test/unittest_dataimport.py
author Rémi Cardona <remi.cardona@logilab.fr>
Thu, 03 Apr 2014 15:36:33 +0200
branchstable
changeset 9695 aa982b7c3f2a
parent 9181 2eac0aa1d3f6
child 9903 d1fdbdbab194
permissions -rw-r--r--
[dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
9181
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     1
from StringIO import StringIO
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     2
from logilab.common.testlib import TestCase, unittest_main
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     3
from cubicweb import dataimport
9695
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
     4
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
     5
9181
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     6
class UcsvreaderTC(TestCase):
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     7
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     8
    def test_empty_lines_skipped(self):
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     9
        stream = StringIO('''a,b,c,d,
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    10
1,2,3,4,
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    11
,,,,
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    12
,,,,
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    13
''')
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    14
        self.assertEqual([[u'a', u'b', u'c', u'd', u''],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    15
                          [u'1', u'2', u'3', u'4', u''],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    16
                          ],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    17
                         list(dataimport.ucsvreader(stream)))
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    18
        stream.seek(0)
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    19
        self.assertEqual([[u'a', u'b', u'c', u'd', u''],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    20
                          [u'1', u'2', u'3', u'4', u''],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    21
                          [u'', u'', u'', u'', u''],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    22
                          [u'', u'', u'', u'', u'']
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    23
                          ],
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    24
                         list(dataimport.ucsvreader(stream, skip_empty=False)))
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    25
9695
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    26
    def test_skip_first(self):
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    27
        stream = StringIO('a,b,c,d,\n'
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    28
                          '1,2,3,4,\n')
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    29
        reader = dataimport.ucsvreader(stream, skipfirst=True,
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    30
                                       ignore_errors=True)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    31
        self.assertEqual(list(reader),
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    32
                         [[u'1', u'2', u'3', u'4', u'']])
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    33
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    34
        stream.seek(0)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    35
        reader = dataimport.ucsvreader(stream, skipfirst=True,
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    36
                                       ignore_errors=False)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    37
        self.assertEqual(list(reader),
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    38
                         [[u'1', u'2', u'3', u'4', u'']])
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    39
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    40
        stream.seek(0)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    41
        reader = dataimport.ucsvreader(stream, skipfirst=False,
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    42
                                       ignore_errors=True)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    43
        self.assertEqual(list(reader),
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    44
                         [[u'a', u'b', u'c', u'd', u''],
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    45
                          [u'1', u'2', u'3', u'4', u'']])
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    46
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    47
        stream.seek(0)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    48
        reader = dataimport.ucsvreader(stream, skipfirst=False,
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    49
                                       ignore_errors=False)
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    50
        self.assertEqual(list(reader),
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    51
                         [[u'a', u'b', u'c', u'd', u''],
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    52
                          [u'1', u'2', u'3', u'4', u'']])
aa982b7c3f2a [dataimport] Prevent ucsvreader from skipping the first line when ignore_errors is True (closes #3705791)
Rémi Cardona <remi.cardona@logilab.fr>
parents: 9181
diff changeset
    53
9181
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    54
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    55
if __name__ == '__main__':
2eac0aa1d3f6 [dataimport] ucsvreader should skip empty lines unless specified otherwise. Closes #3035944
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    56
    unittest_main()