slight csv misbehavior under Windows

[Thomas Womack]

I'm using the csv module in python 2.3 under Windows XP; it seems that
two separate mechanisms try to handle the perverse Windows habit of
ending lines with 0d0a, and so

writer = csv.writer(file('foo.csv','w'))
for i in thing:
writer.write(i)

produces lines ending 0d0d0a, and when you load this into Excel 97 the
data appears double-spaced.

Opening the file with 'wb' solves the problem, but it would be lovely
were this either fixed or documented; it's rather unintuitive that a
file opened with 'wb' still ends lines with 0d0a.

Please tell me if this is not the appropriate place to make this kind
of bug report.

Tom

[achrist]

I think that "wb" is obviously the correct format, and this is not
a bug. CSV fields that contain an end of line within the field contain
a single character end-of-line marker (I can never remember whether
this is \r or \n), and this is not kosher for a Windows text file,
so it must be a binary file.
Al

Thomas Womack wrote:

I'm using the csv module in python 2.3 under Windows XP; it seems that
two separate mechanisms try to handle the perverse Windows habit of
ending lines with 0d0a, and so

writer = csv.writer(file('foo.csv','w'))
for i in thing:
writer.write(i)

produces lines ending 0d0d0a, and when you load this into Excel 97 the
data appears double-spaced.

Opening the file with 'wb' solves the problem, but it would be lovely
were this either fixed or documented; it's rather unintuitive that a
file opened with 'wb' still ends lines with 0d0a.

Please tell me if this is not the appropriate place to make this kind
of bug report.

Tom