Pyteryx/alteryx_runner/expression/functions.py

"""Mapping of Alteryx built-in functions to DuckDB SQL equivalents."""
from __future__ import annotations

# Single-argument function translations (name → SQL template using {0}, {1}, etc.)
FUNCTION_MAP: dict[str, str] = {
    # String
    "Uppercase": "UPPER({0})",
    "Lowercase": "LOWER({0})",
    "Trim": "TRIM({0})",
    "LTrim": "LTRIM({0})",
    "RTrim": "RTRIM({0})",
    "Length": "LENGTH({0})",
    "Left": "LEFT({0}, {1})",
    "Right": "RIGHT({0}, {1})",
    "Substring": "SUBSTR({0}, {1}, {2})",
    "FindString": "INSTR({0}, {1})",
    "ReplaceChar": "REPLACE({0}, {1}, {2})",
    "StringToDate": "STRPTIME({0}, {1})",
    "ToString": "PRINTF('%.' || {1} || 'f', {0})",
    "Contains": "CONTAINS({0}, {1})",
    "StartsWith": "STARTS_WITH({0}, {1})",
    "EndsWith": "ENDS_WITH({0}, {1})",
    "REGEX_Match": "REGEXP_MATCHES({0}, {1})",
    "REGEX_Replace": "REGEXP_REPLACE({0}, {1}, {2})",
    "PadLeft": "LPAD({0}, {1}, {2})",
    "PadRight": "RPAD({0}, {1}, {2})",
    "GetWord": "list_extract(str_split_regex({0}, '\\\\s+'), {1} + 1)",
    "CountWords": "array_length(str_split_regex(TRIM({0}), '\\\\s+'))",
    "CharFromInt": "chr({0}::INTEGER)",
    "IntFromChar": "ascii({0})",
    "ConvertFromCodePage": "{0}",
    "ReverseString": "reverse({0})",
    "DecomposeUnicodeForMatch": "strip_accents(UPPER({0}))",
    # Math
    "ABS": "ABS({0})",
    "Abs": "ABS({0})",
    "CEIL": "CEIL({0})",
    "Ceil": "CEIL({0})",
    "FLOOR": "FLOOR({0})",
    "Floor": "FLOOR({0})",
    "ROUND": "ROUND({0}, {1})",
    "Round": "ROUND({0}, {1})",
    "SQRT": "SQRT({0})",
    "Sqrt": "SQRT({0})",
    "POW": "POWER({0}, {1})",
    "Pow": "POWER({0}, {1})",
    "LOG": "LN({0})",
    "Log": "LN({0})",
    "LOG10": "LOG10({0})",
    "Log10": "LOG10({0})",
    "MOD": "({0} % {1})",
    "Mod": "({0} % {1})",
    "MIN": "LEAST({0}, {1})",
    "Max": "GREATEST({0}, {1})",
    "MAX": "GREATEST({0}, {1})",
    "Min": "LEAST({0}, {1})",
    "RandInt": "FLOOR(RANDOM() * {0})::BIGINT",
    "Random": "RANDOM()",
    "PI": "PI()",
    "SIN": "SIN({0})",
    "COS": "COS({0})",
    "TAN": "TAN({0})",
    "ASIN": "ASIN({0})",
    "ACOS": "ACOS({0})",
    "ATAN": "ATAN({0})",
    "ATAN2": "ATAN2({0}, {1})",
    "EXP": "EXP({0})",
    "Sign": "SIGN({0})",
    # Null handling
    "IsNull": "({0} IS NULL)",
    "IsEmpty": "({0} IS NULL OR {0} = '')",
    "NullConvert": "NULLIF({0}, '')",
    "Null": "NULL",
    # Type conversion
    "ToNumber": "TRY_CAST({0} AS DOUBLE)",
    "ToString_num": "CAST({0} AS VARCHAR)",
    "TOBOOL": "CAST({0} AS BOOLEAN)",
    # Date/Time
    "DateTimeNow": "NOW()",
    "DateTimeToday": "CURRENT_DATE",
    "DateTimeAdd": "({0} + INTERVAL ({1}) {2})",
    "DateTimeDiff": "DATEDIFF({2}, {1}, {0})",
    "DateTimeFormat": "STRFTIME({0}, {1})",
    "ToDate": "CAST({0} AS DATE)",
    "DateTimeYear": "YEAR({0})",
    "DateTimeMonth": "MONTH({0})",
    "DateTimeDay": "DAY({0})",
    "DateTimeHour": "HOUR({0})",
    "DateTimeMinute": "MINUTE({0})",
    "DateTimeSecond": "SECOND({0})",
    "DateTimeFirstOfMonth": "DATE_TRUNC('month', {0})",
    "DateTimeLastOfMonth": "(DATE_TRUNC('month', {0}) + INTERVAL '1 month' - INTERVAL '1 day')::DATE",
    "DateTimeFirstOfYear": "DATE_TRUNC('year', {0})",
    "DateTimeQuarter": "QUARTER({0})",
    "DateTimeTrim": "DATE_TRUNC({1}, {0})",
    # Conditional
    "IIF": "(CASE WHEN {0} THEN {1} ELSE {2} END)",
    "Switch": None,  # handled separately
    # Misc
    "TOPN": None,   # not a scalar function
}


def get_function_sql(name: str, args: list[str]) -> str:
    """Render a function call to DuckDB SQL given evaluated argument SQL strings."""
    # Case-insensitive lookup
    template = FUNCTION_MAP.get(name)
    if template is None:
        canon = name.lower()
        for k, v in FUNCTION_MAP.items():
            if k.lower() == canon:
                template = v
                break

    if template is None:
        # Unknown function — pass through as-is (may work in DuckDB natively)
        args_joined = ", ".join(args)
        return f"{name}({args_joined})"

    if name in ("Switch", "switch"):
        return _render_switch(args)

    try:
        result = template
        for i, arg in enumerate(args):
            result = result.replace(f"{{{i}}}", arg)
        return result
    except Exception:
        args_joined = ", ".join(args)
        return f"{name}({args_joined})"


def _render_switch(args: list[str]) -> str:
    """Switch(val, default, v1, r1, v2, r2, ...) → CASE val WHEN v1 THEN r1 ... ELSE default END"""
    if len(args) < 2:
        return "NULL"
    val = args[0]
    default = args[1]
    pairs = args[2:]
    cases = []
    for i in range(0, len(pairs) - 1, 2):
        cases.append(f"WHEN {pairs[i]} THEN {pairs[i+1]}")
    cases_sql = " ".join(cases)
    return f"CASE {val} {cases_sql} ELSE {default} END"


def titlecase_sql(col: str) -> str:
    """Approximate Titlecase via DuckDB: capitalise first letter of each word."""
    return (
        f"array_to_string(list_transform(str_split({col}, ' '), "
        f"x -> UPPER(LEFT(x,1)) || LOWER(SUBSTR(x,2))), ' ')"
    )