Archive for the ‘Oracle Developer’ tag
PostgreSQL Identity Columns
It’s interesting to see the way different databases implement automatic numbering. Oracle Database 12c is the closest to PostgreSQL in some significant ways. However, its probably more accurate to say Oracle Database 12c copied PostgreSQL’s implementation. At least, that’s my conjecture because Oracle added a way to reset the START WITH value of the indirect sequence. However, I prefer the MySQL approach because the automatic numbering sequence is a property of the table and a simple clause of the CREATE TABLE statement.
Both PostgreSQL and Oracle Database 12c implement automatic numbering as indirect sequences. Indirect sequences are those created by a table when you designate a column as an identity column in Oracle or as a serial column in PostgreSQL. The difference is that PostgreSQL doesn’t provide a syntax version inside the CREATE TABLE semantic.
MySQL provides such syntax. You set an auto numbering column in MySQL by appending the AUTO_INCREMENT clause to the table creation statement when you want it to start with a number other than 1, like this:
CREATE TABLE auto ( id INT UNSIGNED PRIMARY KEY AUTO_INCREMENT , text_field VARCHAR(30) NOT NULL ) ENGINE=InnoDB AUTO_INCREMENT=1001 DEFAULT CHARSET=utf8; |
Oracle disallows you to changing a sequence created as a background activity of the CREATE TABLE statement; and Oracle disallows you dropping an indirect sequence without changing the table that created it, which is exactly how they handle indexes created for unique constraints. Unfortunately, Oracle also disallows altering the START WITH value of any sequence.
If you want to change the START WITH value on an Oracle Database 12c indirect sequence, you must export the table, drop the table, and recreate the table with a new START WITH value before importing the data back into the table. The syntax for setting an IDENTITY column value higher than 1 is:
CREATE TABLE auto ( auto_id NUMBER GENERATED BY DEFAULT AS IDENTITY (START WITH 1001) , text_field VARCHAR2(30) , CONSTRAINT auto_pk PRIMARY KEY (auto_id)); |
You can only create a PostgreSQL table with automatic numbering by using the SERIAL data type, which always sets the initial value to 1. You can reset the SERIAL sequence value in PostgreSQL with the ALTER statement. Unlike Oracle Database 12c, PostgreSQL does let you modify the START WITH value of any sequence. The trick is understanding how to find the sequence name. The name is always the combination of the table name, an underscore, an id string, an underscore, and a seq string. This behavior makes a great case for choosing id as the name of any auto numbering columns in a table.
CREATE TABLE auto ( id SERIAL CONSTRAINT auto_pk PRIMARY KEY , text_field VARCHAR(30)); ALTER SEQUENCE auto_id_seq RESTART WITH 1001; |
You can see the table and assigned sequence with the following command in PostgreSQL:
\d+ auto |
It should display:
Table "public.auto"
Column | Type | Modifiers | Storage | Stats target | Description
------------+-----------------------+---------------------------------------------------+----------+--------------+-------------
id | integer | not null default nextval('auto_id_seq'::regclass) | plain | |
text_field | character varying(30) | | extended | |
Indexes:
"auto_pk" PRIMARY KEY, btree (id)
Has OIDs: no |
As always, I hope this helps those trying to sort through how to start identity columns above the initial value of 1.
Upgrade APEX 4 to 5.1
This blog post shows you how to upgrade APEX Version 4.0.2 on a default Oracle Database 11g XE instance to APEX 5.1.2. Oracle’s APEX t upgrade document was just a bit too short and missed some details. It divided into five parts. The first part confirms your starting point. The second part downloads and positions the extracted software. The third part installs APEX 5. The fourth part configures APEX 5. The fifth part shows you how to access and get to the APEX 5 home page.
