csv.DictWriter.write_header()

[John Machin]

On Aug 13, 1:15 pm, Alan G Isaac <alan.is... at gmail.com> wrote:
> > On Aug 13, 6:45 am, Alan G Isaac <alan.is... at gmail.com> wrote:
> >> Given a csv.DictWriter instance `dw`
> >> I think it would be nice to be able to
> >> say dw.write_header()
> >> instead of
> >> dw.writer.writerow(dw.fieldnames)
>
> >> Good idea?
>
> On 8/12/2009 10:24 PM John Machin apparently wrote:
>
> > Yes, it's a brilliant idea. All you have to do is insert the following
> > lines in your code after importing csv:
>
> > csv.DictWriter.write_header = (
> >     lambda dw: dw.writer.writerow(dw.fieldnames)
> >     )
>
> I do not understand the reason for your silly, sarcastic response.

Duck typing: ask a silly question, get a silly answer.

> I already showed you how to achieve the result,

What result? You said (in effect) "I think it would be nice to be able
to write A instead of B". You didn't show any way of setting things up
so that one could write A instead of B.

I showed one way. Other possibilities are a 2-line patch to the users
own csv.py, and submitting a patch to the dev team. The second is
kludgy, and there could be a wait until e.g. 2.7 before the third
happens. So back to the first way; everybody should have a utility
module that they import to do little fixes like that ... here is a
slightly more robust version:

import csv
try:
    csv.DictWriter.write_header
except AttributeError:
    csv.DictWriter.write_header = (
        lambda dw: dw.writer.writerow(dw.fieldnames)
        )

> so obviously
> I am not just asking for my own personal gain.

???

> I take it you do not realize that many users do not see
> a simple way to get this often needed functionality.

No, I don't. How did you come to that realisation? AFAICT they don't
ask in this forum or on StackOverflow.

> (For example, see the gymnatics suggested by Shai Vaingast
> in his otherwise excellent and useful book.)

I can imagine that one might (without reading the source) make do with
the published APIs:

dw = csv.DictWriter(open(...), my_field_names
dw.writerow(dict((fn, fn) for fn in my_field_names))

> And indeed,
> there is no reason to expect them to.

Indeed not, as for somebody who otherwise sees utility in the
DictWriter class not having a writer_headers method is a glaring
deficiency.

Note that there is a difference between "expecting C to do X" and
"expecting C to be able to do X". I really hope you don't mean that
people should not be encouraged to be able to write a trivial class
like csv.DictWriter and add trivial functionality themselves ... or to
be able to just write the required functionality inline:

wtr = csv.writer(open(...), ...)
wtr.writerow(my_field_names)
for adict in source_of_dicts():
    wtr.writerow(adict[fn] for fn in my_field_names)

or even:

for obj in source_of_objects():
    wtr.writerow(getattr(obj, fn) for fn in my_field_names)

> So my question was, would this improve the class from
> a usability perspective?

Of course.

>  Do you have a useful response?

See above. Also, the csv module appears to be in two minds about
underscores in attribute names (field_size_limit, skipinitialspace,
writerow, writerows) -- it may cause less confusion if it were called
writeheader (consistent with writerow).