Python and Oracle Database: Scripting for the Future

Contents

• Preface
• Connection Information
• Overview
• Using Python cx_Oracle 7 with Oracle Database
• 1. Connecting to Oracle
  • 1.1 Review the connection credentials
  • 1.2 Creating a basic connection
  • 1.3 Indentation indicates code structure
  • 1.4 Executing a query
  • 1.5 Closing connections
  • 1.6 Checking versions
• 2. Connection Pooling
  • 2.1 Session pooling
  • 2.2 Session pool experiments
  • 2.3 Creating a DRCP Connection
  • 2.4 Session pooling and DRCP
  • 2.5 More DRCP investigation
• 3. Fetching Data
  • 3.1 A simple query
  • 3.2 Using fetchone()
  • 3.3 Using fetchmany()
  • 3.4 Scrollable cursors
  • 3.5 Tuning with arraysize
• 4. Binding Data
  • 4.1 Binding in queries
  • 4.2 Binding in inserts
  • 4.3 Batcherrors
  • 4.4 Binding named objects
• 5. PL/SQL
  • 5.1 PL/SQL functions
  • 5.2 PL/SQL procedures
• 6. Type Handlers
  • 6.1 Basic output type handler
  • 6.2 Output type handlers and variable converters
  • 6.3 Input type handlers
• 7. LOBs
  • 7.1 Fetching a CLOB using a locator
  • 7.2 Fetching a CLOB as a string
• 8. Rowfactory functions
  • 8.1 Rowfactory for mapping column names
• 9. Subclassing connections and cursors
  • 9.1 Subclassing connections
  • 9.2 Subclassing cursors
• 10. Advanced Queuing
  • 10.1 Message passing with Oracle Advanced Queuing
• Summary
• Appendix: Python Primer
• Resources

Preface

If you are running this tutorial in your own environment, install the following required software:

1. Python (3.6 preferred but 2.7 should work)
2. cx_Oracle (version 7 preferred but 6.3 or later should work) and Oracle Instant Client Package - Basic (version 18.3 preferred but 12.2 should work)
   • Linux
   • macOS - please note the special instructions for macOS in the link.
   • Windows
3. Oracle Instant Client Package - SQL*Plus.

To create the schema run:

sqlplus sys/yoursyspassword@localhost/orclpdb as sysdba @sql/SetupSamples

Connection Information

The database connection information is set in two files:

• db_config.py which is imported by the other Python modules.
• db_config.sql which is used by the other SQL scripts.

The username is "pythonhol" with the password "welcome". The connect string is "localhost/orclpdb". See sql/SampleEnv.sql.

It is easist to have a local pluggable database with the service 'orclpdb' configured. If your database is not local, or has a different service, you will need to modify the connection information in db_config.py and db_config.sql.

The following sections may need adjusting, depending on how you have set up your environment.

Overview

This tutorial is an introduction to using Python with Oracle Database. It contains beginner and advanced material. Sections can be done in any order. Choose the content that interests you and your skill level.

Follow the steps in this document. The tutorial directory has scripts to run and modify. The tutorial/solutions directory has scripts with the suggested code changes.

Use the Desktop icons to start editors and terminal windows.

If you are new to Python review the Appendix: Python Primer to gain an understanding of the language.

Using Python cx_Oracle 7 with Oracle Database

Python is a popular general purpose dynamic scripting language. The cx_Oracle interface provides Python API to access Oracle Database.

• 1. Connecting to Oracle

  • 1.1 Review the connection credentials

    Review db_config.py and db_config.sql in the tutorial directory. These are included in other Python and SQL files in this tutorial:

    db_config.py

    user = "pythonhol"
    pw = "welcome"
    dsn = "localhost/orclpdb"
    

    db_config.sql

    def user = "pythonhol"
    def pw = "welcome"
    def connect_string = "localhost/orclpdb"
              

    By default they connect to the 'orclpdb' database service on the same machine as Python. You can modify the values in both files to match the connection information for your environment.

  • 1.2 Creating a basic connection

    Review the code contained in connect.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    print("Database version:", con.version)
    

    The cx_Oracle module is imported to provide the API for accessing the Oracle database. Many inbuilt and third party modules can be included in this way in Python scripts.

    The connect() method is passed the username, the password and the connection string that you configured in the db_config.py module. In this case, Oracle's Easy Connect connection string syntax is used. It consists of the hostname of your machine, localhost, and the database service name orclpdb.

    Open a command terminal and change to the tutorial directory:

    cd tutorial

    Run the Python script:

    python connect.py

    The version number of the database should be displayed. An exception is raised if the connection fails. Adjust the username, password or connect string parameters to invalid values to see the exception.

    cx_Oracle also supports "external authentication", which allows connections without needing usernames and passwords to be embedded in the code. Authentication would then instead be performed by, for example, LDAP.

  • 1.3 Indentation indicates code structure

    There are no statement terminators or begin/end keywords or braces to indicate blocks of code.

    Open connect.py in an editor. Indent the print statement with some spaces:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
      print("Database version:", con.version)
    

    Save the script and run it again:

    python connect.py 

    This raises an exception about the indentation. The number of spaces or tabs must be consistent in each block; otherwise, the Python interpreter will either raise an exception or execute code unexpectedly.

    Python may not always be able to identify accidental from deliberate indentation. Check your indentation is correct before running each example. Make sure to indent all statement blocks equally. Note the sample files use spaces, not tabs.

  • 1.4 Executing a query

    Open query.py in an editor. It looks like:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    

    Edit the file and add the code shown in bold below:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    cur = con.cursor()
    cur.execute("select * from dept order by deptno")
    res = cur.fetchall()
    for row in res:
        print(row)
    

    Make sure the print(row) line is indented. This lab uses spaces, not tabs.

    The code executes a query and fetches all data.

    Save the file and run it:

    python query.py

    In each loop iteration a new row is stored in row as a Python "tuple" and is displayed.

    Fetching data is described further in section 3.

  • 1.5 Closing connections

    Connections and other resources used by cx_Oracle will automatically be closed at the end of scope. This is a common programming style that takes care of the correct order of resource closure.

    Resources can also be explicitly closed to free up database resources if they are no longer needed. This may be useful in blocks of code that remain active for some time.