Confirm APEX 4 Installation
- Verify the database version by connecting as the
systemuser through SQL*Plus and running the following query:SELECT banner FROM v$version WHERE banner LIKE 'Oracle Database%';
It should return the following when you’re upgrading the Oracle Database 11g XE:
BANNER -------------------------------------------------------------------------------- Oracle Database 11g Express Edition Release 11.2.0.2.0 - 64bit Production
- Verify the APEX version by connecting as the
systemuser through SQL*Plus and running the following query:COLUMN version_no FORMAT A16 HEADING "Version Number" COLUMN api_compatibility FORMAT A16 HEADING "API|Compatibility" COLUMN patch_applied FORMAT A14 HEADING "Patch Applied" SELECT * FROM apex_release;
It should return the following when you start with the base Oracle Database 11g XE:
API Version Number Compatibility Patch Applied ---------------- ---------------- -------------- 4.0.2.00.09 2010.05.13
- Verify the XML Database version by connecting as the
systemuser through SQL*Plus and running the following query:COLUMN comp_name FORMAT A20 COLUMN version FORMAT A12 COLUMN status FORMAT A8 SELECT comp_name , version , status FROM dba_registry WHERE comp_id = 'XDB';
It should return the following when you’re upgrading the Oracle Database 11g XE:
COMP_NAME VERSION STATUS -------------------- ------------ -------- Oracle XML Database 11.2.0.2.0 VALID
- Verify the
memory_targetof the instance by connecting as the system user through SQL*Plus and running the following query. It should be no smaller than 300 MB.show parameter memory_target
It should return the following when you’re upgrading the Oracle Database 11g XE:
NAME TYPE VALUE ------------------------------------ ----------- ------------------------------ memory_target big integer 1G
- Oracle’s instructions qualify that APEX 5 will install into the
APEX_050000schema. Oracle creates the newAPEX_050000schema with a default of thesysauxandtemptable space. You can verify these as the system user through SQL*Plus by running the following two queries. The first one checks for the tablespaces and the second for available space and auto extensibility.COLUMN default_tablespace FORMAT A22 COLUMN temporary_tablespace FORMAT A22 SELECT default_tablespace , temporary_tablespace FROM dba_users WHERE username = 'APEX_040000';
It should return the following when you’re upgrading the Oracle Database 11g XE:
DEFAULT_TABLESPACE TEMPORARY_TABLESPACE ---------------------- ---------------------- SYSAUX TEMP
COLUMN tablespace_name FORMAT A10 HEADING "Tablespace Name" COLUMN file_name FORMAT A38 HEADING "File Name" COLUMN available_space FORMAT 999,999.00 HEADING "Available|Space MB" COLUMN autoextensible FORMAT A10 HEADING "Auto|Extensible" SELECT tablespace_name , file_name , ((maxbytes - bytes) / 1024) / 1024 AS available_space , autoextensible FROM dba_data_files WHERE tablespace_name IN ('SYSAUX','SYSTEM');
It should return the following when you’re upgrading the Oracle Database 11g XE:
Tablespace Available Auto Name File Name Space MB Extensible ---------- -------------------------------------- ----------- ---------- SYSAUX /u01/app/oracle/oradata/XE/sysaux.dbf 32,027.98 YES SYSTEM /u01/app/oracle/oradata/XE/system.dbf 200.00 YES
Download APEX 5
- Download the APEX software from the Oracle web site. Assuming you download the software as the
studentuser, you can save it in yourDownloadsdirectory.You should open a Terminal session and connect as the
oracleuser. If you’ve setup your instance correctly, you will need to first become therootuser and then theoracleuser. As theoracleuser, you source the Oracle environment and copy theapex_x.x.x.zipfile from the~student/Downloadsdirectory to the/u01/app/oracledirectory.
- You copy the file from the
studentuser’sDownloadsdirectory with the following command:cp /home/student/Downloads/apex_x.x.x.zip /u01/app/oracle
- You unzip the copied
apex_x.x.x.zipfile (version 5.1.2 in this example) with the following command, and it will create a newapexdirectory as a subdirectory of the/u01/app/oracledirectory.unzip apex_5.1.2.zip
Install APEX 5
- You should query the
dba_usersview to check the status of theapex_public_userandanonymoususer accounts, like this:COLUMN username FORMAT A18 HEADING "User Name" COLUMN account_status FORMAT A10 HEADING "Account|Status" SELECT username , account_status FROM dba_users WHERE username IN ('APEX_PUBLIC_USER','ANONYMOUS');
It should return the following when you’re upgrading the Oracle Database 11g XE but the anonymous user name may be open if you’ve previously unlocked it:
Account User Name Status ------------------ ---------- APEX_PUBLIC_USER LOCKED ANONYMOUS LOCKED
- You can unlock the
apex_public_userandanonymousaccounts with the following statements:ALTER USER apex_public_user ACCOUNT UNLOCK; ALTER USER anonymous ACCOUNT UNLOCK;
- The installation uses the
flows_filesschema, which should be installed. You can verify the default and temporary tablespaces with the following query:COLUMN default_tablespace FORMAT A22 COLUMN temporary_tablespace FORMAT A22 SELECT default_tablespace , temporary_tablespace FROM dba_users WHERE username = 'FLOWS_FILES';
DEFAULT_TABLESPACE TEMPORARY_TABLESPACE ---------------------- ---------------------- SYSAUX TEMP
- Open a Terminal session, connect as the
