DOC: better document Dtypes docstrings + avoid sphinx warnings #26067

jorisvandenbossche · 2019-04-12T18:44:44Z

Use class_without_autosummary template for the dtypes (similar to what CategoricalDtype already did) to avoid a bunch of sphinx warnings
Therefore, explicitly list some attributes in the docstrings (for now, I went with only documenting the dtype-specific "metadata")
And while at it, I also gave the docstrings an update (documenting the parameters, etc)
And for doing the above, turned some metadata attributes of the dtypes into properties, so they 1) can be documented (have a docstring) b) cannot be set by the user (xref Dtype attributes (eg PeriodDtype.freq) should not be settable #26096)

…ocstrings

codecov · 2019-04-15T07:02:20Z

Codecov Report

Merging #26067 into master will decrease coverage by 51.2%.
The diff coverage is 85.71%.

@@             Coverage Diff             @@
##           master   #26067       +/-   ##
===========================================
- Coverage   91.95%   40.75%   -51.21%     
===========================================
  Files         175      175               
  Lines       52427    52514       +87     
===========================================
- Hits        48208    21400    -26808     
- Misses       4219    31114    +26895

Flag	Coverage Δ
#multiple	`?`
#single	`40.75% <85.71%> (-0.12%)`	⬇️

Impacted Files	Coverage Δ
pandas/core/arrays/sparse.py	`40.47% <ø> (-51.84%)`	⬇️
pandas/core/arrays/integer.py	`44.18% <100%> (-52.15%)`	⬇️
pandas/core/dtypes/dtypes.py	`73.86% <85%> (-22.49%)`	⬇️
pandas/io/formats/latex.py	`0% <0%> (-100%)`	⬇️
pandas/io/sas/sas_constants.py	`0% <0%> (-100%)`	⬇️
pandas/core/groupby/categorical.py	`0% <0%> (-100%)`	⬇️
pandas/tseries/plotting.py	`0% <0%> (-100%)`	⬇️
pandas/tseries/converter.py	`0% <0%> (-100%)`	⬇️
pandas/io/formats/html.py	`0% <0%> (-99.37%)`	⬇️
pandas/io/sas/sas7bdat.py	`0% <0%> (-91.16%)`	⬇️
... and 135 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d9c9e2f...8818681. Read the comment docs.

codecov · 2019-04-15T07:02:20Z

Codecov Report

Merging #26067 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #26067      +/-   ##
==========================================
- Coverage   91.99%   91.98%   -0.01%     
==========================================
  Files         175      175              
  Lines       52383    52394      +11     
==========================================
+ Hits        48188    48196       +8     
- Misses       4195     4198       +3

Flag	Coverage Δ
#multiple	`90.54% <100%> (ø)`	⬆️
#single	`40.74% <60%> (-0.15%)`	⬇️

Impacted Files	Coverage Δ
pandas/core/arrays/sparse.py	`92.3% <ø> (ø)`	⬆️
pandas/core/dtypes/dtypes.py	`96.42% <100%> (+0.08%)`	⬆️
pandas/core/arrays/integer.py	`96.34% <100%> (+0.01%)`	⬆️
pandas/io/gbq.py	`75% <0%> (-12.5%)`	⬇️
pandas/core/frame.py	`96.9% <0%> (-0.12%)`	⬇️
pandas/util/testing.py	`90.71% <0%> (+0.1%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f53bb06...c8ae0c9. Read the comment docs.

jorisvandenbossche · 2019-04-15T12:22:43Z

@jreback could you take a look at this, if you have some time? (specifically the actual code changes in the dtypes, as this might have some implications for pickling)

jreback

looks fine, just being sure we actually have enough coverage for pickling of these

pandas/core/dtypes/dtypes.py

jreback · 2019-04-15T12:37:36Z

pandas/core/dtypes/dtypes.py

+    def __setstate__(self, state):
+        # for pickle compat.
+        self._freq = state['freq']
+


do we have a round-trip on this test? shouldn't you also need __getstate__? or maybe not as you are setting _freq which is now good.

__getstate__ is implemented in the super class:

pandas/pandas/core/dtypes/dtypes.py

Lines 152 to 154 in 455a2cd

def __getstate__(self):

# pickle support; we don't want to pickle the cache

return {k: getattr(self, k, None) for k in self._metadata}

Added in 154a647

And it's the round trip test that was failing

hmm, what if you move _freq into metdata then? (and so on for the other subclasses)

Yeah, I have been thinking about that as well. _metadata is also used for other things though (that are included in the EA interface, this __getstate__ is only for our internal dtypes). Eg for equality. For equality it would not matter if freq or _freq is included, but if we would want to use it eg also for a default repr, then it need to be the public attributes

hmm, yeah they are coupled. i don't really like having to be explicit about the pickle compat here.....

i don't really like having to be explicit about the pickle compat here.....

