIN RETURN self.TYPE; END to_string; END; / |
As always, I hope this helps those solving problems.
Mac SQL Developer Install
This how you install SQL Developer on Mac OS Yosemite. The first thing you need to do is download and install Java 8, not Java 7 on your Mac OS Yosemite as suggested on some web sites. You can determine whether or not Java is installed by running the following command:
Mac-Pro-3:~ username$ java -version No Java runtime present, requesting install. |
You must accept the Java license to install Java 8 on the Mac OS X operating system:
You have the option of installing the Java SDK or JDK. I’ve opted to install Netbeans 8 with JDK 8u45, as you can tell from the screen capture after you launched the file:
It is a standard Mac OS installation, which is why I didn’t bother showing any dialog messages. After installing the Java JDK or SDK, you should download SQL Developer 4.1 from Oracle’s web site. Below is a screen shot of the Oracle download web page where I’ve accepted the license agreement:
If you attempt to launch the installation and you’ve set your Mac Security to the “Mac App Store and identified developers” setting, you should raise the following exception:
If you reset the Mac Security to an “Anywhere” setting, you can install Oracle SQL Developer on Yosemite. Just make sure you reset it to the “Mac App Store and identified developers” setting after you install SQL Developer.
If you launch SQL Developer with the Security “Anywhere” setting, it displays the following dialog:
After you launch the program, you will see the following progress dialog:
The last step of the installation launches SQL Developer, as shown below:
Click the Connections icon to create an initial connection, like the following:
After connecting to the database, you can write and execute a query as shown in the next screen capture:
As always, I hope that this helps those who require an example to install SQL Server on a Mac OS.
Bash Arrays & Oracle
Last week, I wrote about how to use bash arrays and the MySQL database to create unit and integration test scripts. While the MySQL example was nice for some users, there were some others who wanted me to show how to write bash shell scripts for Oracle unit and integration testing. That’s what this blog post does.
If you don’t know much about bash shell, you should start with the prior post to learn about bash arrays, if-statements, and for-loops. In this blog post I only cover how to implement a bash shell script that runs SQL scripts in silent mode and then queries the database in silent mode and writes the output to an external file.
I’ve copied the basic ERD for the example because of a request from a reader. In their opinion it makes cross referencing the two posts unnecessary.
To run the bash shell script, you’ll need the following SQL files, which you can see by clicking not he title below. There are several differences. For example, Oracle doesn’t support a DROP IF EXISTS syntax and requires you to write anonymous blocks in their PL/SQL language; and you must explicitly issue a QUIT; statement even when running in silent mode unlike MySQL, which implicitly issues an exit.
Setup SQL Files ↓
The actor.sql file:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 | -- Drop actor table and actor_s sequence. BEGIN FOR i IN (SELECT object_name , object_type FROM user_objects WHERE object_name IN ('ACTOR','ACTOR_S')) LOOP IF i.object_type = 'TABLE' THEN EXECUTE IMMEDIATE 'DROP TABLE ' || i.object_name || ' CASCADE CONSTRAINTS'; ELSIF i.object_type = 'SEQUENCE' THEN EXECUTE IMMEDIATE 'DROP SEQUENCE ' || i.object_name; END IF; END LOOP; END; / -- Create an actor table. CREATE TABLE actor ( actor_id NUMBER CONSTRAINT actor_pk PRIMARY KEY , actor_name VARCHAR(30) NOT NULL ); -- Create an actor_s sequence. CREATE SEQUENCE actor_s; -- Insert two rows. INSERT INTO actor VALUES (actor_s.NEXTVAL,'Chris Hemsworth'); INSERT INTO actor VALUES (actor_s.NEXTVAL,'Chris Pine'); INSERT INTO actor VALUES (actor_s.NEXTVAL,'Chris Pratt'); -- Quit session. QUIT; |
