ADOdb for Python

(c) 2004-2005 John Lim ( This software is licensed under a BSD-style license. See LICENSE.txt.

Databases Supported
Bug Reports and Help
Connection Examples
Function Reference
Error Handling
Bind Parameters


ADOdb is a database abstraction library (modelled on Microsoft's database API's). ADOdb was originally developed for PHP, and ported to Python. The Python version implements a subset of the PHP version. ADOdb is licensed using a BSD-style license.

Download:  Python version  PHP version

You might ask why Python needs a database abstraction library when Python provides the official DB API. Unfortunately the DB API does not encapsulate differences in the database implementations. For example, to select a limited number of rows, say 10 rows, you would have to use very different SQL for different databases:

MS SQL select top 10 from table
MySQL and PostgreSQL select * from table limit 10
Oracle select * from (select * from table) where rownum <= 10

These differences are handled by ADOdb (using SelectLimit), but not by the Python DB API. Other important database differences transparently handled by ADOdb include date-formating, associative arrays (records as dictionaries) and LOB-handling.

This class library assumes you are using Python 2.3 or later. Internally it uses the standard Python 2.3 datetime class.

Databases Supported

PHP bundles most database extensions in the official release. In Python, most database extensions are not part of the official release. You have to manually download and install the extensions yourself. The requirements are listed below:

Class Requirements Notes
odbc Download PythonWin extensio

No support for SelectLimit, UpdateBlob, UpdateBlobFile, Insert_ID, RecordCount and Affected_Rows.

access Requires mxodbc.  Only SelectLimit( ) with no offset parameter supported. RecordCount( ) not supported.
mssql Requires mxodbc.  Only SelectLimit( ) with no offset parameter supported. RecordCount( ) not supported.
mysql Download MySQL-python extension  
mxodbc Superior odbc extension. Licensing fee required for commercial us

SelectLimit( ) not supported. RecordCount( ) not supported.

Requires mxodbc. Connect to Oracle using ODBC.
Only SelectLimit( ) with no offset parameter supported. Requires Oracle client installed. RecordCount( ) not supported.
oci8 Download cx_Oracle extension. Also requires Oracle client to be installed.

Despite the name, it works with Oracle 8, 9 and later. SelectLimit( ) does not support the offset parameter. RecordCount( ) not supported.

odbc Download PythonWin extension

SelectLimit( ) only works with Access,VFP and Microsoft SQL Server. The offset parameter is not supported.

No support for UpdateBlob, UpdateBlobFile, Insert_ID, RecordCount and Affected_Rows.

odbc_mssql Download PythonWin extension

Same limitations as adodb_odbc extension, except that Insert_ID and Affected_Rows supported.

postgres Download psycopg extension  
vfp Requires mxodbc.  
Requires pysqlite.
Contributed by Glenn Washburn.


Run from the command prompt:

> python install

This will perform the adodb package installation.

You will need to install database specific extensions to use ADOdb for Python.

Bug Reports and Help

To report bugs, discuss ADOdb, and ask for help, post to the forums at:


The easiest way to learn how to use ADOdb for python is with a few examples. Here's one that contrasts PHP with Python. This example select some data from a table, prints each row, then closes the connection.

PHP Python
include "";
$conn = adodb.NewADOConnection('mysql');
$rs = $conn->Execute('select * from table');

while (!$rs->EOF) {
import adodb;
conn = adodb.NewADOConnection('mysql')
cursor = conn.Execute('select * from table')

while not cursor.EOF:
print cursor.fields


First we create a database connection object, conn. Then we login to the database using Connect( ). We now call Execute( ) to compile and execute the given SQL. The will return a recordset or cursor that will hold the current row in the fields property. Fields are numerically indexed, starting at index zero. When we want to move to the next record, we call MoveNext( ) which also updates the fields property. Lastly, we check to see whether there are any more records left to process by monitoring the EOF property.

