path: root/src/interfaces/python/README

PyGreSQL - v2.4: PostgreSQL module for Python
==============================================

0. Copyright notice
===================

  PyGreSQL, version 2.4
  A Python interface for PostgreSQL database.
  Written by D'Arcy J.M. Cain, darcy@druid.net<BR>
  Based heavily on code written by Pascal Andre, andre@chimay.via.ecp.fr.
  Copyright (c) 1995, Pascal ANDRE (andre@via.ecp.fr)

  Permission to use, copy, modify, and distribute this software and its
  documentation for any purpose, without fee, and without a written agreement
  is hereby granted, provided that the above copyright notice and this
  paragraph and the following two paragraphs appear in all copies or in any 
  new file that contains a substantial portion of this file.

  IN NO EVENT SHALL THE AUTHOR BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, 
  SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, 
  ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN IF THE 
  AUTHOR HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

  THE AUTHOR SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED
  TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 
  PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE 
  AUTHOR HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, 
  ENHANCEMENTS, OR MODIFICATIONS.

  Further modifications copyright 1997, 1998 and 1999 by D'Arcy J.M. Cain
  (darcy@druid.net) subject to the same terms and conditions as above.

1. Presentation
===============

1.1. Introduction
-----------------

PostgreSQL is a database system derived from Postgres4.2. It conforms to
(most of) ANSI SQL and offers many interesting capabilities (C dynamic linking
for functions or type definition, etc.). This package is copyright by the
Regents of the University of California, and is freely distributable.

Python is an interpreted programming language. It is object oriented, simple
to use (light syntax, simple and straightforward statements), and has many
extensions for building GUIs, interfacing with WWW, etc. An intelligent web
browser (HotJava like) is currently under development (November 1995), and
this should open programmers many doors. Python is copyrighted by Stichting S
Mathematisch Centrum, Amsterdam, The Netherlands, and is freely distributable.

PyGreSQL is a python module that interfaces to a PostgreSQL database. It
embeds the PostgreSQL query library to allow easy use of the powerful
PostgreSQL features from a Python script.

PyGreSQL 2.0 was developed and tested on a NetBSD 1.3_BETA system.  It is
based on the PyGres95 code written by Pascal Andre, andre@chimay.via.ecp.fr.
I changed the version to 2.0 and updated the code for Python 1.5 and
PostgreSQL 6.2.1.  While I was at it I upgraded the code to use full ANSI 
style prototypes and changed the order of arguments to connect.


1.2. Distribution files
-----------------------

  README       - this file
  Announce     - announcement of this release
  ChangeLog    - changes that affected this package during its history
  pgmodule.c   - the C python module
  pg.py        - PyGreSQL DB class.
  tutorial/    - demos directory
                 Content: basics.py, syscat.py, advanced.py, func.py and
                 pgtools.py.  The samples here have been taken from the
                 PostgreSQL manual and were used for module testing.  They
                 demonstrate some PostgreSQL features.  Pgtools.py is an
                 add-in used for demonstration.

1.3. Installation
-----------------

* You first have to get and build Python and PostgreSQL.

* PyGreSQL is implemented as two parts, a C module labeled _pg and a
  Python wrapper called pg.py.  This changed between 2.1 and 2.2.  This
  should not affect any existing programs but the installation is slightly
  different.

* Find the directory where your 'Setup' file lives (usually ??/Modules) and
  copy or symlink the 'pgmodule.c' file there.

* Add the following line to your Setup file
    _pg  pgmodule.c -I[pgInc] -L[pgLib] -lpq # -lcrypt # needed on some systems
  where:
    [pgInc] = path of the PostgreSQL include 
    [pgLib] = path of the PostgreSQL libraries 
  Some options may be added to this line:
    -DNO_DEF_VAR  - no default variables support
    -DNO_DIRECT   - no direct access methods
    -DNO_LARGE    - no large object support
    -DNO_PQSOCKET - if running an older PostgreSQL

  Define NO_PQSOCKET if you are using a version of PostgreSQL before 6.4
  that does not have the PQsocket function.  The other options will be
  described in the next sections.

* If you want a shared module, make sure that the "*shared*" keyword is
  uncommented and add the above line below it.  You used to need to install
  your shared modules with "make sharedinstall but this no longer seems
  to be true."

