import from np_datetime instead of src/datetime #18312
Conversation
| @@ -50,6 +50,11 @@ cdef extern from "../src/datetime/np_datetime.h": | |||
| PANDAS_FR_fs | |||
| PANDAS_FR_as | |||
|
|
|||
| int days_per_month_table[2][12] | |||
| int dayofweek(int y, int m, int d) nogil | |||
| int is_leapyear(int64_t year) nogil | |||
There was a problem hiding this comment.
There are a couple of modules that use one or more of these three items but dont need the rest of the relatively heavy-weight src/datetime/np_datetime dependency. Ideally I'd prefer for these to be implemented in a dependency-free cython module.
There was a problem hiding this comment.
sure these are like dateutils
Codecov Report
@@ Coverage Diff @@
## master #18312 +/- ##
==========================================
- Coverage 91.39% 91.38% -0.02%
==========================================
Files 164 164
Lines 49854 49854
==========================================
- Hits 45566 45557 -9
- Misses 4288 4297 +9
Continue to review full report at Codecov.
|
| @@ -50,6 +50,11 @@ cdef extern from "../src/datetime/np_datetime.h": | |||
| PANDAS_FR_fs | |||
| PANDAS_FR_as | |||
|
|
|||
| int days_per_month_table[2][12] | |||
| int dayofweek(int y, int m, int d) nogil | |||
| int is_leapyear(int64_t year) nogil | |||
There was a problem hiding this comment.
sure these are like dateutils
|
thanks! |
Really? I was expecting a "no" because it would be duplicating stuff already in src/datetime/np_datetime, but if that's the go-ahead, then I'll put that together this week. Name-wise I was thinking |
|
well i wouldn’t duplicate stuff rather should be orthogonal - maybe not sure what u r proposing here |
|
No worries, I've got plenty of other stuff to work on, will get distracted by something shiny and forget about this idea. |
There are a handful of things where there is an option to cimport from either tslibs.np_datetime or src/datetime.pxd. This PR updates imports use tslibs.np_datetime where possible.
Exposes NPY_NAT in nattype.pxd so other modules can cimport it from there instead of using util.get_nat(). Thematically
nattypeis the appropriate place for this to come from.I don't think this has any overlaps with #18298.
git diff upstream/master -u -- "*.py" | flake8 --diff