U
|
ý°d* ã @ s` d dl m Z d dlZd dlZd dlZd dlmZmZmZm Z m
|
Z
|
m Z m Z d dl mZ d dlmZ d dlmZ ejfddZejejd ejejejejejejejejejejd ejejejd ejejd ejdZd d
|
Zd d Zd d Zdd Zej dddg¡dd Z dd Z dd Z dd Z d d Z!ej d d d d!d ged"d#d$gd d%fd d&gd'd(d d d d!d d d g ed"d#d$gd d%e d)d*d+ggfd d,gd d d d d!d d d d g ed"d#d$gd d%ed"d#d$gd d%gfg¡ej d-ddg¡d.d/ Z"d0d1 Z#d2d3 Z$d4d5 Z%ej dddg¡ej d6ddg¡ej d7ddg¡d8d9 Z&d:d; Z'd<d= Z(d>d? Z)d@dA Z*dBdC Z+ej dDdEge e,d)d d$gd"d#gdFfe e,d) -dGdH ¡d#gd d"gdFfe e,d'd d$gd"d#gdFfg¡dIdJ Z.dKdL Z/dMdN Z0dOdP Z1ej dQdRdSdRgfdTdUdVgfdWdUdRgfdXdSdVgfg¡dYdZ Z2d[d\ Z3d]d^ Z4d_d` Z5ej dddg¡dadb Z6ej dddg¡dcdd Z7dedf Z8dgdh Z9didj Z:ej dkde e
|
j;e d"d"d#gdldmd"d#d#ggd d&gdnd"d#d$gdodpfdee d"d"d#gdldmd"d#d#gd"d#d$gdqfg¡drds Z<ej dtdej=ej>g¡dudv Z?ej@dwdx ZAej dydzd{g¡d|d} ZBej dydzd{g¡ej d6ddg¡d~d ZCej dde
|
;e dgd) dgd) dldde d
d
ddd
d
ddgdldde dWdXgd) g¡d#d#d)d)d"d"d$d$gfde
|
jDeddgdded
ddgdde dWdXggdddgdnd#d#d)d)ej=ej=d"d"ej=ej=d$d$g fde
|
jDeddgdded
ddgdde dWdXggdddgdnd#d#d)d)ej=ej=d"d"ej=ej=d$d$g fg¡dd ZEdd ZFej dd"d d gd d d gg¡dd ZGdd ZHdd ZIdd ZJdd ZKdd ZLej d6ddg¡dd ZMdd ZNej de jOd"d"d#gfe jPd"d#d#gfg¡d d¡ ZQd¢d£ ZRej d¤dRdTg¡d¥d¦ ZSd§d¨ ZTd©dª ZUd«d¬ ZVdd® ZWd¯d° ZXej d¤dRdTg¡eYeZd±d²d³Z[ej d¤dRdTg¡eYeZd±d´dµZ\d¶d· Z]ej d¤dRdTg¡d¸d¹ Z^dºd» Z_d¼d½ Z`d¾d¿ Zaej dÀdÁdÂdÃg¡ej dddg¡dÄdÅ Zbej dÀdÂdÃg¡ej dddg¡dÆdÇ Zcej dÀdÁdÂdÃg¡ej dÈdÉdÊg¡ej dddg¡dËdÌ Zdej dÀdÁdÂdÃg¡ej dÈd{dzdÍg¡ej dddg¡dÎdÏ Zeej dÀdÁdÂdÃg¡dÐdÑ Zfej d-ddg¡ej dÒdÓgdÓd,gg¡dÔdÕ ZgdS )Öé )ÚdatetimeN)Ú CategoricalÚCategoricalIndexÚ DataFrameÚIndexÚ
|
MultiIndexÚSeriesÚqcut)Ú SeriesGroupBy)Úget_groupby_method_argsc C s. dd }t jt|||d}| j||d ¡ S )zpReindex to a cartesian production for the groupers,
|
preserving the nature (Categorical) of each grouper
|
c S s4 t | ttfr0| j}tjt t|¡|| jd} | S )N©Ú
|
categoriesÚordered) Ú
|
isinstancer r r Ú
|
from_codesÚnpÚarangeÚlenr )Úar © r ú\d:\z\workplace\vscode\pyvenv\venv\Lib\site-packages\pandas/tests/groupby/test_categorical.pyÚf s ÿz)cartesian_product_for_groupers.<locals>.f©Únames©Ú
|
fill_value)r Ú from_productÚmapÚreindexÚ
|
sort_index)ÚresultÚargsr r r Úindexr r r Ú cartesian_product_for_groupers s r# )ÚallÚanyÚcountÚcorrwithÚfirstÚidxmaxÚidxminÚlastÚmaxÚmeanZmedianÚminÚnthÚnuniqueÚprodZquantileZsemÚsizeZskewÚstdÚsumÚvarc C sB t | jd}dd }| j|ddj |¡}|jjd dks>td S )Né c S s | ¡ | ¡ | ¡ | ¡ dS )N©r. r, r& r- r7 )Úgroupr r r Ú get_statsI s
|
üz2test_apply_use_categorical_name.<locals>.get_statsF©Úobservedr ÚC)r r< ÚgroupbyÚDÚapplyr" r ÚAssertionError)ÚdfÚcatsr9 r r r r Ú test_apply_use_categorical_nameF s rC c
|
C sr t dddddddddg ddddgdd} tddddddd d
|
d g | d }ttd ddd}tdddd
|
tjgi|d}|jddd ¡ }t ||¡ t ddddgdddgdd}t ddddgdddgdd}t||ddd d
|
gd}|jddd}tdddgddd} tdt
|
d ddg| di}|j dd}t ||¡ tddgddgddggd d gd }
|
t |
|
j |
|
d < |
|
jd gdd} | d d ¡}t ||
|
d g ¡ |
|
d ¡}|
|
jddg }t ||¡ d!d" } | | ¡}|
|
jddg ¡ }tddgd d#|_|d d$¡|d < t ||¡ tdd d%d&gi}tj|jdd'd(d)d*gd+} |jj| dd t ¡}t ||d ¡ t |jj| dd d,d ¡|d ¡ t |j| dd t ¡|dg ¡ |j| dd}| d-d ¡}t ||dg ¡ t d ¡: | d.d ¡}| t¡}| tjj ¡}| d/d ¡}W 5 Q R X tj ||dg dd0 tj ||dg dd0 t ||dg ¡ t ||dg ¡ t |jj| dd tj ¡|d ¡ t |j| dd tj ¡|¡ tdd d%d&d1gi}tj|jd2dd'd(d)d*gd+} |jj| dd t ¡}t ||d ¡ t |jj| dd d3d ¡|d ¡ t |j| dd t ¡|dg ¡ t |j| dd d4d ¡|dg ¡ tdddddgi}tj|jdddd d