    Open query.py in an editor and add calls to close the cursor and connection like:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    cur = con.cursor()
    cur.execute("select * from dept order by deptno")
    res = cur.fetchall()
    for row in res:
        print(row)
    
    cur.close()
    con.close()
    

    Running the script completes without error:

    python query.py

    If you swap the order of the two close() calls you will see an error.

  • 1.6 Checking versions

    Review the code contained in versions.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    print(cx_Oracle.version)

    Run the script:

    python versions.py

    This gives the version of the cx_Oracle interface.

    Edit the file to print the version of the database, and of the Oracle client libraries used by cx_Oracle:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    print(cx_Oracle.version)
    print("Database version:", con.version)
    print("Client version:", cx_Oracle.clientversion())
    

    When the script is run, it will display:

    7.0.0
    Database version: 18.3.0.0.0
    Client version: (18, 3, 0, 0, 0)
    

    Note the client version is a tuple.

    Any cx_Oracle installation can connect to older and newer Oracle Database versions. By checking the Oracle Database and client versions numbers, the application can make use of the best Oracle features available.

• 2. Connection Pooling

  • 2.1 Session pooling

    Review the code contained in connect_pool.py:

    import cx_Oracle
    import threading
    import db_config
    
    pool = cx_Oracle.SessionPool(db_config.user, db_config.pw, db_config.dsn,
                                 min = 2, max = 5, increment = 1, threaded = True)
    
    def Query():
        con = pool.acquire()
        cur = con.cursor()
        for i in range(4):
            cur.execute("select myseq.nextval from dual")
            seqval, = cur.fetchone()
            print("Thread", threading.current_thread().name, "fetched sequence =", seqval)
    
    thread1 = threading.Thread(name='#1', target=Query)
    thread1.start()
    
    thread2 = threading.Thread(name='#2', target=Query)
    thread2.start()
    
    thread1.join()
    thread2.join()
    
    print("All done!")
    

    The SessionPool() function creates a pool of Oracle "sessions" for the user. Sessions in the pool can be used by cx_Oracle connections by calling pool.acquire(). The initial pool size is 2 sessions. The maximum size is 5 sessions. When the pool needs to grow, 1 new session will be created at a time. The pool can shrink back to the minimum size of 2 when sessions are no longer in use.

    The def Query(): line creates a method that is called by each thread.

    In the method, the pool.acquire() call gets one session from the pool (as long as less than 5 are already in use). This session is used in a loop of 4 iterations to query the sequence myseq. At the end of the method, cx_Oracle will automatically close the cursor and release the session back to the pool for reuse.

    The seqval, = cur.fetchone() line fetches a row and puts the single value contained in the result tuple into the variable seqval. Without the comma, the value in seqval would be a tuple like "(1,)".

    Two threads are created, each invoking the Query() method.

    In a command terminal, run:

    python connect_pool.py

    The output shows interleaved query results as each thread fetches values independently. The order of interleaving may vary from run to run.

  • 2.2 Session pool experiments

    Review connect_pool2.py, which has a loop for the number of threads, each iteration invoking the Query() method:

    import cx_Oracle
    import threading
    import db_config
    
    pool = cx_Oracle.SessionPool(db_config.user, db_config.pw, db_config.dsn,
                                 min = 2, max = 5, increment = 1, threaded = True)
    
    def Query():
        con = pool.acquire()
        cur = con.cursor()
        for i in range(4):
            cur.execute("select myseq.nextval from dual")
            seqval, = cur.fetchone()
            print("Thread", threading.current_thread().name, "fetched sequence =", seqval)
    
    numberOfThreads = 2
    threadArray = []
    
    for i in range(numberOfThreads):
        thread = threading.Thread(name = '#' + str(i), target = Query)
        threadArray.append(thread)
        thread.start()
    
    for t in threadArray:
        t.join()
    
    print("All done!")
    

    In a command terminal, run:

    python connect_pool2.py

    Experiment with different values of the pool parameters and numberOfThreads. Larger initial pool sizes will make the pool creation slower, but the sessions will be available immediately when needed. When numberOfThreads exceeds the maximum size of the pool, the acquire() call will generate an error. Adding the additional argument getmode = cx_Oracle.SPOOL_ATTRVAL_WAIT to the cx_Oracle.SessionPool() call will prevent the exception from taking place, but will cause the thread to wait until a session is available.

    Pool configurations where min is the same as max (and increment = 0) are often recommended as a way to avoid connection storms on the database server.

  • 2.3 Creating a DRCP Connection

    Database Resident Connection Pooling allows multiple Python processes on multiple machines to share a small pool of database server processes.

    Below left is a diagram without DRCP. Every application connection or session has its own 'dedicated' database server process. Application connect and close calls require the expensive create and destroy of those database server processes. To avoid these costs, scripts may hold connections open even when not doing database work: these idle server processes consumes database host resources. Below right is a diagram with DRCP. Scripts can use database servers from a precreated pool of servers and return them when they are not in use.

    Without DRCP

    With DRCP

    DRCP is useful when the database host machine does not have enough memory to handled the number of database server processes required. However, if database host memory is large enough, then the default, 'dedicated' server process model is generally recommended. If DRCP is enabled, it is best used in conjunction with cx_Oracle session pooling.

    Batch scripts doing long running jobs should generally use dedicated connections. Both dedicated and DRCP servers can be used in the same database for different applications.

    Review the code contained in connect_drcp.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn + ":pooled",
                            cclass="PYTHONHOL", purity=cx_Oracle.ATTR_PURITY_SELF)
    print("Database version:", con.version)
    

    This is similar to connect.py but ":pooled" is appended to the connection string, telling the database to use a pooled server. A Connection Class "PYTHONHOL" is also passed into the connect() method to allow grouping of database servers to applications.

    The "purity" of the connection is defined as the ATTR_PURITY_SELF constant, meaning the session state (such as the default date format) might be retained between connection calls, giving performance benefits. Session information will be discarded if a pooled server is later reused by an application with a different connection class name.

    Applications that should never share session information should use a different connection class and/or use ATTR_PURITY_NEW to force creation of a new session. This reduces overall scalability but prevents applications mis-using session information.

    Run connect_drcp.py in a terminal window.

    python connect_drcp.py

    The output is simply the version of the database.

  • 2.4 Session pooling and DRCP

    DRCP works well with session pooling.