As you can see, the PHP and Python code is very similar. The main difference is Execute( ) returns a recordset in PHP, while a cursor is returned in Python. A PHP recordset and Python cursor work identically with SELECT statements. However Python cursors work differently with INSERT/UPDATE/DELETE statements. You can see that below:

PHP Python
$sql = "update table set col='abc'";
$rs = $conn->Execute($sql);
$rows = $conn->Affected_Rows();
sql = "update table set col='abc'"
cursor = conn.Execute($sql)
rows = cursor.Affected_Rows()

In PHP, Affected_Rows( ) runs in the connection. In Python, all information related to an Execute( ) is retrieved from the cursor. This is because Python is multi-threaded, so it is no longer possible to store the affected_rows globally in the connection. Similarly, Insert_ID( ) is called from the cursor in Python.

We support the iterator protocol, which allows you to iterate through the data using a for-loop:

cursor = conn.Execute('select * from table')
for row in cursor: print row for row in conn.Execute('select id from table'): dofunction(row[0])

And we support associative arrays (dictionaries), where the keys are the field names:

cursor = conn.Execute('select id,name from table')
while not cursor.EOF:
arr = cursor.GetRowAssoc(0) # 0 is lower, 1 is upper-case
print 'id=',arr['id'],' name=',arr['name']

Connection Examples

# Oracle connection
import adodb
conn = adodb.NewADOConnection('oci8')
conn.Connect('tns', 'scott', 'tiger')

# Oracle using connection string
import adodb
conn = adodb.NewADOConnection('oci8://scott:tiger@tns/')

import adodb
conn = adodb.NewADOConnection('mysql')
conn.Connect('server', 'user', 'pwd', 'db')

# MySQL using connection string
import adodb
conn = adodb.NewADOConnection('mysql://user:pwd@server/db')

# PostgreSQL
import adodb
conn = adodb.NewADOConnection('postgres')
conn.Connect('server', 'user', 'pwd', 'db')
conn.Connect('host=server user=user password=pwd dbname=db port=4341')

import adodb
conn = adodb.NewADOConnection('access') # mxodbc required
dsn = "Driver={Microsoft Access Driver (*.mdb)};Dbq=d:\\inetpub\\adodb\\northwind.mdb;"

# ODBC for mssql
import adodb
conn = adodb.NewADOConnection('mssql') # mxodbc required
conn.Connect("Driver={SQL Server};Server=localhost;Database=northwind;")

# sqlite
import adodb
conn = adodb.NewADOConnection('sqlite') # pysqlite required
conn.Connect(database = "c:\\sqlite\\mydata.db")

Other drivers such as "vfp" for foxpro are included.

Function Reference

Connection Class Description
Execute(sql, [params]) Execute sql, returning a cursor object. The optional params is a dictionary that contains the bind variables. All blob fields are automatically and transparently retrieved for you.
SelectLimit(sql, limit, [offset]) Execute sql, retrieving only limit rows, an optional offset from the beginning of the recordset, returning a cursor object.

UpdateBlob(table, field, blob,
      whereclause, blobtype='BLOB')

Executes the equivalent following sql statement:

UPDATE table SET field = blob WHERE whereclause

The blobtype field should be set to either 'BLOB' or 'CLOB'. Any special encoding required for the blob is applied transparently.

UpdateBlobFile(table, field, filepath,
      whereclause, blobtype='BLOB')

Loads the binary file filepath into blob. Then calls UpdateBlob( ).