oracleuser, source the Oracle environment file, and change your active directory to the/u01/app/oracle/apexdirectory, and open a SQL*Plus connection as thesysuser. You need superuser privileges, which means you need to connect to the Oracle database with the “sys as sysdba” syntax.sqlplus sys as sysdbaYou can now install APEX 5.x.x by calling the following script with four parameters:
@apexins.sql SYSAUX SYSAUX TEMP /i/
It will take a couple minutes for the installation script to succeed. You will know that it is completed when you see the following message:
Thank you for installing Oracle Application Express 5.1.2.00.09 Oracle Application Express is installed in the APEX_050100 schema. The structure of the link to the Application Express administration services is as follows: http://host:port/pls/apex/apex_admin (Oracle HTTP Server with mod_plsql) http://host:port/apex/apex_admin (Oracle XML DB HTTP listener with the embedded PL/SQL gateway) http://host:port/apex/apex_admin (Oracle REST Data Services) The structure of the link to the Application Express development interface is as follows: http://host:port/pls/apex (Oracle HTTP Server with mod_plsql) http://host:port/apex (Oracle XML DB HTTP listener with the embedded PL/SQL gateway) http://host:port/apex (Oracle REST Data Services) PL/SQL procedure successfully completed. Disconnected from Oracle Database 11g Express Edition Release 11.2.0.2.0 - 64bit Production
- After installing APEX 5, you can re-verify the APEX version by connecting as the
systemuser through SQL*Plus and running the following query:COLUMN version_no FORMAT A16 HEADING "Version Number" COLUMN api_compatibility FORMAT A16 HEADING "API|Compatibility" COLUMN patch_applied FORMAT A14 HEADING "Patch Applied" SELECT * FROM apex_release;
It should return the following after upgrading with APEX 5:
API Version Number Compatibility Patch Applied ---------------- ---------------- -------------- 5.1.2.00.09 2016.08.24 APPLIED
Configure APEX 5
- At this point, you need to set the internal password, which you can do by navigating to the
/u01/app/oracle/apexdirectory. In that directory, you should open a SQL*Plus session as thesysuser with the “sys as sysdba” privilege. Run the following script to set the APEX Administrator’s credentials:@apxchpwd.SQLIt will prompt you for parameters, like so:
================================================================================ This script can be used to change the password of an Application Express instance administrator. If the user does not yet exist, a user record will be created. ================================================================================ Enter the administrator's username [ADMIN] ADMIN User "ADMIN" exists. Enter ADMIN's email [ADMIN] mclaughlinm@byui.edu Enter ADMIN's password [] Changed password of instance administrator ADMIN.
- The next task requires you to run the
apex_epg_config.sqlscript with one directory parameter. You should be connect to thesysuser with the “sys as sysdba” privilege:@apex_epg_config.SQL /u01/app/oracle
It will take a couple minutes to complete this script, and when it is complete it displays:
. Loading images directory: /u01/app/oracle/apex/images
- While the default port for XML DB is 8080, you should confirm it with this query:
SELECT dbms_xdb.gethttpport FROM dual;
It should return the following:
GETHTTPPORT ----------- 8080
Connect to and use APEX 5
- You can type the following URL into your local browser to get to the APEX 5 Administration page:
http://localhost:8080/apex/apex_admin
It should display the following login. The password is the one you entered when you ran the
apxchpwd.sqlscript in the configuration section of this post.
After you enter proper credentials, click the Sign in to Administration button to proceed. It should display the following APEX 5 home page.
You can now work in APEX 5 Administration and setup a individual workspaces.
Cleanup APEX 4
Migrating functionality to APEX 5 is possible but reworking the existing design in the context of new features is better. After you have migrated your applications and upgraded your production instance, you can drop the APEX_040000 user/schema and remove any APEX 4 workspaces. This segment of shows you how to remove an APEX 4 workspace and drop the APEX_040000 user/schema.
- The following anonymous PL/SQL block will remove an APEX 4 workspace from a user schema. It’s designed for you to run it inside the target schema but you can change it to run it as the system user against multiple schemas.
DECLARE /* Cursor for all APEX 4 tables and sequences. */ CURSOR c IS SELECT uo.object_type , uo.object_name FROM user_objects uo WHERE uo.object_name IN ('DEPT' ,'APEX$_WS_WEBPG_SECTION_HISTORY' ,'APEX$_ACL' ,'APEX$_WS_WEBPG_SECTIONS' ,'APEX$_WS_ROWS' ,'EMP' ,'APEX$_WS_FILES' ,'APEX$_WS_TAGS' ,'APEX$_WS_LINKS' ,'APEX$_WS_NOTES' ,'DEMO_USERS' ,'DEMO_CUSTOMERS' ,'DEMO_ORDERS' ,'DEMO_PRODUCT_INFO' ,'DEMO_ORDER_ITEMS' ,'DEMO_STATES' ,'APEX$_WS_HISTORY' ,'DEMO_USERS_SEQ' ,'DEMO_PROD_SEQ' ,'DEMO_ORD_SEQ' ,'DEMO_ORDER_ITEMS_SEQ' ,'DEMO_CUST_SEQ' ,'CUSTOM_HASH' ,'CUSTOM_AUTH') ORDER BY uo.object_type DESC; BEGIN FOR i IN c LOOP IF i.object_type = 'TABLE' THEN EXECUTE IMMEDIATE 'DROP '||i.object_type||' '||i.object_name||' CASCADE CONSTRAINTS'; ELSE EXECUTE IMMEDIATE 'DROP '||i.object_type||' '||i.object_name; END IF; END LOOP; END; /
- You can then connect as the
sysuser with the “sys as sysdba” privilege and drop the APEX_040000 user/schema, like this:DROP USER apex_040000 CASCADE;
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.