|
gt td d5} |j| dd t ¡}t| j j!| j j"d6}t
|
ddddg|d}d|j_#t ||¡ d7d8d9d:g}tj$j%dd
|
d;d<}t j&||dd6} ttj$ 'd;d
|
¡}|j| dd ¡ }|jt (| ¡dd ¡ }t|| j!dd} | )| ¡}t ||¡ |j| dd}| *¡ }| j+ ,¡ }t (| ¡ -|¡}| -|¡}t |dd7d8d9d:gd=}|j|ddd> *¡ }t ||¡ t j&t .d
|
¡ /d?¡|dd6}t|} t 0| 1¡ j 2d¡| ¡ td@dAdBdCdDdEdFdGgd
|
} t 0| 1¡ j 2d¡| ¡ d S )HNr ÚbÚcÚdTr é é é r6 é ©r rD Úabcd©Únamer ©r" Fr: ÚzÚy©ÚAÚBÚvaluesrS rU é r ©Z numeric_onlyz John P. Doez Jane DoveZ person_idÚ person_name©Úcolumnsc S s | S ©Nr ©Úxr r r Ú<lambda>u ó z test_basic.<locals>.<lambda>c S s | d¡jd S )NrX r )Údrop_duplicatesÚilocr\ r r r r | s ztest_basic.<locals>.f©rN Úobjecté é é
|
é é é( )Úbinsc S s
|
t | ¡S r[ ©r r4 ©Úxsr r r r^ r_ c S s t j| ddS ©Nr ©Zaxis©r r, rl r r r r^ r_ c S s t j| ddS rn rp rl r r r r^ r_ c S s t j | ¡S r[ )r ÚmaximumÚreducerl r r r r^ r_ )Z check_dtypeéûÿÿÿiöÿÿÿc S s
|
t | ¡S r[ rk rl r r r r^ ¬ r_ c S s
|
t | ¡S r[ rk rl r r r r^ ° r_ )Úlabels©r ÚfooÚbarÚbazÚquxéd ©r2 )r r ©Úsortr; é r& r- r3 r. ú25%ú50%ú75%r, )3r r r Úlistr Únanr= r- ÚtmÚassert_frame_equalr r4 rX Ú transformr` ra r? Úcopyr r" ÚastypeÚpdÚcutr Úassert_series_equalZassert_produces_warningr, rq rr Úfilterr$ r rU r r rN ÚrandomÚrandintr ÚrandnÚasarrayr ÚdescribeÚcodesÚargsortÚtaker ÚrepeatÚassert_index_equalÚstackÚget_level_values) rB ÚdataÚ exp_indexÚexpectedr Úcat1Úcat2rA ÚgbZexp_idxr] Úgr rE ZgbcÚresult2Zresult3Zresult4Zresult5Úlevelsr ÚgroupedÚ desc_resultÚidxÚ
|
ord_labelsÚord_dataZexp_catsÚexpcÚexpr r r Ú
|
test_basicU sÐ
|
ý þ
|
|
ÿ"
|
$ ÿ" ÿ$
|
|
|
|
ÿ r© c C s¸ t t ddd¡ttddgtdgdgd dgd tdgd d
|
gd d }|jd g| d }t t ddd¡ttddgtdgdgd tdgd d
|
gd d }| d¡}t ||¡ d S )NrH é r rD rf r rJ rG ZIndex1ZIndex2)r¡ r r ©r r" ©Úlevelr; é )
|
r r r r r Úranger= Ú get_groupr r
)r; rA r r r r r r Útest_level_get_groupâ s& ýþ ýþ
|
r± c C sÚ t dgd dgd dddgd t d¡d } t| jdddgd
|
d | _| d ¡d ¡ }|jdd
|
d}ddddddg}t|dddgd
|
d }ddddddgt|g}t j
|
|d d gd}t dgd |dd}t ||¡ d S )NrS é rT ÚhighZmedÚlowr6 g (@)r8 ÚdoseZoutcomesTr r8 rµ r )r Zsort_remainingr rH r& ©r" rN )r r r r rµ r= Z value_countsr r r Ú from_arraysr r r ©rA r r" r r r r Ú(test_sorting_with_different_categoricalsý s ýÿr¹ r TFc C s t td| d}t tdddg| d}t t|¡}t|||d}|jdd gd
|
d }tj||gdd gd }td ddg|dgd}| dd ¡}t
|
||¡ | ¡ }t
|
||¡ | tj ¡}t
|
||¡ tj||gdd gd }td|d}| dd ¡}t
|
||¡ d S )NÚabcru Zaaar rD r )ÚmissingÚdenserU r» r¼ Tr: r r rG ç @rU )r" rZ c S s t j| ddS rn )r r- r\ r r r r^ # r_ z test_apply.<locals>.<lambda>rO c S s dS ©NrG r r\ r r r r^ / r_ )r r r r r r r= r r· r? r r
r- Úaggr r ) r r¼ r» rU rA r¢ r¤ r r r r r Ú
|
test_apply s" rÀ c C s t ddddgdddgdd}t ddddgdddgdd}t||d d
|
d d gd }ddgd
|
|d< |jdddg| d}tj||ddgd
|
gdddgd}tdtd d
|
d d g|di ¡ }| ¡ }| sàt|||ddggt ddd}t
|
||¡ |jddg| d}tj||gddgd}td d
|
d d gddddgd|d}| ¡ }| s\t|||gt ddd}t
|
||¡ t ddddgdddgddd d d
|
d
|
gd d d d gd }t|}|jd!| d} | ¡ }t t d"d!t d#dd$}td%d%gd&d gd'|d}| st t d#d!t d#dd$}
|
| |
|
¡}t
|
||¡ |jd!d(g| d} | d)¡}td*d+d&d,gt ddddgdddgddd d
|
d d
|
gd- d!d(g¡}| s t||jjd d
|
ggd!d(g}t
|
||¡ d.D ]<} | \} }| | ¡}||j| k|j|k@ }t
|
