Upload alpha_factory/personas/expression_compiler.py

alpha_factory/personas/expression_compiler.py

@@ -1,7 +1,7 @@
 """
-Expression Compiler v5 — Fixed per-component sign logic,
+Expression Compiler v6 — Fixed per-component sign logic,
 correct weight semantics (weights inside rank don't linearly combine),
-removed double-negation bugs.
+removed double-negation bugs, WIRED field validation.
 """
 import re
 from jinja2 import Environment, BaseLoader
@@ -30,7 +30,7 @@ ts_decay_linear(group_neutralize({{ comps[0] }}, {{ group_key }}), {{ bp.decay }
 """,
     "vol_scaled_shock": """
 {%- set c = bp.components[0] -%}
-ts_decay_linear(group_neutralize(zscore(ts_delta({{ c.fields[0] }}, {{ c.horizon_days }}) / (ts_std({{ c.fields[0] }}, {{ c.horizon_days * 4 }}) + 0.001)), {{ group_key }}), {{ bp.decay }})
+ts_decay_linear(group_neutralize(zscore(ts_delta({{ c.fields[0] }}, {{ c.horizon_days }}) / (ts_std({{ c.fields[0] }}, {{ c.horizon_days * 4 }}) + 0.001))), {{ group_key }}), {{ bp.decay }})
 """,
     "intraday_mr_decay": """
 {%- set c = bp.components[0] -%}
@@ -148,34 +148,6 @@ def _get_backfill(blueprint: Blueprint) -> int:
     return 10
 
 
-def _apply_component_signs(expression: str, blueprint: Blueprint) -> str:
-    """
-    Apply per-component sign corrections based on field sign conventions.
-    
-    Unlike the old code which blindly negated the entire expression,
-    this only adjusts individual field references based on their
-    registered sign_direction mismatch with the component's declared sign.
-    """
-    # For now, the templates handle sign_direction correctly via Jinja conditionals.
-    # This function is a hook for future per-component sign sweeps.
-    return expression
-
-
-def _apply_post_compile_rules(expression: str, blueprint: Blueprint) -> str:
-    """Mandatory post-compilation transforms."""
-    expr = expression.strip()
-
-    # Apply component-level sign corrections (now handled in templates)
-    expr = _apply_component_signs(expr, blueprint)
-
-    # ALWAYS wrap in ts_decay_linear (min 5, max 10)
-    decay = min(max(blueprint.decay, 5), 10)
-    if not expr.startswith("ts_decay_linear"):
-        expr = f"ts_decay_linear({expr}, {decay})"
-
-    return expr
-
-
 def _validate_expression(expr: str) -> list[str]:
     """Validate expression has no obvious errors. Returns list of issues."""
     issues = []
@@ -196,6 +168,26 @@ def _validate_expression(expr: str) -> list[str]:
     return issues
 
 
+def _apply_post_compile_rules(expression: str, blueprint: Blueprint) -> str:
+    """Mandatory post-compilation transforms."""
+    expr = expression.strip()
+
+    # ALWAYS wrap in ts_decay_linear (min 5, max 10)
+    decay = min(max(blueprint.decay, 5), 10)
+    if not expr.startswith("ts_decay_linear"):
+        expr = f"ts_decay_linear({expr}, {decay})"
+
+    # Validate after all transforms
+    issues = _validate_expression(expr)
+    if issues:
+        # Log but don't crash — some fields may be legitimate but not in our registry yet
+        import logging
+        logger = logging.getLogger(__name__)
+        logger.debug(f"Expression validation issues: {issues}")
+
+    return expr
+
+
 async def compile_expression(blueprint: Blueprint, llm: LLMClient) -> Expression:
     """Convert a Blueprint to a BRAIN expression."""
     
@@ -263,6 +255,13 @@ Output ONLY the expression."""
     if "ts_decay_linear" not in result.operators_used:
         result.operators_used.append("ts_decay_linear")
 
+    # Run field validation on LLM output and surface warnings
+    issues = _validate_expression(result.expression)
+    if issues:
+        import logging
+        logger = logging.getLogger(__name__)
+        logger.warning(f"LLM expression has issues: {issues}")
+
     return result