The film.sql file:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 | -- Drop film table and film_s sequence. BEGIN FOR i IN (SELECT object_name , object_type FROM user_objects WHERE object_name IN ('FILM','FILM_S')) LOOP IF i.object_type = 'TABLE' THEN EXECUTE IMMEDIATE 'DROP TABLE ' || i.object_name || ' CASCADE CONSTRAINTS'; ELSIF i.object_type = 'SEQUENCE' THEN EXECUTE IMMEDIATE 'DROP SEQUENCE ' || i.object_name; END IF; END LOOP; END; / -- Create a film table. CREATE TABLE film ( film_id NUMBER CONSTRAINT film_pk PRIMARY KEY , film_name VARCHAR(30) NOT NULL ); -- Create an actor_s sequence. CREATE SEQUENCE film_s; -- Insert four rows. INSERT INTO film VALUES (film_s.NEXTVAL,'Thor'); INSERT INTO film VALUES (film_s.NEXTVAL,'Thor: The Dark World'); INSERT INTO film VALUES (film_s.NEXTVAL,'Star Trek'); INSERT INTO film VALUES (film_s.NEXTVAL,'Star Trek into Darkness'); INSERT INTO film VALUES (film_s.NEXTVAL,'Guardians of the Galaxy'); -- Quit session. QUIT; |
The movie.sql file:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 | -- Drop movie table and movie_s sequence. BEGIN FOR i IN (SELECT object_name , object_type FROM user_objects WHERE object_name IN ('MOVIE','MOVIE_S')) LOOP IF i.object_type = 'TABLE' THEN EXECUTE IMMEDIATE 'DROP TABLE ' || i.object_name || ' CASCADE CONSTRAINTS'; ELSIF i.object_type = 'SEQUENCE' THEN EXECUTE IMMEDIATE 'DROP SEQUENCE ' || i.object_name; END IF; END LOOP; END; / -- Create an movie table. CREATE TABLE movie ( movie_id NUMBER CONSTRAINT movie_pk PRIMARY KEY , actor_id NUMBER CONSTRAINT movie_nn1 NOT NULL , film_id NUMBER CONSTRAINT movie_nn2 NOT NULL , CONSTRAINT actor_fk FOREIGN KEY (actor_id) REFERENCES actor (actor_id) , CONSTRAINT film_fk FOREIGN KEY (film_id) REFERENCES film(film_id)); -- Create table constraint. CREATE SEQUENCE movie_s; -- Insert translation rows. INSERT INTO movie VALUES ( movie_s.NEXTVAL ,(SELECT actor_id FROM actor WHERE actor_name = 'Chris Hemsworth') ,(SELECT film_id FROM film WHERE film_name = 'Thor')); INSERT INTO movie VALUES ( movie_s.NEXTVAL ,(SELECT actor_id FROM actor WHERE actor_name = 'Chris Hemsworth') ,(SELECT film_id FROM film WHERE film_name = 'Thor: The Dark World')); INSERT INTO movie VALUES ( movie_s.NEXTVAL ,(SELECT actor_id FROM actor WHERE actor_name = 'Chris Pine') ,(SELECT film_id FROM film WHERE film_name = 'Star Trek')); INSERT INTO movie VALUES ( movie_s.NEXTVAL ,(SELECT actor_id FROM actor WHERE actor_name = 'Chris Pine') ,(SELECT film_id FROM film WHERE film_name = 'Star Trek into Darkness')); INSERT INTO movie VALUES ( movie_s.NEXTVAL ,(SELECT actor_id FROM actor WHERE actor_name = 'Chris Pratt') ,(SELECT film_id FROM film WHERE film_name = 'Guardians of the Galaxy')); -- Quit session. QUIT; |
The tables.sql file, lets you verify the creation of the actor, film, and movie tables:
1 2 3 4 5 6 7 8 9 | -- Set Oracle column width. COL table_name FORMAT A30 HEADING "Table Name" -- Query the tables. SELECT table_name FROM user_tables; -- Exit SQL*Plus. QUIT; |
The results.sql file, lets you see join results from actor, film, and movie tables:
1 2 3 4 5 6 7 8 9 10 11 | -- Format query. COL film_actors FORMAT A40 HEADING "Actors in Films" -- Diagnostic query. SELECT a.actor_name || ', ' || f.film_name AS film_actors FROM actor a INNER JOIN movie m ON a.actor_id = m.actor_id INNER JOIN film f ON m.film_id = f.film_id; -- Quit the session. QUIT; |
If you don’t have a sample test schema to use to test this script, you can create a sample schema with the following create_user.sql file. The file depends on the existence of a users and temp tablespace.