||¡ q°d d/d d/d d d gd d d d d0d1d2gddd3ddddgd4}t|}t |d t dd d ¡¡}||d5< |jd5d6gd7| d8}| d)¡}|jd5d6gd| d8}| d)¡ ¡ }t
|
||¡ d S )9Nr rD rP Tr rE rF rQ rG rH rI r6 rR rv rw r< rS rT r: r rU rO ÚABCr r )rU r< ÚABrf rg rh ri )ÚcatÚintsÚvalrà Úabrº )rN r r g ø?ç 4@)rÄ rÅ rÄ r- ç $@g >@g D@)rÅ rà rÄ ))r rG )rD rH )rD rG )r rH r~ é2 é< éF Úe©rv rw rx r¯ rx F©Úas_indexr; )r r r= r r· r r r4 r# r r r
r- r r r¿ Ú set_indexrà rU r° rÄ r r r ÚlinspaceÚ reset_index)r; r r rA r r r r rF Zgroups_single_keyr" Úgroups_double_keyÚkeyrE Úirà ÚgroupsZgroups2r r r Ú test_observed3 sÆ ÿ ÿ ÿ ÿ
|
ÿ
|
|
û ÿ ÿ
|
|
|
|
ÿ
|
ûÿø ÿ
|
ý
|
r× c C sØ ddddgddddgdddd gd
|
}t |}t |d dddd g¡}d |_|j|dg| d}tj|ddddggd dgd}t ddddgddddgd|d}| s¾t||jddddggd dg}| d¡}t
|
||¡ d S )NrI r6 rJ rG rH rf rz éÈ é" )ÚC1ÚC2ÚC3rÚ r² rà rÛ r: r g @g @g @rÈ g Y@g i@g A@)rÚ rÜ rO r- ) r r r rN r= r r· r# rU r¿ r r
)r; rF rA rU rÓ r¤ r r r r r Útest_observed_codes_remap¡ s$ $ ÿ ÿ
|
rÝ c C sÌ t tjjddddtjjddddtjjdddddd} | j t¡ d¡| d< | jdd d
|
gd d }| ¡ }|j j
|
d ¡ | j ¡ kst |j j
|
d ¡ | j ¡ ksªt |j j
|
d ¡ | j ¡ ksÈt d S )Nr éÿ i0u r{ é' )rà Úint_idÚother_idrv Úcategoryrà rà rá Tr: rG rH )r r r r rà r Ústrr= r& r" r¡ r0 r@ rà rá )rA r¢ r r r r Útest_observed_perfµ s üÿ rä c C s t dddgdddgd}t|dddgd}|jd | d
|
}|j}| rftd dgd d tdgd d d}n*td dgd d tg d d tdgd d d}t ||¡ d S )Nr rE rD ©r rG rH rI ©rà Úvalsrà r: r Úint64©Údtype)r rE ©r rD rE )r r r= rÖ r r Úassert_dict_equal)r; rà rA r r r r r r Útest_observed_groupsÊ s "
|
ýrí z,keys, expected_values, expected_index_levelsr rd é rG rH rI rb rD rV r~ r6 rJ r² Úa2Ú test_seriesc
|
C sö t tdddgdddgdtdddgdddgddddgdd d
|
gd d d g¡}d| krd|jdd}|j| dd}|r~|d }| ¡ }t| dkr|}n2dddddddddg ddddg g}t||| d}t d|i|d} |ræ| d } t || ¡ d S )NrG rH rI rå r6 rJ r² rV r~ rî )r rï rD rE r rï rD rY Fr: rE r )r r rO )
|
r r rÐ Údropr= r4 r r r Ú assert_equal)
|
ÚkeysÚexpected_valuesZexpected_index_levelsrð rA r r r" r r r r r Útest_unobserved_in_indexß s6 üÿù $ýrõ c C s t tdtjdgdddgddddgd}|jd | d
|
}|j}| rXdtd dgd d i}n(td dgd d tg d d tg d d d}t ||¡ d S )Nr rD rF rå rG rH rI ræ rà r: r rè ré )r rD rF ) r r r r r= rÖ r r rì )r; rA r r r r r r Ú test_observed_groups_with_nan s þÿ
|
|
ýrö c C sp t dtjtjgdddgd} tdddg}t| |d}|jd d
|
d d d ¡}|d jd g }t ||¡ d S )Nr rD rE rå rG rH rI )rà Úserrà Fr: r÷ r )
|
r r r r r r= r/ ra r r )rà r÷ rA r r r r r Útest_observed_nth* s rø c C s´ t tjdtjdgdddgd}tddddg}t||d }|jd
|
| d ¡ ¡ }| rxtt dgdddgddgd }n,tt dddgdddgddtjtjgd }t ||¡ d S ) Nr rD rE rå rG rH rI r6 )Ús1Ús2rù r: )
|
r r r r r r= r( rÒ r r
)r; rù rú rA r r r r r Ú#test_dataframe_categorical_with_nan5 s ÿ þÿrû r; r} c C sÀ t ddddddgddddg| d}tddddddg}t||d}|jd||dd d
|
¡}t|jjd d }t|j}|sd|| ¡ < t||ks¼d | d| d| d| }ds¼t |d S )NrF r rD r» r )ÚlabelrÅ rü )r; r} rÅ r( rc ré zDLabels and aggregation results not consistently sorted
|
for (ordered=z , observed=z, sort=z
|
)
|
Result:
|
F)
|
r r r r= Ú aggregater" ÚarrayÚisnar$ r@ ) r r; r} rü rÅ rA r ZaggrÚmsgr r r Ú0test_dataframe_categorical_ordered_observed_sortI s
|
ý
|
ÿr c C s t jddd} tjjdddd}tj|| dd}ttj dd¡}|j |d d
|
|
¡ }|j t |¡d d
|
|
¡ }| | ¡}t |j|jdd |_t ||¡ |j |d d
|
}| ¡ }|j ¡ }| |¡} | |¡}
|
|
|
j | d d
|
¡ }t ||¡ t |j|j¡ t |j d¡|j d¡¡ tjt d¡ d ¡| dd} t | } t | ¡ j d¡| ¡ td dddddddgd } t | ¡ j d¡| ¡ d S )Nz
