dataframe-textual 0.3.2__py3-none-any.whl → 1.5.0__py3-none-any.whl

dataframe_textual/common.py

@@ -1,12 +1,20 @@
 """Common utilities and constants for dataframe_viewer."""
 
+import os
 import re
+import sys
 from dataclasses import dataclass
+from io import StringIO
+from pathlib import Path
 from typing import Any
 
 import polars as pl
 from rich.text import Text
 
+# Supported file formats
+SUPPORTED_FORMATS = {"tsv", "csv", "excel", "xlsx", "xls", "parquet", "json", "ndjson"}
+
+
 # Boolean string mappings
 BOOLS = {
     "true": True,
@@ -21,35 +29,59 @@ BOOLS = {
     "0": False,
 }
 
-# itype is used by Input widget for input validation
-# fmt: off
-STYLES = {
-    "Int64": {"style": "cyan", "justify": "right", "itype": "integer", "convert": int},
-    "Float64": {"style": "magenta", "justify": "right", "itype": "number", "convert": float},
-    "String": {"style": "green", "justify": "left", "itype": "text", "convert": str},
-    "Boolean": {"style": "blue", "justify": "center", "itype": "text", "convert": lambda x: BOOLS[x.lower()]},
-    "Date": {"style": "blue", "justify": "center", "itype": "text", "convert": str},
-    "Datetime": {"style": "blue", "justify": "center", "itype": "text", "convert": str},
-}
-# fmt: on
+# Special string to represent null value
+NULL = "NULL"
+NULL_DISPLAY = "-"
 
 
 @dataclass
-class DtypeConfig:
+class DtypeClass:
+    """Data type class configuration.
+
+    Attributes:
+        gtype: Generic, high-level type as a string.
+        style: Style string for display purposes.
+        justify: Text justification for display.
+        itype: Input type for validation.
+        convert: Conversion function for the data type.
+    """
+
+    gtype: str  # generic, high-level type
     style: str
     justify: str
     itype: str
     convert: Any
 
-    def __init__(self, dtype: pl.DataType):
-        dc = STYLES.get(
-            str(dtype), {"style": "", "justify": "", "itype": "text", "convert": str}
-        )
-        self.style = dc["style"]
-        self.justify = dc["justify"]
-        self.itype = dc["itype"]
-        self.convert = dc["convert"]
 
+# itype is used by Input widget for input validation
+# fmt: off
+STYLES = {
+    # str
+    pl.String: DtypeClass(gtype="string", style="green", justify="left", itype="text", convert=str),
+    # int
+    pl.Int8: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.Int16: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.Int32: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.Int64: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.Int128: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.UInt8: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.UInt16: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.UInt32: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    pl.UInt64: DtypeClass(gtype="integer", style="cyan", justify="right", itype="integer", convert=int),
+    # float
+    pl.Float32: DtypeClass(gtype="float", style="yellow", justify="right", itype="number", convert=float),
+    pl.Float64: DtypeClass(gtype="float", style="yellow", justify="right", itype="number", convert=float),
+    pl.Decimal: DtypeClass(gtype="float", style="yellow", justify="right", itype="number", convert=float),
+    # bool
+    pl.Boolean: DtypeClass(gtype="boolean", style="blue", justify="center", itype="text", convert=lambda x: BOOLS[x.lower()]),
+    # temporal
+    pl.Date: DtypeClass(gtype="temporal", style="magenta", justify="center", itype="text", convert=str),
+    pl.Datetime: DtypeClass(gtype="temporal", style="magenta", justify="center", itype="text", convert=str),
+    pl.Time: DtypeClass(gtype="temporal", style="magenta", justify="center", itype="text", convert=str),
+    # unknown
+    pl.Unknown: DtypeClass(gtype="unknown", style="", justify="", itype="text", convert=str),
+}
+# fmt: on
 
 # Subscript digits mapping for sort indicators
 SUBSCRIPT_DIGITS = {
@@ -68,53 +100,147 @@ SUBSCRIPT_DIGITS = {
 # Cursor types ("none" removed)
 CURSOR_TYPES = ["row", "column", "cell"]
 