Oracle SQL Strip Quotes
Somebody wanted to know how to strip double quotes from strings. Obviously, they’re playing with the DBMS_METADATA package. It’s quite simple, the TRIM function does it, like this:
SELECT TRIM(BOTH '"' FROM '"Hello World!"') AS "Message" FROM dual; |
It will print:
Hello World! |
As always, I hope this helps those looking for a solution.
Reset Oracle Password
This blog entry shows you how to reset the system password for an Oracle Database. It uses a Linux image running Oracle Database 11g Express Edition. It assumes the student user is the sudoer user.
After you sign on to the student user account, you open a Terminal session and you should see the following:
[student@localhost python]$ |
The oracle user account should be configured to prevent a login. So, you should use the su command or sudo command to open a terminal shell as the root user.
[student@localhost python]$ sudo sh [sudo] password for student: |
As the root user, you can login as the oracle user with the following command:
su - oracle |
and, you should see the following prompt. You can see the present working directory (pwd) with the pwd command:
-bash-4.2$ pwd /u01/app/oracle |
You need to source the oracle_env.sh shell file created by the installation of the Oracle Database during the installation. You have two approaches to source the environment file, the first approach is with a dot (.), like
. /u01/app/oracle/product/11.2.0/xe/bin/oracle_env.sh |
or, this
source /u01/app/oracle/product/11.2.0/xe/bin/oracle_env.sh |
The oracle_env.sh file contains the following:
export ORACLE_HOME=/u01/app/oracle/product/11.2.0/xe export ORACLE_SID=XE export NLS_LANG=`$ORACLE_HOME/bin/nls_lang.sh` export PATH=$ORACLE_HOME/bin:$PATH |
Now, you can connect to the Oracle Database as the internal user with the following command:
sqlplus / as sysdba |
Once connected as the internal user, you can reset the system user’s password to “cangetin” with this command:
ALTER USER system IDENTIFIED BY cangetin; |
At this point, you can also stop and start the database. You stop the database with this command:
shutdown immediate |
You can then start the database with this command:
startup |
After setting the system user password, sign out of SQL*Plus. Then, you can type two exits to return to the student user account, like this:
-bash-4.2$ exit logout sh-4.2# exit exit [student@localhost python]$ |
As always, I hope this helps those who need to reset the system password when they don’t know what it was to begin with.
Oracle Diagnostic Queries
It’s always a challenge when you want to build your own Oracle SQL Tools. I was asked how you could synchronize multiple cursors into a single source. The answer is quite simple, you write an Oracle object type to represent a record structure, an Oracle list of the record structure, and a stored function to return the list of the record structure.
For this example, you create the following table_struct object type and a table_list collection type:
/* Drop the types from most to least dependent. */ DROP TYPE table_list; DROP TYPE table_struct; /* Create the record type structure. */ CREATE OR REPLACE TYPE table_struct IS OBJECT ( table_name VARCHAR2(30) , column_cnt NUMBER , row_cnt NUMBER ); / /* Create the collection of a record type structure. */ CREATE OR REPLACE TYPE table_list IS TABLE OF table_struct; / |
The following listing function now reads all table names from the user_tables view. A subordinate cursor reads the user_tab_columns view for the number of columns in a table. A Native Dynamic SQL (NDS) cursor counts the number of rows in each tables found in the .
/* Create the listing function. */
CREATE OR REPLACE
FUNCTION listing RETURN table_list IS
/* Variable list. */
lv_column_cnt NUMBER;
lv_row_cnt NUMBER;
/* Declare a statement variable. */
stmt VARCHAR2(200);
/* Declare a system reference cursor variable. */
lv_refcursor SYS_REFCURSOR;
lv_table_cnt NUMBER;
/* Declare an output variable. */
lv_list TABLE_LIST := table_list();
/* Declare a table list cursor that excludes APEX tables. */
CURSOR c IS
SELECT table_name
FROM user_tables
WHERE table_name NOT IN
('DEPT','EMP','APEX$_ACL','APEX$_WS_WEBPG_SECTIONS','APEX$_WS_ROWS'
,'APEX$_WS_HISTORY','APEX$_WS_NOTES','APEX$_WS_LINKS'
,'APEX$_WS_TAGS','APEX$_WS_FILES','APEX$_WS_WEBPG_SECTION_HISTORY'
,'DEMO_USERS','DEMO_CUSTOMERS','DEMO_ORDERS','DEMO_PRODUCT_INFO'
,'DEMO_ORDER_ITEMS','DEMO_STATES');
/* Declare a column count. */
CURSOR cnt
( cv_table_name VARCHAR2 ) IS
SELECT table_name
, COUNT(column_id) AS cnt_columns
FROM user_tab_columns
WHERE table_name = cv_table_name
GROUP BY table_name;
BEGIN
/* Read through the data set of non-environment variables. */
FOR i IN c LOOP
/* Count the columns of a table. */
FOR j IN cnt(i.table_name) LOOP
lv_column_cnt := j.cnt_columns;
END LOOP;
/* Declare a statement. */
stmt := 'SELECT COUNT(*) AS column_cnt FROM '||i.table_name;
/* Open the cursor and write set to collection. */
OPEN lv_refcursor FOR stmt;
LOOP
FETCH lv_refcursor INTO lv_table_cnt;
EXIT WHEN lv_refcursor%NOTFOUND;
lv_list.EXTEND;
lv_list(lv_list.COUNT) := table_struct(
table_name => i.table_name
, column_cnt => lv_column_cnt
, row_cnt => lv_table_cnt );
END LOOP;
END LOOP;
RETURN lv_list;
END;
/ |
The following query pulls the processed data set as the function’s result:
COL table_name FORMAT A20 HEADING "Table Name" COL column_cnt FORMAT 9,999 HEADING "Column #" COL row_cnt FORMAT 9,999 HEADING "Row #" SELECT table_name , column_cnt , row_cnt FROM TABLE(listing); |
It returns the following result set:
Table Name Column # Row # -------------------- -------- ------ SYSTEM_USER 11 5 COMMON_LOOKUP 10 49 MEMBER 9 10 CONTACT 10 18 ADDRESS 10 18 STREET_ADDRESS 8 28 TELEPHONE 11 18 RENTAL 8 4,694 ITEM 14 93 RENTAL_ITEM 9 4,703 PRICE 11 558 TRANSACTION 12 4,694 CALENDAR 9 300 AIRPORT 9 6 ACCOUNT_LIST 8 200 15 rows selected. |
As always, I hope this helps those trying to work with the Oracle database.