    Edit connect_pool2.py, reset any changed pool options, and modify it to use DRCP:

    import cx_Oracle
    import threading
    
    pool = cx_Oracle.SessionPool(db_config.user, db_config.pw, db_config.dsn + ":pooled",
                                 min = 2, max = 5, increment = 1, threaded = True)
    
    def Query():
        con = pool.acquire(cclass = "PYTHONHOL", purity = cx_Oracle.ATTR_PURITY_SELF)
        cur = conn.cursor()
        for i in range(4):
            cur.execute("select myseq.nextval from dual")
            seqval, = cur.fetchone()
            print("Thread", threading.current_thread().name, "fetched sequence =", seqval)
    
    numberOfThreads = 2
    threadArray = []
    
    for i in range(numberOfThreads):
        thread = threading.Thread(name = '#' + str(i), target = Query)
        threadArray.append(thread)
        thread.start()
    
    for t in threadArray:
        t.join()
    
    print("All done!")
    

    The script logic does not need to be changed to benefit from DRCP connection pooling.

    Run the script:

    python connect_pool2.py

    If you get the error "ORA-24418: Cannot open further sessions", it is because connection requests are being made while the pool is starting or growing. Add the argument getmode = cx_Oracle.SPOOL_ATTRVAL_WAIT to the cx_Oracle.SessionPool() call so connection requests wait for pooled sessions to be available.

    Open a new a terminal window and invoke SQL*Plus:

    sqlplus /nolog @drcp_query.sql

    This shows the number of connection requests made to the pool since the database was started ("NUM_REQUESTS"), how many of those reused a pooled server's session ("NUM_HITS"), and how many had to create new sessions ("NUM_MISSES"). Typically the goal is a low number of misses.

    To see the pool configuration you can query DBA_CPOOL_INFO.

  • 2.5 More DRCP investigation

    To explore the behaviors of session and DRCP pooling futher, you could try changing the purity to cx_Oracle.ATTR_PURITY_NEW to see the effect on the DRCP NUM_MISSES statistic.

    Another experiement is to include the time module at the file top:

    import time

    and add calls to time.sleep(1) in the code, for example in the query loop. Then look at the way the threads execute. Use drcp_query.sql to monitor the pool's behavior.

• 3. Fetching Data

  • 3.1 A simple query

    There are a number of functions you can use to query an Oracle database, but the basics of querying are always the same:

    1. Parse the statement for execution.
    2. Bind data values (optional).
    3. Execute the statement.
    4. Fetch the results from the database.

    Review the code contained in query2.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    cur = con.cursor()
    cur.execute("select * from dept order by deptno")
    for deptno, dname, loc in cur:
        print("Department number: ", deptno)
        print("Department name: ", dname)
        print("Department location:", loc)
    

    The cursor() method opens a cursor for statements to use.

    The execute() method parses and executes the statement.

    The loop fetches each row from the cursor and unpacks the returned tuple into the variables deptno, dname, loc, which are then printed.

    Run the script in a terminal window:

    python query2.py

  • 3.2 Using fetchone()

    When the number of rows is large, the fetchall() call may use too much memory.

    Review the code contained in query_one.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    cur.execute("select * from dept order by deptno")
    row = cur.fetchone()
    print(row)
    
    row = cur.fetchone()
    print(row)
    

    This uses the fetchone() method to return just a single row as a tuple. When called multiple time, consecutive rows are returned:

    Run the script in a terminal window:

    python query_one.py

    The first two rows of the table are printed.

  • 3.3 Using fetchmany()

    Review the code contained in query_many.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    cur.execute("select * from dept order by deptno")
    res = cur.fetchmany(numRows = 3)
    print(res)
    

    The fetchmany() method returns a list of tuples. By default the number of rows returned is specified by the cursor attribute arraysize (which defaults to 100). Here the numRows parameter specifies that three rows should be returned.

    Run the script in a terminal window:

    python query_many.py

    The first three rows of the table are returned as a list (Python's name for an array) of tuples.

    You can access elements of the lists by position indexes. To see this, edit the file and add:

    print(res[0])    # first row
    print(res[0][1]) # second element of first row
    

  • 3.4 Scrollable cursors

    Scrollable cursors enable the application to move backwards as well as forwards in query results. They can be used to skip rows as well as move to a particular row.

    Review the code contained in query_scroll.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor(scrollable = True)
    
    cur.execute("select * from dept order by deptno")
    
    cur.scroll(2, mode = "absolute")  # go to second row
    print(cur.fetchone())
    
    cur.scroll(-1)                    # go back one row
    print(cur.fetchone())
    

    Run the script in a terminal window:

    python query_scroll.py

    Edit query_scroll.py and experiment with different scroll options and orders, such as:

    cur.scroll(1)  # go to next row
    print(cur.fetchone())
    
    cur.scroll(mode = "first")  # go to first row
    print(cur.fetchone())
    

    Try some scroll options that go beyond the number of rows in the resultset.

  • 3.5 Tuning with arraysize

    This section demonstrates a way to improve query performance by increasing the number of rows returned in each batch from Oracle to the Python program.

    First, create a table with a large number of rows. Review query_arraysize.sql:

    create table bigtab (mycol varchar2(20));
    begin
      for i in 1..20000
      loop
       insert into bigtab (mycol) values (dbms_random.string('A',20));
      end loop;
    end;
    /
    show errors
    
    commit;
    

    In a terminal window run the script as:

    sqlplus /nolog @query_arraysize.sql

    Review the code contained in query_arraysize.py:

    import cx_Oracle
    import time
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    
    start = time.time()
    
    cur = con.cursor()
    cur.arraysize = 10
    cur.execute("select * from bigtab")
    res = cur.fetchall()
    # print(res)  # uncomment to display the query results
    
    elapsed = (time.time() - start)
    print(elapsed, "seconds")
    

    This uses the 'time' module to measure elapsed time of the query. The arraysize is set to 10. This causes batches of 10 records at a time to be returned from the database to a cache in Python. This reduces the number of "roundtrips" made to the database, often reducing network load and reducing the number of context switches on the database server. The fetchone(), fetchmany() and fetchall() methods will read from the cache before requesting more data from the database.

    In a terminal window, run:

    python query_arraysize.py

    Reload a few times to see the average times.

    Experiment with different arraysize values. For example, edit query_arraysize.py and change the arraysize to:

    cur.arraysize = 2000

    Rerun the script to compare the performance of different arraysize settings.

