incompatible interface with python dictionary.

[korr...@gmail.com]

Debian wheezy package python-dbf 0.88.16-1.

I tried to load dbf to database and convert field types like this:

def row_converter(row, fields):
d = {}
for name in fields:
v = row.get(name, None)
#futher converting field data
...

return d

dbf throws FieldMissing if field doesn't exists in DBF-file.

Traceback (most recent call last):
File "./load_rests.py", line 66, in row_converter
v = row.get(name, None)
File "/usr/lib/pymodules/python2.7/dbf/tables.py", line 212, in __getattr__
raise FieldMissing(name)

It's incompatible behaviour of get with python dictionaries.
IMHO it's not good.
Is it already fixed?
If not, can you fix this in next versions, please?

[Ethan Furman]

On 05/15/2013 12:53 AM, korr...@gmail.com wrote:
> Debian wheezy package python-dbf 0.88.16-1.
>
> I tried to load dbf to database and convert field types like this:
>
> def row_converter(row, fields):
> d = {}
> for name in fields:
> v = row.get(name, None)

Apologies for the delay in responding.

The problem is that a row is not a dictionary, and has none of the dictionary helper methods. It does allow for item
access, like a dict does, but it also allows for index access, like a list does (it's also not a list, though).

The bigger issue is that any helper method (get, setdefault, copy, etc.) is a potential field name, and if the method
exists the field name cannot.

So, no, I won't be adding get, or any of the other methods, to rows.

If you know the fields exist in the row, you can do this:

--> def row_converter(row, fields):
... d = {}
... for name in fields:
... v = row[name]

and if you're not sure the field exists:

--> def row_converter(row, fields):
... d = {}
... existing_fields = dbf.field_names(row)
... fields = [fld for fld in fields if fld in existing_fields]
... for name in fields:
... v = row[name]


You may want to upgrade to the latest version, but be aware there are significant, backwards-incompatible, changes in
the current version.

--
~Ethan~