def test_no_header(self):
args = ['--no-header-row', 'examples/no_header_row3.csv']
output_file = StringIO.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = StringIO.StringIO(output_file.getvalue())
#self.assertEqual(input_file.next(), '|----+---+----|\n')
#self.assertEqual(input_file.next(), '| 1 | 2 | 3 |\n')
#self.assertEqual(input_file.next(), '| 4 | 5 | 6 |\n')
#self.assertEqual(input_file.next(), '|----+---+----|\n')
self.assertEqual(input_file.next(),
'|----------+---------+----------|\n')
self.assertEqual(input_file.next(),
'| column1 | column2 | column3 |\n')
self.assertEqual(input_file.next(),
'|----------+---------+----------|\n')
self.assertEqual(input_file.next(),
'| 1 | 2 | 3 |\n')
self.assertEqual(input_file.next(),
'| 4 | 5 | 6 |\n')
self.assertEqual(input_file.next(),
'|----------+---------+----------|\n')
def test_no_header(self):
args = ['--no-header-row', 'examples/no_header_row3.csv']
output_file = StringIO.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = StringIO.StringIO(output_file.getvalue())
self.assertEqual(input_file.next(), '|----+---+----|\n')
self.assertEqual(input_file.next(), '| 1 | 2 | 3 |\n')
self.assertEqual(input_file.next(), '| 4 | 5 | 6 |\n')
self.assertEqual(input_file.next(), '|----+---+----|\n')
def test_simple(self):
args = ['examples/dummy3.csv']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| a | b | c |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), '| 1 | 4 | 5 |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
def test_simple(self):
args = ['examples/dummy3.csv']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| a | b | c |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), '| 1 | 4 | 5 |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
def test_phabricator(self):
args = ['examples/dummy3.csv', '--phabricator']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----|---|----|\n')
self.assertEqual(next(input_file), '| a | b | c |\n')
self.assertEqual(next(input_file), '|----|---|----|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), '| 1 | 4 | 5 |\n')
self.assertRaises(StopIteration, next, input_file)
def test_no_header(self):
args = ["--no-header-row", "examples/no_header_row3.csv"]
output_file = six.BytesIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.BytesIO(output_file.getvalue())
self.assertEqual(next(input_file), b"|----------+---------+----------|\n")
self.assertEqual(next(input_file), b"| column1 | column2 | column3 |\n")
self.assertEqual(next(input_file), b"|----------+---------+----------|\n")
self.assertEqual(next(input_file), b"| 1 | 2 | 3 |\n")
self.assertEqual(next(input_file), b"| 4 | 5 | 6 |\n")
self.assertEqual(next(input_file), b"|----------+---------+----------|\n")
def test_simple(self):
args = ["examples/dummy3.csv"]
output_file = six.BytesIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.BytesIO(output_file.getvalue())
self.assertEqual(next(input_file), b"|----+---+----|\n")
self.assertEqual(next(input_file), b"| a | b | c |\n")
self.assertEqual(next(input_file), b"|----+---+----|\n")
self.assertEqual(next(input_file), b"| 1 | 2 | 3 |\n")
self.assertEqual(next(input_file), b"| 1 | 4 | 5 |\n")
self.assertEqual(next(input_file), b"|----+---+----|\n")
def test_my_unicode(self):
args = ["examples/customer_sample.csv"]
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), "|-----+------------+-----------+----------|\n")
self.assertEqual(next(input_file), "| id | first_name | last_name | company |\n")
self.assertEqual(next(input_file), "|-----+------------+-----------+----------|\n")
self.assertEqual(next(input_file), "| 1 | Ramon | Jašek | |\n")
self.assertEqual(next(input_file), "|-----+------------+-----------+----------|\n")
def test_no_header(self):
args = ['--no-header-row', 'examples/no_header_row3.csv']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----------+---------+----------|\n')
self.assertEqual(next(input_file), '| column1 | column2 | column3 |\n')
self.assertEqual(next(input_file), '|----------+---------+----------|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), '| 4 | 5 | 6 |\n')
self.assertEqual(next(input_file), '|----------+---------+----------|\n')
self.assertRaises(StopIteration, next, input_file)
def test_unicode(self):
args = ["examples/test_utf8.csv"]
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), "|----+---+----|\n")
self.assertEqual(next(input_file), "| a | b | c |\n")
self.assertEqual(next(input_file), "|----+---+----|\n")
self.assertEqual(next(input_file), "| 1 | 2 | 3 |\n")
self.assertEqual(next(input_file), u"| 4 | 5 | ʤ |\n")
self.assertEqual(next(input_file), "|----+---+----|\n")
def test_unicode(self):
args = ['examples/test_utf8.csv']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| a | b | c |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), u'| 4 | 5 | ʤ |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertRaises(StopIteration, next, input_file)
def test_truncate(self):
args = ['examples/dummy4.csv', '--truncate', '3']
output_file = six.StringIO()
utility = CSVLook(args, output_file)
utility.main()
input_file = six.StringIO(output_file.getvalue())
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| a | b | c |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertEqual(next(input_file), '| 1 | 2 | 3 |\n')
self.assertEqual(next(input_file), '| 1 | 4 | 5 |\n')
self.assertEqual(next(input_file), '| 6 | 7 | 8 |\n')
self.assertEqual(next(input_file), '|----+---+----|\n')
self.assertRaises(StopIteration, next, input_file)
def convert_attachment(attachment):
outfile_path = '/tmp/' + attachment.file.name.split('/')[-1]
outfile = open(outfile_path, 'w')
ext = outfile_path.split('.')[-1].lower()
if ext not in ('xls', 'xlsx', 'csv', 'pdf'):
return
infile_path = attachment.file.path
if ext == 'pdf':
infile_path = tabula_csv(attachment, listify=False)
# in2csv ... can't say the python module utlities are working 100% here
#convert = In2CSV(args=[attachment.file.path], output_file=outfile)
#convert.main()
import subprocess
print 'attachment_id = ', attachment.id
print 'agency = ', get_attachment_agency(attachment)
print 'infile_path = ', infile_path
if ext == 'csv':
# already csv
outfile_path = infile_path
else:
subprocess.call(
[
'in2csv',
#'--format',outfile_path.split('.')[-1].lower(),
infile_path
], #attachment.file.path],
stdout=outfile)
# label column headers by zero-index
headers = get_headers(outfile_path)
if headers:
write_header_indices(outfile_path, headers)
# csvlook file | head
output = StringIO.StringIO()
look = CSVLook(args=[outfile_path, '-l'], output_file=output)
look.main()
input = StringIO.StringIO(output.getvalue())
return input
else:
print 'something wrong with headers. attachment id', attachment.id
