Create clarifier.py

clarifier.py

@@ -0,0 +1,27 @@
+# clarifier.py
+from __future__ import annotations
+from typing import Dict, Any, List
+
+def missing_inputs_questions(plan: Dict[str,Any], binding: Dict[str,Dict[str,Any]]) -> List[str]:
+    """
+    For each required input in plan['requires'], if binding[name]['match'] is None or score too low,
+    produce a specific question asking for the exact data or where it lives.
+    """
+    reqs = [r.get("input") or r.get("name") or "" for r in (plan.get("requires") or [])]
+    out: List[str] = []
+    for r in reqs:
+        b = binding.get(r, {})
+        if (not b) or (b.get("match") is None) or (float(b.get("score", 0.0)) < 0.46):
+            out.append(f"Where can I find **{r}** in your files (exact column name), or please provide its value?")
+    # Always end with planning constraints
+    if out:
+        out.append("Any constraints (time windows, cohorts, or routing/scheduling rules) I should apply?")
+    return out
+
+def render_phase1_markdown(questions: List[str]) -> str:
+    if not questions:
+        return "**Clarification Questions**\n- No data gaps detected. Please confirm to proceed with analysis."
+    out = ["**Clarification Questions**"]
+    for q in questions:
+        out.append(f"- {q}")
+    return "\n".join(out)