"""
Expression Mutator — Evolutionary alpha improvement.
Takes a base alpha expression and generates 5 structural variants.

Mutations:
1. Decay adjustment (±3 days)
2. Horizon change (different lookback window)
3. Neutralization swap (novel group key)
4. Vol-scaling wrapper
5. Sign flip (test opposite direction)
"""
import re
from ..data.brain_groups import ALT_GROUPS, PRODUCTION_GROUPS


def mutate_decay(expression: str, current_decay: int) -> list[dict]:
    """Generate decay variants."""
    variants = []
    for new_decay in [3, 5, 7, 10]:
        if new_decay == current_decay:
            continue
        # Replace the decay value in ts_decay_linear
        new_expr = re.sub(
            r"ts_decay_linear\((.+),\s*\d+\)",
            f"ts_decay_linear(\\1, {new_decay})",
            expression
        )
        if new_expr != expression:
            variants.append({
                "expression": new_expr,
                "mutation": f"decay_{current_decay}_to_{new_decay}",
                "decay": new_decay,
            })
    return variants


def mutate_horizon(expression: str) -> list[dict]:
    """Change lookback windows in ts_rank, ts_mean, ts_delta."""
    variants = []
    horizons = [21, 42, 63, 126, 252]
    
    # Find current horizon
    match = re.search(r"ts_(?:rank|mean|delta)\([^,]+,\s*(\d+)\)", expression)
    if not match:
        return []
    
    current = int(match.group(1))
    for h in horizons:
        if h == current:
            continue
        new_expr = re.sub(
            r"(ts_(?:rank|mean|delta)\([^,]+,\s*)\d+(\))",
            f"\\g<1>{h}\\2",
            expression,
            count=1  # Only change first occurrence
        )
        if new_expr != expression:
            variants.append({
                "expression": new_expr,
                "mutation": f"horizon_{current}_to_{h}",
                "decay": None,  # Keep original
            })
    return variants[:2]  # Max 2 horizon variants


def mutate_neutralization(expression: str) -> list[dict]:
    """Swap neutralization group key."""
    variants = []
    # Find current group key
    match = re.search(r"group_(?:neutralize|zscore|rank)\([^,]+,\s*([a-z0-9_]+)\)", expression)
    if not match:
        return []
    
    current_group = match.group(1)
    # Try novel groups
    novel_groups = [g for g in PRODUCTION_GROUPS if g.id != current_group][:3]
    
    for g in novel_groups:
        new_expr = expression.replace(current_group, g.id)
        if new_expr != expression:
            variants.append({
                "expression": new_expr,
                "mutation": f"group_{current_group}_to_{g.id}",
                "decay": None,
            })
    return variants[:2]


def mutate_vol_scale(expression: str) -> list[dict]:
    """Add volatility scaling to the signal."""
    # If expression has a raw field, wrap it in vol-normalization
    # Pattern: find the innermost field reference and divide by ts_std
    # BRAIN field IDs can contain uppercase (e.g., mdl77_2GlobalDev...)
    match = re.search(r"(ts_(?:rank|mean)\()([a-zA-Z][a-zA-Z0-9_]+)(,\s*\d+\))", expression)
    if not match:
        return []
    
    prefix, field, suffix = match.group(1), match.group(2), match.group(3)
    # Get the window from suffix
    window_match = re.search(r"(\d+)", suffix)
    if not window_match:
        return []
    window = int(window_match.group(1))
    vol_window = min(window * 2, 252)
    
    # Replace field with field / (ts_std(field, vol_window) + 0.001)
    vol_scaled = f"{field} / (ts_std({field}, {vol_window}) + 0.001)"
    new_expr = expression.replace(
        f"{prefix}{field}{suffix}",
        f"{prefix}{vol_scaled}{suffix}"
    )
    
    if new_expr != expression:
        return [{"expression": new_expr, "mutation": "vol_scaled", "decay": None}]
    return []


def mutate_sign_flip(expression: str) -> list[dict]:
    """Flip the sign of the entire expression (test opposite direction)."""
    # Extract inner from ts_decay_linear wrapper
    match = re.match(r"ts_decay_linear\((.+),\s*(\d+)\)$", expression)
    if match:
        inner_expr = match.group(1).strip()
        decay = match.group(2)
        
        if inner_expr.startswith("-(" ) and inner_expr.endswith(")"):
            # Remove negation
            flipped = inner_expr[2:-1]
        else:
            # Add negation
            flipped = f"-({inner_expr})"
        
        new_expr = f"ts_decay_linear({flipped}, {decay})"
        return [{"expression": new_expr, "mutation": "sign_flip", "decay": None}]
    
    return []


def generate_mutations(expression: str, decay: int = 5) -> list[dict]:
    """
    Generate all mutation variants from a base expression.
    Returns list of dicts with 'expression', 'mutation' type, and optional 'decay'.
    """
    all_variants = []
    
    # 1. Decay mutations
    all_variants.extend(mutate_decay(expression, decay))
    
    # 2. Horizon mutations
    all_variants.extend(mutate_horizon(expression))
    
    # 3. Neutralization mutations
    all_variants.extend(mutate_neutralization(expression))
    
    # 4. Vol-scaling
    all_variants.extend(mutate_vol_scale(expression))
    
    # 5. Sign flip
    all_variants.extend(mutate_sign_flip(expression))
    
    return all_variants