    In general, larger array sizes improve performance. Depending on how fast your system is, you may need to use different arraysizes than those given here to see a meaningful time difference.

    The default arraysize used by cx_Oracle 7 is 100. There is a time/space tradeoff for increasing the arraysize. Larger arraysizes will require more memory in Python for buffering the records.

    If you know a query only returns a few records, decrease the arraysize from the default to reduce memory usage.

• 4. Binding Data

  Bind variables enable you to re-execute statements with new data values, without the overhead of reparsing the statement. Bind variables improve code reusability, and can reduce the risk of SQL injection attacks.

  • 4.1 Binding in queries

    Review the code contained in bind_query.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    cur.prepare("select * from dept where deptno = :id order by deptno")
    
    cur.execute(None, id = 20)
    res = cur.fetchall()
    print(res)
    
    cur.execute(None, id = 10)
    res = cur.fetchall()
    print(res)
    

    The statement contains a bind variable ":id" placeholder. The statement is only prepared once but executed twice with different values for the WHERE clause.

    The special symbol "None" is used in place of the statement text argument to execute() because the prepare() method has already set the statement. The second argument to the execute() call can be a sequence (binding by position) or a dictionary (binding by name) or an arbitrary number of named arguments (also binding by name), which is what has been done in this example. In the first execute call, this dictionary has the value 20 for the key of "id". The second execute uses the value 10.

    From a terminal window, run:

    python bind_query.py

    The output shows the details for the two departments.

  • 4.2 Binding in inserts

    Review the code in bind_insert.sql creating a table for inserting data:

    create table mytab (id number, data varchar2(20), constraint my_pk primary key (id));
    

    Run the script as:

    sqlplus /nolog @bind_insert.sql

    Review the code contained in bind_insert.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    rows = [ (1, "First" ), (2, "Second" ),
             (3, "Third" ), (4, "Fourth" ),
             (5, "Fifth" ), (6, "Sixth" ),
             (7, "Seventh" ) ]
    
    cur.executemany("insert into mytab(id, data) values (:1, :2)", rows)
    
    # Now query the results back
    
    cur2 = con.cursor()
    cur2.execute('select * from mytab')
    res = cur2.fetchall()
    print(res)
    

    The 'rows' array contains the data to be inserted.

    The executemany() call inserts all rows. This calls allows "array binding", which is an efficient way to insert multiple records.

    The final part of the script queries the results back and displays them as a list of tuples.

    From a terminal window, run:

    python bind_insert.py

    The new results are automatically rolled back at the end of the script so re-running it will always show the same number of rows in the table.

  • 4.3 Batcherrors

    The Batcherrors features allows invalid data to be identified while allowing valid data to be inserted.

    Edit the data values in bind_insert.py and create a row with a duplicate key:

    rows = [ (1, "First" ), (2, "Second" ),
             (3, "Third" ), (4, "Fourth" ),
             (5, "Fifth" ), (6, "Sixth" ),
             (6, "Duplicate" ),
             (7, "Seventh" ) ]
    
    

    From a terminal window, run:

    python bind_insert.py

    The duplicate generates the error "ORA-00001: unique constraint (PYTHONHOL.MY_PK) violated". The data is rolled back and the query returns no rows.

    Edit the file again and enable batcherrors like:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    rows = [ (1, "First" ), (2, "Second" ),
             (3, "Third" ), (4, "Fourth" ),
             (5, "Fifth" ), (6, "Sixth" ),
             (6, "Duplicate" ),
             (7, "Seventh" ) ]
    
    cur.executemany("insert into mytab(id, data) values (:1, :2)", rows, batcherrors = True)
    
    for error in cur.getbatcherrors():
        print("Error", error.message.rstrip(), "at row offset", error.offset)
    
    # Now query the results back
    
    cur2 = con.cursor()
    cur2.execute('select * from mytab')
    res = cur2.fetchall()
    print(res)
    

    Run the file:

    python bind_insert.py

    The new code shows the offending duplicate row: "ORA-00001: unique constraint (PYTHONHOL.MY_PK) violated at row offset 6". This indicates the 6th data value (counting from 0) had a problem.

    The other data gets inserted and is queried back.

    At the end of the script, cx_Oracle will rollback an uncommitted transaction. If you want to commit results, you can use:

    con.commit()

    To force a rollback in cx_Oracle, use:

    con.rollback()

  • 4.4 Binding named objects

    cx_Oracle can fetch and bind named object types such as Oracle's Spatial Data Objects (SDO).

    In a terminal window, start SQL*Plus using the lab credentials and connection string, such as:

    sqlplus pythonhol/welcome@localhost/orclpdb
    

    Use the SQL*Plus DESCRIBE command to look at the SDO definition:

    desc MDSYS.SDO_GEOMETRY
    

    It contains various attributes and methods. The top level description is:

     Name                                      Null?    Type
     ----------------------------------------- -------- ----------------------------
     SDO_GTYPE                                          NUMBER
     SDO_SRID                                           NUMBER
     SDO_POINT                                          MDSYS.SDO_POINT_TYPE
     SDO_ELEM_INFO                                      MDSYS.SDO_ELEM_INFO_ARRAY
     SDO_ORDINATES                                      MDSYS.SDO_ORDINATE_ARRAY
    

    Review the code contained in bind_sdo.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    # Create table
    cur.execute("""begin
                     execute immediate 'drop table testgeometry';
                     exception when others then
                       if sqlcode <> -942 then
                         raise;
                       end if;
                   end;""")
    cur.execute("""create table testgeometry (
                   id number(9) not null,
                   geometry MDSYS.SDO_GEOMETRY not null)""")
    
    # Create and populate Oracle objects
    typeObj = con.gettype("MDSYS.SDO_GEOMETRY")
    elementInfoTypeObj = con.gettype("MDSYS.SDO_ELEM_INFO_ARRAY")
    ordinateTypeObj = con.gettype("MDSYS.SDO_ORDINATE_ARRAY")
    obj = typeObj.newobject()
    obj.SDO_GTYPE = 2003
    obj.SDO_ELEM_INFO = elementInfoTypeObj.newobject()
    obj.SDO_ELEM_INFO.extend([1, 1003, 3])
    obj.SDO_ORDINATES = ordinateTypeObj.newobject()
    obj.SDO_ORDINATES.extend([1, 1, 5, 7])
    print("Created object", obj)
    
    # Add a new row
    print("Adding row to table...")
    cur.execute("insert into testgeometry values (1, :objbv)", objbv = obj)
    print("Row added!")
    