-# Pagination settings
-INITIAL_BATCH_SIZE = 100  # Load this many rows initially
-BATCH_SIZE = 50  # Load this many rows when scrolling
+# For row index column
+RIDX = "^_ridx_^"
+
+
+@dataclass
+class Source:
+    """Data source representation.
+
+    Attributes:
+        frame: The Polars DataFrame or LazyFrame.
+        filename: The name of the source file.
+        tabname: The name of the tab to display.
+    """
+
+    frame: pl.DataFrame | pl.LazyFrame
+    filename: str
+    tabname: str
+
 
+def DtypeConfig(dtype: pl.DataType) -> DtypeClass:
+    """Get the DtypeClass configuration for a given Polars data type.
 
-def _format_row(vals, dtypes, apply_justify=True) -> list[Text]:
+    Retrieves styling and formatting configuration based on the Polars data type,
+    including style (color), justification, and type conversion function.
+
+    Args:
+        dtype: A Polars data type to get configuration for.
+
+    Returns:
+        A DtypeClass containing style, justification, input type, and conversion function.
+    """
+    if dc := STYLES.get(dtype):
+        return dc
+    elif isinstance(dtype, pl.Datetime):
+        return STYLES[pl.Datetime]
+    elif isinstance(dtype, pl.Date):
+        return STYLES[pl.Date]
+    elif isinstance(dtype, pl.Time):
+        return STYLES[pl.Time]
+    else:
+        return STYLES[pl.Unknown]
+
+
+def format_float(value: float, thousand_separator: bool = False, precision: int = 2) -> str:
+    """Format a float value, keeping integers without decimal point.
+
+    Args:
+        val: The float value to format.
+        thousand_separator: Whether to include thousand separators. Defaults to False.
+
+    Returns:
+        The formatted float as a string.
+    """
+
+    if (val := int(value)) == value:
+        return f"{val:,}" if thousand_separator else str(val)
+    else:
+        if precision > 0:
+            return f"{value:,.{precision}f}" if thousand_separator else f"{value:.{precision}f}"
+        else:
+            return f"{value:,f}" if thousand_separator else str(value)
+
+
+def format_row(vals, dtypes, styles=None, apply_justify=True, thousand_separator=False) -> list[Text]:
     """Format a single row with proper styling and justification.
 
+    Converts raw row values to formatted Rich Text objects with appropriate
+    styling (colors), justification, and null value handling based on data types.
+
     Args:
         vals: The list of values in the row.
         dtypes: The list of data types corresponding to each value.
         apply_justify: Whether to apply justification styling. Defaults to True.
+
+    Returns:
+        A list of Rich Text objects with proper formatting applied.
     """
     formatted_row = []
 
-    for val, dtype in zip(vals, dtypes, strict=True):
+    for idx, (val, dtype) in enumerate(zip(vals, dtypes, strict=True)):
         dc = DtypeConfig(dtype)
 
         # Format the value
         if val is None:
-            text_val = "-"
-        elif str(dtype).startswith("Float"):
-            text_val = f"{val:.4g}"
+            text_val = NULL_DISPLAY
+        elif dc.gtype == "integer" and thousand_separator:
+            text_val = f"{val:,}"
+        elif dc.gtype == "float":
+            text_val = format_float(val, thousand_separator)
         else:
             text_val = str(val)
 
         formatted_row.append(
             Text(
                 text_val,
-                style=dc.style,
+                style=styles[idx] if styles and styles[idx] else dc.style,
                 justify=dc.justify if apply_justify else "",
+                overflow="ellipsis",
+                no_wrap=True,
             )
         )
 
     return formatted_row
 
 
-def _rindex(lst: list, value) -> int:
-    """Return the last index of value in a list. Return -1 if not found."""
+def rindex(lst: list, value) -> int:
+    """Return the last index of value in a list. Return -1 if not found.
+
+    Searches through the list in reverse order to find the last occurrence
+    of the given value.
+
+    Args:
+        lst: The list to search through.
+        value: The value to find.
+
+    Returns:
+        The index (0-based) of the last occurrence, or -1 if not found.
+    """
     for i, item in enumerate(reversed(lst)):
         if item == value:
             return len(lst) - 1 - i
     return -1
 
