If the file is static (i.e not generated specifically for this request) you shouldn't be serving it through django anyway. You should configure some path (like /static/) to be served by your webserver, and save all the django overhead.
If the file is dynamic, there are 2 options:
- Create it in memory and serve it from django.
- Create it on the disk, and return a HttpResponseRedirect to it, so that your webserver deals with the download itself (if the file is very large, you should use this option).
As for serving it dynamically, I've been using the following code (which is a simplified version of ExcelResponse)
class CSVResponse(HttpResponse):
def __init__(self, data, output_name='data', headers=None, encoding='utf8'):
# Make sure we've got the right type of data to work with
valid_data = False
if isinstance(data, ValuesQuerySet):
data = list(data)
elif isinstance(data, QuerySet):
data = list(data.values())
if hasattr(data, '__getitem__'):
if isinstance(data[0], dict):
if headers is None:
headers = data[0].keys()
data = [[row[col] for col in headers] for row in data]
data.insert(0, headers)
if hasattr(data[0], '__getitem__'):
valid_data = True
assert valid_data is True, "CSVResponse requires a sequence of sequences"
import StringIO
output = StringIO.StringIO()
for row in data:
out_row = []
for value in row:
if not isinstance(value, basestring):
value = unicode(value)
value = value.encode(encoding)
out_row.append(value.replace('"', '""'))
output.write('"%s"\n' %
'","'.join(out_row))
mimetype = 'text/csv'
file_ext = 'csv'
output.seek(0)
super(ExcelResponse, self).__init__(content=output.getvalue(),
mimetype=mimetype)
self['Content-Disposition'] = 'attachment;filename="%s.%s"' % \
(output_name.replace('"', '\"'), file_ext)
To use it, just use return CSVResponse(...) passing in a list of lists, a list of dicts (with same keys), a QuerySet, a ValuesQuerySet