Click the link below to see the source code for a script that let’s you create a sample user account as the system user:
Create sample User SQL File ↓
You can use the dbms_metadata.get_ddl function to discover the existence of the tablespaces. The following SQL syntax returns the SQL DDL statement that created a users or temp tablespace:
1 2 | SET LONG 200000 SELECT dbms_metadata.get_ddl('TABLESPACE','USERS') FROM dual; |
You create the sample database with the following SQL statements:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | -- Drop the sample user table. BEGIN FOR i IN (SELECT username FROM dba_users WHERE username = 'SAMPLE') LOOP EXECUTE IMMEDIATE 'DROP USER ' || i.username || ' CASCADE'; END LOOP; END; / -- Create the sample user table. CREATE USER sample IDENTIFIED BY sample DEFAULT TABLESPACE users TEMPORARY TABLESPACE temp QUOTA 50M ON users; -- Grant privileges to sample user. GRANT CREATE CLUSTER, CREATE INDEXTYPE, CREATE OPERATOR , CREATE PROCEDURE, CREATE SEQUENCE, CREATE SESSION , CREATE TABLE, CREATE TRIGGER, CREATE TYPE , CREATE VIEW TO sample; |
The following list_oracle.sh shell script expects to receive the username, password, and fully qualified path in that specific order. The script names are entered manually in the array because this should be a unit test script.
This is an insecure version of the list_oracle.sh script because you provide the password on the command line. It’s better to provide the password as you run the script.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 | #!/usr/bin/bash # Assign user and password username="${1}" password="${2}" directory="${3}" echo "User name:" ${username} echo "Password: " ${password} echo "Directory:" ${directory} # Define an array. declare -a cmd # Assign elements to an array. cmd[0]="actor.sql" cmd[1]="film.sql" cmd[2]="movie.sql" # Call the array elements. for i in ${cmd[*]}; do sqlplus -s ${username}/${password} @${directory}/${i} > /dev/null done # Connect and pipe the query result minus errors and warnings to the while loop. sqlplus -s ${username}/${password} @${directory}/tables.sql 2>/dev/null | # Read through the piped result until it's empty. while IFS='\n' read actor_name; do echo $actor_name done # Connect and pipe the query result minus errors and warnings to the while loop. sqlplus -s ${username}/${password} @${directory}/result.sql 2>/dev/null | # Read through the piped result until it's empty. while IFS='\n' read actor_name; do echo $actor_name done |
The IFS (Internal Field Separator) works with whitespace by default. The IFS on lines 29 and 37 sets the IFS to a line return ('\n'). That’s the trick to display the data, and you can read more about the IFS in this question and answer post.
You can run the shell script with the following syntax:
./list_oracle.sh sample sample /home/student/Code/bash/oracle > output.txt |
You can then display the results from the output.txt file with the following command:
cat output.txt command: |
It will display the following output:
User name: sample Password: sample Directory: /home/student/Code/bash/oracle Table Name ------------------------------ MOVIE FILM ACTOR Actors in Films ---------------------------------------- Chris Hemsworth, Thor Chris Hemsworth, Thor: The Dark World Chris Pine, Star Trek Chris Pine, Star Trek into Darkness Chris Pratt, Guardians of the Galaxy |
As always, I hope this helps those looking for a solution.
Leaf node queries
A reader posted A dynamic level limiting hierarchical query about Oracle’s hierarchical queries. They wanted to know how to capture only the hierarchy above the level where the first leaf node occurs. They gave me the following hierarchy map as an example:
1 2
+-------------+ +-----------+
| | | |
3 5 4 6
+---------+ +-----------+ +-----+ +------+
| | | | | | | |
7 9 11 13 8 10 12 14
+-----+ +-----+ +--+ +-------+ +-----+
| | | | | | | | |
15 17 19 21 23 27 29 16 18
+---+
|
20 |
You can find the node values and hierarchical level with the following query:
SELECT tt.child_id , LEVEL FROM test_temp tt WHERE CONNECT_BY_ISLEAF = 1 START WITH tt.parent_id IS NULL CONNECT BY PRIOR tt.child_id = tt.parent_id ORDER BY 2; |
We really don’t need the node values to solve the problem. We only need the lowest LEVEL value returned by the query, which is 3. The combination of the MIN and CONNECT_BY_ISLEAF functions let us solve this problem without writing a PL/SQL solution. The subquery returns the lowest level value, which is the first level where a leaf node occurs.