 
-def _next(lst: list[Any], current, offset=1) -> Any:
-    """Return the next item in the list after the current item, cycling if needed."""
+def get_next_item(lst: list[Any], current, offset=1) -> Any:
+    """Return the next item in the list after the current item, cycling if needed.
+
+    Finds the current item in the list and returns the item at position (current_index + offset),
+    wrapping around to the beginning if necessary.
+
+    Args:
+        lst: The list to cycle through.
+        current: The current item (must be in the list).
+        offset: The number of positions to advance. Defaults to 1.
+
+    Returns:
+        The next item in the list after advancing by the offset.
+
+    Raises:
+        ValueError: If the current item is not found in the list.
+    """
     if current not in lst:
         raise ValueError("Current item not in list")
     current_index = lst.index(current)
@@ -122,83 +248,452 @@ def _next(lst: list[Any], current, offset=1) -> Any:
     return lst[next_index]
 
 
-def parse_filter_expression(
-    expression: str, df: pl.DataFrame, current_col_idx: int
-) -> str:
-    """Parse and convert a filter expression to Polars syntax.
+def parse_placeholders(template: str, columns: list[str], current_cidx: int) -> list[str | pl.Expr]:
+    """Parse template string into a list of strings or Polars expressions
+
+    Supports multiple placeholder types:
+    - `$_` - Current column (based on current_cidx parameter)
+    - `$#` - Row index (1-based, requires '^__ridx__^' column to be present)
+    - `$1`, `$2`, etc. - Column index (1-based)
+    - `$name` - Column name (e.g., `$product_id`)
+
+    Args:
+        template: The template string containing placeholders and literal text
+        columns: List of column names in the dataframe
+        current_cidx: 0-based index of the current column for `$_` references in the columns list
+
+    Returns:
+        A list of strings (literal text) and Polars expressions (for column references)
+
+    Raises:
+        ValueError: If invalid column index or non-existent column name is referenced
+    """
+    if "$" not in template or template.endswith("$"):
+        return [template]
+
+    # Regex matches: $_ or $\d+ or $\w+ (column names)
+    placeholder_pattern = r"\$(_|#|\d+|[a-zA-Z_]\w*)"
+    placeholders = re.finditer(placeholder_pattern, template)
+
+    parts = []
+    last_end = 0
+
+    # Get current column name for $_ references
+    try:
+        col_name = columns[current_cidx]
+    except IndexError:
+        raise ValueError(f"Current column index {current_cidx} is out of range for columns list")
+
+    for match in placeholders:
+        # Add literal text before this placeholder
+        if match.start() > last_end:
+            parts.append(template[last_end : match.start()])
+
+        placeholder = match.group(1)  # Extract content after '$'
+
+        if placeholder == "_":
+            # $_ refers to current column (where cursor was)
+            parts.append(pl.col(col_name))
+        elif placeholder == "#":
+            # $# refers to row index (1-based)
+            parts.append((pl.col(RIDX)))
+        elif placeholder.isdigit():
+            # $1, $2, etc. refer to columns by 1-based position index
+            col_idx = int(placeholder) - 1  # Convert to 0-based
+            try:
+                col_ref = columns[col_idx]
+                parts.append(pl.col(col_ref))
+            except IndexError:
+                raise ValueError(f"Invalid column index: ${placeholder} (valid range: $1 to ${len(columns)})")
+        else:
+            # $name refers to column by name
+            if placeholder in columns:
+                parts.append(pl.col(placeholder))
+            else:
+                raise ValueError(f"Column not found: ${placeholder} (available columns: {', '.join(columns)})")
+
+        last_end = match.end()
+
+    # Add remaining literal text after last placeholder
+    if last_end < len(template):
+        parts.append(template[last_end:])
 
