#!/usr/bin/env python3
"""WP44 — live verification of the landmark records against the OpenAlex API (single-record lookups are free):
for each landmark id, fetch the live record, compare the normalised title with the snapshot title, and record the live
cited_by_count and publication_year. Records whose live title differs (ids merged or re-assigned since the snapshot, or
snapshot data errors) are flagged live_match = False and excluded from Part IX landmark statistics (they stay in the file).
Output: literature_synthesis/landmarks/landmarks_verified.csv + verification_report.json. Politeness: 8 requests/s max."""
import argparse, json, os, re, time, datetime as dt, unicodedata, requests, pandas as pd

UA = "WP44-UQO-systematic-mapping/0.1 (mailto:spbou4@icloud.com)"
norm = lambda s: re.sub(r"[^a-z0-9]+", "", unicodedata.normalize("NFKD", str(s or "")).encode("ascii", "ignore").decode().lower())


def main():
    ap = argparse.ArgumentParser(); ap.add_argument("--root", required=True); a = ap.parse_args(); R = a.root
    d = pd.read_csv(f"{R}/literature_synthesis/landmarks/landmarks.csv"); s = requests.Session(); rows = []
    for i, r in enumerate(d.itertuples()):
        for attempt in range(3):
            try:
                resp = s.get(f"https://api.openalex.org/works/W{int(r.work_id)}", params={"mailto": "spbou4@icloud.com"}, headers={"User-Agent": UA}, timeout=30); break
            except requests.RequestException:
                time.sleep(5)
        if resp.status_code == 200:
            j = resp.json(); lt = j.get("title") or j.get("display_name") or ""; lid = j.get("id", "").rsplit("W", 1)[-1]
            match = norm(lt) == norm(r.title) or (norm(lt) and norm(r.title) and (norm(lt) in norm(r.title) or norm(r.title) in norm(lt)))
            rows.append({"work_id": int(r.work_id), "live_status": 200, "live_id": lid, "live_title": lt, "live_year": j.get("publication_year"), "live_cited_by_count": j.get("cited_by_count"), "live_match": bool(match), "id_changed": lid != str(int(r.work_id))})
        else:
            rows.append({"work_id": int(r.work_id), "live_status": resp.status_code, "live_id": None, "live_title": None, "live_year": None, "live_cited_by_count": None, "live_match": False, "id_changed": None})
        time.sleep(0.13)
        if (i + 1) % 200 == 0:
            print(i + 1, "checked", flush=True)
    v = d.merge(pd.DataFrame(rows), on="work_id", how="left"); v.to_csv(f"{R}/literature_synthesis/landmarks/landmarks_verified.csv", index=False)
    rep = {"n": int(len(v)), "live_match": int(v.live_match.sum()), "mismatch": int((~v.live_match).sum()), "id_changed": int(v.id_changed.fillna(False).sum()), "http_not_200": int((v.live_status != 200).sum()),
           "when": dt.datetime.now(dt.UTC).isoformat(), "mismatch_examples": v[~v.live_match][["work_id", "title", "live_title", "c36"]].head(15).to_dict("records")}
    json.dump(rep, open(f"{R}/literature_synthesis/landmarks/verification_report.json", "w"), indent=1, default=str); print({k: rep[k] for k in ("n", "live_match", "mismatch", "id_changed", "http_not_200")})


if __name__ == "__main__":
    main()