    # Query the row
    print("Querying row just inserted...")
    cur.execute("select id, geometry from testgeometry");
    for row in cur:
        print(row)
    

    This uses gettype() to get the database types of the SDO and its object attributes. The newobject() calls create Python representations of those objects. The python object atributes are then set. Oracle VARRAY types such as SDO_ELEM_INFO_ARRAY are set with extend().

    Run the file:

    python bind_sdo.py

    The new SDO is shown as an object, similar to:

    (1, <cx_Oracle.Object MDSYS.SDO_GEOMETRY at 0x104a76230>)

    To show the attribute values, edit the the query code section at the end of the file. Add a new method that traverses the object. The file below the existing comment "# (Change below here)") should look like:

    # (Change below here)
    
    # Define a function to dump the contents of an Oracle object
    def dumpobject(obj, prefix = "  "):
        if obj.type.iscollection:
            print(prefix, "[")
            for value in obj.aslist():
                if isinstance(value, cx_Oracle.Object):
                    dumpobject(value, prefix + "  ")
                else:
                    print(prefix + "  ", repr(value))
            print(prefix, "]")
        else:
            print(prefix, "{")
            for attr in obj.type.attributes:
                value = getattr(obj, attr.name)
                if isinstance(value, cx_Oracle.Object):
                    print(prefix + "  " + attr.name + " :")
                    dumpobject(value, prefix + "    ")
                else:
                    print(prefix + "  " + attr.name + " :", repr(value))
            print(prefix, "}")
    
    # Query the row
    print("Querying row just inserted...")
    cur.execute("select id, geometry from testgeometry")
    for id, obj in cur:
        print("Id: ", id)
        dumpobject(obj)
    

    Run the file again:

    python bind_sdo.py

    This shows

    Querying row just inserted...
    Id:  1
       {
        SDO_GTYPE : 2003
        SDO_SRID : None
        SDO_POINT : None
        SDO_ELEM_INFO :
           [
             1
             1003
             3
           ]
        SDO_ORDINATES :
           [
             1
             1
             5
             7
           ]
       }
    

    To explore further, try setting the SDO attribute SDO_POINT, which is of type SDO_POINT_TYPE.

    The gettype() and newobject() methods can also be used to bind PL/SQL Records and Collections.

• 5. PL/SQL

  PL/SQL is Oracle's procedural language extension to SQL. PL/SQL procedures and functions are stored and run in the database. Using PL/SQL lets all database applications reuse logic, no matter how the application accesses the database. Many data-related operations can be performed in PL/SQL faster than extracting the data into a program (for example, Python) and then processing it.

  • 5.1 PL/SQL functions

    Review plsql_func.sql which creates a PL/SQL stored function myfunc() to insert a row into a new table named ptab and return double the inserted value:

    create table ptab (mydata varchar(20), myid number);
    
    create or replace function myfunc(d_p in varchar2, i_p in number) return number as
      begin
        insert into ptab (mydata, myid) values (d_p, i_p);
        return (i_p * 2);
      end;
    /
    

    Run the script using:

    sqlplus /nolog @plsql_func.sql

    Review the code contained in plsql_func.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    res = cur.callfunc('myfunc', int, ('abc', 2))
    print(res)
    

    This uses callfunc() to execute the function. The second parameter is the type of the returned value. It should be one of the types supported by cx_Oracle or one of the type constants defined by cx_Oracle (such as cx_Oracle.NUMBER). The two PL/SQL function parameters are passed as a tuple, binding them to the function parameter arguments.

    From a terminal window, run:

    python plsql_func.py

    The output is a result of the PL/SQL function calculation.

  • 5.2 PL/SQL procedures

    Review plsql_proc.sql which creates a PL/SQL procedure myproc() to accept two parameters. The second parameter contains an OUT return value.

    create or replace procedure myproc(v1_p in number, v2_p out number) as
    begin
      v2_p := v1_p * 2;
    end;
    /
    

    Run the script with:

    sqlplus /nolog @plsql_proc.sql

    Review the code contained in plsql_proc.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    myvar = cur.var(int)
    cur.callproc('myproc', (123, myvar))
    print(myvar.getvalue())
    

    This creates an integer variable myvar to hold the value returned by the PL/SQL OUT parameter. The input number 123 and the output variable name are bound to the procedure call parameters using a tuple.

    To call the PL/SQL procedure, the callproc() method is used.

    In a terminal window, run:

    python plsql_proc.py

    The getvalue() method displays the returned value.

• 6. Type Handlers

  • 6.1 Basic output type handler

    Output type handlers enable applications to change how data is fetched from the database. For example, numbers can be returned as strings or decimal objects. LOBs can be returned as string or bytes.

    A type handler is enabled by setting the outputtypehandler attribute on either a cursor or the connection. If set on a cursor it only affects queries executed by that cursor. If set on a connection it affects all queries executed on cursors created by that connection.

    Review the code contained in type_output.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    print("Standard output...")
    for row in cur.execute("select * from dept"):
        print(row)
    

    In a terminal window, run:

    python type_output.py

    This shows the department number represented as digits like 10.

    Add an output type handler to the bottom of the file:

    def ReturnNumbersAsStrings(cursor, name, defaultType, size, precision, scale):
        if defaultType == cx_Oracle.NUMBER:
            return cursor.var(str, 9, cursor.arraysize)
    
    print("Output type handler output...")
    cur = con.cursor()
    cur.outputtypehandler = ReturnNumbersAsStrings
    for row in cur.execute("select * from dept"):
        print(row)
    

    This type handler converts any number columns to strings with maxium size 9.

    Run the script again:

    python type_output.py

    The new output shows the department numbers are now strings within quotes like '10'.

  • 6.2 Output type handlers and variable converters

    When numbers are fetched from the database, the conversion from Oracle's decimal representation to Python's binary format may need careful handling. To avoid unexpected issues, the general recommendation is to do number operations in SQL or PL/SQL, or to use the decimal module in Python.

    Output type handlers can be combined with variable converters to change how data is fetched.