ErrorMsg( ) Returns last database error message. This function is not thread-safe.
IsConnected( ) Returns boolean indicating whether connected.
qstr(str) Quotes a varchar or char variable.
quote(str) Synonym for qstr( )
GetAll(sql) Execute sql and return 2-dimensional array of tuples, the data recordset.
GetArray(sql) Synonym for GetAll(sql).
GetRow(sql) Execute sql and return first row of recordset as a tuple.
GetOne(sql) Execute sql and return 1 element of first row of recordset.
GetAssoc(sql) Returns a dictionary, with the first columns as the keys to the dictionary. If more than 2 columns are returned, then the dictionary values is a tuple of the 2nd to last columns. If 2 columns are returned, then the 2nd column becomes the dictionary values. If one column is returned, then the values are set to None.
GetDict(sql) Synonym for GetAssoc().
GetCol(sql) Returns the first column of each row as an array.
MetaType(fieldtype) Returns the ADOdb metatype of a native field type.
  • C: character fields that fit into a text input field.
  • X: larger character fields that fit into a textarea.
  • B: Blobs, or Binary Large Objects. Typically images.
  • D: Date field
  • T: Timestamp field
  • L: Logical field (boolean or bit-field)
  • I:  Integer field
  • N: Numeric field. Includes autoincrement, numeric, floating point, real and integer.
MetaColumns(table) Returns a 2-dimensional array containing information on the fields of a table. Each row contains [fieldname, fieldtype, maxlength]. Maxlength can be -1, indicating that the maximum length is not known.

Note that some databases return more information in each row.

DBDate(datetime) Given a Python 2.3 datetime object, convert into a date string acceptable to the database.
DBTimeStamp(datetime) Given a Python 2.3 datetime object, convert into a timestamp string acceptable to the database.
Date(field) Converts a date returned by a select statement into a Python 2.3 datetime object
TimeStamp(field) Converts a timestamp returned by a select statement into a Python 2.3 datetime object
BeginTrans( ) ADOdb defaults to auto-commit mode. Call BeginTrans( ) to start a transaction. This might not be thread-safe.
RollbackTrans( ) Rollback transaction initiated by BeginTrans( ).
CommitTrans( ) Commit transaction initiated by BeginTrans( ).
Close( ) Close database connection. This is optional, as the connection is closed when the object is freed.
Module( ) Returns the DB module object.
Conn( ) Returns DB connection object.
DriverInfo( ) Returns the threadsafety, apilevel and paramstyle values

Connection Class Properties Description
debug Set to 1 to output SQL generated to stdout. Set to 2 to output to stdout as HTML. Set to a filepath (a string) if you want the debug output to be logged into a file.
getLOBs Boolean that determines whether LOBs (large data objects) are loaded automatically. Default is True, autoload. For best performance (especially for cursors with no LOBs), set this to False.
sysDate SQL to generate current date.
sysTimeStamp SQL to generate current timestamp.

Cursor Class Functions Description
RecordCount( ) Number of rows returned by SELECT statement, or number of rows affected by INSERT/UPDATE/DELETE. Returns -1 if not supported.
Affected_Rows( ) Synonym for RecordCount( ).
MoveNext( ) Move to next row of recordset. Returns current EOF value.
FetchRow( ) Retrieves the current row of the recordset, then moves to the next row. The row retrieved is returned as a tuple.
GetRowAssoc(upper=1) Returns the current row as a dictionary, with the key's being the field names. Setting upper = 0 will lower-case the keys. Setting upper=1 will upper-case the keys. Setting upper to any other value, and the keys will be left in the natural case.
Insert_ID( ) Returns last insert id generated by an auto-incrementing field. Only supported by mysql and odbc_mssql drivers currently.

Returns field information from a SELECT statement. The fieldoffset is zero-based, so to retrieve info on the 1st field use FetchField(0). A tuple is returned, consisting of:

(name, type_code,display_size, internal_size, precision, scale,null_ok).

Close( ) Close cursor. This is optional, as the cursor is closed when the object is freed.
Cursor( ) Returns DB cursor object.

Cursor Class Properties Description
fields Property that holds the current row of the recordset as a tuple (or list).
EOF When moving true the recordset, EOF is set to True after we pass the last row.

