3
���h�N � @ s d dl Zd dlZd dlZd dlmZmZmZmZm Z m
Z
m Z m Z m
Z
mZmZmZmZmZmZ d dljZd dlmZ d dljj j Z dd� Z dd� Z dd � Z!d
d � Z"d d
� Z#dd� Z$dd� Z%dd� Z&dd� Z'ej(j)deddd�e
j*e
d�e
j+ge
j,e
j*e
d�fe
d�e
j+fg�fd�d dgej-ej.ej/�j*d ej.ej/�j+gd d �e
j,ej.ej/�j*d fd ej.ej/�j+fg�fej0d��ej0d �ej0d�gej-ej0ej.ej/�j+ �ej0d �ej0ej.ej/�j+�g�e
j,ej0ej.ej/�j+ �ej0d �fej0d �ej0ej.ej/�j+�fg�fg�d d � �Z1d d!� Z2ej(j)d"g d#d$fdd#dgd%d&fg�d'd(� �Z3ej(j)d)d#ej4d#�eej4d#��g�ej(j)d*eeg�d+d,� ��Z5ej(j)d-d dd#dd.ej6gej6 d dd#dd.gej6 d dd#dd.ej6gg�d/d0� �Z7d1d2� Z8ej(j)d3d4d�d6d%d7dgd8fd9d d6d%d7d�gd:fg�d;d<� �Z9d=d>� Z:d?d@� Z;ej(j)dAdd9g�dBdC� �Z<dDdE� Z=dFdG� Z>ej(j)dHdIdJ� dKdJ� fdLdJ� dMdJ� fg�dNdO� �Z?dPdQ� Z@ej(j)dRejAdS�ejAdS�dT g�dUdV� �ZBej(j)dWd�d�d�d�g�d\d]� �ZCd^d_� ZDd`da� ZEej(j)dbeFdcdd�dfeF� defeFdfdd�defeFdgdd�dhfg�didj� �ZGej(j)d-dkd�dlg�ej(j)dmdd#g�dndo� ��ZHej(j)dpd�d�d�g�dqdr� �ZIej(j)dsdtdJ� dudJ� dvdJ� dwdJ� g�dxdy� �ZJej(j)d-ee dzd{d|g��ejKdz�ejKd{�ejKd|�gej-ejKdz�ejKd{�ejKd|�g�edzd{d|g�g�d}d~� �ZLej(j)dde
d��e
d��e
d��e
d��gg�ej(j)d�eMej-ee g�d�d�� ��ZNd�d�� ZOd�d�� ZPej(j)d�dd�d�g�d�d�� �ZQd�d�� ZRej(j)dd�d�g�ej(j)d�e ejSfej-ejTfeMejUfg�d�d�� ��ZVej(j)dAdgdd4g�d�d�� �ZWej(j)ddd d�d�gg�ej(j)d8d4d9g�ej(j)d�d4d9g�d�d�� ���ZXej(j)d�d�d�d�gd�d�d�d�gd�d�d�gd dd gd�d�gfddd�gd d#d.d�d�gd#d dd#gd#d dgd dd#gfg�d�d�� �ZYej(j)d�d�d�d�gd�d�d�d�gd�d�d�gd dd#gd�d�d�gfddd�gd d#d.d�d�gdd dd#gd dd#gdd dd#gfg�d�d�� �ZZd�d�� Z[d�d�� Z\dS )�� N)� Categorical� DataFrame�
DatetimeIndex�Index�Interval�
IntervalIndex�Series�TimedeltaIndex� Timestamp�cut�
date_range�isna�qcut�timedelta_range� to_datetime)�CategoricalDtypec C sD t jddd�} t| ddd�}t jdddddg�}tj||dd� d S ) N� �int64)�dtype� F)�labels� )Z check_dtype)�npZonesr �array�tm�assert_numpy_array_equal)�data�result�expected� r �?/tmp/pip-build-5_djhm0z/pandas/pandas/tests/reshape/test_cut.py� test_simple s r! c C s� t jddddddg�} t| ddd �\}}tj|jd��}|jd
d
d
d d d
g�}t|dd
�}tj ||� tj
|t jddddg�� d S )Ng�������?gffffff�?g @g������@gffffff#@g������ @� T)�retbinsr r � )�orderedg����Mb�?g>wa���
@g���!""@) r r r r � from_breaks�round�taker r �assert_categorical_equal�assert_almost_equal)r r �bins� intervalsr r r r � test_bins# s r- c C s� t jdddddddg�} t| dd d d
�\}}tj|jd ��}t|d d �}|jd
d
d
dd d
d
g�}tj ||� tj
|t jdddddg�� d S )Ng�������?gffffff�?g @g������@gffffff#@g������ @g������@r T)�rightr# r" )r% r r$ g����Mb�?g������@g�����L @) r r r r r& r' r r( r r) r* )r r r+ r, r r r r �
