#!/usr/bin/env python3
"""Append collected runs to responses.jsonl, flagging pre-registered discriminators."""
import json, sys, os

BASE = "/home/claude/audit"
QS = json.load(open(os.path.join(BASE, "questions.json"), encoding="utf-8"))
INSTR = QS["instruction"]
QMAP = {q["id"]: q for q in QS["questions"]}

def _normalise(s):
    s = s.lower().replace(" ", " ").replace(" ", " ")
    s = s.replace(" per cent", "%").replace(" percent", "%").replace("percent", "%")
    return " ".join(s.split())

def flag(text, discs):
    norm = _normalise(text)
    return [d for d in discs if _normalise(d["match"]) in norm]

def add(rec):
    q = QMAP[rec["question_id"]]
    rec.setdefault("question_class", q["class"])
    rec.setdefault("prompt", f"{q['text']} {INSTR}")
    rec.setdefault("surface", "api")
    rec.setdefault("collected_via", "dataforseo/ai_optimization")
    rec.setdefault("error", None)
    rec.setdefault("note", None)
    rec["discriminator_hits"] = flag(rec["answer_text"], q.get("discriminators", [])) if rec.get("answer_text") else None
    order = ["question_id","question_class","prompt","system","model_reported","run","started_utc",
             "surface","collected_via","search_used","money_spent","error","note","answer_text",
             "citations","discriminator_hits"]
    out = {k: rec[k] for k in order if k in rec}
    for k in rec:
        if k not in out: out[k] = rec[k]
    with open(os.path.join(BASE, "responses.jsonl"), "a", encoding="utf-8") as fh:
        fh.write(json.dumps(out, ensure_ascii=False) + "\n")
    hits = "  ".join(f"[{h['match']} -> {h['diagnoses']}]" for h in (out["discriminator_hits"] or []))
    print(f"ok {out['question_id']} run {out['run']} {out['system']}: {len(out.get('citations') or [])} citations  {hits}")

if __name__ == "__main__":
    for path in sys.argv[1:]:
        for rec in json.load(open(path, encoding="utf-8")):
            add(rec)