Error Handling

ADOdb for Python by default relies on the standard Python exception mechanism. Here's how to capture an error:

curs = conn.Execute('select * from badtable'); # table does not exist
print sys.exc_info()[1]; # retrieve the error message returned by database

Alternatively, you can use PHP style ErrorMsg( ) by setting the connection.useExceptions flag to True. ErrorMsg( ) is not thread-safe.

conn.useExceptions = False
curs = conn.Execute('select * from badtable'); # table does not exist
if curs == None:
print conn.ErrorMsg()

Bind Parameters

Python drivers do not use a consistent bind parameter convention. Here is a brief summary of some drivers, obtained by running Connection.DriverInfo( ), which outputs the following to the console:

Driver       = mysql
API Level = 2.0
Param Style = format
Thread Safety= 1 (0=none, 1=module, 2=connections, 3=cursors)

Driver = oci8
API Level = 2.0
Param Style = named
Thread Safety= 2 (0=none, 1=module, 2=connections, 3=cursors)

Driver = postgres
API Level = 2.0
Param Style = pyformat
Thread Safety= 2 (0=none, 1=module, 2=connections, 3=cursors)

Driver = mssql (and all odbc drivers)
API Level = 2.0
Param Style = qmark
The bind format you use is defined in Param Style. From the Python DB API docs:

String constant stating the type of parameter marker
formatting expected by the interface. Possible values are

'qmark' Question mark style,
e.g. '...WHERE name=?'
'numeric' Numeric, positional style,
e.g. '...WHERE name=:1'
'named' Named style,
e.g. '...WHERE name=:name'
'format' ANSI C printf format codes,
e.g. '...WHERE name=%s'
'pyformat' Python extended format codes,
e.g. '...WHERE name=%(name)s'

So for mysql:

	cur_name = "snake"
new_name = "turtle"
connection.Execute ("UPDATE animal SET name = %s WHERE name = %s", (new_name, cur_name))

For oci8, the cx_oracle docs say:

Parameters may be passed as a dictionary or sequence or as keyword arguments. If the arguments are a dictionary, the values will be bound by name and if the arguments are a sequence the values will be bound by position.

For odbc, the paramstyle is qmark, eg.

	connection.Execute('select * from table where id=?',(100,));

Change Log

Todo: add support for sequences. Need to add SelectLimit support for mxodbc derived classes.

2.00 Nov 2005
Added sqlite support, thanks to Glenn Washburn. Improved installation with

1.13 6 May 2005

Fixed oci8 blob bugs. Also updateclob( ) was not working.

Close calls close( ) before freeing connectoin object.

1.12  24 March 2005

Added mxoracle driver (mxodbc for oracle)..

Fixed cursor iterator bug in mxodbc driver.

1.11 23 Feb 2005

Fixed commit code for mysql driver.

1.10 21 Dec 2004

Added sysDate and sysTimeStamp to mysql driver.

Added support for connection strings of the form driver://user:pwd@server/database in NewADOConnection.

DriverInfo( ) now checks paramStyle before thread safety, as some drivers do not implement thread safety.

MySQL insert_id() changed to lastrowid as insert_id() deprecated. Thx Alex Verstraeten.

GetRow() now closes cursor. Thx Guoda Rugeviciute

Postgresql metacolumns() did not work - not a tuple. Fixed. Thx Simon Hedberg.

1.01 30 July 2004
A debugging print in oci8 was removed.

1.00 12 July 2004

GetRowAssoc(), upper = False not implemented correctly. Fixed.

Tested with psyco. Works fine.

0.90 15 Apr 2004

Added ADONewConnection and NewADOConnection.

Added GetAssoc, GetDict, GetArray and GetCol.

Added mxodbc driver. Also vfp, access and mssql classes that inherit from mxodbc. Tested only on Windows.

Added MetaColumns and MetaType support for Oracle, mxODBC, MySQL

Perform explicit close for Get* functions.

0.03 31 Jan 2004

Implemented installer.

Some postgres bugs fixed.

0.02 28 Jan 2004

Revised GetRowAssoc( ).

Fixed use of false - should be False.

0.01 25 Jan 2004
First Release. MySQL, Oracle, PostgreSQL, generic ODBC and ODBC for MSSQL drivers.