test_right/ s r/ c C s� t jdddddddg�} t| dd d
d �\}}tj|jd �d
d�}|jddddd ddg�}t|d
d�}tj ||� tj
|t jdddddg�� d S )Ng�������?gffffff�?g @g������@gffffff#@g������ @g������@r FT)r. r# r" �left)�closedr r$ r )r% g������@g�����L @g%��Ck#@) r r r r r& r' r( r r r) r* )r r r+ r, r r r r �
test_no_right; s r2 c C s~ ddddddg} t | ddd �\}}tj|jd��}|jd
d
d
d d d
g�}t|dd
�}tj||� tj|t j
ddddg�� d S )Ng�������?gffffff�?g @g������@gffffff#@g������ @r" T)r# r r r$ )r% g����Mb�?g>wa���
@g���!""@) r r r& r' r( r r r) r* r r )r r r+ r, r r r r �test_array_likeG s r3 c C sn t td�d�} | }t td�|jd�}tj||� tjtj| j d�| jdd�}t td�|jd�}tj||� d S ) Nr r" )r+ r T)�
categoriesr% � ���)
r �ranger4 r r) r �
from_codesr �append�codes)�cr r r r r � test_bins_from_interval_indexS s r< c
C s� t jddddddddd g �} t| d
d d d
gd�}tjdddg�}tj|j|� tdddg|jd�}tj|j|� tj|j t jdddgdd�� d S )N�
� �
� � � � �; �< r � �# �F )r+ � �2 r r$ Zint8)r )r rF )rF rG )rG rH )
r r r r � from_tuplesr �assert_index_equalr4 r r: )Zagesr; r r r r r �)test_bins_from_interval_index_doc_example` s rM c C sB d} t jd d
dg�}tjt| d�� td d
g|d � W d Q R X d S )Nz)Overlapping IntervalIndex is not acceptedr r= r$ r@ r � )�matchr r5 )r+ )r r= )r$ r@ )r rN )r rK �pytest�raises�
ValueErrorr )�msg�iir r r �-test_bins_not_overlapping_from_interval_indexl s rU c
C sD d} ddddddg}t jt| d�� t|d d
d d g� W d Q R X d S )
Nz bins must increase monotonicallyg�������?gffffff�?g @g������@gffffff#@g������ @)rO g�������?g �?r r= )rP rQ rR r )rS r r r r �test_bins_not_monotonicu s rV zx, bins, expectedz
2017-12-31r" )�periodsz
2018-01-01r r )r c C s t | |�}tj|j|� d S )N)r r rL r4 )�xr+ r r r r r �#test_bins_monotonic_not_overflowing} s *
rY c
C sL d} ddddddg}t jt| d�� t|d d
d gd d
dgd� W d Q R X d S )Nz9Bin labels must be one fewer than the number of bin edgesg�������?gffffff�?g @g������@gffffff#@g������ @)rO r r r= �foo�barZbaz)r )rP rQ rR r )rS r r r r �test_wrong_num_labels� s r\ z
x,bins,msgr$ zCannot cut empty arrayg �?z#`bins` should be a positive integerc
C s( t jt|d�� t| |� W d Q R X d S )N)rO )rP rQ rR r )rX r+ rS r r r �test_cut_corner� s r] �arg�cut_funcc
C s, d}t jt|d�� || d� W d Q R X d S )Nz!Input array must be 1 dimensional)rO r$ )rP rQ rR )r^ r_ rS r r r �test_cut_not_1d_arg� s r` r r c C s. d}t jt|d�� t| dd� W d Q R X d S )Nz?cannot specify integer `bins` when input data contains infinity)rO r" )r+ )rP rQ rR r )r rS r r r �test_int_bins_with_inf� s