* Copy pg.py to the lib directory where the rest of your modules are.  For
  example, that's /usr/local/lib/Python on my system.

* Do 'make -f Makefile.pre.in boot' and do 'make && make install'

* For more details read the documentation at the top of Makefile.pre.in

* If you are on NetBSD, look in the packages directory under databases.  If
  it isn't there yet, it should be there shortly.  You can also pick up the
  package files from ftp://ftp.druid.net/pub/distrib/pygresql.pkg.tgz.
  There is also a package in the FreeBSD ports collection but as I write
  this it is at version 2.1.  I will try to get that updated as well.

* For Linux installation look at README.linux


1.4. Where to get ... ?
-----------------------

The home sites of the different packages are:

  - Python:     http://www.python.org/
  - PosgreSQL:  http://www.PostgreSQL.org/
  - PyGreSQL:   http://www.druid.net/pygresql/

A Linux RPM can be picked up from ftp://www.eevolute.com/pub/python/.
A NetBSD package thould be in the distribution soon and is available
at ftp://ftp.druid.net/pub/distrib/pygresql.pkg.tgz.

1.5. Information and support
----------------------------

If you need information about these packages please check their web sites:

  - Python:     http://www.python.org/
  - PostgreSQL: http://www.postgresql.org/
  - PyGres95:   http://www.via.ecp.fr/via/products/pygres.html
  - PyGreSQL:   http://www.druid.net/pygresql/

For support:

  - Python:      newgroup comp.lang.python
  - PostgreSQL:  mailing list (see package documentation for information)
  - PyGres95:    contact me (andre@via.ecp.fr) for bug reports, ideas, remarks
                 I will try to answer as long as my free time allow me to do 
                 that.
  - PyGreSQL:    contact me (darcy@druid.net) concerning the changes to 2.x.


2. Programming information
==========================

This module defines three objects: the pgobject that handles the connection 
and all the requests to the database, the pglargeobject that handles
all the accesses to Postgres large objects and pgqueryobject that handles
query results.

If you want to see a simple example of the use of some of these functions,
see http://www.druid.net/rides/ where I have a link at the bottom to the
actual Python code for the page.

2.1. pg module description
----------------------------

The module defines only a few methods that allow to connect to a database and
to allow to define "default variables" that override the environment variables
used by PostgreSQL. 

These "default variables" were designed to allow you to handle general 
connection parameters without heavy code in your programs. You can prompt the
user for a value, put it in the default variable, and forget it, without 
having to modify your environment. The support for default variables can be
disabled by setting the -DNO_DEF_VAR option in the Python Setup file. Methods
relative to this are specified by te tag [DV].

All variables are set to None at module initialization, specifying that 
standard environment variables should be used.

  2.1.1. connect - opens a pg connection
  ----------------------------------------

  Syntax:      
    connect(dbname, host, port, opt, tty, user, passwd)
  Parameters: 
    dbname        - name of connected database (string/None)
    host          - name of the server host (string/None)
    port          - port used by the database server (integer/-1)
    opt           - connection options (string/None)
    tty           - debug terminal (string/None)
	user          - PostgreSQL user (string/None)
    passwd        - password for user (string/None)
  Return type:
    pgobject      - the object handling the connection
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments
    SyntaxError   - duplicate argument definition 
    pg.error      - some error occurred during pg connection definition
    (+ all exceptions relative to object allocation)
  Description:
    This method opens a connection to a specified database on a given
    PostgreSQL server. You can use keywords here, as described in the
    Python tutorial; 
    the names of the keywords are the name of the parameters given in the 
    syntax line. For a precise description of the parameters, please refer to 
    the PostgreSQL user manual.

  2.1.2. get_defhost, set_defhost - default server host name handling [DV]
  ------------------------------------------------------------------------

  Syntax: get_defhost()
  Parameters: 
    none
  Return type:
    string, None  - default host specification
  Exceptions raised:
    SyntaxError   - too many arguments
  Description:
    This method returns the current default host specification, or None if the
    environment variables should be used. Environment variables won't be looked
    up.

  Syntax: set_defhost(host)
  Parameters:
    host          - new default host (string/None)
  Return type:
    string, None  - previous default host specification
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments
  Description:
    This methods sets the default host value for new connections. If None is
    supplied as parameter, environment variables will be used in future 
    connections. It returns the previous setting for default host.

