You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Support quoted column identifiers for scan row_filter (#1863)
# Rationale for this change
Our data lake uses old-school Kimball style quoted column names ("User
ID", "Customer Name" etc). The string parser for `row_filter` was unable
to parse this. Now it is.
example:
```python
# before
>> parser.parse(' "User Name" = 'ted')
ParseException: Expected '"', found ' '
# after
>> parser.parse(' "User Name" = 'ted')
EqualTo("User Name", "ted")
# Are these changes tested?
Yes a new test was added.
```
>[!NOTE]
> The `quoted_column_with_dots` previously errored `with "Expected '"',
found '.'"` _when using **double quotes only**_. It now raises error
text expecting an `'or'` value; I didn't toil over finding where the
exception is clobbered, because the error message between single and
double quote exceptions is inconsistent and I didn't really consider
this a polished/first-class error message. If this change is an issue, I
can dig further to try and revert the wording change; IMO raising the
same exception type is more than reasonable to consider the change
non-breaking.
# Are there any user-facing changes?
Yes quoted identifiers are now supported
0 commit comments