|
2014-01-01r6 )Úperiodsr rz r{ Tru Fr: r r~ r& r- r3 r. r r r r, rG )r Ú
|
date_ranger r r r r r r r= r- r r r r" r r
r r r r r r r r r r ) r¡ r rB r r r r¢ r£ r¤ r¥ r¦ r§ r¨ r r r Ú test_datetimel s<
|
ÿ
|
|
|
|
|
ÿ r c C s( t j d¡} ddddg}| jdddd }tj||d
|
d }tt t d¡d¡ d d¡t
|
d d}||d< | d¡j ddd ¡ }|t
|
d j |jdd ¡ }ttjddddg|d
|
d dd|_t ||¡ |j ddd ¡ }|t
|
d j |jdd ¡ }ttjddddg|d
|
d dd|_t ||¡ d S )Ni90 rv rw rx ry r r6 rg r{ Tru éÿÿÿÿrL rY rB Fr¬ r: rG rH rI rb )r r Z RandomStater r r r r r Zreshaper rÐ r= r4 r r r" r r
)Úsr¡ r rB rA r r r r r Útest_categorical_index s( & ÿ ÿr c C sz t ddddgddddgdd} ttj dd¡| d }| d
|
d d dgd ¡ ¡ }t | ¡ j
|
| ¡ t | ¡ j
|
j | j ¡ d S )Nry rv rx rw Tr rg r6 rY rG rH rI rJ ) r r r r r r= r r r r rZ Úassert_categorical_equalrU )rB rA r r r r Ú!test_describe_categorical_columns¬ s
|
|
ýr c C sÊ t tdddgd tdd d} | d d ¡| d< | jd
|
dgd d d ¡ ¡ }| ¡ }tddgd dd}t
|
|j |¡ t |j j |j ¡ |d |d }tddgtddgd
|
dd}t ||¡ d S )Nrf rS rT rJ ZXYXXYrH )r ÚmediumÚartistr
|
râ r Fr: r ©r rN r² r6 ÚXÚYrb rO )r r¯ r r r= r& Úunstackr r r r rZ r rU r r r )rA Zgcatr Z exp_columnsr r r r Útest_unstack_categoricalº s ÿ r c C s^ t tjtjddddddddg
|
} t | ¡ jd¡}tjt dd |
|
|¡ ¡ W 5 Q R X d S )NrG rH rI r6 z$Grouper and axis must be same length©Úmatch) r r r r r ÚdropnarU ÚpytestÚraisesÚ
|
ValueErrorr= r- )Úseriesrj r r r Útest_bins_unequal_lenÍ s r r r ©rS rT c C s | d S r¾ r )r¤ r r r r^ Þ r_ r^ c C sD | ttddd¡}| t¡}t|t| ¡ d}t ||¡ d S )NZABBArâ ré rO )r= r r rý r ró r r )r r r= r r r r r Útest_categorical_series× s
|
r c s t tdddgdddgdddgdddgd jd
|
d gd d d ¡ } t tddg jjjdddgddgd d
|
d dgd}t | |¡ fdd} jd
|
|gd d d ¡ } t tddg jjjdddgddgd d
|
d dgd}t | |¡ tdddgd
|
d} jd
|
|gd d d ¡ } t | |¡ d
|
d g}t tddg jjjdddgddgd d
|
d dgd}dD ]8}t t
|
d|d _ j|d d d ¡ } t | |¡ q`d S )NrG rH rI rf é ée éf ég )rà rS rT rà rS FTrÎ rå éÍ rT rY c s j | df S )NrS )Úloc)Úr©rA r r r^ r_ z test_as_index.<locals>.<lambda>rª r rD rb )Nr rT rº ) r r r= r4 rà r r r
r r r r" )r r r r Z group_columnsrN r r" r Ú test_as_indexí sP ýÿýú ýú ýú r# c C s t d} tdtt d| ddi}t| | ddd}tt d| ddd}t |jdddd ¡ j|¡ t |jdddd ¡ j|¡ tdtt d| ddi}t| | ddd}tt dt dddd}t |jdddd ¡ j|¡ t |jdddd ¡ j|¡ d S )
|
Nrº rS ÚbaTr r ÚbacFr| ) r r r r r r r= r( r" )r rA r Z nosort_indexr r r Útest_preserve_categories% s. ÿ ÿ ÿ ÿr& c C sø t dddddgdddddgttdtd d
|
d ttdtd d d d } t ddtjgddtjgttd td d
|
d ttd td d d d }dD ]^}| j|d
|
d
|
djd d}| j|d d
|
djd d ¡ }|j|j d}t
|
||¡ t
|
||¡ qd S )NrG rH rf é rª é rÙ Zabaabr% Fr T)rS rT rÚ rÛ r½ g ð?g 9@rÇ )rÚ rÛ ©ÚbyrÏ r; rW rY ) r r r r r r= r- rÒ r rZ r r
)rA Zexp_fullÚcolZresult1r r r r r Ú test_preserve_categorical_dtypeC s2 üÿ
|
|
üÿÿÿÿ r, z func, valuesr( Úsecondr+ ÚfourthÚthirdr. r, c C s t ddddgdd}tddddg|d }| d
|
¡}t|| }tddgt||jd d d
|
¡}t ||¡ | d
|
¡d }t|| }|d }t ||¡ d S ) Nr( r- r/ r. Tru r éþÿÿÿ)Úpayloadr+ r1 ré r+ )
|
r r r= Úgetattrr rê rÐ r r
r )ÚfuncrU rE rA r r r Zsgbr r r Ú test_preserve_on_ordered_opsd s
|
ÿþ r4 c C s t tj d¡} t dddddddddg ¡}tj|dddgdd}| j|dd ¡ }| j|dd ¡ }t |j
|
|j |j d |_
|
t ||¡ t ddddddd
|
d
|
d
|
g ¡}tj|dddd
|
gdd}| j|dd ¡ }| j|dd ¡ |j ¡}t |j
|
|j |j d |_
|