SELECT LPAD(' ', 2*(LEVEL - 1)) || tt.child_id AS child_id FROM test_temp tt WHERE LEVEL <= (SELECT MIN(LEVEL) FROM test_temp tt WHERE CONNECT_BY_ISLEAF = 1 START WITH tt.parent_id IS NULL CONNECT BY PRIOR tt.child_id = tt.parent_id) START WITH tt.parent_id IS NULL CONNECT BY PRIOR tt.child_id = tt.parent_id; |
It returns:
1 2
+-------------+ +-----------+
| | | |
3 5 4 6
+---------+ +-----------+ +-----+ +------+
| | | | | | | |
7 9 11 13 8 10 12 14 |
While I answered the question in a comment originally, it seemed an important trick that should be shared in its own post.
SQL Developer – Fedora
This is the continuation of my efforts to stage an awesome Fedora developer’s instance. It shows you how to install Java 1.8 software development kit, which is nice to have. Though you can’t use Java 1.8 officially with Oracle SQL Developer 4.0.3 it is required for Oracle SQL Developer 4.1. Fortunately, the Oracle Product Manager, Jeff Smith has advised us that you can use Java 1.8 JDK with Oracle SQL Developer 4.0.3, and he’s written a comment to the blog post that it runs better with the Java 1.8 SDK.
After you install Oracle SQL Developer 4.0.3 or Oracle SQL Developer 4.1, you can watch Jeff Smith’s YouTube Video on SQL Developer 3.1 to learn how to use the basics of SQL Developer. I couldn’t find an updated version of the video for SQL Developer 4 but I didn’t try too hard.
You use yum as the root user to install Java SDK 1.8, much like my earlier Installing the Java SDK 1.7 and Java-MySQL Sample Program. The following command installs Java 8:
yum install -y java-1.8* |
It produces the following output:
Loaded plugins: langpacks, refresh-packagekit
fedora/20/x86_64/metalink | 18 kB 00:00
mysql-connectors-community | 2.5 kB 00:00
mysql-tools-community | 2.5 kB 00:00
mysql56-community | 2.5 kB 00:00
pgdg93 | 3.6 kB 00:00
updates/20/x86_64/metalink | 16 kB 00:00
updates | 4.9 kB 00:00
(1/2): mysql-tools-community/20/x86_64/primary_db | 21 kB 00:00
(2/2): updates/20/x86_64/primary_db | 13 MB 00:09
updates/20/x86_64/pkgtags
updates
(1/2): updates/20/x86_64/pkgtags | 1.4 MB 00:02
(2/2): updates/20/x86_64/updateinfo | 1.9 MB 00:04
Package 1:java-1.8.0-openjdk-headless-1.8.0.31-1.b13.fc20.x86_64 already installed and latest version
Package 1:java-1.8.0-openjdk-javadoc-1.8.0.31-1.b13.fc20.noarch already installed and latest version
Resolving Dependencies
--> Running transaction check
---> Package java-1.8.0-openjdk.x86_64 1:1.8.0.31-1.b13.fc20 will be installed
---> Package java-1.8.0-openjdk-accessibility.x86_64 1:1.8.0.31-1.b13.fc20 will be installed
---> Package java-1.8.0-openjdk-demo.x86_64 1:1.8.0.31-1.b13.fc20 will be installed
---> Package java-1.8.0-openjdk-devel.x86_64 1:1.8.0.31-1.b13.fc20 will be installed
---> Package java-1.8.0-openjdk-src.x86_64 1:1.8.0.31-1.b13.fc20 will be installed
--> Finished Dependency Resolution
Dependencies Resolved
================================================================================
Package Arch Version Repository
Size
================================================================================
Installing:
java-1.8.0-openjdk x86_64 1:1.8.0.31-1.b13.fc20 updates 201 k
java-1.8.0-openjdk-accessibility x86_64 1:1.8.0.31-1.b13.fc20 updates 12 k
java-1.8.0-openjdk-demo x86_64 1:1.8.0.31-1.b13.fc20 updates 1.9 M
java-1.8.0-openjdk-devel x86_64 1:1.8.0.31-1.b13.fc20 updates 9.2 M
java-1.8.0-openjdk-src x86_64 1:1.8.0.31-1.b13.fc20 updates 45 M
Transaction Summary
================================================================================
Install 5 Packages
Total download size: 56 M
Installed size: 92 M
Downloading packages:
(1/5): java-1.8.0-openjdk-accessibility-1.8.0.31-1.b13.fc20 | 12 kB 00:00
(2/5): java-1.8.0-openjdk-1.8.0.31-1.b13.fc20.x86_64.rpm | 201 kB 00:02
(3/5): java-1.8.0-openjdk-demo-1.8.0.31-1.b13.fc20.x86_64.r | 1.9 MB 00:03