    Review type_converter.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    for value, in cur.execute("select 0.1 from dual"):
        print("Value:", value, "* 3 =", value * 3)
    

    Run the file:

    python type_converter.py

    The output is like:

    Value: 0.1 * 3 = 0.30000000000000004

    Edit the file and add a type handler that uses a Python decimal converter:

    import cx_Oracle
    import decimal
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    def ReturnNumbersAsDecimal(cursor, name, defaultType, size, precision, scale):
        if defaultType == cx_Oracle.NUMBER:
            return cursor.var(str, 9, cursor.arraysize, outconverter = decimal.Decimal)
    
    cur.outputtypehandler = ReturnNumbersAsDecimal
    
    for value, in cur.execute("select 0.1 from dual"):
        print("Value:", value, "* 3 =", value * 3)
    

    The Python decimal.Decimal converter gets called with the string representation of the Oracle number. The output from decimal.Decimal is returned in the output tuple.

    Run the file again:

    python type_converter.py

    Output is like:

    Value: 0.1 * 3 = 0.3

    Although the code demonstrates the use of outconverter, in this particular case, the variable can be created simply by using the following code to replace the outputtypehandler function defined above:

    def ReturnNumbersAsDecimal(cursor, name, defaultType, size, precision, scale):
        if defaultType == cx_Oracle.NUMBER:
            return cursor.var(decimal.Decimal, arraysize = cursor.arraysize)
    

  • 6.3 Input type handlers

    Input type handlers enable applications to change how data is bound to statements, or to enable new types to be bound directly without having to be converted individually.

    Review type_input.py, which is similar to the final bind_sdo.py from section 4.4, with the addition of a new class and converter (shown in bold):

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    # Create table
    cur.execute("""begin
                     execute immediate 'drop table testgeometry';
                     exception when others then
                       if sqlcode <> -942 then
                         raise;
                       end if;
                   end;""")
    cur.execute("""create table testgeometry (
                   id number(9) not null,
                   geometry MDSYS.SDO_GEOMETRY not null)""")
    
    # Create a Python class for an SDO
    class mySDO(object):
    
        def __init__(self, gtype, elemInfo, ordinates):
            self.gtype = gtype
            self.elemInfo = elemInfo
            self.ordinates = ordinates
    
    # Get Oracle type information
    objType = con.gettype("MDSYS.SDO_GEOMETRY")
    elementInfoTypeObj = con.gettype("MDSYS.SDO_ELEM_INFO_ARRAY")
    ordinateTypeObj = con.gettype("MDSYS.SDO_ORDINATE_ARRAY")
    
    # Convert a Python object to MDSYS.SDO_GEOMETRY
    def SDOInConverter(value):
        obj = objType.newobject()
        obj.SDO_GTYPE = value.gtype
        obj.SDO_ELEM_INFO = elementInfoTypeObj.newobject()
        obj.SDO_ELEM_INFO.extend(value.elemInfo)
        obj.SDO_ORDINATES = ordinateTypeObj.newobject()
        obj.SDO_ORDINATES.extend(value.ordinates)
        return obj
    
    def SDOInputTypeHandler(cursor, value, numElements):
        if isinstance(value, mySDO):
            return cursor.var(cx_Oracle.OBJECT, arraysize = numElements,
                    inconverter = SDOInConverter, typename = objType.name)
    
    sdo = mySDO(2003, [1, 1003, 3], [1, 1, 5, 7])  # Python object
    cur.inputtypehandler = SDOInputTypeHandler
    cur.execute("insert into testgeometry values (:1, :2)", (1, sdo))
    
    # Define a function to dump the contents of an Oracle object
    def dumpobject(obj, prefix = "  "):
        if obj.type.iscollection:
            print(prefix, "[")
            for value in obj.aslist():
                if isinstance(value, cx_Oracle.Object):
                    dumpobject(value, prefix + "  ")
                else:
                    print(prefix + "  ", repr(value))
            print(prefix, "]")
        else:
            print(prefix, "{")
            for attr in obj.type.attributes:
                value = getattr(obj, attr.name)
                if isinstance(value, cx_Oracle.Object):
                    print(prefix + "  " + attr.name + " :")
                    dumpobject(value, prefix + "    ")
                else:
                    print(prefix + "  " + attr.name + " :", repr(value))
            print(prefix, "}")
    
    # Query the row
    print("Querying row just inserted...")
    cur.execute("select id, geometry from testgeometry")
    for (id, obj) in cur:
        print("Id: ", id)
        dumpobject(obj)
    

    In the new file, a Python class mySDO is defined, which has attributes corresponding to each Oracle MDSYS.SDO_GEOMETRY attribute. The mySDO class is used lower in the code to create a Python instance:

    sdo = mySDO(2003, [1, 1003, 3], [1, 1, 5, 7])

    which is then directly bound into the INSERT statement like:

    cur.execute("insert into testgeometry values (:1, :2)", (1, sdo))

    The mapping between Python and Oracle objects is handled in SDOInConverter which uses the cx_Oracle newobject() and extend() methods to create an Oracle object from the Python object values. The SDOInConverter method is called by the input type handler SDOInputTypeHandler whenever an instance of mySDO is inserted with the cursor.

    To confirm the behavior, run the file:

    python type_input.py

• 7. LOBs

  • 7.1 Fetching a CLOB using a locator

    Review the code contained in clob.py:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    print("Inserting data...")
    cur.execute("truncate table testclobs")
    longString = ""
    for i in range(5):
        char = chr(ord('A') + i)
        longString += char * 250
        cur.execute("insert into testclobs values (:1, :2)",
                       (i + 1, "String data " + longString + ' End of string'))
    con.commit()
    
    print("Querying data...")
    cur.prepare("select * from testclobs where id = :id")
    cur.execute(None, {'id': 1})
    (id, clob) = cur.fetchone()
    print("CLOB length:", clob.size())
    clobdata = clob.read()
    print("CLOB data:", clobdata)
    

    This inserts some test string data and then fetches one record into clob, which is a cx_Oracle LOB Object. Methods on LOB include size() and read().

    To see the output, run the file:

    python clob.py

    Edit the file and experiment reading chunks of data by giving start character position and length, such as clob.read(1,10)

  • 7.2 Fetching a CLOB as a string

    For CLOBs small enough to fit in the application memory, it is much faster to fetch them directly as strings.