t ||¡ td d d d d d d d d g d d d dgdd }tddddddd
|
ddg |d} | jd dd ¡ }|d j}t dddtjg¡}t ||¡ d S )Nrî r rG rH Tru Fr: r rI r rD rE rF r6 rJ rK )r r r r rþ r r r= r- r r" r r r r r r rU r Úassert_numpy_array_equal)r r rB r r¨ r r r Ú test_categorical_no_compress s> ÿ ÿ
|
ý
|
r6 c C sd t d gd tdddgd} | d¡ ¡ d }ttg ddgdtg ddd dd
|
}t ||¡ d S ) NrI ZtrainÚtestr rS rT rå rc ©rê rN r¶ )r r r= r( r r r ©rA r r r r r Ú test_groupby_empty_with_category¥ s ýr: c C sº t dtj ddd¡i} dd tdddD }t||}| jdgdd } tj| j tdd
|
dd |d | d < | j
|
d gd dd ¡ }|t |j dd d }t|j |j jd|_ t ||¡ d S )NÚvaluer rß rz c S s g | ]}| d |d qS )z - ió r )Ú.0rÕ r r r Ú
|
<listcomp>» s z test_sort.<locals>.<listcomp>iô T)r* Z ascendingi) F)Úrightrt Z value_groupr: c S s t | ¡ d S )Nr )ÚfloatÚsplitr\ r r r r^ Ä r_ ztest_sort.<locals>.<lambda>)rÔ rb )r r r r r¯ r Z sort_valuesr r r; r= r& Úsortedr" r rN r r )rA rt Z
|
cat_labelsÚresr¨ r r r Ú test_sort³ s
|
|
ÿ
|
rC c
|
C sô t dddgdddgdddgdd d
|
gdd d gd ddgdddggdddgd}t|d |d|d< |jd| dd ¡ }| r ddgddgd d
|
gddgg}d dddg}n(ddgddgd d
|
gddgg}dddd g}t |ddgt|d|dd}t ||¡ d S )Nz (7.5, 10]rf r~ rg z(2.5, 5]rJ rh z(5, 7.5]r² ri r6 rÉ z(0, 2.5]rG rÊ rV rË r¯ rv rw rY ru Fr| rM ©rZ r" )r r r= r( r r r
)r} r rA r Ú data_valuesÚ index_valuesr r r r Ú
|
test_sort2É s0 ù ö ýrG c C sf t tdddtdddtdddtdddtdddtdddtdddgddddd ddgdd
|
d d d ddgddddgd}t|d |d|d< | rÞddgdd gdd gddgg}tdddtdddtdddtdddg}nHddgdd gdd gddgg}tdddtdddtdddtdddg}t |ddgt|d|dd}|jd| dd ¡ }t ||¡ d S )NiÛ rV rG rH rJ rf r~ r² r6 rg rh ri rÉ rÊ rË )Údtrv rw rH rv rw rY ru rM rD Fr| )r r r r r= r( r r
)r} r rA rE rF r r r r r Útest_sort_datetimelikeì sF
|
|
|
|
|
|
|
ù õ ò
|
|
|
|
ü
|
|
|
|
ü ýrI c C s t tdddgdddgddddgd} tdddgdd }| jdd
|
d j ¡ }td dd g|dd }t ||¡ | jdd
|
d jjd d}td dd g|dd }t ||¡ | jdd
|
d jjdd}td dt j
|
g|dd }t ||¡ | jdd
|
d jjdd}td t j
|
t j
|
g|dd }t ||¡ d S )Nr rD rE rå rG rH r rS rb Fr: rI r rT ©Z min_count) r r r r= rT r4 r r r r r ©rA Z expected_idxr r r r r Útest_empty_sum s "ÿ rL c C sä t tdddgdddgddddgd} tdddgdd }| jdd
|
d j ¡ }tdddg|d d }t ||¡ | jdd
|
d jjd d}tdddg|d d }t ||¡ | jdd
|
d jjdd}tddt j
|
g|d d }t ||¡ d S )Nr rD rE rå rG rH r rS rb Fr: rT r rJ ) r r r r= rT r1 r r r r r rK r r r Útest_empty_prod; s "ÿ rM c C s¬ t ttdtttjddddd t d¡d} | dd g¡ ¡ }t j
|
td
|
d d gttjddddgdd gd }t ddddddddtj dg i|d}t ||¡ d S )NZ abcbabcbaz 2018-06-01 00Z1TrI )Úfreqr rî )Úkey1Úkey2rU rO rP r rD rE r rU r r6 r~ rJ r² rH rO )r r r r r r r r= r- r r r r r
)rA r r¤ r r r r Ú,test_groupby_multiindex_categorical_datetimeS s"
|
ÿûÿ þû$rQ zas_index, expectedrâ ré r r] )r" r rN ©r rD r] c C sV t tdddgdddddgdddgd}|jddg| d d
|
d ¡ }t ||¡ d S ) NrG rH râ ré rI rR r rD TrÎ r] )r r r= r4 r rò )rÏ r rA r r r r Ú,test_groupby_agg_observed_true_single_columnl s
|
$ÿ rS r c C sZ t ddddgddddgdd}t d dddgddddgdd}|jd| d}t ||¡ d S ) Nr rD rE rF Fr rG r )r Úshiftr rò )r Úctr rB r r r Ú
|
test_shift s
|
|
ÿ
|
|
ÿrV c C s\ | ¡ dd
}|d d¡|d< |d d¡|d< tddddg|d < |jd
|
gdd }|S ) a¥
|
DataFrame with multiple categorical columns and a column of integers.
|
Shortened so as not to contain all possible combinations of categories.
|
Useful for testing `observed` kwarg functionality on GroupBy objects.
