import os as PI_OS
import polars as PI_POLARS


# Import
def FC_IMPORT_TEXT(
        ZVFCI_ST_SOURCE_FILE,
        ZVFCI_ST_FOLDER,
        ZVFCI_ST_DELIMITER,
        ZVFCI_ST_PREFIX,
        ZVFCI_LI_NUM_COLS
    ):
    ZV_ST_FILE_PATH = PI_OS.path.join(ZVFCI_ST_FOLDER, ZVFCI_ST_SOURCE_FILE)

    # Try encodings in order (utf8 first, then common Windows/SAP exports, then latin1 as last resort)
    ZV_LI_ENCODINGS = ["utf8", "utf-16", "cp1252", "windows-1252", "latin1"]

    ZV_DF_HEADER = None
    ZV_ST_ENCODING_USED = None
    ZV_LAST_ERR = None

    for ZV_ST_ENC in ZV_LI_ENCODINGS:
        try:
            ZV_DF_HEADER = PI_POLARS.read_csv(
                ZV_ST_FILE_PATH,
                separator=ZVFCI_ST_DELIMITER,
                encoding=ZV_ST_ENC,
                has_header=True,
                n_rows=1,
                infer_schema_length=0,
                quote_char=None
            )
            ZV_ST_ENCODING_USED = ZV_ST_ENC
            break
        except Exception as ZV_E:
            ZV_LAST_ERR = ZV_E

    if ZV_DF_HEADER is None:
        raise RuntimeError(f"Could not read file with any encoding. Last error: {ZV_LAST_ERR}")

    ZV_LI_COLS = ZV_DF_HEADER.columns
    ZV_DI_SCHEMA_OVERRIDES = {ZV_COL: PI_POLARS.Utf8 for ZV_COL in ZV_LI_COLS}

    # Read full file with the detected encoding (as Utf8 for safe cleaning)
    ZV_DF = (
        PI_POLARS.read_csv(
            ZV_ST_FILE_PATH,
            separator=ZVFCI_ST_DELIMITER,
            encoding=ZV_ST_ENCODING_USED,
            ignore_errors=True,
            has_header=True,
            schema_overrides=ZV_DI_SCHEMA_OVERRIDES,
            infer_schema_length=0,
            null_values=[''],
            quote_char=None
        )
        .fill_null('')
    )

    # Prefix all columns in one go (instead of renaming inside a loop)
    ZV_DF = ZV_DF.rename({ZV_COL: f"{ZVFCI_ST_PREFIX}{ZV_COL}" for ZV_COL in ZV_DF.columns})

    # Clean + cast numeric columns (handles quotes, thousands separators, spaces, trailing minus, blanks)
    # Note: ZVFCI_LI_NUM_COLS should contain the *prefixed* names (e.g. "REGUH_RBETR"),
    # because we already prefixed columns above.
    for ZV_COL in ZVFCI_LI_NUM_COLS:
        if ZV_COL in ZV_DF.columns:
            ZV_DF = (
                ZV_DF
                .with_columns(
                    (
                        PI_POLARS.when((PI_POLARS.col(ZV_COL) == '') | (PI_POLARS.col(ZV_COL).is_null()))
                        .then(0.0)
                        .when(PI_POLARS.col(ZV_COL).str.ends_with('-'))
                        .then(
                            PI_POLARS.col(ZV_COL)
                            .cast(PI_POLARS.Utf8)
                            .str.replace_all('"', '')       # remove quotes
                            .str.replace_all(',', '')       # remove thousands separators
                            .str.replace_all(' ', '')       # remove spaces
                            .str.replace(r'-$', '')         # remove trailing minus
                            .cast(PI_POLARS.Float64, strict=False) * -1
                        )
                        .otherwise(
                            PI_POLARS.col(ZV_COL)
                            .cast(PI_POLARS.Utf8)
                            .str.replace_all('"', '')
                            .str.replace_all(',', '')
                            .str.replace_all(' ', '')
                            .cast(PI_POLARS.Float64, strict=False)
                        )
                        .fill_null(0.0)
                    )
                    .alias(ZV_COL)
                )
            )

    return ZV_DF