    Review the code contained in clob_string.py. The differences from clob.py are shown in bold:

    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    print("Inserting data...")
    cur.execute("truncate table testclobs")
    longString = ""
    for i in range(5):
        char = chr(ord('A') + i)
        longString += char * 250
        cur.execute("insert into testclobs values (:1, :2)",
                    (i + 1, "String data " + longString + ' End of string'))
    con.commit()
    
    def OutputTypeHandler(cursor, name, defaultType, size, precision, scale):
        if defaultType == cx_Oracle.CLOB:
            return cursor.var(cx_Oracle.LONG_STRING, arraysize = cursor.arraysize)
    
    con.outputtypehandler = OutputTypeHandler
    
    print("Querying data...")
    cur.prepare("select * from testclobs where id = :id")
    cur.execute(None, {'id': 1})
    (id, clobdata) = cur.fetchone()
    print("CLOB length:", len(clobdata))
    print("CLOB data:", clobdata)
    

    The OutputTypeHandler causes cx_Oracle to fetch the CLOB as a string. Standard Python string functions such as len() can be used on the result.

    The output is the same as for clob.py. To check, run the file:

    python clob_string.py

• 8. Rowfactory functions

  Rowfactory functions enable queries to return objects other than tuples. They can be used to provide names for the various columns or to return custom objects.

  • 8.1 Rowfactory for mapping column names

    Review the code contained in rowfactory.py:

    import collections
    import cx_Oracle
    import db_config
    
    con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
    cur = con.cursor()
    
    cur.execute("select deptno, dname from dept")
    rows = cur.fetchall()
    
    print('Array indexes:')
    for row in rows:
        print(row[0], "->", row[1])
    
    print('Loop target variables:')
    for c1, c2 in rows:
        print(c1, "->", c2)
    

    This shows two methods of accessing result set items from a data row. The first uses array indexes like row[0]. The second uses loop target variables which take the values of each row tuple.

    Run the file:

    python rowfactory.py

    Both access methods gives the same results.

    To use a rowfactory function, edit rowfactory.py and add this code at the bottom:

    print('Rowfactory:')
    cur.execute("select deptno, dname from dept")
    cur.rowfactory = collections.namedtuple("MyClass", ["DeptNumber", "DeptName"])
    
    rows = cur.fetchall()
    for row in rows:
        print(row.DeptNumber, "->", row.DeptName)
    

    This uses the Python factory function namedtuple() to create a subclass of tuple that allows access to the elements via indexes or the given field names.

    The print() function shows the use of the new named tuple fields. This coding style can help reduce coding errors.

    Run the script again:

    python rowfactory.py

    The output results are the same.

• 9. Subclassing connections and cursors

  Subclassing enables application to "hook" connection and cursor creation. This can be used to alter or log connection and execution parameters, and to extend cx_Oracle functionality.

  • 9.1 Subclassing connections

    Review the code contained in subclass.py:

    import cx_Oracle
    import db_config
    
    class MyConnection(cx_Oracle.Connection):
    
        def __init__(self):
            print("Connecting to database")
            return super(MyConnection, self).__init__(db_config.user, db_config.pw, db_config.dsn)
    
    con = MyConnection()
    cur = con.cursor()
    
    cur.execute("select count(*) from emp where deptno = :bv", (10,))
    count, = cur.fetchone()
    print("Number of rows:", count)
    

    This creates a new class "MyConnection" that inherits from the cx_Oracle Connection class. The __init__ method is invoked when an instance of the new class is created. It prints a message and calls the base class, passing the connection credentials.

    In the "normal" application, the application code:

    con = MyConnection()

    does not need to supply any credentials, as they are embedded in the custom subclass. All the cx_Oracle methods such as cursor() are available, as shown by the query.

    Run the file:

    python subclass.py

    The query executes successfully.

  • 9.2 Subclassing cursors

    Edit subclass.py and extend the cursor() method with a new MyCursor class:

    import cx_Oracle
    import db_config
    
    class MyConnection(cx_Oracle.Connection):
    
        def __init__(self):
            print("Connecting to database")
            return super(MyConnection, self).__init__(db_config.user, db_config.pw, db_config.dsn)
    
        def cursor(self):
            return MyCursor(self)
    
    class MyCursor(cx_Oracle.Cursor):
    
       def execute(self, statement, args):
           print("Executing:", statement)
           print("Arguments:")
           for argIndex, arg in enumerate(args):
               print("  Bind", argIndex + 1, "has value", repr(arg))
               return super(MyCursor, self).execute(statement, args)
    
       def fetchone(self):
           print("Fetchone()")
           return super(MyCursor, self).fetchone()
    
    con = MyConnection()
    cur = con.cursor()
    
    cur.execute("select count(*) from emp where deptno = :bv", (10,))
    count, = cur.fetchone()
    print("Number of rows:", count)
    

    When the application gets a cursor from the MyConnection class, the new cursor() method returns an instance of our new MyCursor class.

    The "application" query code remains unchanged. The new execute() and fetchone() methods of the MyCursor class get invoked. They do some logging and invoke the parent methods to do the actual statement execution.

    To confirm this, run the file again:

    python subclass.py

• 10. Advanced Queuing

  • 10.1 Message passing with Oracle Advanced Queuing

  Review aq.py:

  import cx_Oracle
  import decimal
  import db_config
  
  con = cx_Oracle.connect(db_config.user, db_config.pw, db_config.dsn)
  cur = con.cursor()
  
  BOOK_TYPE_NAME = "UDT_BOOK"
  QUEUE_NAME = "BOOKS"
  QUEUE_TABLE_NAME = "BOOK_QUEUE_TABLE"
  
  # Cleanup
  cur.execute(
      """begin
           dbms_aqadm.stop_queue('""" + QUEUE_NAME + """');
           dbms_aqadm.drop_queue('""" + QUEUE_NAME + """');
           dbms_aqadm.drop_queue_table('""" + QUEUE_TABLE_NAME + """');
           execute immediate 'drop type """ + BOOK_TYPE_NAME + """';
           exception when others then
             if sqlcode <> -24010 then
               raise;
             end if;
         end;""")
  
  # Create type
  print("Creating books type UDT_BOOK...")
  cur.execute("""
          create type %s as object (
              title varchar2(100),
              authors varchar2(100),
              price number(5,2)
          );""" % BOOK_TYPE_NAME)
  