(4/5): java-1.8.0-openjdk-devel-1.8.0.31-1.b13.fc20.x86_64. | 9.2 MB 00:07
(5/5): java-1.8.0-openjdk-src-1.8.0.31-1.b13.fc20.x86_64.rp | 45 MB 05:05
--------------------------------------------------------------------------------
Total 187 kB/s | 56 MB 05:05
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction (shutdown inhibited)
Installing : 1:java-1.8.0-openjdk-1.8.0.31-1.b13.fc20.x86_64 1/5
Installing : 1:java-1.8.0-openjdk-devel-1.8.0.31-1.b13.fc20.x86_64 2/5
Installing : 1:java-1.8.0-openjdk-demo-1.8.0.31-1.b13.fc20.x86_64 3/5
Installing : 1:java-1.8.0-openjdk-accessibility-1.8.0.31-1.b13.fc20.x86 4/5
Installing : 1:java-1.8.0-openjdk-src-1.8.0.31-1.b13.fc20.x86_64 5/5
Verifying : 1:java-1.8.0-openjdk-devel-1.8.0.31-1.b13.fc20.x86_64 1/5
Verifying : 1:java-1.8.0-openjdk-demo-1.8.0.31-1.b13.fc20.x86_64 2/5
Verifying : 1:java-1.8.0-openjdk-1.8.0.31-1.b13.fc20.x86_64 3/5
Verifying : 1:java-1.8.0-openjdk-accessibility-1.8.0.31-1.b13.fc20.x86 4/5
Verifying : 1:java-1.8.0-openjdk-src-1.8.0.31-1.b13.fc20.x86_64 5/5
Installed:
java-1.8.0-openjdk.x86_64 1:1.8.0.31-1.b13.fc20
java-1.8.0-openjdk-accessibility.x86_64 1:1.8.0.31-1.b13.fc20
java-1.8.0-openjdk-demo.x86_64 1:1.8.0.31-1.b13.fc20
java-1.8.0-openjdk-devel.x86_64 1:1.8.0.31-1.b13.fc20
java-1.8.0-openjdk-src.x86_64 1:1.8.0.31-1.b13.fc20
Complete! |
Then, you go to Oracle’s SQL Developer 4.0.3 web page or Oracle’s Beta SQL Developer 4.1 web page and download the SQL Developer RPM. At the time of writing, you download the following SQL Developer 4.0.3 RPM:
sqldeveloper-4.0.3.16.84-1.noarch.rpm |
Assuming you download the sqldeveloper-4.0.3.16.84-1.noarch.rpm file to the student user’s account. It will download into the /home/student/Downloads directory. You run the SQL Developer RPM file with the following syntax as the root user:
rpm -Uhv /home/student/Downloads/sqldeveloper-4.0.3.16.84-1.noarch.rpm |
Running the SQL Developer RPM produces the following output:
Preparing... ################################# [100%] Updating / installing... 1:sqldeveloper-4.0.3.16.84-1 ################################# [100%] |
You can now run the sqldeveloper.sh file as the root user with the following syntax:
/opt/sqldeveloper/sqldeveloper.sh |
At this point, it’s important to note that my download from the Oracle SQL Developer 4.1 page turned out to be SQL Developer 4.0.3. It prompts you for the correct Java JDK, as shown below. You may opt to enter the path to the Java JDK 1.8 for SQL Developer 4.1 because until today you downloaded the Oracle SQL Developer 4.0.3 version from the Oracle SQL Developer 4.1 page. Naturally, the Oracle SQL Developer 4.1 instructions say to use the Java 1.8 JDK on the RPM for Linux Installation Notes web page, as shown below:
If you assume from the instructions on the Oracle instruction page above that Oracle SQL Developer 4.0.3 and Oracle SQL Developer 4.1 support Java 1.8 JDK, you may enter the location for the Java JDK 1.8 when prompted. Jeff Smith, the Product Manager wrote this blog post on Oracle SQL Developer 4: Windows and the JDK. Unfortunately, you’ll see the following message if you attempt to run Oracle SQL Developer 4.0.3 with the Java 1.8 SDK at the command-line:
Oracle SQL Developer Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved. Type the full pathname of a JDK installation (or Ctrl-C to quit), the path will be stored in /root/.sqldeveloper/4.0.0/product.conf /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.31.x86_64 OpenJDK 64-Bit Server VM warning: ignoring option MaxPermSize=256M; support was removed in 8.0 |
It also raises the following error message dialog:
Text version of Unsupported JDK Version error message:
Running this product is supported with a minimum Java version of 1.7.0_51 and a maximum version less than 1.8.