Yep, I agree. Although it is consistent with how it is already done in the other dtypes (Categorical actually has the same), it would be nice to clean it up. I could also add a different _metadata_pickles list, so we can handle both getstate and setstate in the PandasExtensionDtype superclass. Of course that doesn't change the implementation, but it does reduce some duplication in the code.

pandas/core/dtypes/dtypes.py

jorisvandenbossche · 2019-04-15T12:56:40Z

why exactly is this needed? e

It's also doing the same as what is already done in DatetimeTZDtype:

pandas/pandas/core/dtypes/dtypes.py

Lines 720 to 723 in 455a2cd

    
           def __setstate__(self, state): 
        
               # for pickle compat. 
        
               self._tz = state['tz'] 
        
               self._unit = state['unit']

I was only wondering if we need to worry about old pickle compat, as the above code I am showing was done in the initial implementation of DatetimeTZDtype, while for PeriodDtype that I am changing here, we already have released versions without this.

jreback · 2019-04-15T12:57:49Z

right, see my comment above, I think that this might just work if you move the private attributes into _metadata (and leave the accessors as propertys)

jorisvandenbossche · 2019-04-19T13:36:20Z

@jreback are you fine with merging this as is?
It is consistent with how the existing CategoricalDtype and DatetimeTZDtype handle pickling compat, so I think it is fine. We could try to clean it up a bit by moving some part to the base PandasExtensionDtype, but that could also be done separate.

jreback

see my comment otherwise lgtm

jreback · 2019-04-19T13:52:02Z

pandas/core/dtypes/dtypes.py

            return is_dtype_equal(self.subtype, other.subtype)

+    def __setstate__(self, state):
+        # for pickle compat.


i think would add a bit more comment here. (in all of the setstates to indicate why this is needed, maybe reference this issue)

OK, expanded the comment!

…ocstrings

jreback · 2019-04-19T14:23:14Z

something faiking merge on green

jreback · 2019-04-20T16:50:44Z

thanks @jorisvandenbossche

…s-dev#26067)

jorisvandenbossche added 3 commits April 12, 2019 12:07

Use class_without_autosummary template for all Dtype docstring pages

acc47b9

make PeriodDtype.freq and IntervalDtype.subdtype into properties

e79d793

update docstrings

6620d84

jorisvandenbossche added Docs Dtype Conversions Unexpected or buggy dtype conversions labels Apr 12, 2019

jorisvandenbossche added this to the 0.25.0 milestone Apr 12, 2019

jorisvandenbossche added 2 commits April 12, 2019 20:49

fix subdtype -> subtype

44b8d09

Merge remote-tracking branch 'upstream/master' into doc-dtype-class-d…

8818681

…ocstrings

jorisvandenbossche added 2 commits April 15, 2019 09:12

linter (order of sections)

f61ebc0

add pickle compat

d43ee0f

jreback reviewed Apr 15, 2019

View reviewed changes

jorisvandenbossche mentioned this pull request Apr 15, 2019

Dtype attributes (eg PeriodDtype.freq) should not be settable #26096

Closed

jreback reviewed Apr 19, 2019

View reviewed changes

jorisvandenbossche added 2 commits April 19, 2019 15:57

add extra comments

82e53a5

Merge remote-tracking branch 'upstream/master' into doc-dtype-class-d…

84baf36

…ocstrings

empty

c8ae0c9

jreback merged commit 3face07 into pandas-dev:master Apr 20, 2019

yhaque1213 pushed a commit to yhaque1213/pandas that referenced this pull request Apr 22, 2019

DOC: better document Dtypes docstrings + avoid sphinx warnings (panda…

fd4d6d1

…s-dev#26067)

ryanreh99 pushed a commit to ryanreh99/pandas that referenced this pull request Apr 22, 2019

DOC: better document Dtypes docstrings + avoid sphinx warnings (panda…

f6c97ea

…s-dev#26067)

jorisvandenbossche deleted the doc-dtype-class-docstrings branch April 22, 2019 07:41

	def __getstate__(self):
	# pickle support; we don't want to pickle the cache
	return {k: getattr(self, k, None) for k in self._metadata}

Uh oh!

DOC: better document Dtypes docstrings + avoid sphinx warnings #26067

DOC: better document Dtypes docstrings + avoid sphinx warnings #26067

Uh oh!

Conversation

jorisvandenbossche commented Apr 12, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Apr 15, 2019

Codecov Report

Uh oh!

codecov bot commented Apr 15, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

jorisvandenbossche commented Apr 15, 2019

Uh oh!

jreback left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jorisvandenbossche commented Apr 15, 2019

Uh oh!

jreback commented Apr 15, 2019

Uh oh!

jorisvandenbossche commented Apr 19, 2019

Uh oh!

jreback left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jreback commented Apr 19, 2019

Uh oh!

jreback commented Apr 20, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jorisvandenbossche commented Apr 12, 2019 •

edited

Loading

codecov bot commented Apr 15, 2019 •

edited

Loading