Owg,ddlmZddlmZddlZddlmZddlmZddl m Z ddl m Z m Z ddlmZdd lmZdd lmZddlZdd lmZdd lmZmZdd lmZerddlmZGddeZy)) annotations) TYPE_CHECKINGN)using_pyarrow_string_dtype)lib)import_optional_dependency) ParserError ParserWarning)find_stack_level) pandas_dtype) is_integer) DataFrame)_arrow_dtype_mappingarrow_string_types_mapper) ParserBase) ReadBuffercLeZdZdZdfd Zd dZd dZd dZd dZd dZ xZ S) ArrowParserWrapperz7 Wrapper for the pyarrow engine for read_csv() c `t||||_||_|j yN)super__init__kwdssrc _parse_kwds)selfrr __class__s ]/var/www/horilla/myenv/lib/python3.12/site-packages/pandas/io/parsers/arrow_parser_wrapper.pyrzArrowParserWrapper.__init__%s*   c|jjd}|dn||_|jd}t|tr t dt |jd|_y)z? Validates keywords before passing to pyarrow. encodingNzutf-8 na_valuesz?The pyarrow engine doesn't support passing a dict for na_values)rgetr isinstancedict ValueErrorlistr!)rr r!s rrzArrowParserWrapper._parse_kwds,sc $yy}}Z8#+#3 IIk* i &Q dii 45rc0ddddddd}|jD]X\}}||jvs|jj|1|jj||j|<Z|j}t |t r|g}nd}||jd <|jjDcic]\}}||d vr||c}}|_|jjd }|t|r||jd <n|tjjk(rd|jd <n^|tjjk(rdd }||jd <n-|tjjk(rd|jd <|jjDcic]\}}||dvr||c}}|_d|jdv|jd<|j;d|jvr-|jdD cgc]} d|  c} |jd<|jdu|j |jn|jd|j d|_ycc}}wcc}}wcc} w)z: Rename some arguments to pass to pyarrow include_columns null_values escape_charignore_empty_lines decimal_point quote_char)usecolsr! escapecharskip_blank_linesdecimal quotecharNtimestamp_parsers) delimiterr-r*r+ on_bad_linesinvalid_row_handlerctjd|jd|jd|jt t y)Nz Expected z columns, but found z: ) stacklevelskip)warningswarnexpected_columnsactual_columnstextr r ) invalid_rows rhandle_warningz?ArrowParserWrapper._get_pyarrow_options..handle_warningksLMM#K$@$@#AAU&556b9I9I8JL%#3#5  "rcy)Nr9)_s rz9ArrowParserWrapper._get_pyarrow_options..vsr)r(r) true_values false_valuesr,r3strings_can_be_nullfskiprows)autogenerate_column_names skip_rowsr )returnstr)itemsrr"pop date_formatr#rN parse_optionscallablerBadLineHandleMethodERRORWARNSKIPconvert_optionsheaderr read_options) rmapping pandas_name pyarrow_namerQ option_name option_valuer5r@ns r_get_pyarrow_optionsz'ArrowParserWrapper._get_pyarrow_options:s )&' 4&%  *1 E %Kdii'DIIMM+,F,R*.))-- *D ,' E&& k3 '&-KK)4 %&.2YY__-> ) \'OP  % yy}}^4  # % ) \'  % 79DIIm "7sJ?J $ Jc0t|j}d}|j|j|jt ||_t|j|k7r>t t |t|jz |jz|_d}|j|_|j |j|\}}|jl|jj}t|jD]\}}t|r|j|||<n||jvrtd|d|jN|jj|||jj|fn6|j||jj|j|f\}} | ||j| ||<|j|=|j|dd|j4|s2dgt|j jz|j _|jt#|jt$rK|jj'D cic]\} } | |jvr | t)| !c} } |_ nt)|j|_ |j|j}|S|Scc} } w#t*$r} t| d} ~ wwxYw)z Processes data read in based on kwargs. Parameters ---------- frame: DataFrame The DataFrame to process. Returns ------- DataFrame The processed DataFrame. TNFzIndex z invalid)dropinplace)lencolumnsrYnamesranger&_do_date_conversions index_colcopy enumerater r%dtyper"astype set_indexindexr#r$rOr TypeError) rframenum_colsmulti_index_namedrC index_to_setiitemkey new_dtypekves r_finalize_pandas_outputz*ArrowParserWrapper._finalize_pandas_outputsu}}%  ;; zz!;;&!