-    Supports:
+    # If no placeholders found, treat entire template as literal
+    if not parts:
+        parts = [template]
+
+    return parts
+
+
+def parse_polars_expression(expression: str, columns: list[str], current_cidx: int) -> str:
+    """Parse and convert an expression to Polars syntax.
+
+    Replaces column references with Polars col() expressions:
     - $_ - Current selected column
-    - $1, $2, etc. - Column by 1-based index
-    - $col_name - Column by name
-    - Comparison operators: ==, !=, <, >, <=, >=
-    - Logical operators: &&, ||
-    - String literals: 'text', "text"
-    - Numeric literals: integers and floats
+    - $# - Row index (1-based, requires '^__ridx__^' column to be present)
+    - $1, $2, etc. - Column index (1-based)
+    - $col_name - Column name (valid identifier starting with _ or letter)
 
     Examples:
     - "$_ > 50" -> "pl.col('current_col') > 50"
+    - "$# > 10" -> "pl.col('^__ridx__^') > 10"
     - "$1 > 50" -> "pl.col('col0') > 50"
     - "$name == 'Alex'" -> "pl.col('name') == 'Alex'"
-    - "$1 > 3 && $name == 'Alex'" -> "(pl.col('col0') > 3) & (pl.col('name') == 'Alex')"
     - "$age < $salary" -> "pl.col('age') < pl.col('salary')"
 
     Args:
-        expression: The filter expression as a string.
-        df: The DataFrame to validate column references.
+        expression: The input expression as a string.
+        columns: The list of column names in the DataFrame.
+        current_cidx: The index of the currently selected column (0-based). Used for $_ reference.
+
+    Returns:
+        A Python expression string with $references replaced by pl.col() calls.
+
+    Raises:
+        ValueError: If a column reference is invalid.
+    """
+    # Early return if no $ present
+    if "$" not in expression:
+        if "pl." in expression:
+            # This may be valid Polars expression already
+            return expression
+        else:
+            # Return as a literal string
+            return f"pl.lit({expression})"
+
+    parts = parse_placeholders(expression, columns, current_cidx)
+
+    result = []
+    for part in parts:
+        if isinstance(part, pl.Expr):
+            col = part.meta.output_name()
+
+            result.append(f"pl.col('{col}')")
+        else:
+            result.append(part)
+
+    return "".join(result)
+
+
+def tentative_expr(term: str) -> bool:
+    """Check if the given term could be a Polars expression.
+
+    Heuristically determines whether a string might represent a Polars expression
+    based on common patterns like column references ($) or direct Polars syntax (pl.).
+
+    Args:
+        term: The string to check.
+
+    Returns:
+        True if the term appears to be a Polars expression, False otherwise.
+    """
+    if "$" in term and not term.endswith("$"):
+        return True
+    if "pl." in term:
+        return True
+    return False
+
+
+def validate_expr(term: str, columns: list[str], current_col_idx: int) -> pl.Expr | None:
+    """Validate and return the expression.
+
+    Parses a user-provided expression string and validates it as a valid Polars expression.
+    Converts special syntax like $_ references to proper Polars col() expressions.
+
+    Args:
+        term: The input expression as a string.
+        columns: The list of column names in the DataFrame.
         current_col_idx: The index of the currently selected column (0-based). Used for $_ reference.
 
     Returns:
-        A Python expression string that can be eval'd with Polars symbols.
+        A valid Polars expression object if validation succeeds.
 
