# -*- coding: utf-8 -*-
# =============================================================================
# Nazev:   backfill_send_failed_v1.0.py
# Verze:   1.0
# Datum:   2026-06-17
# Popis:   Jednorazovy backfill. Pro Sent-Items dokumenty bez Internet
#          Message-ID (_id zacina "filename:") v Mongo emaily."vbuzalka@its.jnj.com"
#          precte z odpovidajici .msg na Toweru (/mnt/user/JNJEMAILS) MAPI
#          PrimarySendAccount (0x0E28) + SentRepresenting (0x0065) a nastavi:
#             send_account = hodnota 0x0E28
#             send_failed  = True, kdyz obsahuje buzalka.cz (SendAs past)
#             send_error   = "SendAs buzalka.cz (PrimarySendAccount)"
#          -> rychly dotaz na NEODESLANE:  { send_failed: true }
# Pouziti: python backfill_send_failed_v1.0.py            (dry-run / nahled)
#          python backfill_send_failed_v1.0.py --apply     (zapise do Mongo)
# Pozn.:   Cte .msg pres SFTP (paramiko) primo do pameti (BytesIO + olefile).
# =============================================================================

import io
import sys
import paramiko
import olefile
from pymongo import MongoClient

MONGO_URI = "mongodb://192.168.1.76:27017"
DBN, COL = "emaily", "vbuzalka@its.jnj.com"
TOWER_HOST, TOWER_USER, TOWER_PASS = "192.168.1.76", "root", "7309208104"
REMOTE_DIR = "/mnt/user/JNJEMAILS"


def ole_str(ole, tag4):
    for t in (tag4 + "001F", tag4 + "001E"):
        name = "__substg1.0_" + t
        if ole.exists(name):
            b = ole.openstream(name).read()
            if t.endswith("001F"):
                try:
                    return b.decode("utf-16-le").strip()
                except Exception:
                    pass
            for enc in ("cp1250", "cp1252", "utf-8", "latin-1"):
                try:
                    return b.decode(enc).strip()
                except Exception:
                    continue
    return ""


def main():
    apply = "--apply" in sys.argv
    cli = MongoClient(MONGO_URI)
    col = cli[DBN][COL]

    docs = list(col.find(
        {"jnj_folder": {"$regex": "Sent Items"}, "_id": {"$regex": "^filename:"}},
        {"_id": 1, "filename": 1, "to": 1, "subject": 1}))
    print(f"Sent-Items bez Message-ID (filename:): {len(docs)}")

    ssh = paramiko.SSHClient()
    ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
    ssh.connect(TOWER_HOST, username=TOWER_USER, password=TOWER_PASS, timeout=30)
    sftp = ssh.open_sftp()

    n_buz = n_other = n_missing = 0
    updates = []
    for d in docs:
        fn = d.get("filename")
        if not fn:
            n_missing += 1
            continue
        try:
            raw = sftp.open(f"{REMOTE_DIR}/{fn}", "rb").read()
            ole = olefile.OleFileIO(io.BytesIO(raw))
        except Exception as e:
            print(f"  !! {fn}: {e}")
            n_missing += 1
            continue
        try:
            send_acct = ole_str(ole, "0E28")
            sentrep = ole_str(ole, "0065")
        finally:
            ole.close()

        blob = (send_acct + " " + sentrep).lower()
        is_buz = "buzalka.cz" in blob
        if is_buz:
            n_buz += 1
            err = "SendAs buzalka.cz (PrimarySendAccount)"
        else:
            n_other += 1
            err = None
        updates.append((d["_id"], send_acct, is_buz, err, d.get("to")))

    sftp.close(); ssh.close()

    print(f"  buzalka.cz (send_failed=True): {n_buz}")
    print(f"  jine send-account (False):     {n_other}")
    print(f"  chybi soubor/cteni:            {n_missing}")
    print()
    for _id, acct, is_buz, err, to in updates[:10]:
        print(f"   [{'FAIL' if is_buz else 'ok  '}] {to}  | acct={acct[:60]}")
    if len(updates) > 10:
        print(f"   ... a dalsich {len(updates)-10}")

    if not apply:
        print("\n>>> DRY-RUN. Pro zapis spust s --apply")
        return

    n = 0
    for _id, acct, is_buz, err, to in updates:
        res = col.update_one({"_id": _id}, {"$set": {
            "send_account": acct,
            "send_failed": bool(is_buz),
            "send_error": err,
        }})
        n += res.modified_count
    print(f"\n>>> ZAPSANO (modified): {n}")


if __name__ == "__main__":
    main()