REF: Consistent optional dependency handling #26802
Conversation
| @@ -275,12 +272,7 @@ def _interpolate_scipy_wrapper(x, y, new_x, method, fill_value=None, | |||
| raise ImportError("Your version of Scipy does not support " | |||
| "PCHIP interpolation.") | |||
| elif method == 'akima': | |||
| try: | |||
There was a problem hiding this comment.
Akima1DInterpolater was added in SciPy 0.14, and we now require 0.19, so this additional check isn't needed anymore.
doc/source/install.rst
Outdated
| Dependency Minimum Version Notes | ||
| ========================= ================== ============================================================= | ||
| matplotlib 2.2.2 Visualization | ||
| xarray 0.8.2 pandas-like API for N-dimensional data |
There was a problem hiding this comment.
are these meant to be in some ordering?
I would maybe put them in sections: io, numeric, other i think
There was a problem hiding this comment.
I roughly grouped the I/O, HTML parsers, and just threw the others wherever.
With headers, it doesn't look as nice I think. What do you think?
There was a problem hiding this comment.
yeah ok maybe alphabetize then?
|
This is a great idea! Just to be clear did you purposely only select a subset of optional dependencies to update? I opened #26782 which could either be included here or done as a follow up |
pandas/compat/_optional.py
Outdated
|
|
||
|
|
||
| def import_optional_dependency( | ||
| name, extra="", raise_on_missing=True, on_version="raise" |
There was a problem hiding this comment.
Can you add annotations here? Should be straightforward I think
I grepped for I seem to have just missed the xlrd one. Do you want me to handle it here? |
|
Looks like I missed bottleneck too, sorry. |
Up to you |
Codecov Report
@@ Coverage Diff @@
## master #26802 +/- ##
==========================================
+ Coverage 91.76% 91.85% +0.08%
==========================================
Files 178 179 +1
Lines 50751 50707 -44
==========================================
+ Hits 46574 46579 +5
+ Misses 4177 4128 -49
Continue to review full report at Codecov.
|
|
small comment, otherwise lgtm. |
|
Great change @TomAugspurger ! |
| # xlrd uses a capitalized attribute name | ||
| version = module.__VERSION__ | ||
| version = getattr(module, '__VERSION__', None) |
There was a problem hiding this comment.
should we not encode the 'version' (or 'VERSION') with the list of deps directly? (or have a way to override); in fact we already do this in pandas/util/_print_versions. so actually I would update that to use this list of deps (and also to add the version accessor)
|
actually @TomAugspurger see my comments. This duplicates a ton of code in _print_versions. Can you (or others) do a followup (or create an issue) about de-duplicating. |
|
see #26814 |
|
FYI @pandas-dev/pandas-core you may want to glance through http://pandas-docs.github.io/pandas-docs-travis/development/contributing.html#optional-dependencies, when we get PRs adding a new optional dependency. |
|
Nice! |
Standardizes how we do our optional dependency checking.
import_optional_dependencyhelper that handles optional importing, raising with a nice message, and warning / raising if the version is not correct.try / except ImportErrorto useimport_optional_dependency>>> pd.read_gbq('') Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/Users/taugspurger/sandbox/pandas/pandas/io/gbq.py", line 140, in read_gbq pandas_gbq = _try_import() File "/Users/taugspurger/sandbox/pandas/pandas/io/gbq.py", line 14, in _try_import extra=msg, File "/Users/taugspurger/sandbox/pandas/pandas/compat/_optional.py", line 82, in import_optional_dependency raise ImportError(message.format(name=name, extra=extra)) from None ImportError: Missing optional dependency 'pandas_gbq'. pandas-gbq is required to load data from Google BigQuery. See the docs: https://pandas-gbq.readthedocs.io. Use pip or conda to install pandas_gbq.