&xDJ4::(* "%3tzz?(B"CDtzzQ $)! JJEM,,U]]EB5 >> %>>..0L$T^^4 ,4d#&+mmD&9LO.$vdV8%<==::) ::>>$/;tzz~~d34#mmD14::>>%--PTBU3VW#C !,%*3Z%6%6y%Ac  JJsO! ,$ OOLtTO B{{"+<%)FS1B1B-C$C ! :: !$**d+!% 0 0 21EMM)|A& *$**5  $ TZZ0 u  $ m# $s$K6K<< L LLctj|rtd|Ds tdt |r tdy)Nc3<K|]}t|tywrr#rN.0xs r z7ArrowParserWrapper._validate_usecols..s0UAs1C0UzwThe pyarrow engine does not allow 'usecols' to be integer column positions. Pass a list of string column names instead.z=The pyarrow engine does not allow 'usecols' to be a callable.)r is_list_likeallr%rS)rr.s r_validate_usecolsz$ArrowParserWrapper._validate_usecolssO   G $S0UW0U-UP g O rc td}td}|j |jd i|j} |j|j|jd i|j|jd i|j |}|j&d }|tj(ur|j*} |j-} t/|j*j0D]Q\} } |j0j3| s"| j| | j5| j7| } S|j9| }|dk(r!|j;t<j> } n|d k(rLtA}t=jB||jE<|j;|j } n5tGr|j;tI } n|j;} |jK| S#t$r|jj dd}||j ||jj dt}tj|rtd|Ds t dwxYw#|j"$r}t%||d}~wwxYw) z Reads the contents of a CSV file into a DataFrame and processes it according to the kwargs passed in the constructor. Returns ------- DataFrame The DataFrame created from the CSV file. pyarrowz pyarrow.csvr(Nr)c3<K|]}t|tywrrrs rrz*ArrowParserWrapper.read..s6'( 1c"6rz9The 'pyarrow' engine requires all na_values to be strings)rZrRrX dtype_backend) types_mappernumpy_nullablerB)&rraConvertOptionsrXrqr"rsetrrrread_csvr ReadOptionsrZ ParseOptionsrR ArrowInvalidrr no_defaultschemafloat64rltypesis_nullfield with_typecast to_pandaspd ArrowDtyper Int64Dtypenullrrr})rpa pyarrow_csvrXincludenullstabler|r new_schemanew_typerv arrow_typerr dtype_mappings rreadzArrowParserWrapper.reads( 20?  !!# 8k88P4;O;OPO  (((4[44It7H7HI6k66L9K9KL / )E /2  CNN *Jzz|H!*5<<+=+=!>  :88##J/!+:++A.88B"J  JJz*E I %OOO?E . .12M')}}M"'') $OO1B1BOCE ' )OO1J1LOMEOO%E++E22k **../@$GG"&&w/((,,]CEBE##E*#6,163 O  , (a.a ' (s%HAJ)BJ&)K 8 KK )rzReadBuffer[bytes]rMNone)rMr)rrr rMr )rMr ) __name__ __module__ __qualname____doc__rrrar}rr __classcell__)rs@rrr s+ 6Y vGR F3rr) __future__rtypingrr:pandas._configr pandas._libsrpandas.compat._optionalr pandas.errorsrr pandas.util._exceptionsr pandas.core.dtypes.commonr pandas.core.dtypes.inferencer pandasrr pandas.io._utilrrpandas.io.parsers.base_parserrpandas._typingrrrBrrrsK" 5>5235)O3O3r