I probably am doing something wrong, but you have yet to simulate the case
originally described, which is a loop that fails after about 10 iterations
with data that is not discernibly different.
On Wednesday, August 22, 2012 4:32:39 PM UTC-4, Niphlod wrote:
>
> You are doing something wrong....a different model, some strange value
> type...etc.
> Tested with dict automatic unpacking.
>
> >>> movie = dict(name='alien4', genres=[1,2])
> >>> db.movies.validate_and_insert(**movie)
> <Row {'errors': <Row {}>, 'id': 4}>
> >>> movie = dict(name='alien4', genres=[1,7])
> >>> db.movies.validate_and_insert(**movie)
> <Row {'errors': <Row {'genres': 'value not in database'}>, 'id': None}>
>
>
> Could you at least print one movie you're trying to insert that fails in
> your script ? standard print, simplejson.dump()ed and pprint.pprint()ed,
> directly before "validate_and_insert(**movie)". Just to pinpoint the issue.
>
> On Wednesday, August 22, 2012 10:10:40 PM UTC+2, Mike Girard wrote:
>>
>> Yeah, I get the references directly from the genre/person tables before
>> inserting them in movie. I also test to make sure the item being inserted
>> is a list with data. Printing contents after the error shows nothing odd.
>>
>> I have tested this in both Postgres and SQLlite. Same thing happens in
>> both.
>>
>>
>>
>> On Wednesday, August 22, 2012 3:45:31 PM UTC-4, villas wrote:
>>>
>>> Did you try checking that all the references exist?
>>> So, maybe it is a referential integrity problem.
>>> You are prob not using Sqlite but a proper DB.
>>> Just ideas, D
>>>
>>>
>>> On Wednesday, August 22, 2012 8:23:23 PM UTC+1, Mike Girard wrote:
>>>>
>>>> Yeah. A result of switching it off and then back on.
>>>>
>>>> Good catch.
>>>>
>>>> On Wed, Aug 22, 2012 at 3:19 PM, Anthony <[email protected]> wrote:
>>>>
>>>>> I assume this is a typo too, but just to be sure:
>>>>>
>>>>> Field('genres','string','list:reference genre', ...
>>>>>
>>>>> You have both 'string' and 'list:reference' there.
>>>>>
>>>>> Anthony
>>>>>
>>>>>
>>>>> On Wednesday, August 22, 2012 2:49:29 PM UTC-4, Mike Girard wrote:
>>>>>
>>>>>> I have written a script to parse a large xml file and insert the
>>>>>> contents in to my app db. I am using lxml.
>>>>>>
>>>>>> After about 10 records get inserted, the script fails with
>>>>>> <type 'exceptions.TypeError'> argument of type 'NoneType' is not
>>>>>> iterable
>>>>>>
>>>>>> Troubleshooting determined the following:
>>>>>>
>>>>>> 1. This error is associated with 3 list:reference fields. When I
>>>>>> remove them from the script, the script executes uneventfully. If any
>>>>>> one
>>>>>> of them is included, it fails.
>>>>>> 2. This only happens after 10 records have been successfully inserted.
>>>>>> 3. There is no discernible difference between the records that get
>>>>>> successfully added and those that don't. The error happens even when I
>>>>>> hard
>>>>>> code the lists for the list:reference field. It seems to be associated
>>>>>> with
>>>>>> number of records, rather than which records.
>>>>>> 4. The script executes successfully when I change the field types
>>>>>> from 'list:reference' to 'string' and insert strings instead of lists.
>>>>>> You
>>>>>> should not assume from this that there is a data issue. As I said,
>>>>>> hardcoded lists get rejected also. I am 99% certain valid data is not
>>>>>> the
>>>>>> issue.
>>>>>> 5. This happens in both SQLLite and Postgres
>>>>>>
>>>>>> Here is the model declaration for one of the three fields. They are
>>>>>> all analogous:
>>>>>>
>>>>>> Field('genres','string','list:**reference genre',
>>>>>> requires=IS_IN_DB(db, 'genre.id', '%(name)s [%(id)s]',
>>>>>> multiple=True))
>>>>>>
>>>>>> Here is how I update each new row in the the database:
>>>>>>
>>>>>> db.movies.validate_and_insert(****movie) (movie is a dict)
>>>>>>
>>>>>> Here is how I hardcoded values into the fields: movie['genre'] =
>>>>>> {456, 368, 239]
>>>>>>
>>>>>> Now, if someone doesn't have a solution, can they tell me if I can
>>>>>>
>>>>>> 1.Programmatically remove the list:reference from the model prior to
>>>>>> data updates and programmatically restore it afterwards?
>>>>>> 2. Retain all the functionality of these fields by toggling this way?
>>>>>>
>>>>>> Seriously considering going the join table route and skipping the
>>>>>> list reference fields. Are there any gotchas there?
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
--