  2.1.3. get_defport, set_defport - default server port handling [DV]
  -------------------------------------------------------------------

  Syntax: get_defport()
  Parameters: none
  Return type:
    integer, None - default port specification
  Exceptions raised:
    SyntaxError   - too many arguments
  Description: 
    This method returns the current default port specification, or None if
    the environment variables should be used. Environment variables won't
    be looked up.

  Syntax: set_defport(port)
  Parameters:
    port          - new default port (integer/-1)
  Return type:
    integer, None - previous default port specification
  Description:
    This methods sets the default port value for new connections. If -1 is
    supplied as parameter, environment variables will be used in future 
    connections. It returns the previous setting for default port.

  2.1.4. get_defopt, set_defopt - default connection options handling [DV]
  ------------------------------------------------------------------------

  Syntax: get_defopt()
  Parameters: none
  Return type:
    string, None  - default options specification
  Exceptions raised:
    SyntaxError   - too many arguments
  Description:
    This method returns the current default connection options  specification,
    or None if the environment variables should be used. Environment variables 
    won't be looked up.

  Syntax: set_defopt(options)
  Parameters:
    options       - new default connection options (string/None)
  Return type:
    string, None  - previous default options specification
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments
  Description:
    This methods sets the default connection options value for new connections.
    If None is supplied as parameter, environment variables will be used in 
    future connections. It returns the previous setting for default options.

  2.1.5. get_deftty, set_deftty - default connection debug tty handling [DV]
  --------------------------------------------------------------------------

  Syntax: get_deftty()
  Parameters: none
  Return type:
    string, None  - default debug terminal specification
  Exceptions raised:
    SyntaxError   - too many arguments
  Description:
    This method returns the current default debug terminal specification, or 
    None if the environment variables should be used. Environment variables 
    won't be looked up.

  Syntax: set_deftty(terminal)
  Parameters:
    terminal      - new default debug terminal (string/None)
  Return type:
    string, None  - previous default debug terminal specification
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments
  Description:
    This methods sets the default debug terminal value for new connections. If
    None is supplied as parameter, environment variables will be used in future
    connections. It returns the previous setting for default terminal.

  2.1.6. get_defbase, set_defbase - default database name handling [DV]
  ---------------------------------------------------------------------

  Syntax: get_defbase()
  Parameters: none
  Return type:
    string, None  - default database name specification
  Exceptions raised:
    SyntaxError   - too many arguments
  Description:
    This method returns the current default database name specification, or 
    None if the environment variables should be used. Environment variables 
    won't be looked up.

  Syntax: set_defbase(base)
  Parameters:
    base          - new default base name (string/None)
  Return type:
    string, None  - previous default database name specification
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments
  Description:
    This method sets the default database name value for new connections. If 
    None is supplied as parameter, environment variables will be used in 
    future connections. It returns the previous setting for default host.

  2.1.7. Module constants
  -----------------------

  Some constants are defined in the module dictionary. They are intended to be
used as parameters for methods calls. You should refer to PostgreSQL user 
manual for more information about them. These constants are:

  - large objects access modes, used by (pgobject.)locreate and 
    (pglarge.)open: (pg.)INV_READ, (pg.)INV_WRITE, (pg.)INV_ARCHIVE
  - positional flags, used by (pglarge.)seek: (pg.)SEEK_SET, 
    (pg.)SEEK_CUR, (pg.)SEEK_END.
  - version and __version__ constants that give the current version.

  2.1.9. 
  2.1.10. Miscellaneous attributes

  The following methods return information about the current connection.

  - 
2.2. pgobject description
---------------------------

  This object handle a connection to a PostgreSQL database. It embeds and 
hides all the parameters that define this connection, thus just leaving really
significant parameters in function calls.
  Some methods give direct access to the connection socket. They are specified
by the tag [DA]. DO NOT USE THEM UNLESS YOU REALLY KNOW WHAT YOU ARE DOING. If
you prefer disabling them, set the -DNO_DIRECT option in the Python Setup file.
  Some other methods give access to large objects (refer to PostgreSQL user
manual for more information about these). if you want to forbid access to these
from the module, set the -DNO_LARGE option in the Python Setup file. These 
methods are specified by the tag [LO].

