inserting/retriving dates in psycopg

Re: inserting/retriving dates in psycopg


Michele Simionato wrote:[color=blue]
> Look at this example:
>[color=green][color=darkred]
> >>> import psycopg
> >>> psycopg.__versi on__[/color][/color]
> '1.1.19'[color=green][color=darkred]
> >>> import datetime
> >>> today = datetime.dateti me.today()
> >>> co = psycopg.connect ('')
> >>> cu = co.cursor()[/color][/color]
>[color=green][color=darkred]
> >>> cu.execute('CRE ATE TABLE example (date date)')
> >>> cu.execute("INS ERT into example VALUES (%s)", (today,))[/color][/color]
> Traceback (most recent call last):
> File "/usr/lib/python2.4/doctest.py", line 1243, in __run
> compileflags, 1) in test.globs
> File "<doctest __main__[6]>", line 1, in ?
> cu.execute("INS ERT into example VALUES (%s)", (today,))
> ProgrammingErro r: ERROR: syntax error at or near "17" at character 40
>
> INSERT into example VALUES (2006-01-04 17:19:03.354615 )
>
> Is there a smart way of solving this, without stripping the date by
> hand?
> Also, if I insert the date as a string
>[color=green][color=darkred]
> >>> cu.execute("INS ERT into example VALUES ('2006-01-04')")
> >>> co.commit()[/color][/color]
>
> I get back a DateTime object (I think a mxDateTime object)
>[color=green][color=darkred]
> >>> cu.execute("SEL ECT * from example")
> >>> d = cu.fetchone()[0]
> >>> print type(d)[/color][/color]
> <type 'DateTime'>
>
> and NOT a datetime.dateti me.
>
> So I need an adaptation mechanism; alternatively it would be enough for
> me to be able
> to redefine the __str__ representation of psycopg DateTime objects (
> which are defined
> at C level, so I cannot just override the __str__ method).
>
> Any hints? I am sure there is a custom way to do this.
>
> Michele Simionato[/color]

I am no expert, but as I have not seen any other replies, I will
explain how I do it.

Firstly, I seem to recall that when you install psycopg, it tries to
guess whether to compile it for mx.DateTime or the built-in datetime
module, based on version of Python, whether mx is installed, etc. I am
sure it is possible to force it to choose your preferred one by
modifying the .cfg file (I think).

Assuming it is set up correctly for the datetime module, this is how I
insert and retrieve.

If I have a datetime.dateti me instance called dat, I insert it using
str(dat).

I found retrieving a bit trickier, as I have standardised on using the
datetime.dateti me type throughout my app, but I have a mixture of DATE
and TIMESTAMP types in my database. Therefore I convert the result as
follows -

import datetime as dt

def DbToDate(dat):
if isinstance(dat, dt.datetime):
return dat # already in datetime format
if isinstance(dat, dt.date):
return dt.datetime.com bine(dat,dt.tim e(0)) # convert to
datetime

HTH

Frank Millman

Re: inserting/retriving dates in psycopg

Frank Millman:
[color=blue]
> import datetime as dt[/color]
[color=blue]
> def DbToDate(dat):
> if isinstance(dat, dt.datetime):
> return dat # already in datetime format
> if isinstance(dat, dt.date):
> return dt.datetime.com bine(dat,dt.tim e(0)) # convert to datetime[/color]

This is exactly the type checking I would like to avoid :-/

Michele Simionato

Re: inserting/retriving dates in psycopg


Michele Simionato wrote:[color=blue]
> Frank Millman:
>[color=green]
> > import datetime as dt[/color]
>[color=green]
> > def DbToDate(dat):
> > if isinstance(dat, dt.datetime):
> > return dat # already in datetime format
> > if isinstance(dat, dt.date):
> > return dt.datetime.com bine(dat,dt.tim e(0)) # convert to datetime[/color]
>
> This is exactly the type checking I would like to avoid :-/
>
> Michele Simionato[/color]

psycopg returns a datetime.dateti me object from a TIMESTAMP column, and
a datetime.date object from a DATE column. You should not have to do
any type checking unless you are doing something odd, like I am, and
wanting to convert them all into datetime.dateti me objects.

Perhaps if you explain what you are trying to do, I may be able to
suggest something.

Frank

Re: inserting/retriving dates in psycopg

Frank Millan:[color=blue]
> Perhaps if you explain what you are trying to do, I may be able to
> suggest something.[/color]

I am looking for an adaptation/type cast mechanism and looking at the
sources I think I have
found it in doc/examples/usercast.py. I am doing some experiment now
....

Michele Simionato