AHjKdZddlZddlZddlmZddlmZddlmZddlm Z m Z ddl m Z ejeZej!d e ee d eezdzd ed ededef dZd edzd ed edededzf dZed d edefdZdedefdZdddedej0edej0ed edededeefdZdedede defd Zee d!ej6edej0ed"ed ed#ed$ededzfd%Zd&ed'ed edefd(Zdede dedzfd)Zd*eezdzd+edefd,Zd#eezdzdefd-Z d$eezdzdefd.Z!y)/z/ Filters for date parsing and date validators. N)Counter)datetime) lru_cache) CACHE_SIZEMIN_DATE) Extractorzminimum date setting: %s)maxsize date_input outputformatearliestlatestreturnc |yt|tr|}nN |dk(r2tt|ddt|ddt|dd}ntj||}|j |j cxkr|j kr:nn7|j |j cxkr|j kry tjd |y#t$rYywxYw) zEValidate a string w.r.t. the chosen outputformat and basic heuristicsNF%Y-%m-%d Tzdate not valid: %s) isinstancerintstrptime ValueErroryear timestampLOGGERdebug)r r r r dateobjects K/root/tools/cai/cai_env/lib/python3.12/site-packages/htmldate/validators.py is_valid_dater!s*h'  z)% 2A'Z!_)=s:aPRCS?T &..z<H  7FKK7    J$8$8$: Pf>N>N>P P Q LL%z2   sA C C+*C+c|6t||||r( tjd||j|Sy#t$r!}tj d||Yd}~yd}~wwxYw)z5Robust validation and conversion for plausible dates.Nzcustom parse result: %sz$value error during conversion: %s %s)r!rrstrftimererror)r r r rerrs r validate_and_convertr&:sp-L(F# R LL2J ?&&|4 4  R LL?S Q Q  Rs&: A$AA$ctddddd} |j|d|vrtj d |yy #ttf$r!}tj d||Yd}~yd}~wwxYw) z*Validate the output format in the settingsi rrz"wrong output format or type: %s %sNF%zmalformed output format: %sT)rr# TypeErrorrrr$)r rr%s r is_valid_formatr,Lss$1a+JL)  , 2LA  z " 9<Ms>A. A))A.rc(|dkr ||dk\rdndz }|S)z!Adapt year from YY to YYYY formatdZili)rs r correct_yearr1]s  cz  , KF) incomplete htmlstringpatternyearpatr3ct|j|}|j|j}}t|D]}} |j | } | t j d| || =0| d} |s t| } ntt| } || cxkr|krcnt j d| || =|S)z5Filter the date patterns to find plausible years onlyznot a year pattern: %srzno potential year: %s) rfindallrlistsearchrrrr1) r4r5r6r rr3 occurrencesmin_yearmax_yearitem year_match lastdigitspotential_years r plausible_year_filterrBds'//*56K! hH[!"^^D)   LL14 8D! ]  _N)#j/:N>5X5 LL0$ 7D!"" r2 referenceattemptoptionsc, ttj||jj }|jr|rt||}|S|}|St||}|S#t $r#}t jd|||cYd}~Sd}~wwxYw)z*Compare the date expression to a referencez-datetime.strptime exception: %s for string %sN) rrrformatr Exceptionrroriginalminmax)rCrDrErr%s r compare_valuesrLs))'7>>BLLNO 1:C 9-  AJ   9-   Dc7Ss7A'' B0BBB bestmatchcopyearmin_datemax_datec|fdj|d|d|dg}t|d||r:|dk(st|d|k\r$tj d ||t |d|Sy) z-Filter free text candidates in the YMD formatN-rrr rrzdate found for pattern "%s": %s)joinr!rrr convert_date)rMr5rNr rOrPpagedates r filter_ymd_candidaterYss88Yq\9Q<1FG : R qLC ! -8 LL:GX N*lC C r2 datestring inputformatc||k(r|St|tr|j|Stj||}|j|S)z.Parse date and return string in desired format)rrr#r)rZr[r rs r rWrWsPl"*h'""<00"":{;J   | ,,r2c|dkDr_tj|}|j|j}t ||j|j |j r|Sy)z4Test if the extracted reference date can be returnedrrUN)r fromtimestampr#rGr!rJrK)rCrEr converteds r check_extracted_referencer`sU1}++I6 ''7  w~~ GKK   r2 date_objectdefaultct|tr|St|tr tj|S|S#t$rt j d|Y|SwxYw)zTCheck if the input is a usable datetime or ISO date string, return default otherwisezinvalid datetime string: %s)rrstr fromisoformatrrwarning)rarbs r check_date_inputrgsa+x(+s# G))+6 6 N G NN8+ F N Gs;AAc"t|tS)zEValidates the minimum date and/or defaults to earliest plausible date)rgr)rOs r get_min_dateris Hh //r2chtjjdddd}t||S)a4Validates the maximum date and/or defaults to the end of the current day. A day-granular default stays stable across calls (unlike datetime.now()), which lets the date-validation caches be reused from one document to the next in batch processing, and accepts dates published earlier the same day.;i?B)hourminutesecond microsecond)rnowreplacerg)rP end_of_todays r get_max_daterts7 <<>)) 26*L Hl 33r2)"__doc__loggingre collectionsrr functoolsrsettingsrrutilsr getLogger__name__rrrdboolr!r&r,rr1PatternrBrLMatchrYrWr`rgrirtr0r2r rs *   8 $ '2 :"3%"""  "  ""J4    4Z $ 2 # $   ssZZ_ZZ_      S\B c C )   :xx} ZZ_      4Z& -S -s -# -# -  y S4Z  (S.4"7 ( x 08c>D00X0 48c>D04X4r2