  2.2.1. query - executes a SQL command string
  --------------------------------------------

  Syntax: query(command)
  Parameters:
    command       - SQL command (string)
  Return type:
    pgqueryobject, None    - result values
  Exceptions raised:
    TypeError     - bad argument type, or too many arguments.
    ValueError    - empty SQL query
    pg.error      - error during query processing, or invalid connection
  Description:
    This method simply sends a SQL query to the database. If the query is
    an insert statement, the return value is the OID of the newly
    inserted row.  If it is otherwise a query that does not return a result
    (ie. is not a some kind of SELECT statement), it returns None.
    Otherwise, it returns a pgqueryobject that can be accessed via the
    getresult method or printed.

  pgqueryobject methods
  ---------------------

    2.2.1.1. getresult - gets the values returned by the query
    -------------------------------------------------------------

    Syntax: getresult()
    Parameters: none
    Return type:
      list          - result values
    Exceptions raised:
      SyntaxError   - too many parameters
      pg.error      - invalid previous result
    Description:
      This method returns the list of the values returned by the query.
      More information about this result may be get using listfields,
      fieldname and fiednum methods.

    2.2.1.2. dictresult - like getresult but returns list of dictionaries
    ---------------------------------------------------------------------

    Syntax: dictresult()
    Parameters: none
    Return type:
      list          - result values as a dictionary
    Exceptions raised:
      SyntaxError   - too many parameters
      pg.error      - invalid previous result
    Description:
      This method returns the list of the values returned by the query
      with each tuple returned as a dictionary with the field names
      used as the dictionary index.


    2.2.1.3. listfields - lists the fields names of the previous query result
    -----------------------------------------------------------------------

    Syntax: listfields()
    Parameters: none
    Return type:
      list          - fields names
    Exceptions raised:
      SyntaxError   - too many parameters
      pg.error      - invalid previous result, or invalid connection
    Description:
      This method returns the list of names of the fields defined for the
      query result. The fields are in the same order as the result values.
  
    2.2.1.4. fieldname, fieldnum - field name-number conversion
    ---------------------------------------------------------

    Syntax: fieldname(i)
    Parameters:
      i              - field number (integer)
    Return type:
      string         - field name
    Exceptions raised:
      TypeError      - bad parameter type, or too many parameters
      ValueError     - invalid field number
      pg.error       - invalid previous result, or invalid connection
    Description:
      This method allows to find a field name from its rank number. It can be 
      useful for displaying a result. The fields are in the same order than the
      result values.

    Syntax: fieldnum(name)
    Parameters:
      name           - field name (string)
    Return type:
      integer        - field number
    Exceptions raised:
      TypeError      - bad parameter type, or too many parameters
      ValueError     - unknown field name
      pg.error       - invalid previous result, or invalid connection
    Description:
      This method returns a field number from its name.  It can be used to
      build a function that converts result list strings to their correct
      type, using a hardcoded table definition.  The number returned is the
      field rank in the result values list.

    2.2.1.5 ntuples - return number of tuples in query object
    ---------------------------------------------------------

    Syntax: ntuples()
    Parameters: None
    Return type: integer
    Description:
      This method returns the number of tuples found in a query.


  2.2.2. reset - resets the connection
  ------------------------------------

  Syntax: reset()
  Parameters: None
  Return type: None
  Exceptions raised:
    TypeError     -  too many (any) arguments
  Description:
    This method resets the current database.


  2.2.3. close - close the database connection
  --------------------------------------------

  Syntax: close()
  Parameters: none
  Return type: None
  Exceptions raised:
    TypeError     -  too many (any) arguments
  Description:
    This method closes the database connection.  The connection will
    be closed in any case when the connection is deleted but this
    allows you to explicitly close it.  It is mainly here to allow
    the DB-SIG API wrapper to implement a close function.


  2.2.4. fileno - returns the socket used to connect to the database
  ------------------------------------------------------------------

  Syntax: fileno()
  Parameters: none
    Exceptions raised:
    TypeError     -  too many (any) arguments
  Description:
    This method returns the underlying socket id used to connect
    to the database.  This is useful for use in select calls, etc.
    Note:  This function depends on having a recent version of the
    database.  See "-DNO_PQSOCKET" described above.


