LMh UddlZddlZddlmZmZmZmZddlZddl m Z m Z de de dej fdZidejd ejd ejd ejd ejd ejdededededededejdejdejdejdejZee efed<ejejejejejdZee efed<d ddiiZee ee ee efffed<dZe j dd d!e de d"e d#edee deej eee!e fej fff d$Z"d%e dee d"e de#ee ffd&Z$d'ede dee de%fd(Z&d%e d#edefd)Z'dS)*N)AnyCallableOptionalUnion)PolarsFrameTypedataset_download sql_querypathreturnctj|5}tj||}|cdddS#1swxYwYdSN)sqlite3connectpl read_database)r r conndfs Y/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/kagglehub/polars_datasets.pywrapped_read_databasers   $  i . .s 9==.csv.tsvz.json.jsonl.parquet.featherz.sqlitez.sqlite3z.dbz.db3z.s3dbz.dl3z.xlsz.xlsxz.xlsbz.xlsmz.ods%SUPPORTED_READ_FUNCTIONS_BY_EXTENSION)rrrrr%SUPPORTED_SCAN_FUNCTIONS_BY_EXTENSION separator STATIC_KWARGS_BY_EXTENSIONz/Loading from a SQLite file requires a SQL query)polars_frame_type polars_kwargshandler r!cb|in|}tj|d}t|||\}}t ||} |t |||it ||} n'#t$r} d| } t| | d} ~ wwxYw|tj ur|tj urut| trL| D]6\} } t| tjr| | | <7n| } | S)aCreates polars LazyFrame(s) or DataFrame(s) from a file in the dataset Args: handle: (string) The dataset handle path: (string) Path to a file within the dataset polars_frame_type: (PolarsFrameType) Optional control for which Frame to return: LazyFrame or DataFrame. The default is PolarsFrameType.LAZY_FRAME. PolarsFrameType.LAZY_FRAME: We attempt to use a scan_* method if it's available for the provided file extension. Otherwise, we use a read_* method to produce a DataFrame and return the result after calling .lazy() on it. This satisfies the requested polars_frame_type as a LazyFrame, but does require loading the file in memory. PolarsFrameType.DATA_FRAME: We use whatever read_* method corresponds to the provided file extension and return the resulting DatFrame. polars_kwargs: (dict) Optional set of kwargs to pass to the polars `read_*` method while constructing the DataFrame(s) sql_query: (string) Argument to be used for SQLite files. Required when reading a SQLite file. See polars documentation for details: https://docs.pola.rs/api/python/stable/reference/api/polars.read_database.html Returns: - dict[int | str, LazyFrame] or dict[int | str, DataFrame] for Excel-like files with multiple sheets - A polars LazyFrame or DataFrame for all others Raises: ValueError: If the file extension is not supported or the file fails to read NzError reading file: )osr splitext_validate_io_functionr _build_args _build_kwargs Exception ValueErrorr DATA_FRAME LAZY_FRAME isinstancedictitemsr DataFramelazy)r"r r r!r file_extension io_function io_frame_typefilepathresulteread_error_messagekeyvalues rload_polars_datasetr<HsZJ(/BB]MW%%d++A.N!6~yRc!d!dK --H4 h : : NM::   4447A77+,,!34 2227HOLf7f7f fd # # #$llnn / / UeR\22/"'**,,F3K  / [[]]F Ms%A77 BBBr3cl|tvrBd|ddt}t|dt|}|tur|stt |t jus |tvr|t jfSt|t j fS)NzUnsupported file extension: 'z"'. Supported file extensions are: z, ) rjoinkeysr+rMISSING_SQL_QUERY_ERROR_MESSAGErr,rr-)r3r r extension_error_message read_functions rr'r'sBBB hN h h.2ii8]8b8b8d8d.e.e h h 011t;9.IM---i-8999O666.Pu:u:u9:: 1. A?C] ^^rBc&|tkr|gn||gSr )r)rBr r s rr(r(s"&;;;D66)TARRrCc>|tvrin t|}i||Sr )r)r3r! static_kwargss rr)r)s2$>>>D^_mDn .m -} --rC)(r%rtypingrrrrpolarsrkagglehub.datasetsrrstrr1rread_csv read_json read_ndjson read_parquetread_ipc read_excelrr/__annotations__scan_csv scan_ndjson scan_parquetscan_ipcrrboolr@r-intr<tupler'listr(r)rCrr[sf 111111111111@@@@@@@@ S > BK> BK> R\> bn >  >  >$>%> > !> "> !>$ BM%>& R]'>( R])>. R]/>2 BM3>%tCM':B K Kn >>%tCM':GM{\`NaEbDd3c4i0@+@&A!ABbbb"S*9)C @@@ @ @' @  @ } @ 2<eCHor|;< <=@@@@F__$,SM_FU_ 8_ $%____*SxSsSx}SQUSSSS .#.c.d......rC