#!/usr/bin/env python3
# orange-audit.py — Phase 1 du ménage Orange : AUDIT EN LECTURE SEULE de la boîte
# thierry.tallagrand@wanadoo.fr par IMAP. Équivalent de gmail_audit.py (juin 2026).
#
# Ne modifie RIEN : sélection de la boîte en lecture seule, en-têtes lus avec
# BODY.PEEK (ne marque pas comme lu). Sorties dans le workspace :
#   orange-audit.md        synthèse par expéditeur / domaine / année, base de orange-rules.md
#   orange-audit-raw.json  une ligne par message (uid, from, date, lu, désabonnement)
#
# Le secret vient de /home/node/.openclaw/.env (ORANGE_IMAP_PASSWORD), jamais en clair ici.
# Usage (dans le conteneur) : python3 /home/node/.openclaw/workspace/orange-audit.py [dossier=INBOX]
import imaplib, re, json, sys, time, email, email.utils
from collections import defaultdict
from datetime import datetime

USER = "thierry.tallagrand@wanadoo.fr"
HOST, PORT = "imap.orange.fr", 993
WS = "/home/node/.openclaw/workspace"
FOLDER = sys.argv[1] if len(sys.argv) > 1 else "INBOX"
CHUNK = 1000

env = dict(l.strip().split("=", 1) for l in open("/home/node/.openclaw/.env") if "=" in l and not l.startswith("#"))
PWD = env["ORANGE_IMAP_PASSWORD"].strip().strip('"').strip("'")

t0 = time.time()
m = imaplib.IMAP4_SSL(HOST, PORT)
m.login(USER, PWD)
typ, n = m.select(f'"{FOLDER}"', readonly=True)
total = int(n[0])
typ, data = m.uid("search", None, "ALL")
uids = data[0].split()
print(f"{FOLDER} : {total} messages, {len(uids)} UID", flush=True)

RX_META = re.compile(rb'UID (\d+).*?FLAGS \((.*?)\).*?INTERNALDATE "([^"]+)"', re.S)
rows = []
for i in range(0, len(uids), CHUNK):
    chunk = uids[i:i + CHUNK]
    rng = f"{chunk[0].decode()}:{chunk[-1].decode()}"
    for attempt in range(3):
        try:
            typ, resp = m.uid("fetch", rng, "(UID FLAGS INTERNALDATE BODY.PEEK[HEADER.FIELDS (FROM DATE LIST-UNSUBSCRIBE)])")
            break
        except imaplib.IMAP4.abort:
            time.sleep(5); m = imaplib.IMAP4_SSL(HOST, PORT); m.login(USER, PWD); m.select(f'"{FOLDER}"', readonly=True)
    for part in resp:
        if not isinstance(part, tuple): continue
        meta, hdr = part
        mm = RX_META.search(meta)
        if not mm: continue
        uid, flags, idate = int(mm.group(1)), mm.group(2).decode(errors="replace"), mm.group(3).decode()
        msg = email.message_from_bytes(hdr)
        name, addr = email.utils.parseaddr(str(email.header.make_header(email.header.decode_header(msg.get("From", "")))) if msg.get("From") else "")
        try: d = email.utils.parsedate_to_datetime(idate.replace('"', ''))
        except Exception: d = None
        rows.append({"uid": uid, "from": addr.lower(), "name": name[:60], "date": d.strftime("%Y-%m-%d") if d else idate[:11],
                     "seen": "\\Seen" in flags, "unsub": bool(msg.get("List-Unsubscribe"))})
    print(f"  {min(i + CHUNK, len(uids))}/{len(uids)}  ({time.time() - t0:.0f}s)", flush=True)
m.logout()

json.dump(rows, open(f"{WS}/orange-audit-raw.json", "w"), ensure_ascii=False)

# --- agrégation ---
by_addr = defaultdict(lambda: {"n": 0, "unread": 0, "unsub": 0, "first": "9999", "last": "0000", "name": ""})
by_dom = defaultdict(lambda: {"n": 0, "unread": 0})
by_year = defaultdict(lambda: {"n": 0, "unread": 0})
for r in rows:
    a = by_addr[r["from"] or "(vide)"]; a["n"] += 1; a["unread"] += (not r["seen"]); a["unsub"] += r["unsub"]
    a["first"] = min(a["first"], r["date"]); a["last"] = max(a["last"], r["date"]); a["name"] = a["name"] or r["name"]
    dom = r["from"].split("@")[-1] if "@" in r["from"] else "(vide)"
    by_dom[dom]["n"] += 1; by_dom[dom]["unread"] += (not r["seen"])
    by_year[r["date"][:4]]["n"] += 1; by_year[r["date"][:4]]["unread"] += (not r["seen"])

unread_total = sum(1 for r in rows if not r["seen"])
L = [f"# Audit Orange — {FOLDER} — {datetime.now().strftime('%Y-%m-%d %H:%M')}", "",
     f"_Lecture seule, {len(rows)} messages lus en {time.time() - t0:.0f} s. Source : orange-audit-raw.json._", "",
     f"- Total : **{len(rows)}** messages, dont **{unread_total}** non lus ({100 * unread_total // max(1, len(rows))} %)",
     f"- Expéditeurs distincts : {len(by_addr)} ; domaines distincts : {len(by_dom)}",
     f"- Messages avec en-tête de désabonnement (newsletters) : {sum(1 for r in rows if r['unsub'])}", "",
     "## Par année", "", "| Année | Messages | Non lus |", "|---|---:|---:|"]
for y in sorted(by_year): L.append(f"| {y} | {by_year[y]['n']} | {by_year[y]['unread']} |")
L += ["", "## Top 40 domaines", "", "| Domaine | Messages | Non lus |", "|---|---:|---:|"]
for dom, v in sorted(by_dom.items(), key=lambda kv: -kv[1]["n"])[:40]: L.append(f"| {dom} | {v['n']} | {v['unread']} |")
L += ["", "## Top 150 expéditeurs", "", "| Expéditeur | Nom | Messages | Non lus | Newsletter | Premier | Dernier |", "|---|---|---:|---:|---:|---|---|"]
for addr, v in sorted(by_addr.items(), key=lambda kv: -kv[1]["n"])[:150]:
    L.append(f"| {addr} | {v['name'].replace('|', '/')} | {v['n']} | {v['unread']} | {v['unsub']} | {v['first']} | {v['last']} |")
open(f"{WS}/orange-audit.md", "w").write("\n".join(L) + "\n")
print(f"écrit : {WS}/orange-audit.md et orange-audit-raw.json — {len(rows)} messages, {len(by_addr)} expéditeurs")