  2.2.5. getnotify - gets the last notify from the server
  -------------------------------------------------------

  Syntax: getnotify()
  Parameters: none
  Return type:
    tuple, None    - last notify from server
  Exceptions raised:
    SyntaxError    - too many parameters
    pg.error       - invalid connection
  Description:
    This methods try to get a notify from the server (from the SQL statement 
    NOTIFY). If the server returns no notify, the methods returns None. 
    Otherwise, it returns a tuple (couple) (relname, pid), where relname is the
    name of the notify and pid the process id of the connection that triggered 
    the notify.  Remember to do a listen query first otherwise getnotify
    will always return None.

  2.2.6. inserttable - insert a list into a table
  -----------------------------------------------

  Syntax: inserttable(table, values)
  Parameters:
    table          - the table name (string)
    values         - list of rows values (list)
  Return type:
    None
  Exception raised:
    pg.error       - invalid connection
    TypeError      - bad argument type, or too many arguments
  Description:
    This method allow to quickly insert large blocks of data in a table: it
    inserts the whole values list into the given table. The list is a list of
    tuples/lists that define the values for each inserted row. The rows values
    may contain string, integer, long or double (real) values. 
    BE VERY CAREFUL: this method doesn't typecheck the fields according to the
    table definition; it just look whether or not it knows how to handle such
    types.

  2.2.7. putline - writes a line to the server socket [DA]
  --------------------------------------------------------

  Syntax: putline(line)
  Parameters:
    line           - line to be written (string)
  Return type:
    None
  Exceptions raised:
    pg.error       - invalid connection
    TypeError      - bad parameter type, or too many parameters
  Description: 
    This method allows to directly write a string to the server socket.

  2.2.8. getline - gets a line from server socket [DA]
  ----------------------------------------------------

  Syntax: getline()
  Parameters: none
  Return type:
    string         - the line read
  Exceptions raised:
    pg.error       - invalid connection
    SyntaxError    - too many parameters
  Description:
    This method allows to directly read a string from the server socket.

  2.2.9. endcopy - synchronizes client and server [DA]
  ----------------------------------------------------

  Syntax: endcopy()
  Parameters: none
  Return type:
    None
  Exceptions raised:
    pg.error       - invalid connection
    SyntaxError    - too many parameters
  Description:
    The use of direct access methods may desynchonize client and server. This
    method ensure that client and server will be synchronized.

  2.2.10. locreate - creates of large object in the database [LO]
  ---------------------------------------------------------------

  Syntax: locreate(mode)
  Parameters:
    mode           - large object create mode
  Return type:
    pglarge        - object handling the postgres large object
  Exceptions raised:
    pg.error       - invalid connection, or creation error
    TypeError      - bad parameter type, or too many parameters
  Description:
    This method creates a large object in the database. The mode can be defined
    by OR-ing the constants defined in the pg module (INV_READ, INV_WRITE and
    INV_ARCHIVE). Please refer to PostgreSQL user manual for a description of
    the mode values.

  2.2.11. getlo - builds a large object from given oid [LO]
  ---------------------------------------------------------

  Syntax: getlo(oid)
  Parameters:
    oid            - oid of the existing large object (integer)
  Return type:
    pglarge        - object handling the postgres large object
  Exceptions raised:
    pg.error       - invalid connection
    TypeError      - bad parameter type, or too many parameters
    ValueError     - bad oid value (0 is invalid_oid)
  Description:
    This method allows to reuse a formerly created large object through the 
    pglarge interface, providing the user have its oid.

  2.2.12. loimport - import a file to a postgres large object [LO]
  ----------------------------------------------------------------

  Syntax: loimport(name)
  Parameters:
    name           - the name of the file to be imported (string)
  Return type:
    pglarge        - object handling the postgres large object
  Exceptions raised:
    pg.error       - invalid connection, or error during file import
    TypeError      - bad argument type, or too many arguments
  Description: 
    This methods allows to create large objects in a very simple way. You just 
    give the name of a file containing the data to be use.

  2.2.13. pgobject attributes
  -----------------------------

  Every pgobject defines a set of read-only attributes that describe the 
connection and its status. These attributes are:
  host             - the hostname of the server (string)
  port             - the port of the server (integer)
  db               - the selected database (string)
  options          - the connection options (string)
  tty              - the connection debug terminal (string)
  user             - the username on the database system (string)
  status           - the status of the connection (integer: 1 - OK, 0 - BAD)
  error            - the last warning/error message from the server (string)

2.3. pglarge description
--------------------------

  This object handles all the request concerning a postgres large object. It 
embeds and hides all the 'recurrent' variables (object oid and connection), 
exactly in the same way pgobjects do, thus only keeping significant 
parameters in function calls. It keeps a reference to the pgobject used for
its creation, sending requests though with its parameters. Any modification but
dereferencing the pgobject will thus affect the pglarge object. 
Dereferencing the initial pgobject is not a problem since Python won't 
deallocate it before the large object dereference it.
  All functions return a generic error message on call error, whatever the 
exact error was. The 'error' attribute of the object allow to get the exact 
error message.