     Raises:
-        ValueError: If the expression contains invalid column references.
-        SyntaxError: If the expression has invalid syntax.
+        ValueError: If the expression is invalid, contains non-existent column references, or cannot be evaluated.
     """
-    # Tokenize the expression
-    # Pattern matches: $_, $index, $identifier, strings, operators, numbers, etc.
-    token_pattern = r'\$_|\$\d+|\$\w+|\'[^\']*\'|"[^"]*"|&&|\|\||<=|>=|!=|==|[+\-*/%<>=()]|\d+\.?\d*|\w+|.'
-
-    tokens = re.findall(token_pattern, expression)
-
-    if not tokens:
-        raise ValueError("Expression is empty")
-
-    # Convert tokens to Polars expression syntax
-    converted_tokens = []
-    for token in tokens:
-        if token.startswith("$"):
-            # Column reference
-            col_ref = token[1:]
-
-            # Special case: $_ refers to the current selected column
-            if col_ref == "_":
-                col_name = df.columns[current_col_idx]
-            # Check if it's a numeric index
-            elif col_ref.isdigit():
-                col_idx = int(col_ref) - 1  # Convert to 0-based index
-                if col_idx < 0 or col_idx >= len(df.columns):
-                    raise ValueError(f"Column index out of range: ${col_ref}")
-                col_name = df.columns[col_idx]
-            else:
-                # It's a column name
-                if col_ref not in df.columns:
-                    raise ValueError(f"Column not found: ${col_ref}")
-                col_name = col_ref
+    term = term.strip()
+
+    try:
+        # Parse the expression
+        expr_str = parse_polars_expression(term, columns, current_col_idx)
+
+        # Validate by evaluating it
+        try:
+            expr = eval(expr_str, {"pl": pl})
+            if not isinstance(expr, pl.Expr):
+                raise ValueError(f"Expression evaluated to `{type(expr).__name__}` instead of a Polars expression")
+
+            # Expression is valid
+            return expr
+        except Exception as e:
+            raise ValueError(f"Failed to evaluate expression `{expr_str}`: {e}") from e
+    except Exception as ve:
+        raise ValueError(f"Failed to validate expression `{term}`: {ve}") from ve
+
+
+def load_dataframe(
+    filenames: list[str],
+    file_format: str | None = None,
+    has_header: bool = True,
+    infer_schema: bool = True,
+    comment_prefix: str | None = None,
+    quote_char: str | None = '"',
+    skip_lines: int = 0,
+    skip_rows_after_header: int = 0,
+    null_values: list[str] | None = None,
+    ignore_errors: bool = False,
+) -> list[Source]:
+    """Load DataFrames from file specifications.
+
+    Handles loading from multiple files, single files, or stdin. For Excel files,
+    loads all sheets as separate entries. For other formats, loads as single file.
 
-            converted_tokens.append(f"pl.col('{col_name}')")
+    Args:
+        filenames: List of filenames to load. If single filename is "-", read from stdin.
+        file_format: Optional format specifier for input files (e.g., 'csv', 'excel').
+        has_header: Whether the input files have a header row. Defaults to True.
+        infer_schema: Whether to infer data types for CSV/TSV files. Defaults to True.
+        comment_prefix: Character(s) indicating comment lines in CSV/TSV files. Defaults to None.
+        quote_char: Quote character for reading CSV/TSV files. Defaults to '"'.
+        skip_lines: Number of lines to skip when reading CSV/TSV files. Defaults to 0.
+        skip_rows_after_header: Number of rows to skip after header. Defaults to 0.
+        null_values: List of values to interpret as null when reading CSV/TSV files. Defaults to None.
+        ignore_errors: Whether to ignore errors when reading CSV/TSV files. Defaults to False.
 
