Q/Ph2 ddlZddlZddlmZddlmZmZmZmZm Z ddl Z ddl m Z ddlmZmZddlmZddlmZddlmZd ed ed efd Zd eded efdZd eded e eeffdZd eded e eeeffdZd eeded efdZdeed dfdZ deed dfdZ!dee"d e"fdZ# d'deedeed eee"eeffdZ$deed eee"eeffdZ%de eeeefdeed dfdZ&deded efdZ'ded ed efd!Z(d"ed efd#Z) d(de eeeefdeed%ed efd&Z*dS))N)asdict)AnyListOptionalTupleUnion) from_dict) CorrelationSettings)BaseDescription)Alert) ProfileReportv1v2returnct|ttfrdSt|tjr/t|tjr||St|tjr/t|tjr||S ||kS#t$rYdSwxYwNF) isinstancelistdictpd DataFrameequalsSeries ValueError)rrs _/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/ydata_profiling/compare_reports.py _should_wraprs"tTl##u"bl## 2r|(D(Dyy}}"bi  ZBI%>%>yy}}Rx uus2B88 CCd1d2c>ifdhhzDS)Nci|]K}|t||r||gnt||LS)r_update_merge_mixed).0krrs r z&_update_merge_dict..'sj    BqE2a5))31r!u~~$RUBqE22   r"rrs``r_update_merge_dictr)s\          rUUrU]      r'c,t|trt|tr||fSt|trt|trg||RSgt|tr|n|gt|tr|n|gSN)rrtupler(s r_update_merge_seqr-0s "d  2t 4 4  2v B   :b$#7#7 yRyy r4((2bbrd r4((2bbrd  r'ct|tr%t|trt||St||Sr+)rrr)r-r(s rr#r#AsE"d) 2t 4 4)!"b))) R(((r'c||St|trt|tstdt||S)NzLBoth arguments need to be of type dictionary (ProfileReport.description_set))rr TypeErrorr)r(s r _update_merger1HsV z b$   z"d';';  Z    b" % %%r'reportscd|D}d|D}|D]b}|D]9}|D]4}||jvr i|j|<||j|vrd|j||<5:|D]!}||jdvrd|jd|<"cdS)z?Generates placeholders in the dataset descriptions where neededc&h|]}|jD]}|Sr")scatterr$rvs r z _placeholders..Xs' 2 2 2! 2 21A 2 2 2 2r'c2h|]}|jdD]}|S)types)tabler6s rr9z _placeholders..Ys,>>>qQWW-=>>>>>>r'r;rN)r5r<)r2keys type_keysreportk1k2type_keys r _placeholdersrDUs 3 2w 2 2 2D>>G>>>I 4 4 0 0B 0 0V^++)+FN2&V^B///-/FN2&r*  0" 4 4Hv|G44423 W%h/ 4 4 4r'ct|D]4\}}|jjdkrdtd|z|j_5dS)z/Redefine title of reports with the default one.zYData Profiling ReportzDataset AN) enumerateconfigtitlechr)r2idxr@s r_update_titlesrLisU ))== V = ": : :"z!_compare_title..qs, 6 6%6!9  6 6 6 6 6 6r'rz, zComparing z and )alljoin)rMrIs` r_compare_titlerVpsn 6 6 6 66!"": 6 6 666Gay &"+&&FUFF&*FFFr'rHc"d|D}|tt|djjjjdkrFt |D]6\}}|jjjj|g|jjj_7n\t|jjjdkr:t |D]*\}}|jjj|jjj_+d|D}t ||D]\}}||j_||fS)Nc&g|]}|jjSr")rHrIr$r@s r z6_compare_profile_report_preprocess..}s 8 8 8ffm! 8 8 8r'rrRc6g|]}|Sr")get_descriptionrYs rrZz6_compare_profile_report_preprocess..s$CCCF**,,CCCr') lenrHhtmlstyleprimary_colorsrGzipanalysisrI)r2rHlabelsrKr@ descriptionslabel descriptions r"_compare_profile_report_preprocessrgxs( 9 8 8 8 8F~ wqz %+: ; ;a ? ?(11   VM&,;C@; "(77 v{ / 0 01 4 4(11   VK%4 "(77 DC7CCCL!&,77++{%* "" < r'c"d|D}||fS)Nc&g|]}|jjSr")rbrIrYs rrZz;_compare_dataset_description_preprocess..s : : :fo# : : :r'r")r2rcs r'_compare_dataset_description_preprocessrjs!; :' : : :F 7?r'configscdt|dkrtdt|dkrtjdd|D}t |t |krtdt |dtr*d|D}t |stdt |dtr d |Dn d |Dt fd Dstjd d Sd S)zValidate if the reports are comparable. Args: reports: two reports to compare input may either be a ProfileReport, or the summary obtained from report.get_description() z5At least two reports are required for this comparisonzsComparison of more than two reports is not supported. Reports may be produced, but may yield unexpected formatting.c0g|]}|jjjSr")vars timeseriesactive)r$cs rrZz$validate_reports..s >>>AF%,>>>r'zCComparison between timeseries and tabular reports is not supported.rc g|] }|jdu Sr+)dfr$r7s rrZz$validate_reports..s===14t+===r'z/Reports where not initialized with a DataFrame.c@g|]}t|jjSr")setrtcolumnsrus rrZz$validate_reports..s$777!C %%777r'cZg|](}t|j)Sr")rw variablesr>rus rrZz$validate_reports..s.===C ((**++===r'c30K|]}d|kVdSrPr")r$xfeaturess rrQz#validate_reports..s,22Ax{a222222r'zkThe datasets being profiled have a different set of columns. Only the left side profile will be calculated.N)r]rwarningswarnrTanyrr)r2rk report_typesis_df_availabler}s @rvalidate_reportsrsq 7||aPQQQ 7||a  L   ?