|
|
Parameters
|
----------
|
df: DataFrame
|
Non-categorical, longer DataFrame from another fixture, used to derive
|
this one
|
|
Returns
|
-------
|
df_cat: DataFrame
|
Nr6 rS râ rT rG rH rI r< r> ro )r r r rñ )rA Údf_catr r r rW s rW Ú operationr¿ r? c C s t ddddg| d jdd}t ddddg| d jdd}t ||g¡}td d
|
d d g|d d ¡ }| jddgddd }t||t}t
|
||¡ d S )Nrw rv rS r8 ÚoneÚthreeÚtworT rH r6 rG rI r< ©r r" rN Tr: ) r rê r r· r r r= r2 r4 r r )rW rX Zlev_aZlev_br" r r¢ r r r r Ú test_seriesgroupby_observed_trueµ s r] c C s t jtddgddtdddgddgdd gd
|
¡ \}}td d tjd tjdg|dd}|dkrl|jddd}| jdd g|dd }t ||t
|
}t ||¡ d S )Nrw rv Fru rY rZ r[ rS rT r rH r6 rG rI r< r\ r¿ r Zinfer)Zdowncastr: ) r r r Z sortlevelr r r Zfillnar= r2 r4 r r )rW r; rX r" Ú_r r¢ r r r r Ú)test_seriesgroupby_observed_false_or_noneà s þû r_ zobserved, index, datarw rv rS r8 rY rZ r[ rT ru c C s> t ||dd}| jddg|dd dd ¡}t ||¡ d S )Nr< r\ rS rT r: c S s | ¡ | ¡ dS )N©r. r, r` r\ r r r r^ r_ z8test_seriesgroupby_observed_apply_dict.<locals>.<lambda>)r r= r? r r )rW r; r" r r r r r r Ú&test_seriesgroupby_observed_apply_dictØ s
|
.ÿra c C s< | ddg¡d ¡ }| ddg¡ ¡ d }t ||¡ d S )NrS rT r< )r= r- r r )rW r r r r r Ú4test_groupby_categorical_series_dataframe_consistent s rb Úcodec C sr t ddddgddddgd d
|
d d gd }tj| tdd}|j|dd ¡ }|jj|dd ¡ j}t ||¡ d S )NrG rH rI r6 r r0 éýÿÿÿéüÿÿÿrJ r² rV r~ rë rº rå ro r ) r r r r r= r- ÚTr r
)rc rA rà r r r r r Ú test_groupby_categorical_axis_1 s
|
(rg c C s\ t tddg|dddgdddgd }| ¡ }|jd| d
|
jt jd d ¡ }t ||¡ d S ) NZBobZGregru rG rH )ÚNameÚItemrh ri rY r: T)Zskipna) r r r r= r¿ r4 rÒ r r
)r; r rA r r r r r Ú$test_groupby_cat_preserves_structure s þ ÿÿrj c C sL t ddddgtdd} tjtdd | d¡ dd ¡ W 5 Q R X d S )
|
Nr rD r6 ©r5 rÅ z'vau'r r5 c S s& t | jd d g| jd d gdS )Nr r5 Zvaurk )r ra )Úrowsr r r r^ 4 s ÿz/test_get_nonexistent_category.<locals>.<lambda>)r r¯ r r ÚKeyErrorr= r? r" r r r Ú test_get_nonexistent_category/ s
|
|
ÿrn c
|
C sÀ | dkrt d¡ | dkr4t jjdd}|j |¡ tttdtddttd d
|
tddd gd d }t | |}|rd nd}|j
|
ddg|dd }t || }|| } t | |ks¼t d S )NÚngroupú ngroup is not truly a reductionr' ú6TODO: implemented SeriesGroupBy.corrwith. See GH 32293©ÚreasonÚAABBÚABCDrå r rH ç¹?r6 ©Úcat_1Úcat_2r; r' rx ry r: r; )r ÚskipÚmarkÚxfailÚnodeÚ
|
add_markerr r r r r= r2 r r@ )
|
Úreduction_funcr; Úrequestr{ rA r! Zexpected_lengthÚseries_groupbyr¿ r r r r Ú0test_series_groupby_on_2_categoricals_unobserved: s&
|
ÿ ýÿ
|
|
r c C s, | dkrt d¡ | dkr4t jjdd}|j |¡ tttdtddttd d
|
tddd gd d }t dt dt dt dt dg}t
|
| |}|j ddgddd }t || }|| }t | } |D ].}
|
|j|
|
} t | ¡rðt | ¡sÎ| | ksÎtqÎ| dkr(| dkr(t |jtj¡s(td S )Nro rp r' rq rr rt rÁ rå r rH rv r6 rw ZACZBCÚCAZCBÚCCrx ry Fr: r; r r4 )r rz r{ r| r} r~ r r r Útupler r= r2 Ú-_results_for_groupbys_with_missing_categoriesr r rÿ r@ r Z
|
issubdtyperê Úinteger) r r r{ rA Z
|
unobservedr! r r¿ r Z zero_or_nanr¤ rÅ r r r Ú?test_series_groupby_on_2_categoricals_unobserved_zeroes_or_nansW s0
|
ÿ ýÿ"
|
|
|
"r c C s | dkrt d¡ tttdtddttdtddddddgd }d
|
d d d g}|jddgdd}t| |}t|| | }|D ]}||jkst qd S )Nro ú2ngroup does not return the Categories on the indexrt rÁ rå Ú1111Ú12rv rw ©rS Ú2©rT r ©r< Ú1©r< r rx ry Tr: )
|
r rz r r r r= r r2 r" r@ )r rA Úunobserved_catsÚdf_grpr! rB rà r r r Ú>test_dataframe_groupby_on_2_categoricals_when_observed_is_true s
|
|
ýÿ
|
r c C sÌ | dkrt d¡ tttdtddttdtddddddgd }d
|
d d d g}|jddg|d}t| |}t|| | }t| }|t j
|
kr®|j | ¡ ¡ ¡ sÈtn|j | |k ¡ ¡ sÈtd S )Nro r rt rÁ rå r r rv rw r r r r rx ry r: )r rz r r r r= r r2 r r r r Zisnullr$ r@ )r r; rA r r r! rB r r r r Ú?test_dataframe_groupby_on_2_categoricals_when_observed_is_false s
|
|
ýÿ
|
|
r c C s ddddgddddgdd dd gd
|