  # Create queue table and queue and start the queue
  print("Creating queue table...")
  cur.callproc("dbms_aqadm.create_queue_table",
               (QUEUE_TABLE_NAME, BOOK_TYPE_NAME))
  cur.callproc("dbms_aqadm.create_queue", (QUEUE_NAME, QUEUE_TABLE_NAME))
  cur.callproc("dbms_aqadm.start_queue", (QUEUE_NAME,))
  
  # Enqueue a few messages
  booksType = con.gettype(BOOK_TYPE_NAME)
  book1 = booksType.newobject()
  book1.TITLE = "The Fellowship of the Ring"
  book1.AUTHORS = "Tolkien, J.R.R."
  book1.PRICE = decimal.Decimal("10.99")
  book2 = booksType.newobject()
  book2.TITLE = "Harry Potter and the Philosopher's Stone"
  book2.AUTHORS = "Rowling, J.K."
  book2.PRICE = decimal.Decimal("7.99")
  options = con.enqoptions()
  messageProperties = con.msgproperties()
  for book in (book1, book2):
      print("Enqueuing book", book.TITLE)
      con.enq(QUEUE_NAME, options, messageProperties, book)
  con.commit()
  
  # Dequeue the messages
  options = con.deqoptions()
  options.navigation = cx_Oracle.DEQ_FIRST_MSG
  options.wait = cx_Oracle.DEQ_NO_WAIT
  while con.deq(QUEUE_NAME, options, messageProperties, book):
      print("Dequeued book", book.TITLE)
  con.commit()
  

  This file sets up Advanced Queuing using Oracle's DBMS_AQADM package. The queue is used for passing Oracle UDT_BOOK objects.

  Run the file:

  python aq.py

  The output shows messages being queued and dequeued.

  To experiment, split the code into three files: one to create and start the queue, and two other files to queue and dequeue messages. Experiment running the queue and dequeue files concurrently in separate terminal windows. If you are stuck, look in the solutions directory at the aq-dequeue.py, aq-enqueue.py and aq-queuestart.py files.

  Try changing the dequeue options and mode. For example change the dequeue options.wait value to cx_Oracle.DEQ_WAIT_FOREVER.

Summary

In this tutorial, you have learned how to:

• Create connections
• Use sessions pooling and Database Resident Connection Pooling
• Execute queries and fetch data
• Use bind variables
• Use PL/SQL stored functions and procedures
• Extend cx_Oracle classes
• Use Oracle Advanced Queuing

Appendix: Python Primer

Python is a dynamically typed scripting language. It is most often used to run command-line scripts but is also used in Web applications.

Running Python

You can either:

• Create a file of Python commands, such as myfile.py. This can be run with:

  python myfile.py

• Alternatively run the Python interpreter by executing the python command in a terminal, and then interactively enter commands. Use Ctrl-D to exit back to the operating system prompt.

When you run scripts, Python automatically creates bytecode versions of them in a folder called __pycache__. These improve performance of scripts that are run multiple times. They are automatically recreated if the source file changes.

Indentation

Whitespace indentation is significant in Python. When copying examples, use the same column alignment as shown. The samples in this lab use spaces, not tabs.

The following indentation prints 'done' once after the loop has completed:

for i in range(5):
    print(i)
print('done')

But this indentation prints 'done' in each iteration:

for i in range(5):
    print(i)
    print('done')

Strings

Python strings can be enclosed in single or double quotes:

'A string constant'
"another constant"

Multi line strings use a triple-quote syntax:

"""
SELECT *
FROM EMP
"""

Variables

Variables do not need types declared:

count = 1
ename = 'Arnie'

Comments

Comments are either single line:

# a short comment

They can be multi-line using the triple-quote token to create a string that does nothing:

"""
a longer
comment
"""

Printing

Strings and variables can be displayed with a print() function:

print('Hello, World!')
print('Value:', count)

Note the print syntax and output is different in Python 2. Examples in this lab use from __future__ import print_function so that they run with Python 2 and Python 3.

Data Structures

Associative arrays are called 'dictionaries':

a2 = {'PI':3.1415, 'E':2.7182}

Ordered arrays are called 'lists':

a3 = [101, 4, 67]

Lists can be accessed via indexes.

print(a3[0])
print(a3[-1])
print(a3[1:3])

Tuples are like lists but cannot be changed once they are created. They are created with parentheses:

a4 = (3, 7, 10)

Individual values in a tuple can be assigned to variables like:

v1, v2, v3 = a4

Now the variable v1 contains 3, the variable v2 contains 7 and the variable v3 contains 10.

The value in a single entry tuple like "(13,)"can be assigned to a variable by putting a comma after the variable name like:

v1, = (13,)

If the assignment is:

v1 = (13,)

then v1 will contain the whole tuple "(13,)"

Objects

Everything in Python is an object. As an example, given the of the list a3 above, the append() method can be used to add a value to the list.

a3.append(23)

Now a3 contains [101, 4, 67, 23]

Flow Control

Code flow can be controlled with tests and loops. The if/elif/else statements look like:

if sal > 900000:
    print('Salary is way too big')
elif sal > 500000:
    print('Salary is huge')
else:
    print('Salary might be OK')

This also shows how the clauses are delimited with colons, and each sub block of code is indented.

Loops

A traditional loop is:

for i in range(10):
    print(i)

This prints the numbers from 0 to 9. The value of i is incremented in each iteration.

The 'for' command can also be used to iterate over lists and tuples:

a5 = ['Aa', 'Bb', 'Cc']
for v in a5:
    print(v)

This sets v to each element of the list a5 in turn.

Functions

A function may be defined as:

def myfunc(p1, p2):
    "Function documentation: add two numbers"
    print(p1, p2)
    return p1 + p2

Functions may or may not return values. This function could be called using:

v3 = myfunc(1, 3)

Function calls must appear after their function definition.

Functions are also objects and have attributes. The inbuilt __doc__ attribute can be used to find the function description:

print(myfunc.__doc__)

Modules

Sub-files can be included in Python scripts with an import statement.

import os
import sys

Many predefined modules exist, such as the os and the sys modules.

Resources

• Python 3 Documentation
• Python cx_Oracle Documentation
• Python cx_Oracle Source Code Repository

Copyright © 2017, Oracle and/or its affiliates. All rights reserved