o ̳i @sUddlZddlZddlmZddlmZddlZddlm Z m Z de de dej fdd Z id ejd ejd ejd ejdejdejde de de de de de dejdejdejdejdejZee efed<ejejejejejdZee efed<d ddiiZee ee e eBffed <d!Ze jdd"d#e de d$e d%ede dBdej ee e Bej fBf d&d'Z!d(e de dBd$e de"ee ffd)d*Z#d+ede de dBde$fd,d-Z%d(e d%edefd.d/Z&dS)0N)Callable)Any)PolarsFrameTypedataset_download sql_querypathreturncCs>t|}t||}|WdS1swYdSN)sqlite3connectpl read_database)rrconndfrM/home/ubuntu/.local/lib/python3.10/site-packages/kagglehub/polars_datasets.pywrapped_read_databases  $r.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_kwargshandlerrc Cs|durin|}tj|d}t|||\}}t||}z|t|||it||} WntyC} z d| } t| | d} ~ ww|t j urn|t j urnt | t rj| D]\} } t | tjrg| | | <qW| S| } | 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: )osrsplitext_validate_io_functionr _build_args _build_kwargs Exception ValueErrorr DATA_FRAME LAZY_FRAME isinstancedictitemsr DataFramelazy)rrrrrfile_extension io_function io_frame_typefilepathresulteread_error_messagekeyvaluerrrload_polars_datasetIs0%        r8r/cCsr|tvrd|ddt}t|dt|}|tur$|s$tt|tjus-|tvr2|tjfSt|tj fS)NzUnsupported file extension: 'z"'. Supported file extensions are: z, ) rjoinkeysr'rMISSING_SQL_QUERY_ERROR_MESSAGErr(rr))r/rrextension_error_message read_functionrrrr#s    r#r=cCs|tkr|gS||gSr )r)r=rrrrrr$sr$cCs |tvrint|}i||Sr )r)r/r static_kwargsrrrr%s r%)'r!r collections.abcrtypingrpolarsr kagglehub.datasetsrrstrr-rread_csv read_json read_ndjson read_parquetread_ipc read_excelrr+__annotations__scan_csv scan_ndjson scan_parquetscan_ipcrrboolr;r)intr8tupler#listr$r%rrrrs        !(  C