From 6140bd50242ee0b67a05504be1baccf0988cb9e1 Mon Sep 17 00:00:00 2001
From: Bert Palm <bert.palm@ufz.de>
Date: Tue, 10 Mar 2020 14:21:48 +0100
Subject: [PATCH] notna isna dropna
---
dios/dios.py | 46 ++++++++++++++++++++++++++++++++--------------
1 file changed, 32 insertions(+), 14 deletions(-)
diff --git a/dios/dios.py b/dios/dios.py
index 65a3c4f..cd5e7e1 100644
--- a/dios/dios.py
+++ b/dios/dios.py
@@ -101,7 +101,7 @@ class DictOfSeries:
self._data = data
else:
# it is significantly faster, to provide an index and fill it,
- # than to successively build the index by adding series's
+ # than to successively build the index by adding data
self._data = pd.Series(dtype='O', index=columns)
else:
@@ -416,13 +416,10 @@ class DictOfSeries:
if `method` == `indexed` (default):
every column is represented by a own index and corresponding values
- if `method` == `aligned`:
+ if `method` == `aligned` [2]:
one(!) global index is generated and values from a column appear at
the corresponding index-location.
- Note: the (common-) params are directly passed to pd.DataFrame.to_string(..)
- under the hood
-
Common params :
- max_cols: not more column than `max_cols` are printed [1]
- max_rows: see `min_rows` [1]
@@ -440,7 +437,10 @@ class DictOfSeries:
that this should differ from `na_rep`, otherwise you cannot differ missing- from NaN- values.
Notes:
- [1] defaults to the corresponding value in `dios_options`
+ [1]: defaults to the corresponding value in `dios_options`
+ [2]: the common-params are directly passed to pd.DataFrame.to_string(..)
+ under the hood, if method is `aligned`
+
"""
if self.empty:
return _empty_repr(self)
@@ -505,18 +505,36 @@ class DictOfSeries:
raise ValueError(axis)
return new
+ def _fast_apply(self, attr, **kwargs):
+ data = pd.Series(dtype='O', index=self.columns)
+ call = callable(getattr(pd.Series, attr))
+ for c in self.columns:
+ s = self._data.at[c]
+ data.at[c] = getattr(s, attr)(**kwargs) if call else getattr(s, attr)
+ return data
+
@property
def dtypes(self):
- s = pd.Series(index=self.columns, dtype='O')
- for k in self.columns:
- s.at[k] = self._data.at[k].dtype
- return s
+ return self._fast_apply('dtype')
def astype(self, dtype, copy=True, errors='raise'):
- new = self.copy_empty(columns=True) if copy else self
- for k in self.columns:
- new._data.at[k] = self._data.at[k].astype(dtype=dtype, copy=copy, errors=errors)
- return new
+ data = self._fast_apply('astype', dtype=dtype, copy=copy, errors=errors)
+ return DictOfSeries(data=data, itype=self.itype, cast_policy=self._policy, fastpath=True)
+
+ def isna(self):
+ data = self._fast_apply('isna')
+ return DictOfSeries(data=data, itype=self.itype, cast_policy=self._policy, fastpath=True)
+
+ def notna(self):
+ data = self._fast_apply('notna')
+ return DictOfSeries(data=data, itype=self.itype, cast_policy=self._policy, fastpath=True)
+
+ def dropna(self, inplace=False):
+ data = self._fast_apply('dropna', inplace=False) # never pass inplace=True
+ if inplace:
+ self._data = data
+ else:
+ return DictOfSeries(data=data, itype=self.itype, cast_policy=self._policy, fastpath=True)
def memory_usage(self, index=True, deep=False):
mem = 0
--
GitLab