Install cx_Oracle for Python
This shows you how to install the cx_Oracle library for Python 2.7 on Fedora Linux. If Fedora has it on the server you can download it with the following yum command:
yum install -y cx_Oracle-5.2.1-11g-py27-1.x86_64.rpm |
Currently, you’ll get the following failure because it’s not available in the Fedora repository:
Loaded plugins: langpacks, refresh-packagekit mysql-connectors-community | 2.5 kB 00:00:00 mysql-tools-community | 2.5 kB 00:00:00 mysql56-community | 2.5 kB 00:00:00 pgdg93 | 3.6 kB 00:00:00 updates/20/x86_64/metalink | 2.3 kB 00:00:00 No package cx_Oracle-5.2.1-11g-py27-1.x86_64.rpm available. Error: Nothing to do |
You can download the cx_Oracle library from the Python web site. The cx_Oracle documentation qualifies module interfaces, objects, and connections. Assuming your Linux user’s name is student, you download the cx_Oracle library into the /home/student/Downloads directory. Then, you use the su or sudo command to become the root user.
As the root user, run the following yum command:
yum install -y ~student/Downloads/cx_Oracle-5.2.1-11g-py27-1.x86_64.rpm |
You should see the following output:
Loaded plugins: langpacks, refresh-packagekit Examining /home/student/Downloads/cx_Oracle-5.2.1-11g-py27-1.x86_64.rpm: cx_Oracle-5.2.1-1.x86_64 Marking /home/student/Downloads/cx_Oracle-5.2.1-11g-py27-1.x86_64.rpm to be installed Resolving Dependencies --> Running transaction check ---> Package cx_Oracle.x86_64 0:5.2.1-1 will be installed --> Finished Dependency Resolution Dependencies Resolved ======================================================================================= Package Arch Version Repository Size ======================================================================================= Installing: cx_Oracle x86_64 5.2.1-1 /cx_Oracle-5.2.1-11g-py27-1.x86_64 717 k Transaction Summary ======================================================================================= Install 1 Package Total size: 717 k Installed size: 717 k Downloading packages: Running transaction check Running transaction test Transaction test succeeded Running transaction (shutdown inhibited) Installing : cx_Oracle-5.2.1-1.x86_64 1/1 Verifying : cx_Oracle-5.2.1-1.x86_64 1/1 Installed: cx_Oracle.x86_64 0:5.2.1-1 Complete! |
After you install the cx_Oracle-5.2.1-1.x86_64 package, you can find the installed files with this rpm command:
rpm -ql cx_Oracle-5.2.1-1.x86_64 |
It lists:
/usr/lib64/python2.7/site-packages/cx_Oracle-5.2.1-py2.7.egg-info /usr/lib64/python2.7/site-packages/cx_Oracle-5.2.1-py2.7.egg-info/PKG-INFO /usr/lib64/python2.7/site-packages/cx_Oracle-5.2.1-py2.7.egg-info/SOURCES.txt /usr/lib64/python2.7/site-packages/cx_Oracle-5.2.1-py2.7.egg-info/dependency_links.txt /usr/lib64/python2.7/site-packages/cx_Oracle-5.2.1-py2.7.egg-info/top_level.txt /usr/lib64/python2.7/site-packages/cx_Oracle.so /usr/share/doc/cx_Oracle-5.2.1 /usr/share/doc/cx_Oracle-5.2.1/BUILD.txt /usr/share/doc/cx_Oracle-5.2.1/README.txt /usr/share/doc/cx_Oracle-5.2.1/samples /usr/share/doc/cx_Oracle-5.2.1/samples/DatabaseChangeNotification.py /usr/share/doc/cx_Oracle-5.2.1/samples/DatabaseShutdown.py /usr/share/doc/cx_Oracle-5.2.1/samples/DatabaseStartup.py /usr/share/doc/cx_Oracle-5.2.1/samples/ReturnLongs.py /usr/share/doc/cx_Oracle-5.2.1/samples/ReturnUnicode.py /usr/share/doc/cx_Oracle-5.2.1/samples/RowsAsInstance.py /usr/share/doc/cx_Oracle-5.2.1/test /usr/share/doc/cx_Oracle-5.2.1/test/3kArrayDMLBatchError.py /usr/share/doc/cx_Oracle-5.2.1/test/3kNumberVar.py /usr/share/doc/cx_Oracle-5.2.1/test/3kStringVar.py /usr/share/doc/cx_Oracle-5.2.1/test/ArrayDMLBatchError.py /usr/share/doc/cx_Oracle-5.2.1/test/BooleanVar.py /usr/share/doc/cx_Oracle-5.2.1/test/Connection.py /usr/share/doc/cx_Oracle-5.2.1/test/Cursor.py /usr/share/doc/cx_Oracle-5.2.1/test/CursorVar.py /usr/share/doc/cx_Oracle-5.2.1/test/DateTimeVar.py /usr/share/doc/cx_Oracle-5.2.1/test/IntervalVar.py /usr/share/doc/cx_Oracle-5.2.1/test/LobVar.py /usr/share/doc/cx_Oracle-5.2.1/test/LongVar.py /usr/share/doc/cx_Oracle-5.2.1/test/NCharVar.py /usr/share/doc/cx_Oracle-5.2.1/test/NumberVar.py /usr/share/doc/cx_Oracle-5.2.1/test/ObjectVar.py /usr/share/doc/cx_Oracle-5.2.1/test/SessionPool.py /usr/share/doc/cx_Oracle-5.2.1/test/SetupTest.sql /usr/share/doc/cx_Oracle-5.2.1/test/StringVar.py /usr/share/doc/cx_Oracle-5.2.1/test/TestEnv.py /usr/share/doc/cx_Oracle-5.2.1/test/TimestampVar.py /usr/share/doc/cx_Oracle-5.2.1/test/test.py /usr/share/doc/cx_Oracle-5.2.1/test/test3k.py /usr/share/doc/cx_Oracle-5.2.1/test/test_dbapi20.py /usr/share/doc/cx_Oracle-5.2.1/test/uArrayDMLBatchError.py /usr/share/doc/cx_Oracle-5.2.1/test/uConnection.py /usr/share/doc/cx_Oracle-5.2.1/test/uCursor.py /usr/share/doc/cx_Oracle-5.2.1/test/uCursorVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uDateTimeVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uIntervalVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uLobVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uLongVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uNumberVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uObjectVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uSessionPool.py /usr/share/doc/cx_Oracle-5.2.1/test/uStringVar.py /usr/share/doc/cx_Oracle-5.2.1/test/uTimestampVar.py |
After you installed the software, you can test whether inside Python’s IDLE environment with the import command, like this:
Python 2.7.5 (default, Apr 10 2015, 08:09:05) [GCC 4.8.3 20140911 (Red Hat 4.8.3-7)] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import cx_Oracle Traceback (most recent call last): File "<stdin>", line 1, in <module> ImportError: libclntsh.so.11.1: cannot open shared object file: No such file or directory |
This error indicates that Oracle Client software isn’t installed, which is true in this case. I only installed the Oracle Database 11g Express Edition. You need to download the Oracle Client software and install it as the root user.
You download the Oracle Client software from the Oracle web site. Assuming your Linux user’s name is student, you download the cx_Oracle library into the /home/student/Downloads directory. Then, you use the su or sudo command to become the root user.
