Python 3.4.3 (v3.4.3:9b73f1c3e601, Feb 24 2015, 22:44:40) [MSC v.1600 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import csv
>>> s = 'x = "a", y = "b, c"'
>>> s
'x = "a", y = "b, c"'
>>> for row in csv.reader([s]): print(row)
...
['x = "a"', ' y = "b', ' c"']
>>> len(row)
3
>>> for row1 in csv.reader([s], skipinitialspace=True):print(row1)
...
['x = "a"', 'y = "b', 'c"']
>>> len(row1)
3
>>> s2 = 'x = "a",y="b,c"'
>>> s2
'x = "a",y="b,c"'
>>> for row2 in csv.reader([s]): print(row2)
...
['x = "a"', ' y = "b', ' c"']
>>> len(row2)
3
>>> for row3 in csv.reader([s], skipinitialspace=True): print(row3)
...
['x = "a"', 'y = "b', 'c"']
>>> len(row3)
3

>>> b = io.StringIO()
>>> w = csv.writer(b)
>>> w.writerow(['x = "a"', 'y = "b, c"']) 
28
>>> b.getvalue()
'"x = ""a""","y = ""b, c"""\r\n'


In other words, your input was not validly quoted csv.

You wrote ['x = "a"', 'y = "b, c"']
I wrote ['x = "a", y = "b, c"']

...muss

On Sun, Dec 13, 2015 at 7:08 PM, R. David Murray <report@bugs.python.org>
wrote:

>
> R. David Murray added the comment:
>
> >>> b = io.StringIO()
> >>> w = csv.writer(b)
> >>> w.writerow(['x = "a"', 'y = "b, c"'])
> 28
> >>> b.getvalue()
> '"x = ""a""","y = ""b, c"""\r\n'
>
>
> In other words, your input was not validly quoted csv.
>
> ----------
> nosy: +r.david.murray
> resolution:  -> not a bug
> stage:  -> resolved
> status: open -> closed
>
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue25857>
> _______________________________________
>

You may be right.  I just saved it from LibreOffice Calc and I got
[x=”a”,"y=”b, c”"].  I thought the original was saved from a spreadsheet
program.

...muss

On Sun, Dec 13, 2015 at 7:51 PM, Ioan Fintescu <ifintescu@gmail.com> wrote:

> You wrote ['x = "a"', 'y = "b, c"']
> I wrote ['x = "a", y = "b, c"']
>
>
> ...muss
>
>
> On Sun, Dec 13, 2015 at 7:08 PM, R. David Murray <report@bugs.python.org>
> wrote:
>
>>
>> R. David Murray added the comment:
>>
>> >>> b = io.StringIO()
>> >>> w = csv.writer(b)
>> >>> w.writerow(['x = "a"', 'y = "b, c"'])
>> 28
>> >>> b.getvalue()
>> '"x = ""a""","y = ""b, c"""\r\n'
>>
>>
>> In other words, your input was not validly quoted csv.
>>
>> ----------
>> nosy: +r.david.murray
>> resolution:  -> not a bug
>> stage:  -> resolved
>> status: open -> closed
>>
>> _______________________________________
>> Python tracker <report@bugs.python.org>
>> <http://bugs.python.org/issue25857>
>> _______________________________________
>>
>
>

Well, since there's no real standard for csv, it might have been.  Since it is iherently ambiguous according to "normal" csv rules, though, I'd say that if that is the case the originating spreadsheet is the one with the bug.  If you can prove there is a spreadsheet that uses this format, perhaps an enhancement request for csv would be in order.  I don't *think* there's any way to parse that with the existing dialect support, though I could be wrong.

There seems to be a CSV specification, namely IETF RFC 4180, and, as far as
I can tell, it indicates you are correct, and I am wrong.  Especially
points 5., 6., and 7 on page 3.  Here is a quote from 5 (RFC 4180, page 2
<https://tools.ietf.org/html/rfc4180#page-2>).

If fields are not enclosed with double quotes, then double quotes may not
> appear inside the fields.

It gets more specific in 6., and 7.

So the csv module is doing the right thing.  An error message may help,
though.

...muss

On Mon, Dec 14, 2015 at 9:05 AM, R. David Murray <report@bugs.python.org>
wrote:

>
> R. David Murray added the comment:
>
> Well, since there's no real standard for csv, it might have been.  Since
> it is iherently ambiguous according to "normal" csv rules, though, I'd say
> that if that is the case the originating spreadsheet is the one with the
> bug.  If you can prove there is a spreadsheet that uses this format,
> perhaps an enhancement request for csv would be in order.  I don't *think*
> there's any way to parse that with the existing dialect support, though I
> could be wrong.
>
> ----------
>
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue25857>
> _______________________________________
>

There's no place to generate an error message, the csv module parsed the line according to the rules.

I'm glad that there's an RFC now.  There wasn't when the module was written (which was well before my time on this project...)

Oh, maybe there is.  We could add an RFC-strict dialect that would raise an error, if I'm understanding your quotes from it correctly.  That would be a new feature, though.

If you consider that, you should take a look at the RFC; it also contains a
BNF like grammar plus some pointers to another RFC (2234).  It may require
more effort that it is worth, absent some demand for it.

...muss

On Mon, Dec 14, 2015 at 9:34 AM, R. David Murray <report@bugs.python.org>
wrote:

>
> R. David Murray added the comment:
>
> Oh, maybe there is.  We could add an RFC-strict dialect that would raise
> an error, if I'm understanding your quotes from it correctly.  That would
> be a new feature, though.
>
> ----------
>
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue25857>
> _______________________________________
>

Yeah, we'll leave it alone until someone actually submits an enhancment request...and the only way it'll get done is if they do it, I suspect :)