-        elif token in ("&&", "||"):
-            # Convert logical operators and wrap surrounding expressions in parentheses
-            if token == "&&":
-                converted_tokens.append(") & (")
-            else:
-                converted_tokens.append(") | (")
+    Returns:
+        List of `Source` objects.
+    """
+    data: list[Source] = []
+    prefix_sheet = len(filenames) > 1
+
+    for filename in filenames:
+        if filename == "-":
+            source = StringIO(sys.stdin.read())
+            file_format = file_format or "tsv"
+
+            # Reopen stdin to /dev/tty for proper terminal interaction
+            try:
+                tty = open("/dev/tty")
+                os.dup2(tty.fileno(), sys.stdin.fileno())
+            except (OSError, FileNotFoundError):
+                pass
+        else:
+            source = filename
+
+        # Load from file
+        # Determine file format if not specified
+        if not file_format:
+            ext = Path(filename).suffix.lower()
+            if ext == ".gz" or ext == ".bz2" or ext == ".xz":
+                ext = Path(filename).with_suffix("").suffix.lower()
+            fmt = ext.removeprefix(".")
+
+            # Default to TSV
+            file_format = fmt if fmt in SUPPORTED_FORMATS else "tsv"
+
+        # Load the file
+        data.extend(
+            load_file(
+                source,
+                prefix_sheet=prefix_sheet,
+                file_format=file_format,
+                has_header=has_header,
+                infer_schema=infer_schema,
+                comment_prefix=comment_prefix,
+                quote_char=quote_char,
+                skip_lines=skip_lines,
+                skip_rows_after_header=skip_rows_after_header,
+                null_values=null_values,
+                ignore_errors=ignore_errors,
+            )
+        )
+
+    return data
+
+
+RE_COMPUTE_ERROR = re.compile(r"at column '(.*?)' \(column number \d+\)")
+
+
+def handle_compute_error(
+    err_msg: str,
+    file_format: str | None,
+    infer_schema: bool,
+    schema_overrides: dict[str, pl.DataType] | None = None,
+) -> tuple[bool, dict[str, pl.DataType] | None]:
+    """Handle ComputeError during schema inference and determine retry strategy.
+
+    Analyzes the error message and determines whether to retry with schema overrides,
+    disable schema inference, or exit with an error.
+
+    Args:
+        err_msg: The error message from the ComputeError exception.
+        file_format: The file format being loaded (tsv, csv, etc.).
+        infer_schema: Whether schema inference is currently enabled.
+        schema_overrides: Current schema overrides, if any.
+
+    Returns:
+        A tuple of (infer_schema, schema_overrides):
+
+    Raises:
+        SystemExit: If the error is unrecoverable.
+    """
+    # Already disabled schema inference, cannot recover
+    if not infer_schema:
+        print(f"Error loading even with schema inference disabled:\n{err_msg}", file=sys.stderr)
+
+        if "CSV malformed" in err_msg:
+            print(
+                "\nSometimes quote characters might be mismatched. Try again with `-q` or `-E` to ignore errors",
+                file=sys.stderr,
+            )
 