Update the SetJavaHome in “/root/.sqldeveloper/4.0.0/product.conf” to point to another Java.
This produce will not be supported, and may not run correctly if you proceed. Continue anyway?
The error dialog message tells us that the instructions on the RPM for Linux Installation Notes web page can be misleading. You really need to use the Java JDK 1.7 to be supported officially, but you can safely ignore the error.
If you want a certified component, leave the “Skip This Message Next Time” checkbox unchecked and click the “No” button to continue. At this point, there’s no automatic recovery. You need to open the following file:
/root/.sqldeveloper/4.0.0/product.conf |
You need to change the SetJavaHome parameter in the file to the following:
# SetJavaHome /path/jdk SetJavaHome /usr/lib/jvm/java-1.7.0-openjdk-1.7.0.79-2.5.5.0.fc20.x86_64 |
After making the change, you can re-run the sqldeveloper.sh shell as follows:
/opt/sqldeveloper/sqldeveloper.sh |
It launches the following dialog message:
The installation pauses to ask you if you want to transfer an existing SQL Developer configuration by raising the following dialog. Assuming this is a new installation, the installer won’t find a prior configuration file. You need to click the “No” button to proceed.
The installation continues and launches SQL Developer. The first time launch shows you the following Oracle Usage Tracking dialog. If you don’t want your use monitored, uncheck the “Allow automated usage reporting to Oracle” checkbox. Click the “OK” button to continue.
After dismissing the Oracle Usage Tracking dialog, you see the SQL Developer environment:
After installing SQL Developer in the root account, you can install it as the student user. You use this command as the student user:
/opt/sqldeveloper/sqldeveloper.sh |
It returns the following error because it’s the second installation and SQL Developer doesn’t prompt you to configure the user’s product.conf file with the working JDK location:
Oracle SQL Developer Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved. Type the full pathname of a JDK installation (or Ctrl-C to quit), the path will be stored in /home/student/.sqldeveloper/4.0.0/product.conf Error: Unable to get APP_JAVA_HOME input from stdin after 10 tries |
You need to edit the /home/student/.sqldeveloper/4.0.0/product.conf file, and add the following line to the file:
# SetJavaHome /path/jdk SetJavaHome /usr/lib/jvm/java-1.7.0-openjdk-1.7.0.79-2.5.5.0.fc20.x86_64 |
Now, you can launch SQL Developer with the following command:
/opt/sqldeveloper/sqldeveloper.sh |
Alternatively, you can add the following alias to the student user’s .bashrc file:
# Set alias for SQL Developer tool. alias sqldeveloper="/opt/sqldeveloper/sqldeveloper.sh" |
You can now launch the SQL Developer tool, like this as the student user:
sqldeveloper |
You see the following when SQL Developer launches:
As always, I hope this helps those trying to sort out installing SQL Developer on a Fedora server.
Oracle Cleanup a Schema
Back in January 2014, I wrote a script to cleanup an Oracle student schema. It worked well until I started using APEX 4 in my student schema. You create the following 75 objects when you create an APEX 4 schema.