} t | }t |d t d dd ¡¡}||d< |jddgddd}|d d¡}| d¡d }t ||¡ d S )Nrf r~ r6 rG rg rh ri rF rE rÍ rv r rJ r¯ rx T)rÏ r} r- ) r r r r rÑ r= r¿ r r )rF rA rà rÖ r r r r r Ú3test_series_groupby_categorical_aggregation_getitem¾ s $r zfunc, expected_valuesc C sv t dddddgdddddgtdddddgd d¡}| d¡ | ¡}t d |itdddgdd
|
d }t ||¡ d S ) Nr rG rH rI r6 )ÚidrÖ r; r rÖ r; rb rO )r r rÐ r= r¿ r r r
)r3 rô rA r r r r r Ú$test_groupby_agg_categorical_columnsÊ s ýÿú ÿr c C s t dtdddgdddgdi} t dddgit ddg¡d}| dddg¡ tj¡}t ||¡ | dddg¡ ¡ }t ||¡ d S ) NrS r rD rE rå rH rG rO )
|
r r r rþ r= r¿ r r0 r r
©rA r r r r r Ú test_groupby_agg_non_numericß s r r3 c C sl t dgtdgddj ¡ d}| d¡d }t|| }tdgtdgddd|d jd }t ||¡ d S )
|
NiÍ rD râ ré r rS rT rb ©r" rN rê )
|
r r rà Z
|
as_orderedr= r2 r rê r r )r3 rA Z
|
df_groupedr r r r r Ú<test_groupby_first_returned_categorical_instead_of_dataframeê s ÿr c C s~ t ddg¡} d| j_tddddgtddddgt| dd}td d
|
d git| d d d}|jd dd ¡ }t
|
||¡ d S )NrG rH FrI rJ rV rå rK r r½ g @rD rb r« ©r} ) r rþ ÚflagsZ writeabler r r r r= r- r r
)rB rA r r r r r Ú test_read_only_category_no_sortù s $ÿ r c
|
C sÖ t ddddddddgddddddddgd} | d d¡jjd dddgd
|
d | d< t d d d d dd ddd ddd d}|jddd}td dddgd dddgd
|
ddd|_| ddg¡ ¡ ¡ }t
|
||¡ d S )NÚsmallÚlarger
|
r< rS )rv rw rv râ ÚtinyTru r )rS r< rG rI rH )r¢ r r
|
r¡ rw r" ro )r r rN rê ) r r rà Zset_categoriesZ rename_axisr rZ r= r2 r r r
r r r r Ú#test_sorted_missing_category_values sJ ø
|
õÿÿ
|
þÿüÿ
|
|
ûr£ c C s t dddddgi} | d d¡| d< | d¡j ¡ }tdddgtdddgddd| d jd}t ||¡ | d¡
|
dd i¡}| ¡ }t ||¡ d S )
|
NZcol_numrG rH rI râ Úcol_catrb r r( ) r r r= r¤ r( r r rê r r r¿ Úto_framer
r9 r r r Ú1test_agg_cython_category_not_implemented_fallback2 s ü r¦ c C s® t ddddgddddgddtjdgddddgddddgd } | d
|
d i¡} | d d g¡ dd ¡}tjddgddggdd}t ddgddgddgd|d}t ||¡ d S )NrG rH rv gÉ?g333333Ó?rv rw Zfee)rS rT Ú numerical_colÚ
|
object_colÚcategorical_colr© râ rS rT c S s | ¡ ¡ S r[ )rÿ r4 r" r r r r^ V r_ z7test_aggregate_categorical_with_isnan.<locals>.<lambda>r r r )r§ r¨ r© r« )
|
r r r r r= r¿ r r· r r
r¸ r r r Ú%test_aggregate_categorical_with_isnanH s&
|
|
|
|
ûÿ
|
ýúrª c C sÔ t ddddddgddddddgd} tjdddgdd }| d
|
|¡| d
|
< | d ¡d
|
t¡| d < | ¡ }t ddddddgddddddgddddddgd }|d
|
|¡|d
|
< |d |¡|d < t ||¡ d S )NrG rH rI ZWaitingZOnTheWayZ Delivered)Ú
|
package_idÚstatusTr r¬ r« Ú last_status)r« r¬ r )
|
r r ÚCategoricalDtyper r= r r, r r r
)rA Zdelivery_status_typer r r r r Útest_categorical_transformc sN úþÿ ÿú úöÿr¯ )r3 r; c C sÒ t ddddg}ddddg}t|||d}t ddg}tj||gddgd}tdtjtjdg|ddtdtjtjdg|ddd }|| }|r | ¡ tj ¡}|j
|
ddg|d
|
d } t | | }
|
t |
|
|¡ d S © Nr rG rë r rD r rE rb )r( r+ r: )r r r r r r ÚNaNr r rè r= r2 r r ) r3 r; rà rÅ rA r r¤ Ú expected_dictr Zsrs_grpr r r r ÚFtest_series_groupby_first_on_categorical_col_grouped_on_2_categoricals s þ r³ c C sÒ t ddddg}ddddg}t|||d}t ddg}tj||gddgd}tdtjtjdg|ddtdtjtjdg|ddd }|| ¡ }|r¤| ¡ tj
|
¡}|j ddg|d
|
} t | | }
|
t |
|
|¡ d S r° )r r r r r r r± r¥ r r rè r= r2 r r
) r3 r; rà rÅ rA r r¤ r² r r r r r r ÚBtest_df_groupby_first_on_categorical_col_grouped_on_2_categoricals± s þ r´ c C s¤ t tdddgdddgdtdd} | jddd }|j}tjd
|
d gd d tjdgd d tjg d d d}| ¡ | ¡ ks~t| ¡ D ]}t
|
|| || ¡ qd S )NrD r rE rå rI )rÔ r+ rÔ Fr r rG Zintpré rH )rD r rE ) r r r¯ r= Úindicesr rþ ró r@ r r5 )rA r¢ r r rÔ r r r Ú2test_groupby_categorical_indices_unused_categoriesÊ s þÿ ý r¶ c C sp t ddddgi}|d d¡|d< t| d¡d | }ttdddgdtdddgddd}t ||¡ d S ) Nr rG rH rI râ rD rb )rN r" ) r r r2 r= r r r r r )r3 rA r r r r r Ú1test_groupby_last_first_preserve_categoricaldtypeÞ s ÿr· c C s t ddgddgddgd} | jdddd} | jd d
|
gd d ¡ d }tddgt tddgd dtddgd
|
dg¡d d}t ||¡ d S )NrG rH rf r rë râ rK ré r rD Tr: rE rb r¶ )
|
r r r= r0 r r r· r r r r9 r r r Ú)test_groupby_categorical_observed_nuniqueê s ÿûr¸ c C s t jddgdd} tddgddgddggddgd d| i¡}| d¡d ¡ }tddgtddgdd
|
dt jddgddd }t ||¡ d S ) Nr ÚbigTr rG rH ÚgrpÚ descriptionrY rb r )
