BUG: replace iterrows with itertuples in sql insert (GH6509) #6591
Conversation
| df.to_sql("test_read_write", self.conn) | ||
| df2 = sql.read_table("test_read_write", self.conn) | ||
|
|
||
| tm.assert_equal(df['s1'].values, df2['s1'].values) |
There was a problem hiding this comment.
@jreback I couldn't use:
tm.assert_frame_equal(df, df2.set_index('pandas_index'),
check_dtype=False, check_names=False)
because it seems this uses assert_almost_equal, and it seems that 33554433 is 'almost' equal to 33554432 ...
Is this also OK?
There was a problem hiding this comment.
I see....
maybe add an argument to assert_frame_equal (and also to assert_series_equal) that is check_exact=False; which is the default, but if True, then use assert_numpy_array_equal?
in this case you REALLY do care that they are exactly equal
There was a problem hiding this comment.
even better
self.assertTrue(df.equals(df2) (though it has to match exacty including the index, names, dtype, everything)
There was a problem hiding this comment.
for this last one, does it also have to match the dtype? Because in this case, I want to check the values while the dtypes don't match (int32 vs int64)
There was a problem hiding this comment.
some if these may not match dtype and hence are not round trippable
eg the values will match but u lose the original dtype
nothing to do about that (and should be a gotcha in the docs)
| if check_exact: | ||
| if not np.array_equal(left.values, right.values): | ||
| raise AssertionError('{0} is not equal to {1}.'.format(left.values, | ||
| right.values)) |
There was a problem hiding this comment.
@jreback Is there a reason that assert_numpy_array_equal is only avaiable in the TestCase class and not in the rest of the testing module? (so I cannot use that function here)
|
@jreback I added a |
|
@jorisvandenbossche perfect. assert_numpy_array_equal should prob be a module level function too (it just calls np.array_equal in any event), but no biggie. in theory all functions should actually be on the TestCase, but too much work to fix pls add a release note then merge |
|
Should there be a release note? As this fixes something (the new sqlalchemy based sql functions) that is not yet released? |
|
oh right.....though like to reference the closed issue (maybe just add it to an existing one) |
|
Actually, it was also fixed in the legacy code. Anyway, added a line to release notes. |
BUG: replace iterrows with itertuples in sql insert (GH6509)
Fixes #6509.
Should still add a test