Support casting Utf8 to Boolean
#1738
Conversation
|
Should I submit a related issue? |
|
@MazterQyou Yea, we prefer to have related issue submitted for tracking purpose. Thank you. |
Codecov Report
@@ Coverage Diff @@
## master #1738 +/- ##
==========================================
+ Coverage 83.27% 83.30% +0.02%
==========================================
Files 195 196 +1
Lines 55896 55998 +102
==========================================
+ Hits 46549 46649 +100
- Misses 9347 9349 +2
Continue to review full report at Codecov.
|
|
Submitted and mentioned the related issue. |
| @@ -280,6 +280,8 @@ pub fn can_cast_types(from_type: &DataType, to_type: &DataType) -> bool { | |||
| /// | |||
| /// Behavior: | |||
| /// * Boolean to Utf8: `true` => '1', `false` => `0` | |||
| /// * Utf8 to boolean: `true`, `yes`, `on`, `1` => `true`, `false`, `no`, `off`, `0` => `false`, | |||
There was a problem hiding this comment.
I just looked at C++ Arrow: https://github.com/apache/arrow/blob/b8431fba68e2540b3e57def0bd0ad718652c4b98/cpp/src/arrow/util/value_parsing.h#L92. Seems it only converts "0", "1", "true", "false". Should we follow it and remove "on", "off", "yes", "no"?
There was a problem hiding this comment.
I haven't come across "on", "off", "yes", "no" in practice. They're language specific, unlike T and F which are computing primitives.
So I'd stick to "1", "true" and their inverses.
There was a problem hiding this comment.
I guess removing those rare variants makes sense.
unlike T and F which are computing primitives
Should t, f shorter equivalents be left in then? If yes, just t and f, or all shorter variants?
What about whitespace trimming?
There was a problem hiding this comment.
I haven't come across "on", "off", "yes", "no" in practice
Actually, I just remembered that YES, NO variants are used in ANSI information_schema views.
There was a problem hiding this comment.
Thank you @MazterQyou -- this is looking good. A little more polishing and hopefully this can make it into the next release in a day or two
MazterQyou
force-pushed
the
upstream-patch/utf8-to-boolean-cast
branch
2 times, most recently
from
aeb564c
to
ac19755
Compare
There was a problem hiding this comment.
I think the code looks good -- thank you @MazterQyou
I left some style comments, but I think this PR is looking ready to go now
MazterQyou
force-pushed
the
upstream-patch/utf8-to-boolean-cast
branch
from
ac19755
to
f2d2e41
Compare
MazterQyou
force-pushed
the
upstream-patch/utf8-to-boolean-cast
branch
from
f2d2e41
to
65b2d7c
Compare
There was a problem hiding this comment.
I think this looks good . thanks @MazterQyou
Are we all happy to keep the on/off/yes/no?
|
I prefer to stick with "1", "true", "0", "false" only, but not strong opinion. |
I echo this, but also not a strong opinion |
Which issue does this PR close?
Closes #1740.
Rationale for this change
Casting Utf8 to Boolean is useful for comparison between the two types.
Different Utf8 inputs might be implied as true or false Boolean type; the change implements conversion from strings supported by PostgreSQL.
What changes are included in this PR?
This PR implements Utf8 to Boolean cast, respecting the
cast_options.safevalue, and adds related tests.Are there any user-facing changes?