def test_readcsv_fromfile_with_upd_props_mtable():
filename = '__tbl__.csv'
A1 = mg.read_csv(path_for_A)
A1.to_csv(filename)
A1.set_key('ID')
A2 = mg.read_csv(filename, key='ID')
assert_equal(A2.get_key(), 'ID')
assert_equal(len(A1), len(A2))
def test_readcsv_fromfile_with_upd_props_mtable():
filename = '__tbl__.csv'
A1 = mg.read_csv(path_for_A)
A1.to_csv(filename)
A1.set_key('ID')
A2 = mg.read_csv(filename, key='ID')
assert_equal(A2.get_key(), 'ID')
assert_equal(len(A1), len(A2))
def test_tocsv_mtable_wo_props_2():
filename = '__tbl__.csv'
A = mg.read_csv(path_for_A, key='ID')
A.to_csv(filename, suppress_properties=True)
A1 = mg.read_csv(filename, key='ID')
try:
os.remove(filename)
except OSError:
pass
assert_equal(A1.get_key(), "ID")
assert_equal(len(A1), len(A))
def test_readcsv_fromfile_with_props_mtable():
A1 = mg.read_csv(path_for_A, key='ID')
filename = '__tbl__.csv'
A1.to_csv(filename)
A2 = mg.read_csv(filename)
try:
os.remove(filename)
except OSError:
pass
assert_equal(A2.get_key(), 'ID')
assert_equal(len(A1), len(A2))
def test_tocsv_mtable_wi_props():
filename = '__tbl__.csv'
A = mg.read_csv(path_for_A, key='ID')
A.to_csv(filename)
A1 = mg.read_csv(filename)
try:
os.remove(filename)
except OSError:
pass
assert_equal(A1.get_key(), A.get_key())
assert_equal(len(A1), len(A))
def test_readcsv_fromfile_with_props_mtable():
A1 = mg.read_csv(path_for_A, key='ID')
filename = '__tbl__.csv'
A1.to_csv(filename)
A2 = mg.read_csv(filename)
try:
os.remove(filename)
except OSError:
pass
assert_equal(A2.get_key(), 'ID')
assert_equal(len(A1), len(A2))
def test_bl_combiner_wi_no_tuples():
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'name', 'name')
C = mg.combine_block_outputs_via_union([C1, C1, C1])
s = ['_id', 'ltable.ID', 'rtable.ID']
assert_equal(sorted(s) == sorted(C.columns), True)
assert_equal(C.get_key(), '_id')
assert_equal(C.get_property('foreign_key_ltable'), 'ltable.ID')
assert_equal(C.get_property('foreign_key_rtable'), 'rtable.ID')
def test_bl_combiner_wi_no_tuples():
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'name', 'name')
C = mg.combine_block_outputs_via_union([C1, C1, C1])
s = ['_id', 'ltable.ID', 'rtable.ID']
assert_equal(sorted(s) == sorted(C.columns), True)
assert_equal(C.get_key(), '_id')
assert_equal(C.get_property('foreign_key_ltable'), 'ltable.ID')
assert_equal(C.get_property('foreign_key_rtable'), 'rtable.ID')
def test_bl_combiner_wi_no_fields():
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'zipcode', 'zipcode')
C2 = ab.block_tables(A, B, 'hourly_wage', 'hourly_wage')
C3 = ab.block_tables(A, B, 'birth_year', 'birth_year')
C = mg.combine_block_outputs_via_union([C1, C2, C3])
s = ['_id', 'ltable.ID', 'rtable.ID']
assert_equal(sorted(s) == sorted(C.columns), True)
assert_equal(C.get_key(), '_id')
assert_equal(C.get_property('foreign_key_ltable'), 'ltable.ID')
assert_equal(C.get_property('foreign_key_rtable'), 'rtable.ID')
def test_bl_combiner_wi_no_fields():
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'zipcode', 'zipcode')
C2 = ab.block_tables(A, B, 'hourly_wage', 'hourly_wage')
C3 = ab.block_tables(A, B, 'birth_year', 'birth_year')
C = mg.combine_block_outputs_via_union([C1, C2, C3])
s = ['_id', 'ltable.ID', 'rtable.ID']
assert_equal(sorted(s) == sorted(C.columns), True)
assert_equal(C.get_key(), '_id')
assert_equal(C.get_property('foreign_key_ltable'), 'ltable.ID')
assert_equal(C.get_property('foreign_key_rtable'), 'rtable.ID')
def test_tocsv_vtable_wi_props():
filename = '__tbl__.csv'
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'zipcode', 'zipcode')
C1.to_csv(filename, suppress_properties=False)
C2 = mg.read_csv(filename, ltable=A, rtable=B)
try:
os.remove(filename)
except OSError:
pass
assert_equal(C2.get_key(), C1.get_key())
assert_equal(C2.get_property('foreign_key_ltable'), C1.get_property('foreign_key_ltable'))
assert_equal(C2.get_property('foreign_key_rtable'), C1.get_property('foreign_key_rtable'))
def test_readcsv_fromfile_with_props_vtable():
filename = '__tbl__.csv'
A = mg.read_csv(path_for_A, key='ID')
B = mg.read_csv(path_for_B, key='ID')
ab = mg.AttrEquivalenceBlocker()
C1 = ab.block_tables(A, B, 'zipcode', 'zipcode')
C1.to_csv(filename, suppress_properties=False)
C2 = mg.read_csv(filename, ltable=A, rtable=B)
try:
os.remove(filename)
except OSError:
pass
assert_equal(C2.get_key(), C1.get_key())
assert_equal(C2.get_property('foreign_key_ltable'), C1.get_property('foreign_key_ltable'))
assert_equal(C2.get_property('foreign_key_rtable'), C1.get_property('foreign_key_rtable'))
def test_readcsv_valid_id():
A = mg.read_csv(path_for_A, key='ID')
assert_equal(A.get_key(), 'ID')
def test_readcsv_key_not_present():
mg.read_csv(path_for_A, key='ID1')
def test_readcsv_no_id():
A = mg.read_csv(path_for_A)
assert_equal(A.get_key(), '_id')
def test_readcsv_invalid_id():
mg.read_csv(path_for_A, key='zipcode')
def test_readcsv_no_id():
A = mg.read_csv(path_for_A)
assert_equal(A.get_key(), '_id')
def test_readcsv_key_not_present():
mg.read_csv(path_for_A, key='ID1')
def test_readcsv_invalid_id():
mg.read_csv(path_for_A, key='zipcode')
def test_readcsv_valid_id():
A = mg.read_csv(path_for_A, key='ID')
assert_equal(A.get_key(), 'ID')