+        sys.exit(1)
+
+    # Schema mismatch error
+    if "found more fields than defined in 'Schema'" in err_msg:
+        print(f"Input might be malformed:\n{err_msg}.\nTry again with `-E` to ignore errors", file=sys.stderr)
+        sys.exit(1)
+
+    # ComputeError: could not parse `n.a. as of 04.01.022` as `dtype` i64 at column 'PubChemCID' (column number 16)
+    if file_format in ("tsv", "csv") and (m := RE_COMPUTE_ERROR.search(err_msg)):
+        col_name = m.group(1)
+
+        if schema_overrides is None:
+            schema_overrides = {}
+        schema_overrides.update({col_name: pl.String})
+    else:
+        infer_schema = False
+
+    return infer_schema, schema_overrides
+
+
+def load_file(
+    source: str | StringIO,
+    first_sheet: bool = False,
+    prefix_sheet: bool = False,
+    file_format: str | None = None,
+    has_header: bool = True,
+    infer_schema: bool = True,
+    comment_prefix: str | None = None,
+    quote_char: str | None = '"',
+    skip_lines: int = 0,
+    skip_rows_after_header: int = 0,
+    schema_overrides: dict[str, pl.DataType] | None = None,
+    null_values: list[str] | None = None,
+    ignore_errors: bool = False,
+) -> list[Source]:
+    """Load a single file.
+
+    For Excel files, when `first_sheet` is True, returns only the first sheet. Otherwise, returns one entry per sheet.
+    For other files or multiple files, returns one entry per file.
+
+    If a ComputeError occurs during schema inference for a column, attempts to recover
+    by treating that column as a string and retrying the load. This process repeats until
+    all columns are successfully loaded or no further recovery is possible.
+
+    Args:
+        filename: Path to file to load.
+        first_sheet: If True, only load first sheet for Excel files. Defaults to False.
+        prefix_sheet: If True, prefix filename to sheet name as the tab name for Excel files. Defaults to False.
+        file_format: Optional format specifier (i.e., 'tsv', 'csv', 'excel', 'parquet', 'json', 'ndjson') for input files.
+                     By default, infers from file extension.
+        has_header: Whether the input files have a header row. Defaults to True.
+        infer_schema: Whether to infer data types for CSV/TSV files. Defaults to True.
+        comment_prefix: Character(s) indicating comment lines in CSV/TSV files. Defaults to None.
+        quote_char: Quote character for reading CSV/TSV files. Defaults to '"'.
+        skip_lines: Number of lines to skip when reading CSV/TSV files. The header will be parsed at this offset. Defaults to 0.
+        skip_rows_after_header: Number of rows to skip after header when reading CSV/TSV files. Defaults to 0.
+        schema_overrides: Optional dictionary of column name to Polars data type to override inferred schema.
+        null_values: List of values to interpret as null when reading CSV/TSV files. Defaults to None.
+        ignore_errors: Whether to ignore errors when reading CSV/TSV files.
+
+    Returns:
+        List of `Source` objects.
+    """
+    data: list[Source] = []
+    filename = f"stdin.{file_format}" if isinstance(source, StringIO) else source
+    filepath = Path(filename)
+
+    # Load based on file format
+    if file_format in ("tsv", "csv"):
+        lf = pl.scan_csv(
+            source,
+            separator="\t" if file_format == "tsv" else ",",
+            has_header=has_header,
+            infer_schema=infer_schema,
+            comment_prefix=comment_prefix,
+            quote_char=quote_char,
+            skip_lines=skip_lines,
+            skip_rows_after_header=skip_rows_after_header,
+            schema_overrides=schema_overrides,
+            null_values=null_values,
+            ignore_errors=ignore_errors,
+        )
+        data.append(Source(lf, filename, filepath.stem))
+    elif file_format in ("xlsx", "xls", "excel"):
+        if first_sheet:
+            # Read only the first sheet for multiple files
+            lf = pl.read_excel(source).lazy()
+            data.append(Source(lf, filename, filepath.stem))
         else:
-            # Keep as-is (operators, numbers, strings, parentheses)
-            converted_tokens.append(token)
+            # For single file, expand all sheets
+            sheets = pl.read_excel(source, sheet_id=0)
+            for sheet_name, df in sheets.items():
+                tabname = f"{filepath.stem}_{sheet_name}" if prefix_sheet else sheet_name
+                data.append(Source(df.lazy(), filename, tabname))
+    elif file_format == "parquet":
+        lf = pl.scan_parquet(source)
+        data.append(Source(lf, filename, filepath.stem))
+    elif file_format == "json":
+        lf = pl.read_json(source).lazy()
+        data.append(Source(lf, filename, filepath.stem))
+    elif file_format == "ndjson":
+        lf = pl.scan_ndjson(source, schema_overrides=schema_overrides)
+        data.append(Source(lf, filename, filepath.stem))
+    else:
+        raise ValueError(f"Unsupported file format: {file_format}. Supported formats are: {SUPPORTED_FORMATS}")
+
+    # Attempt to collect, handling ComputeError for schema inference issues
+    try:
+        data = [Source(src.frame.collect(), src.filename, src.tabname) for src in data]
+    except pl.exceptions.ComputeError as ce:
+        # Handle the error and determine retry strategy
+        infer_schema, schema_overrides = handle_compute_error(str(ce), file_format, infer_schema, schema_overrides)
+
+        # Retry loading with updated schema overrides
+        if isinstance(source, StringIO):
+            source.seek(0)
+
+        return load_file(
+            source,
+            file_format=file_format,
+            has_header=has_header,
+            infer_schema=infer_schema,
+            comment_prefix=comment_prefix,
+            quote_char=quote_char,
+            skip_lines=skip_lines,
+            skip_rows_after_header=skip_rows_after_header,
+            schema_overrides=schema_overrides,
+            null_values=null_values,
+            ignore_errors=ignore_errors,
+        )
+
+    return data
+
+
+def now() -> str:
+    """Get the current local time as a formatted string."""
+    import time
+
+    return time.strftime("%m/%d/%Y %H:%M:%S", time.localtime())
+
+
+async def sleep_async(seconds: float) -> None:
+    """Async sleep to yield control back to the event loop.
+
+    Args:
+        seconds: The number of seconds to sleep.
+    """
+    import asyncio
 
-    # Join tokens with space to ensure proper separation
-    result = "(" + " ".join(converted_tokens) + ")"
-    return result
+    await asyncio.sleep(seconds)