  2.3.1. open - opens a large object
  ----------------------------------

  Syntax: open(mode)
  Parameters:
    mode           - open mode definition (integer)
  Return type:
    None
  Exceptions raised:
    pg.error       - invalid connection
    TypeError      - bad parameter type, or too many parameters
    IOError        - already opened object, or open error
  Description:
    This method opens a large object for reading/writing, in the same way than
    the UNIX open() function. The mode value can be obtained by OR-ing the 
    constants defined in the pgmodule (INV_READ, INV_WRITE).

  2.3.2. close - closes a large object
  ------------------------------------

  Syntax: close()
  Parameters: none
  Return type:
    None
  Exceptions raised:
    pg.error       - invalid connection
    SyntaxError    - too many parameters
    IOError        - object is not opened, or close error
  Description:
    This method closes a previously opened large object, in the same way than 
    the UNIX close() function.

  2.3.4. read, write, tell, seek, unlink - file like large object handling
  ------------------------------------------------------------------------

  Syntax: read(size)
  Parameters:
    size           - maximal size of the buffer to be read
  Return type:
    sized string   - the read buffer
  Exceptions raised:
    pg.error       - invalid connection or invalid object
    TypeError      - bad parameter type, or too many parameters
    IOError        - object is not opened, or read error
  Description:
    This function allows to read data from a large object, starting at current
    position.

  Syntax: write(string)
  Parameters:
    (sized) string - buffer to be written
  Return type:
    None
  Exceptions raised:
    pg.error       - invalid connection or invalid object
    TypeError      - bad parameter type, or too many parameters
    IOError        - object is not opened, or write error
  Description:
    This function allows to write data to a large object, starting at current 
    position.

  Syntax: seek(offset, whence)
  Parameters:
    offset          - position offset
    whence          - positional parameter
  Return type:
    integer         - new position in object
  Exception raised:
    pg.error        - invalid connection or invalid object
    TypeError       - bad parameter type, or too many parameters
    IOError         - object is not opened, or seek error
  Description:
    This method allows to move the position cursor in the large object. The 
    whence parameter can be obtained by OR-ing the constants defined in the 
    pg module (SEEK_SET, SEEK_CUR, SEEK_END).

  Syntax: tell()
  Parameters: none
  Return type:
    integer         - current position in large object
  Exception raised:
    pg.error        - invalid connection or invalid object
    SyntaxError     - too many parameters
    IOError         - object is not opened, or seek error
  Description:
    This method allows to get the current position in the large object.

  Syntax: unlink()
  Parameter: none
  Return type:
    None
  Exception raised:
    pg.error        - invalid connection or invalid object
    SyntaxError     - too many parameters
    IOError         - object is not closed, or unlink error
  Description:
    This methods unlinks (deletes) the postgres large object.

  2.3.5. size - gives the large object size
  -----------------------------------------

  Syntax: size()
  Parameters: none
  Return type:
    integer         - large object size
  Exceptions raised:
    pg.error        - invalid connection or invalid object
    SyntaxError     - too many parameters
    IOError         - object is not opened, or seek/tell error
  Description:
    This (composite) method allows to get the size of a large object. Currently
    the large object needs to be opened. It was implemented because this 
    function is very useful for a WWW interfaced database.

  2.3.6. export - saves a large object to a file
  ----------------------------------------------

  Syntax: export(name)
  Parameters:
    name            - file to be created
  Return type:
    None
  Exception raised:
    pg.error        - invalid connection or invalid object
    TypeError       - bad parameter type, or too many parameters
    IOError         - object is not closed, or export error
  Description:
    This methods allows to dump the content of a large object in a very simple
    way. The exported file is created on the host of the program, not the 
    server host.

