From: | Adrian Klaver <adrian(dot)klaver(at)aklaver(dot)com> |
---|---|
To: | Filipe Pina <filipe(dot)pina(at)impactzero(dot)pt> |
Cc: | Postgresql General <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: pl/python composite type array as input parameter |
Date: | 2015-06-02 22:00:16 |
Message-ID: | 556E2770.4050105@aklaver.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On 06/02/2015 12:33 PM, Filipe Pina wrote:
> Basically, in an (maybe-over)simplified example:
>
> CREATE OR REPLACE FUNCTION add_transaction(transaction core_transaction)
> RETURNS integer AS $$
> DECLARE
> transaction2 core_transaction;
> BEGIN
> transaction.field1 := 'lapse’;
> transaction2.field2 := transaction.field2;
> transaction2.field1 := 'lapse2’;
> INSERT INTO core_transaction VALUES(transaction.*);
> INSERT INTO core_transaction VALUES(transaction2.*);
> RETURN 1;
> END
> $$
> LANGUAGE plpgsql;
>
> So, I wanted to do the same in plpython…
>
> CREATE OR REPLACE FUNCTION add_transaction(transaction core_transaction)
> RETURNS integer AS $$
> transaction['field1'] = ‘lapse’
> transaction2 = { ‘field1’: ‘lapse2’, ‘field2’: transaction[‘field1’] }
>
> # not this but something that would work without enumericating all
> columns/fields
> pl = plpy.execute('INSERT INTO core_transaction VALUES(transaction.*)’)
> pl = plpy.execute('INSERT INTO core_transaction
> VALUES(transaction2.*)')
> return 1
> END
> $$
> LANGUAGE plpythonu;
Yea, I do not see a way of doing that. plpgsql is more tightly coupled
to Postgres then plpythonu, so you get a lot more shortcuts. This why I
tend to use plpgsql even though I prefer programming in Python. That
being said, the feature set of plpythonu has been extended a good deal
over the last couple of Postgres versions and I would expect that to
continue.
>
>
>
>> On 02/06/2015, at 15:51, Adrian Klaver <adrian(dot)klaver(at)aklaver(dot)com
>> <mailto:adrian(dot)klaver(at)aklaver(dot)com>> wrote:
>>
>> On 06/02/2015 03:10 AM, Filipe Pina wrote:
>>> HI Adrian,
>>>
>>> I had a typo in the email:
>>>
>>> INSERT INTO my_table VALUES(my_table.*);
>>>
>>> was actually
>>>
>>> INSERT INTO my_table VALUES(my_var.*);
>>
>> Aah, that is different:)
>>
>>>
>>> So I meant to insert the variable I had in memory (dict representing
>>> a row), not the rows from the table..
>>
>> So where is the variable getting its data?
>>
>> Or can we see a simple example of what you are trying to do?
>>
>>>
>>>> On 02/06/2015, at 01:44, Adrian Klaver <adrian(dot)klaver(at)aklaver(dot)com
>>>> <mailto:adrian(dot)klaver(at)aklaver(dot)com>> wrote:
>>>>
>>>> On 06/01/2015 07:42 AM, Filipe Pina wrote:
>>>>> Thanks for the reply anyway, it's a pity though, it'd be useful..
>>>>>
>>>>> Another bump I've found along the pl/python road: insert ROWTYPE in
>>>>> table..
>>>>> Maybe you have some hint on that? :)
>>>>>
>>>>> So, in PLPGSQL I can:
>>>>>
>>>>> DECLARE
>>>>> my_var my_table;
>>>>> BEGIN
>>>>> my_var.col1 := 'asd';
>>>>> INSERT INTO my_table VALUES(my_table.*);
>>>>> END;
>>>>>
>>>>> How would I do something like that in pl/python?
>>>>>
>>>>> First, how to declare a ROW-TYPE variable, as they're all python
>>>>> mappings?
>>>>>
>>>>> my_var = { 'col1': 'asd' } enough? it'd would miss all the other
>>>>> columns...
>>>>>
>>>>> Second, how to insert it?
>>>>>
>>>>> plpy.prepare and .execute say they don't support composite types, so I
>>>>> cannot simply pass
>>>>>
>>>>> pl = plpy.prepare('INSERT INTO core_customer VALUES ($1)',
>>>>> ['my_table'])
>>>>>
>>>>> Any workarounds for this? (meaning I wouldn't have to specify any
>>>>> columns in the insert statement)
>>>>
>>>> http://www.postgresql.org/docs/9.4/interactive/sql-insert.html
>>>>
>>>> pl = plpy.prepare('INSERT INTO core_table SELECT * FROM my_table')
>>>>
>>>>>
>>>>> Thanks
>>>>>
>>>>> On Sex, Mai 29, 2015 at 2:00 , Peter Eisentraut <peter_e(at)gmx(dot)net>
>>>>> wrote:
>>>>>> On 5/18/15 10:52 AM, Filipe Pina wrote:
>>>>>>
>>>>>> But one of the functions I need to create needs to accept an array
>>>>>> of records.
>>>>>>
>>>>>> PL/Python doesn't support that. Some more code needs to be written to
>>>>>> support that. You did everything correctly. I don't know of a good
>>>>>> workaround.
>>>>
>>>>
>>>> --
>>>> Adrian Klaver
>>>> adrian(dot)klaver(at)aklaver(dot)com
>>>
>>>
>>
>>
>> --
>> Adrian Klaver
>> adrian(dot)klaver(at)aklaver(dot)com <mailto:adrian(dot)klaver(at)aklaver(dot)com>
>
--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com
From | Date | Subject | |
---|---|---|---|
Next Message | Gavin Flower | 2015-06-02 22:24:53 | Re: TRIGGER TRUNCATE -- CASCADE or RESTRICT |
Previous Message | Fabio Ugo Venchiarutti | 2015-06-02 21:30:16 | Re: Minor revision downgrade (9.2.11 -> 9.2.10) |