OBJECT TYPE TOTAL ------------ ------- TABLE 17 INDEX 28 SEQUENCE 5 TRIGGER 14 LOB 9 FUNCTION 2 |
Here’s the modified script that ignores the objects created automatically by Oracle APEX when you create a student workspace:
BEGIN FOR i IN (SELECT object_name , object_type , last_ddl_time FROM user_objects WHERE object_name NOT IN ('APEX$_WS_WEBPG_SECTION_HISTORY','APEX$_WS_WEBPG_SECTIONS_T1' ,'APEX$_WS_WEBPG_SECTIONS_PK','APEX$_WS_WEBPG_SECTIONS' ,'APEX$_WS_WEBPG_SECHIST_IDX1','APEX$_WS_TAGS_T1' ,'APEX$_WS_TAGS_PK','APEX$_WS_TAGS_IDX2','APEX$_WS_TAGS_IDX1' ,'APEX$_WS_TAGS','APEX$_WS_ROWS_T1','APEX$_WS_ROWS_PK' ,'APEX$_WS_ROWS_IDX','APEX$_WS_ROWS','APEX$_WS_NOTES_T1' ,'APEX$_WS_NOTES_PK','APEX$_WS_NOTES_IDX2','APEX$_WS_NOTES_IDX1' ,'APEX$_WS_NOTES','APEX$_WS_LINKS_T1','APEX$_WS_LINKS_PK' ,'APEX$_WS_LINKS_IDX2','APEX$_WS_LINKS_IDX1','APEX$_WS_LINKS' ,'APEX$_WS_HISTORY_IDX','APEX$_WS_HISTORY','APEX$_WS_FILES_T1' ,'APEX$_WS_FILES_PK','APEX$_WS_FILES_IDX2','APEX$_WS_FILES_IDX1' ,'APEX$_WS_FILES','APEX$_ACL_T1','APEX$_ACL_PK','APEX$_ACL_IDX1' ,'APEX$_ACL','CUSTOM_AUTH','CUSTOM_HASH','DEPT','EMP' ,'UPDATE_ORDER_TOTAL') AND NOT ((object_name LIKE 'DEMO%' OR object_name LIKE 'INSERT_DEMO%' OR object_name LIKE 'BI_DEMO%') AND object_type IN ('TABLE','INDEX','SEQUENCE','TRIGGER')) AND NOT (object_name LIKE 'SYS_LOB%' AND object_type = 'LOB') AND NOT (object_name LIKE 'SYS_C%' AND object_type = 'INDEX') ORDER BY object_type DESC) LOOP /* Drop types in descending order. */ IF i.object_type = 'TYPE' THEN /* Drop type and force operation because dependencies may exist. Oracle 12c also fails to remove object types with dependents in pluggable databases (at least in release 12.1). Type evolution works in container database schemas. */ EXECUTE IMMEDIATE 'DROP '||i.object_type||' '||i.object_name||' FORCE'; /* Drop table tables in descending order. */ ELSIF i.object_type = 'TABLE' THEN /* Drop table with cascading constraints to ensure foreign key constraints don't prevent the action. */ EXECUTE IMMEDIATE 'DROP '||i.object_type||' '||i.object_name||' CASCADE CONSTRAINTS PURGE'; /* Oracle 12c ONLY: Purge the recyclebin to dispose of system-generated sequence values because dropping the table doesn't automatically remove them from the active session. CRITICAL: Remark out the following when working in Oracle Database 11g. */ EXECUTE IMMEDIATE 'PURGE RECYCLEBIN'; ELSIF i.object_type = 'LOB' OR i.object_type = 'INDEX' THEN /* A system generated LOB column or INDEX will cause a failure in a generic drop of a table because it is listed in the cursor but removed by the drop of its table. This NULL block ensures there is no attempt to drop an implicit LOB data type or index because the dropping the table takes care of it. */ NULL; ELSE dbms_output.put_line('DROP '||i.object_type||' '||i.object_name||';'); /* Drop any other objects, like sequences, functions, procedures, and packages. */ EXECUTE IMMEDIATE 'DROP '||i.object_type||' '||i.object_name; END IF; END LOOP; END; / |
As always, I hope this helps others.