ra c C sZ d} t ddddd g| d�}t|ddgdd�}t tjtjtjdtjg| d�}tj||� d S )
NrX r r r" )�nameF)r r6 ���)r r r �nanr �assert_series_equal)rb �ser�ind�expr r r �test_cut_out_of_range_more� s
ri zright,breaks,closedTg����MbP?g �?g �?r. Fr0 c C sJ t jt jddd�d�}t|dd| d�\}}tj||d�}tj|j|� d S )Nr g)\��(�?g�������?r T)r# r. )r1 ) r �tile�aranger r r& r rL r4 )r. Zbreaksr1 �arrr r+ � ex_levelsr r r � test_labels� s rn c C s4 d} t tjjd�| d�}t|d�}|j| ks0t�d S )NrZ �d )rb r )r r �random�randnr rb �AssertionError)rb rf Zfactorr r r �#test_cut_pass_series_name_to_factor� s
rs c C sB t jddd�} t| ddd�}tjd dd d
d g�}tj|j|� d S )
Nr g\��(\�?g{�G�z�?r r$ )� precisiong�9̗G?g
ףp=
�?g
ףp=
�?gH�z�G�?g
ףp=
�?g�9̗G�)r rk r r r& r rL r4 )rl r rm r r r �test_label_precision� s ru r c C sZ t jddd�}t j|d d d�< t|d| d�}t j|�}t jt|�t j|�}tj||� d S )Nr g �?g{�G�z�?r" r )r ) r rk rd r Zasarray�wherer
r r* )r rl r r r r r �test_na_handling s
rw c C s� t jd�} t| dd�}t j ddt jg}t| |�}t||�}tj|�}tj|j |� |d t
dt j�kslt �|d t
t j d�ks�t �|d t
dt j�ks�t �|d t
t j d�ks�t �d S )Nr5 r )r r$ r r r ) r rk r �infr r r& r rL r4 r rr )r Zdata_serr+ r Z
result_serZ
ex_uniquesr r r �test_inf_handling s
ry c C sD t jjd�} t| dddg�}t|�}| dk | dkB }tj||� d S )Nro r r r6 r6 )r rp rq r r
r r )rl r �maskZex_maskr r r �test_cut_out_of_bounds s
r{ zget_labels,get_expectedc C s | S )Nr )r r r r �<lambda>* s r| c C s" t dgddg ddg | dd�S )N�Mediumr �Small�LargeT)r4 r% )r )r r r r r| + s c C s t jdddg| �S )Nr r r$ )r r8 )r r r r r| 2 s c C s t jdgddg ddg | �S )Nr r r r$ )r r8 )r r r r r| 3 s c C sN ddddg}dddddd d
g}d d d
g}t ||| |�d�}tj|||�� d S )Nr rB rJ ro r r= r> rI � rH r~ r} r )r )r r r) )Z
get_labelsZ get_expectedr+ rl r r r r r �test_cut_pass_labels& s
r� c C sR dddddddg} dd d
g}t | d |d �}t | d t||d
d�d �}tj||� d S )NrJ r r= r> rI r� rH ZGoodr} ZBadr" )r T)r4 r% )r r r r) )rl r r rh r r r �test_cut_pass_labels_compat@ s
r� rX g &@g _�Bc C s t | d� d S )Nr$ )r )rX r r r �test_round_frac_just_worksJ s r� zval,precision,expected眢#��]@�v �ݔ �~. ?�i UMu ?c C s t j| |d�}||kst�d S )N)rt )�tmodZ _round_fracrr )�valrt r r r r r �test_round_fracP s r� c C s� t ddddddddd g �} t| d�}tjdd dd
�jd�}|d d 8 < t tj|d d
�jdddddddddg ��jt dd��}t
j ||� d S )Nr r r$ r" r r r5 � � )�numg����Mb�?r. )r1 T)r% ) r r r Zlinspacer' r r& r( �astype�CDTr re )rf r Zexp_binsr r r r �test_cut_return_intervals_ s
