DictRow should behave like an OrderedDict?

[psycoteam]

Originally submitted by: Laurent Fasnacht

I think RowDict should behave like an OrderedDict. In particular, I would expect the following to be True:

[row_dict[k] for k in row_dict.keys()] == list(row_dict)

[dvarrazzo]

It's a sort of reasonable expectation, yes. If you provide the implementation for it we would merge it to devel. The best way to contribute is to open a feature branch on a github clone.

Current devel supports Python from 2.5 to 3.3. The branch should include tests and documentation.

I expect the new feature to not slow down the current implementation.

[auvipy]

Dict's are Ordered from 3.6

[kumatty]

I encountered the same issue when I tried to export csv from a postgres database. I used the first dict row's keys as the csv header and I realized the csv header and the data did not match. At least it worked with python's native sqlite3 adapter when I used sqlite3, i.e., keys of a row and actual data of a row are in the same order.

I would like to suggest to change a little bit DictRow#keys implementation as per the following:

def keys(self):
  - return self._index.keys()
  + return sorted(self._index, key=self._index.get)

Or at least it would be nice to have a method like ordered_keys. What do you think?

(For now I made a super class of DictRow that overrides keys() as workaround.)

[dvarrazzo]

You can use [d[0] for d in cur.description] as keys: that will work and is available even when your dataset is empty.

[kumatty]

Thanks for your quick reply, I actually didn't know that but the thing is I do not have a cursor object when I actually process the data. I know this is just our implementation problem but if DictRow has the information of keys I think it would be nice to know the order of the keys too.
I also think our situation is not quite limited because in many code "how we get the data" and "how we process the data" are separated.

[dvarrazzo]

Sure. For the moment you'll have to use your own subclass of DictRow. If you have an improvement to propose patches are welcome.