As the root user, run the following yum command:
yum install -y ~student/Downloads/oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64.rpm |
You should see the following output:
Loaded plugins: langpacks, refresh-packagekit
Examining /home/student/Downloads/oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64.rpm: oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64
Marking /home/student/Downloads/oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64.rpm to be installed
Resolving Dependencies
--> Running transaction check
---> Package oracle-instantclient11.2-basic.x86_64 0:11.2.0.4.0-1 will be installed
--> Finished Dependency Resolution
Dependencies Resolved
================================================================================
Package
Arch Version
Repository Size
================================================================================
Installing:
oracle-instantclient11.2-basic
x86_64 11.2.0.4.0-1
/oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64 179 M
Transaction Summary
================================================================================
Install 1 Package
Total size: 179 M
Installed size: 179 M
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction (shutdown inhibited)
Installing : oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64 1/1
Verifying : oracle-instantclient11.2-basic-11.2.0.4.0-1.x86_64 1/1
Installed:
oracle-instantclient11.2-basic.x86_64 0:11.2.0.4.0-1
Complete! |
You can create a Python program that checks your ability to connect to the Oracle database, like the following oracleConnection.py file:
# Import the Oracle library. import cx_Oracle try: # Create a connection. db = cx_Oracle.connect("student/student@xe") # Print a message. print "Connected to the Oracle " + db.version + " database." except cx_Oracle.DatabaseError, e: error, = e.args print >> sys.stderr, "Oracle-Error-Code:", error.code print >> sys.stderr, "Oracle-Error-Message:", error.message finally: # Close cursor. db.close() |
You can run this from the Linux command line with the following syntax:
python oracleConnection.py |
It should return the following string:
Connected to the Oracle 11.2.0.2.0 database. |
Now, you can create a Python program that reads data from the Oracle database. The following oracleString.py file reads a string literal from the pseudo table dual:
# Import the Oracle library. import cx_Oracle try: # Create a connection. db = cx_Oracle.connect("student/student@xe") # Create a cursor. cursor = db.cursor() # Execute a query. cursor.execute("SELECT 'Hello world!' FROM dual") # Read the contents of the cursor. for row in cursor: print (row[0]) except cx_Oracle.DatabaseError, e: error, = e.args print >> sys.stderr, "Oracle-Error-Code:", error.code print >> sys.stderr, "Oracle-Error-Message:", error.message finally: # Close cursor and connection. cursor.close() db.close() |
You can run this from the Linux command line with the following syntax:
python oracleString.py |
It should return the following string:
Hello world! |
Now, you can create a Python program that reads actual table data from the Oracle database (assuming you have a copy of my video store database). The following oracleTable.py file reads a string literal from the pseudo table dual:
# Import the Oracle library. import cx_Oracle try: # Create a connection. db = cx_Oracle.connect("student/student@xe") # Create a cursor. cursor = db.cursor() # Execute a query. cursor.execute("SELECT item_title, item_subtitle FROM item") # Read the contents of the cursor. for row in cursor: print (row[0], row[1]) except cx_Oracle.DatabaseError, e: error, = e.args print >> sys.stderr, "Oracle-Error-Code:", error.code print >> sys.stderr, "Oracle-Error-Message:", error.message finally: # Close cursor and connection. cursor.close() db.close() |
You can run this from the Linux command line with the following syntax:
python oracleTable.py |
It should return the following strings (only a subset of the returned values):