r� c C sZ t tjd��} t| ddd�\}}t tjd ddgdd �jd��jtdd
��}t j
||� d S ) Nr r$ T)r# g�~j�t�h?g �?r" r. )r1 )r% g�~j�t�h�) r r rk r r r& �repeatr� r� r re )rf r r+ r r r r �test_series_ret_binsn s r� z
kwargs,msg�drop)�
duplicateszBin edges must be unique�raiserZ z(invalid value for 'duplicates' parameterc
C s� ddddddg}t tjdddd d
g�d d d
ddgd�}|d k rhtjt|d�� t||f| � W d Q R X n*t||f| �}t|tj|��}t j
||� d S )Nr r$ r r5 r= r r" r r� � �a�br; �d�e)�index)rO ) r r r rP rQ rR r �pd�uniquer re )�kwargsrS r+ �valuesr r r r r �test_cut_duplicates_biny s &r� g "@g �lengthc C s: t | g| �}t|ddd�}t dg| �}tj||� d S )Nr F)r r )r r r re )r r� rf r r r r r �test_single_bin� s r� z#array_1_writeable,array_2_writeablec C sR t jddd�}| |j_t jddd�}||j_t jd�}tjt||�t||�� d S )Nr ro r= )r rk �flagsZ writeabler r) r )Zarray_1_writeableZarray_2_writeableZarray_1Zarray_2Zhundred_elementsr r r �test_cut_read_only� s
r� �convc C s t | �S )N)r
)�vr r r r| � s c C s t | �S )N)r )r� r r r r| � s c C s
t j| �S )N)r �
datetime64)r� r r r r| � s c C s t | �j� S )N)r
Z
to_pydatetime)r� r r r r| � s c s� t jd�t jd�g}dddg}tttt|d �t|d ��tt|d �t|d ��g��jtd d
��}� fd d �|D �}tt||d
��}t j
||� d S )Nz
2012-12-13z
2012-12-15z
2012-12-12z
2012-12-14z
2012-12-16r r r$ T)r% c s g | ] }� |��qS r r )�.0r� )r� r r �
<listcomp>� s z%test_datetime_bin.<locals>.<listcomp>)r+ ) r r� r r r r
r� r� r r re )r� r Zbin_datar r+ r r )r� r �test_datetime_bin� s
r� z
2013-01-01z
2013-01-02z
2013-01-03c C sn t | ddd�\}}ttttd�td��ttd�td��ttd�td��g��jtdd��}tjt|�|� d S ) Nr" T)r# z2012-12-31 23:57:07.200000z2013-01-01 16:00:00z2013-01-02 08:00:00z2013-01-03 00:00:00)r% ) r r r r r
r� r� r re )r r �_r r r r �test_datetime_cut� s r� r+ z2013-01-01 04:57:07.200000z2013-01-01 21:00:00z2013-01-02 13:00:00z2013-01-03 05:00:00�boxc
C s� d}t tdd|d��}t| t�s(|| �} t|| �}t tttd|d�td|d��ttd|d�td|d��ttd|d�td |d��g��jt d
d ��}t
j ||� d S ) Nz
US/Eastern�20130101r" )rW �tzz2012-12-31 23:57:07.200000)r� z2013-01-01 16:00:00z2013-01-02 08:00:00z2013-01-03 00:00:00T)r% ) r r �
isinstance�intr r r r
r� r� r re )r+ r� r� �sr r r r r �test_datetime_tz_cut� s$
r� c C s< d} t jt| d�� ttddd�dddgd � W d Q R X d S )
Nz bins must be of datetime64 dtype)rO r� r" )rW r r$ r )r+ )rP rQ rR r r )rS r r r �test_datetime_nan_error s r� c C sb t tddd�tddd�d�} | jj� }tj|tjdg�� | j� }tj|tjdddddg�� d S ) NZ20130102r )rW r� r$ )r+ FT)r r r4 r