>g>>>L <C ---- Q   '!*m,,P==W===?## PNOO O'!*m,,>77w777==W=== 2222222 2 2   =       r'rfcfd|jD|_fd|jD|_jjjjjjg}d|D}t|r|jng|_j jdkr|j nd|_ j j r|j ni|_ |S)aIApply the configuration for visualilzation purposes. This handles the cases in which the report description was computed prior to comparison with a different config Args: description: report summary config: the settings object for the ProfileReport Returns: the updated description c6i|]\}}j|||Sr")missing_diagramsr$r%r8rHs rr&z!_apply_config..s=A8OPQ8R 1r'cxi|]6\}}j|tdj3||7S)F) calculate) correlationsgetr rrs rr&z!_apply_config..sV    Aq   " "1kE&B&B&B&L M M  1   r'cg|]}|dk S)rr")r$ss rrZz!_apply_config..s&&&q1u&&&r'rN) missingitemsrsamplesheadtailrandomrsample duplicates interactions continuousr5)rfrHrs ` r _apply_configrs $,2244K    ,2244   K ~"FN$79NOG&&g&&&G/27||C++K"("3"81"<"< $281D1OW+--UWK r'alert alert_listc:tfd|DS)Nc3XK|]$}|jjko|jjkV%dSr+) column_name alert_type)r$ars rrQz$_is_alert_present..sP  **Oq|u?O/Or')r)rrs` r_is_alert_presentrs;   r' report_alertsc`ddlm}d|D}t|D]~\}}|D]v}|||t|D]I\}}||kr t|||s'||}d|_||Jwt |S)Nr)copycg|]}gSr"r")r$_s rrZz-_create_placehoder_alerts..s---!2---r'T)rrGappendr _is_emptyr,) rrfixedrKalertsrifix empty_alerts r_create_placehoder_alertsrs--}---E // , , V , ,E #J  e $ $ $#E** , ,388( a0@AA,"&$u++K,0K)JJ{+++  , , <<r'Fcomputec"t|dkrtdd|D}tt|dkrtdt |dt r d|D}nKd|D}g}|D]:}t }||}||;t||t |dt rp|dj j }|dd D]+fd |D} j j d d | f_ ,d |D}t|dkr|dSnt|dj}d} |dd D](fd |D} t| dkr| dz } )| dkr&t d |d } |d| _| Sd ||dn|t |dt rk|D]hjjjj} jj} |_| j_| jjj_|rd _it/d|Dr#t1|t3|\}}n;t/d|Drt5|\}}ntd|jj_t=|fd|D}t?d |d}|dd D]}t?||}tA|dd|dd<tC|d|d<tE|dsd |d<t d  } tGtH|| _| S)a Compare Profile reports Args: reports: two reports to compare input may either be a ProfileReport, or the summary obtained from report.get_description() config: the settings object for the merged ProfileReport compute: recompute the profile report using config or the left report config recommended in cases where the reports were created using different settings rz$No reports available for comparison.c,g|]}t|Sr")typerus rrZzcompare..s...T!WW...r'rRzThe input must have the same data type for all reports. Comparing ProfileReport objects to summaries obtained from the get_description() method is not supported.cg|] }|j Sr"rHrus rrZzcompare..s111Aqx111r'cLg|]!}tj|jd"S)ydata_profiling_config)jsonloadspackagerus rrZzcompare..s7   @ADJqy!9: ; ;   r')r2rkNc0g|]}|jjv|Sr")rtrxr$colr@s rrZzcompare..)s(WWWccVYEV>V>Vc>V>V>Vr'c*g|]}|jj|Sr")rtemptyrus rrZzcompare..+s!888QTZ81888r'cdg|],}|tjv*|-Sr")rrzr>rs rrZzcompare..2sAtF.Ks, C C:fm , , C C C C C Cr'c3@K|]}t|tVdSr+)rr rYs rrQzcompare..Ps, G GVZ 0 0 G G G G G Gr'cJg|]}tt| Sr")rr)r$d_configs rrZzcompare..[s+QQQq a 9 9::QQQr'rbrIrtime_index_analysis) data_classdata)%r]rrwr0rrr updaterrrtrxlocrrzr>_description_setrrHrorprqrIrTrLrgrjr^r__labelsrDr1rVrrr r )r2rHr report_dtypes all_configs configs_strc_str c_setting base_featurescols_2_comparenon_empty_reportsprofiletsmoderIrcrddescriptions_dictresr7rr@s @@rcomparers  7||q?@@@..g...M 3}  "" p   '!*m,, *11111   EL      * *E I!((//I   y ) ) ) )Wk::::'!*m,, - abbk 9 9FWWWW]WWWN aaa&78FII88g888 w<<1  1:  WQZ1668899 abbk ' 'F,N>""Q&&!Q&!  ! !#DQ@@@G'.qzG $NG ~a.%%''++-- gaj- 0 0 3! 3 3+6= + &  &+ #7= "-43.2F+ C C7 C C CCC   wA'7SS G Gw G G G G G FwOO  p   "(GL,QQQQLQQQd$5a$899C qrr "$$C##-c*og.FGGC OG-c(m<rs 4444444444444488888888111111......888888 S c d    3CD$ # 3 5u+=    ")C)S)U4u3D-E)))) &htn &$ &4 & & & &4404T4444(=D/=D====G49GGGGG"&  -  X   49d?+ +,    : / " 49d?+ +,& 4 &_(== >& IMd&  & & & & R_DUUu&"&hh 4 &_(== >h X hh hhhhhhr'