("Harry Potter and the Sorcer's Stone", 'Two-Disc Special Edition')
('Harry Potter and the Chamber of Secrets', 'Two-Disc Special Edition')
('Harry Potter and the Prisoner of Azkaban', 'Two-Disc Special Edition')
('Harry Potter and the Chamber of Secrets', None)
('Harry Potter and the Goblet of Fire', 'Widescreen Edition')
('Harry Potter and the Goblet of Fire', 'Two-Disc Special Edition')
('Harry Potter and the Order of the Phoenix', 'Widescreen Edition')
('The Lord of the Rings - Fellowship of the Ring', 'Widescreen Edition')
('The Lord of the Rings - Fellowship of the Ring', 'Platinum Series Special Extended Edition')
('The Lord of the Rings - Two Towers', 'Widescreen Edition')
('The Lord of the Rings - Two Towers', 'Platinum Series Special Extended Edition')
('The Lord of the Rings - The Return of the King', 'Widescreen Edition')
('The Lord of the Rings - The Return of the King', 'Platinum Series Special Extended Edition')
('Star Wars - Episode I', 'The Phantom Menace')
('Star Wars - Episode II', 'Attack of the Clones')
('Star Wars - Episode III', 'Revenge of the Sith')
('Star Wars - Episode IV', 'A New Hope')
('Star Wars - Episode V', 'The Empire Strikes Back')
('Star Wars - Episode VI', 'Return of the Jedi') |
As always, I hope this helps others who want to work with Python and the Oracle database.
Oracle Segment Fails
The instance that I’ve built for my students in a Fedora VM is quite stable except for one feature. The feature is the hibernation process of the base operating system. Sometimes when the base operating system hibernates, it causes the Oracle shared memory segment to fail. When that happens you get the following error:
ERROR: ORA-01034: ORACLE NOT available ORA-27101: shared memory realm does NOT exist Linux-x86_64 Error: 2: No such FILE OR DIRECTORY Process ID: 0 SESSION ID: 0 Serial NUMBER: 0 |
I created the master sudoer account as the student user. The oracle user is configured so that you can’t log in to the Linux OS with it. To restart the instance you can do the following in a default Oracle 11g XE installation:
su - root |
or, you can do this:
sudo sh |
Then as the root user, you can sign on to the oracle user’s account by using the su command without a password, like:
su - oracle |
As the user who installed the Oracle instance, you can connect to the database without a password after you source the environment file. The standard Oracle 11g XE environment file can be sources like this:
. /u01/app/oracle/product/11.2.0/xe/bin/oracle_env.sh |
Alternatively, for my students there is a .bashrc file that they can manually source. It contains the following:
# Source global definitions if [ -f /etc/bashrc ]; then . /etc/bashrc fi # Uncomment the following line if you don't like systemctl's auto-paging feature: # export SYSTEMD_PAGER= # User specific aliases and functions . /u01/app/oracle/product/11.2.0/xe/bin/oracle_env.sh # Wrap sqlplus with rlwrap to edit prior lines with the # up, down, left and right keys. sqlplus () { # Discover the fully qualified program name. path=`which rlwrap 2>/dev/null` file='' # Parse the program name from the path. if [ -n ${path} ]; then file=${path##/*/} fi; # Wrap when there is a file and it is rewrap. if [ -n ${file} ] && [[ ${file} = "rlwrap" ]]; then rlwrap sqlplus "${@}" else echo "Command-line history unavailable: Install the rlwrap package." $ORACLE_HOME/bin/sqlplus "${@}" fi } # Set vi as a command line editor. set -o vi |
You can source the oracle user’s .bashrc account, like this:
. .bashrc |
After you’ve sourced the environment, you can connect as the internal user with the following syntax:
sqlplus / AS sysdba |
Connected as the internal user, run these two commands in sequence:
shutdown IMMEDIATE
startup |
Then, you should be able to connect as the student user or another ordinary user with the following syntax:
CONNECT student/student |
Hope this helps my students and those who want to know how to restart the Oracle instance.