  2.3.7. Object attributes
  ------------------------

  pglarge objects define a read-only set of attributes that allow to get some
information about it. These attributes are:
  oid                - the oid associated with the object
  pgcnx              - the pgobject associated with the object
  error              - the last warning/error message of the connection
BE CAREFUL:  in multithreaded environments, 'error' may be modified by another
thread using the same pgobject. Remember these object are shared, not 
duplicated. You should provide some locking to be able if you want to check 
this.
  The oid attribute is very interesting because it allow you reuse the oid 
later, creating the pglarge object with a pgobject getlo() method call.


3. The pg wrapper
================

The previous functions are wrapped in a module called pg.  The module
has a class called DB.  The above functions are also included in the
name space so it isn't necessary to import both modules.  The preferred
way to use this module is as follows.

from pg import DB
db = DB(...) # See description of the initialization method below.

The following describes the methods and variables of this class.


 3.1. Initialization
 -------------------
 The DB class is initialized with the same arguments as the connect
 method described in section 2.  It also initializes a few internal
 variables.  The statement 'db = DB()' will open the local database
 with the name of the user just like connect() does.

 3.2. pkey
 ---------
 Syntax:
   pkey(table)
 Parameters:
   table - name of table
 Returns:
   Name of field which is the primary key of the table.
 Description:
   This method returns the primary key of a table.  Note that this raises
   an exception if the table doesn't have a primary key.

 3.3. get_databases - get list of databases in the system
 --------------------------------------------------------
 Syntax: get_databases()
 Parameters: none
 Returns: list of databases in the system
 Description:
   Although you can do this with a simple select, it is added here for
   convenience

 3.4. get_tables - get list of tables in connected database
 ----------------------------------------------------------
 Syntax: get_tables() 
 Parameters: none
 Returns: list of tables in connected database

 3.5. get_attnames
 -----------------
 Syntax: 
   get_attnames(table)       
 Parameters: 
   table - name of table
 Returns:
   List of attribute names
 Description:
   Given the name of a table, digs out the list of attribute names.

 3.6. get - get a tuple from a database table
 --------------------------------------------
 Syntax: 
   get(table, arg, [keyname])
 Parameters:
   table - name of table
   arg - either a dictionary or the value to be looked up
   keyname - name of field to use as key (optional)
 Returns:
   A dictionary mapping attribute names to row values.
 Description:
   This method is the basic mechanism to get a single row.  It assumes
   that the key specifies a unique row.  If keyname is not specified
   then the primary key for the table is used.  If arg is a dictionary
   then the value for the key is taken from it and it is modified to
   include the new values, replacing existing values where necessary.
   The oid is also put into the dictionary but in order to allow the
   caller to work with multiple tables, the attribute name is munged
   to make it unique.  It consists of the string "oid_" followed by
   the name of the table.


 3.7. insert - insert a tuple into a database table
 --------------------------------------------------
 Syntax:
   insert(table, a)
 Parameters:
   table - name of table
   a - a dictionary of values
 Returns:
   The OID of the newly inserted row.
 Description:
   This method inserts values into the table specified filling in the
   values from the dictionary.  It then reloads the dictionary with the
   values from the database.  This causes the dictionary to be updated
   with values that are modified by rules, triggers, etc.


 3.8. update
 -----------
 Syntax:
   update(table, a)             
 Parameters:
   table - name of table
   a - a dictionary of values
 Returns:
   A dictionary with the new row
 Description:
   Similar to insert but updates an existing row.  The update is based
   on the OID value as munged by get.  The array returned is the
   one sent modified to reflect any changes caused by the update due
   to triggers, rules, defaults, etc.

 3.9. clear
 ----------
 Syntax:
   clear(table, [a])
 Parameters:
   table - name of table
   a - a dictionary of values
 Returns:
   A dictionary with an empty row
 Description:
   This method clears all the attributes to values determined by the types.
   Numeric types are set to 0, dates are set to 'TODAY' and everything
   else is set to the empty string.  If the array argument is present,
   it is used as the array and any entries matching attribute names
   are cleared with everything else left unchanged.

 3.8. delete
 -----------
 Syntax:
   delete(table, a)
 Parameters:
   table - name of table
   a - a dictionary of values
 Returns:
   None
 Description:
   This method deletes the row from a table.  It deletes based on the OID
   as munged as described above.


4. Future directions
====================

The large object and direct access functions need much more attention.

I want to add a DB-SIG API wrapper around the underlying module.  This
will be in 3.0.