r r r r )r rz r r r �test_datetime_nan_mask$ s
r� r� �UTCz
US/Pacificc C sb t tdd| d��}t|ddd�\}}t||�}tj||� tddd g�}|j| �}tj||� d S )
NZ20180101r" )rW r� r$ T)r# z2017-12-31 23:57:07.200000z2018-01-02 00:00:00z2018-01-03 00:00:00)r r r r re r Z tz_localizerL )r� rf r � result_binsr �
expected_binsr r r �test_datetime_cut_roundtrip0 s
r� c C sV t tddd��} t| ddd�\}}t| |�}tj||� tddd g�}tj||� d S )
NZ1dayr" )rW r$ T)r# z0 days 23:57:07.200000z2 days 00:00:00z3 days 00:00:00)r r r r re r rL )rf r r� r r� r r r � test_timedelta_cut_roundtrip@ s
r� r5 r� z box, comparec C sV |dddddgd �}|dddddgd �}t || dd�}t || dd�}|||� d S )Nr r r= FTr� )r� )r )r+ r� �compareZ
data_expectedZ data_resultr r r r r � test_cut_bool_coercion_to_intN s
r� c C s8 t d�}d}tjt|d�� t|d| d� W d Q R X d S )Nr zJBin labels must either be False, None or passed in as a list-like argument)rO r )r )r7 rP rQ rR r )r r� rS r r r �test_cut_incorrect_labels` s r� r r> �include_lowestc C sb t jjdddd�jt�}t j|d d d�< ttj|dd�| ||d�}t|| ||d�}t j
||� d S ) Nr r= rJ )�sizer$ ZInt64)r )r. r� ) r rp �randintr� �floatrd r r� r r r) )r+ r. r� r� r r r r r �test_cut_nullable_integeri s r� z3data, bins, labels, expected_codes, expected_labels� � rN � rF rI �A�Br� c C s0 t | ||dd�}tj||dd�}tj||� d S )NF)r+ r r% )r4 r% )r r r8 r r) )r r+ r �expected_codes�expected_labelsr r r r r �test_cut_non_unique_labelsv s r� �Cc C s0 t | ||dd�}tj||dd�}tj||� d S )NF)r+ r r% )r4 r% )r r r8 r r) )r r+ r r� r� r r r r r �test_cut_unordered_labels� s r� c C s: d} t jt| d�� tddgdddgdd � W d Q R X d S )
Nz.'labels' must be provided if 'ordered = False')rO g �?r" r r r$ F)r+ r% )rP rQ rR r )rS r r r �3test_cut_unordered_with_missing_labels_raises_error� s r� c C sp t jdddddg�} t jddddg�}t jdd d
g�}t j| ||d d �}t jddd d d
gd
d�}tj||� d S )Nr r$ r" r r r r5 r� r� r; F)r+ r r% �category)r )r� r r r re )r� r+ r r r r r r �%test_cut_unordered_with_series_labels� s r� r6 r6 g����MbP�gj�t��?眢#��]����)r� r" r� )r� r" r� )r� r$ r� )r� r$ r� g "�)TT)TF)FF)]Znumpyr rP Zpandasr� r r r r r r r r r
r r r
r r r Zpandas._testingZ_testingr Zpandas.api.typesr r� Zpandas.core.reshape.tile�coreZreshaperj r� r! r- r/ r2 r3 r< rM rU rV �markZ parametrize�min�maxrK r Ziinfor Z timedelta64rY r\ r] Zeyer` rx ra ri rn rs ru rw ry r{ r� r� rk r� r� r� r� �dictr� r� r� r� r� r� �listr� r� r� r� r� re r) Z assert_equalr� r� r� r� r� r� r� r r r r �<module> s D
"& 0 " $
*
"
*> ,@ |