Oracle EBS 12.2 & Ruby
As does seem to occur from time-to-time, I’m out there in the weeds again and sorting out a solution that fits a customer’s need. They wanted to know if they could write Oracle EBS 12.2 Concurrent Manager Programs in Ruby. They don’t want to write Java.
I checked the documentation, which as is too common, didn’t answer the question. I’m sure if I downloaded the PDF and searched it for one of the languages I knew Oracle supported, I would have found the list of supported languages.
It was simply quicker to query the Oracle EBS 12.2 FND_LOOKUPS table like so:
SELECT lookup_type , lookup_code , SUBSTR(meaning,1,30) AS meaning FROM fnd_lookups WHERE lookup_type = 'CP_EXECUTION_METHOD_CODE' ORDER BY meaning; |
It returns the list of possible types of Oracle EBS 12.2 Concurrent Manager Programs:
LOOKUP_TYPE LOOKUP_CODE MEANING -------------------------- ------------ ------------------------------ CP_EXECUTION_METHOD_CODE X FlexRpt CP_EXECUTION_METHOD_CODE F FlexSql CP_EXECUTION_METHOD_CODE H Host CP_EXECUTION_METHOD_CODE S Immediate CP_EXECUTION_METHOD_CODE K Java Concurrent Program CP_EXECUTION_METHOD_CODE J Java Stored Procedure CP_EXECUTION_METHOD_CODE M Multi Language Function CP_EXECUTION_METHOD_CODE P Oracle Reports CP_EXECUTION_METHOD_CODE I PL/SQL Stored Procedure CP_EXECUTION_METHOD_CODE E Perl Concurrent Program CP_EXECUTION_METHOD_CODE B Request Set Stage Function CP_EXECUTION_METHOD_CODE L SQL*Loader CP_EXECUTION_METHOD_CODE Q SQL*Plus CP_EXECUTION_METHOD_CODE R SQL*Report CP_EXECUTION_METHOD_CODE Z Shutdown Callback CP_EXECUTION_METHOD_CODE A Spawned |
That gave me some of the answer. You can’t call Ruby programs directly. However, Perl lets you use Inline::Ruby. You can use Inline:Ruby to call your Ruby programs. So, if you use Perl to wrap Ruby you don’t have to use Java.
DB_LINK w/o tnsnames.ora
A question popped up, which I thought was interesting. How can you create a DB_LINK in Oracle without the DBA changing the tnsnames.ora file? It’s actually quite easy, especially if the DBA sets the TNS address name the same as the instance’s service name or in older databases SID value.
- Do the following with the
tnspingutility:tnsping mohawk
It should return this when the server’s
hostnameismohawkand domain name istechtinker.com:TNS Ping Utility for Linux: Version 11.2.0.2.0 - Production on 26-JUL-2016 16:55:58 Copyright (c) 1997, 2011, Oracle. All rights reserved. Used parameter files: Used TNSNAMES adapter to resolve the alias Attempting to contact (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = mohawk.techtinker.com)(PORT = 1521)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = ORCL))) OK (10 msec)
- You can now create a
DB_LINKin another Oracle instance without atnsnames.oraentry by referencing the type of server connection and service name with the following syntax (please note that you should remove extraneous white space):CREATE DATABASE LINK test CONNECT TO student IDENTIFIED BY student USING '(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=mohawk.techtinker.com)(PORT=1521))(CONNECT_DATA=(SERVER=DEDICATED)(SERVICE_NAME=ORCL)))'
In an older database version, you may need to refer to the
SID, like this:CREATE DATABASE LINK test CONNECT TO student IDENTIFIED BY student USING '(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=mohawk.techtinker.com)(PORT=1521))(CONNECT_DATA=(SERVER=DEDICATED)(SID=ORCL)))'
Then, you can query a contact table in the remote instance like this:
SELECT COUNT(*) FROM contact@test;
As always, I hope this helps somebody trying to solve a problem.
Subscribe via RSS