# -----------------------------------------------------------------------------
# Home-Ka — US real-estate aggregator (Groupe KA)
# Author: Simon-Pierre Boucher — contact@spboucher.ai
# connectors/sftp/drop.py : SFTP drop connector — brokerages/MLS vendors that
# push periodic CSV/XML exports to an SFTP folder. Downloads the newest
# matching file then dispatches to the csv/xml parser.
#
#   {"id": "acme_sftp", "connector_type": "sftp", "config": {
#       "host": "sftp.acmefeeds.com", "port": 22,
#       "username": "$ENV:ACME_SFTP_USER", "password": "$ENV:ACME_SFTP_PASS",
#       "key_path": "",                       # or private key file
#       "remote_dir": "/exports", "pattern": "listings_*.csv",
#       "format": "csv",                      # csv | xml
#       ... then the csv/xml family options (mapping, static, ...)}}
#
# Requires `paramiko` (in requirements.txt).
# -----------------------------------------------------------------------------
from __future__ import annotations

import fnmatch
import io

from ..base import BaseConnector
from ...schema import Listing


class SFTPDropConnector(BaseConnector):
    family = "sftp"

    def _download(self) -> bytes:
        try:
            import paramiko
        except ImportError:
            raise RuntimeError("paramiko missing — pip install paramiko")
        cfg = self.config
        client = paramiko.SSHClient()
        client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
        kw: dict = {"username": cfg.get("username", "")}
        if cfg.get("key_path"):
            kw["key_filename"] = cfg["key_path"]
        else:
            kw["password"] = cfg.get("password", "")
        client.connect(cfg["host"], port=int(cfg.get("port", 22)),
                       timeout=30, **kw)
        try:
            sftp = client.open_sftp()
            remote_dir = cfg.get("remote_dir", ".")
            pattern = cfg.get("pattern", "*")
            names = [n for n in sftp.listdir(remote_dir)
                     if fnmatch.fnmatch(n, pattern)]
            if not names:
                raise RuntimeError(f"{self.source_id}: no file matching "
                                   f"{pattern} in {remote_dir}")
            # newest file by mtime
            newest = max(names, key=lambda n: sftp.stat(f"{remote_dir}/{n}").st_mtime)
            buf = io.BytesIO()
            sftp.getfo(f"{remote_dir}/{newest}", buf)
            return buf.getvalue()
        finally:
            client.close()

    def fetch(self) -> list[Listing]:
        data = self._download()
        fmt = self.config.get("format", "csv")
        if fmt == "csv":
            from ..csv.feed import parse_csv
            return parse_csv(self.source_id,
                             data.decode(self.config.get("encoding", "utf-8-sig"),
                                         errors="replace"),
                             self.config)
        if fmt == "xml":
            from ..xml.feed import XMLFeedConnector
            conn = XMLFeedConnector(self.source_id,
                                    {**self.config, "path": None})
            import xml.etree.ElementTree as ET
            root = ET.fromstring(data)
            item_tag = self.config.get("item_tag", "listing")
            mapping = self.config.get("mapping") or {}
            static = self.config.get("static") or {}
            out = []
            for el in root.iter():
                if el.tag.rsplit("}", 1)[-1] != item_tag:
                    continue
                lst = conn._to_listing(el, mapping, static)
                if lst is not None:
                    out.append(lst)
            return out
        raise RuntimeError(f"{self.source_id}: unknown sftp format {fmt!r}")