|
r r® r r r= r, r r r r )rê rA r r r r r Ú,test_groupby_categorical_aggregate_functionsù s ÿþür¼ c C s° t ddgdddgd}tt ddgdddgdddgd}|jd| |d}| ¡ }| rltd ddgi|d
|
}n,tdddgdddg}td ddd gi|d
|
}d|j_t ||¡ d S ) NrG rH rI rå r6 )r] rQ r] )r; r rQ rO r ) r r r= r4 r r" rN r r
)r; r rà rA r r r r" r r r Ú test_groupby_categorical_dropna s $r½ Ú
|
index_kindr¯ ÚsingleÚmultic C s| |dkr0|s0|dkr0d}| j tjj|d¡ n@|dkrX|sXd}| j tjj|d¡ n|dkrp|sptjdd ttd d
|
d d gd
|
d d d g|d td d}|dkr¬dg} n@|dkrÆdg} | | ¡}n&|dkrìddg} |d |d< | | ¡}t
|
||}
|
|j | |||d} t | ||
|
} |r,| j d¡j} n | d jj} td
|
d d d g}t | |¡ |dkrx| j d¡j} t | |¡ d S )N)r) r* rÀ z1GH#10694 - idxmax/min fail with unused categoriesrr r' zDGH#49950 - corrwith with as_index=False may not have grouping columnr¯ ú/Result doesn't have categories, nothing to testrH rG rI r6 r rK r r¿ rï ©rÏ r} r; )r} r~ r r{ r| rz r r r¯ rÐ r r= r2 r" r r rà r r r )r rÏ r} r; r r¾ r r rA ró r! r Ú op_resultr r r r r Útest_category_order_reducer sL ÿþý þÿ
|
|
|
rÄ c C sæ t tddddgddddg|dtdd}|dkrFdg}| |¡}n&|d krldd
|
g}|d |d
|
< | |¡}t||}|j|| ||d } t| || }
|
|
|
j d¡j } t
|
ddddg} t | | ¡ |d krâ|
|
j d
|
¡j } t | | ¡ d S ) NrH rG rI r6 r rK r¿ r rÀ rï r ) r r r¯ rÐ r r= r2 r" r r r r r ) rÏ r} r; Ztransformation_funcr¾ r rA ró r! r rà r r r r r Ú test_category_order_transformerL s* þÿ
|
|
rÅ ÚmethodÚheadÚtailc C s t tddddgddddg|dtdd}|dkr<dg}n@|d krVdg}| |¡}n&|d
|
kr|dd g}|d |d < | |¡}|j|| ||d }t|| } |dkr°| d jj}
|
n| j d¡j}
|
t
|
ddddg} t |
|
| ¡ |d
|
krü| j d ¡j}
|
t |
|
| ¡ d S ) NrH rG rI r6 r rK r¯ r r¿ rÀ rï r ) r r r¯ rÐ r= r2 rà r r" r r r r © rÏ r} r; rÆ r¾ r rA ró r rà r r r r r Ú test_category_order_head_tailk s0 þÿ
|
rÊ r c C s: |dkr|dks | s&|dkr&t d¡ ttddddgddddg|dtdd }|dkrbd
|
g}n@|d kr|d
|
g}| |¡}n&|d kr¢d
|
d g}|d
|
|d < | |¡}|j|| ||d}t||dd } |dksÒ| sè|dkrè| d
|
jj }
|
n| j
|
d
|
¡j }
|
t ddddg} t |
|
| ¡ |d kr6| j
|
d ¡j }
|
t |
|
| ¡ d S )Nr r¯ z(No categories in result, nothing to testrH rG rI r6 r rK r r¿ rÀ rï r c S s | j ddS )NTrW )r4 r\ r r r r^ ¨ r_ z+test_category_order_apply.<locals>.<lambda>)r rz r r r¯ rÐ r= r2 rà r r" r r r r rÉ r r r Útest_category_order_apply s< ÿÿ
|
þÿ
|
|
rË c C sd |dkr| st jdd t ddd¡}tddddg||d }t|td
|
d }|dkr^d g}n@|d krxd g}| |¡}n&|dkrd dg}|d |d< | |¡}|j|| |dd}| ¡ } |rÆdddgndddg}
|
t
|
|
|
|j |d d} | r td|
|
i} |dkrt t| | d¡| _n| | _n6|dkrBtt| t| |
|
d} ntt| |
|
d } t | | ¡ d S )Nr¯ rÁ rr i' r rH rG rI r r6 rK r r¿ rÀ rï Tr )r r rN rD )r rï )r rï rD )r rz r r r r r¯ rÐ r= r4 r r r Z
|
from_framer" r r r
) rÏ r} r¾ r r ZgrouperrA ró r r r r" r r r r Útest_many_categoriesµ s@
|
ÿ
|
|
rÌ ró Úa1c C s |r&|dkr&t tdrtt d¡ n^|dkrJd}| j tjj|d¡ n:|dkr|st |dkr|s|sd}| j tjj|d¡ t
|
dddgd d
|
d
|
gd d d gd}| ddd¡}d|krÊ|j dd}|j |||d}|ræ|d }t||} |j|gf| }
|
t||| } |r0|s&|dkr0| |¡} |sf|sF| |¡} tdg|ggdgdggd| _n|sx||g | _t |
|
| ¡ d S )Nr' z*corrwith not implemented for SeriesGroupByz1GH#32293: attempts to call SeriesGroupBy.corrwithrr r0 rG z GH#52848 - raises a ValueErrorr rH rI r6 rJ r² )rÍ rï rD râ )rÍ rï rï rY r) rD r2 )r¡ r )Úhasattrr
|
r@ r rz r} r~ r{ r| r r r rñ r= r r¿ r2 r¥ rÐ r rZ r rò ) r rÏ r; r rð ró r rA r r! r r r r r Ú test_agg_listÝ sP ÿþ
|
ýüû"
|
|
|
|
|
ÿ
|
rÏ )hr Únumpyr r Zpandasr r r r r r r r Zpandas._testingZ_testingr Zpandas.core.groupby.genericr
|
Zpandas.tests.groupbyr r± r# r rC r© r± r¹ r{ Z parametrizerÀ r× rÝ rä rí rõ rö rø rû r r r r r r r¯ Úrenamer r# r& r, r4 r6 r: rC rG rI rL rM rQ r· rS r ZNaTrV ZfixturerW r] r_ r ra rb rg rj rn r r r r r r0 r& r r r r r£ r¦ rª r¯ rã Úboolr³ r´ r¶ r· r¸ r¼ r½ rÄ rÅ rÊ rË rÌ rÏ r r r r Ú<module> sÈ $ ä
|
n ýþýùþ " '
|
$ ùþ
|
|
8 !
|
|
|
|
üþ
|
%
|
"
|
2 ÿûþ ýÿþõþ
|
|
|
|
ýùÿ ó
|
ýú"ö
|
ýú"öäþ,
|
|
+
|
! þ
|
|
-5 ÿ ÿ
|
. !#
|
'
|