OUT Parameter Trick
Raja asked a question but unfortunately, I was buried in the final aspects of the write of the new Oracle Database 12c PL/SQL Programming book. He wanted to know how to pass an object type as an OUT-only mode parameter from a procedure.
That’s a great question, and it’s actually simple once you understand the difference between Oracle object types and other data types. Oracle object types must always be initiated before you use them, which means you must initialize any OUT-only mode parameters at the top of your execution section, like this:
1 2 3 4 5 6 7 8 9 10 11 12 | CREATE OR REPLACE PROCEDURE reset_troll ( pv_troll OUT TROLL_OBJECT ) IS /* Troll default name. */ lv_troll_name VARCHAR2(20) := 'Bert'; BEGIN /* Initialize the incoming parameter by allocating memory to it. */ pv_troll := troll_object(); /* Set the name to something other than the 'Tom' default value. */ pv_troll.set_troll(lv_troll_name); END reset_troll; / |
Line 7 shows you the trick, initialize the incoming parameter because there isn’t an incoming parameter for an OUT-only mode parameter. The calling parameter to an OUT-only mode parameter is only a reference where PL/SQL will copy the internal object reference. While the calling parameter has been initialized, the reference to the call parameter’s object is where the internal object will be copied. The local program must first ensure a new memory location for a new instance of the object type before it can act on or return an object instance to the external reference. More or less, the internal object is copied to the calling object instance’s memory location when the procedure completes its execution.
Here’s the source code for the troll_object object type and body:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 | CREATE OR REPLACE TYPE troll_object IS OBJECT ( troll VARCHAR2(20) , CONSTRUCTOR FUNCTION troll_object RETURN SELF AS RESULT , CONSTRUCTOR FUNCTION troll_object ( troll VARCHAR2 ) RETURN SELF AS RESULT , MEMBER FUNCTION get_troll RETURN VARCHAR2 , MEMBER PROCEDURE set_troll (troll VARCHAR2) , MEMBER FUNCTION to_string RETURN VARCHAR2) INSTANTIABLE NOT FINAL; / CREATE OR REPLACE TYPE BODY troll_object IS /* Default no-argument constructor. */ CONSTRUCTOR FUNCTION troll_object RETURN SELF AS RESULT IS troll TROLL_OBJECT := troll_object('Tom'); BEGIN SELF := troll; RETURN; END troll_object; /* Single argument constructor. */ CONSTRUCTOR FUNCTION troll_object (troll VARCHAR2) RETURN SELF AS RESULT IS BEGIN SELF.troll := troll; RETURN; END troll_object; /* A getter function. */ MEMBER FUNCTION get_troll RETURN VARCHAR2 IS BEGIN RETURN SELF.troll; END get_troll; /* A setter procedure. */ MEMBER PROCEDURE set_troll (troll VARCHAR2) IS BEGIN SELF.troll := troll; END set_troll; /* A function that returns the formatted object type's contents. */ MEMBER FUNCTION to_string RETURN VARCHAR2 IS BEGIN RETURN 'Hello '||SELF.troll; END to_string; END; / |
You can test the reset_troll procedure with the following anonymous block:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | /* Enable printing from a PL/SQL block. */ SET SERVEROUTPUT ON SIZE UNLIMITED /* Anonymous testing block. */ DECLARE lv_troll TROLL_OBJECT := troll_object('Bill'); BEGIN dbms_output.put_line('--------------------'); /* Prints 'Hello William' */ dbms_output.put_line(lv_troll.to_string()); dbms_output.put_line('--------------------'); reset_troll(lv_troll); /* Prints 'Hello Bert' */ dbms_output.put_line(lv_troll.to_string()); dbms_output.put_line('--------------------'); END; / |
If you remark out line 7 from the reset_troll procedure, you’d raise the following exception by the call on line 10 because the local object hasn’t been instantiated (given life). It means there’s no memory location allocated for the instantiated (instance of an object type).
-------------------- Hello Bill -------------------- DECLARE * ERROR at line 1: ORA-30625: method dispatch ON NULL SELF argument IS disallowed ORA-06512: at "VIDEO.RESET_TROLL", line 10 ORA-06512: at line 8 |
Hope this helps those trying to solve the same problem.
Subscribe via RSS