From 4fef7fbf0042379989e62f70bfc1d82a20124113 Mon Sep 17 00:00:00 2001
From: grymphen <grymphen@gmail.com>
Date: Sun, 10 May 2026 17:42:24 +0200
Subject: [PATCH] sync errors

---
 sftp.py | 59 +++++++++++++++++++++++++++++----------------------------
 1 file changed, 30 insertions(+), 29 deletions(-)

diff --git a/sftp.py b/sftp.py
index 117ef9d..0033307 100644
--- a/sftp.py
+++ b/sftp.py
@@ -1,8 +1,7 @@
 import io
 import logging
-import posixpath
+import shlex
 import socket
-import stat
 from dataclasses import dataclass
 from pathlib import Path
 
@@ -81,22 +80,21 @@ def test_connection(cfg: SFTPConfig) -> tuple[bool, str]:
 
 def list_new_zips(cfg: SFTPConfig, max_results: int | None = None) -> list[RemoteZip]:
     transport = _make_transport(cfg)
-    sftp = paramiko.SFTPClient.from_transport(transport)
     try:
-        log.info("Walking remote directory tree from %s ...", cfg.remote_path)
+        all_zips = _find_remote_zips(transport, cfg.remote_path)
+        log.info("Remote scan done: %d zip(s) found", len(all_zips))
+
         new_zips: list[RemoteZip] = []
-        total_seen = 0
-        for zip_info in _walk_zips_iter(sftp, cfg.remote_path):
-            total_seen += 1
+        for zip_info in all_zips:
             if not db.is_zip_processed(zip_info.remote_path):
                 new_zips.append(zip_info)
                 if max_results and len(new_zips) >= max_results:
-                    log.info("Reached limit of %d — stopping walk early", max_results)
+                    log.info("Reached limit of %d", max_results)
                     break
-        log.info("Remote walk done: %d zip(s) seen, %d new", total_seen, len(new_zips))
+
+        log.info("%d new zip(s) to process", len(new_zips))
         return new_zips
     finally:
-        sftp.close()
         transport.close()
 
 
@@ -115,24 +113,27 @@ def download(cfg: SFTPConfig, remote_zip: RemoteZip, dest_dir: str) -> Path:
     return local_path
 
 
-def _walk_zips_iter(sftp: paramiko.SFTPClient, remote_dir: str):
-    log.info("Listing %s ...", remote_dir)
-    try:
-        entries = sftp.listdir_attr(remote_dir)
-    except IOError as e:
-        log.warning("Cannot list %s: %s", remote_dir, e)
-        return
+def _find_remote_zips(transport: paramiko.Transport, remote_path: str) -> list[RemoteZip]:
+    """Single SSH exec: find all .zip files server-side. Vastly faster than per-directory SFTP calls."""
+    channel = transport.open_session()
+    cmd = f"find {shlex.quote(remote_path)} -type f -iname '*.zip' -printf '%s\\t%p\\n'"
+    log.info("Running remote find under %s ...", remote_path)
+    channel.exec_command(cmd)
 
-    subdirs = []
-    zips_here = 0
-    for entry in entries:
-        full_path = posixpath.join(remote_dir, entry.filename)
-        if stat.S_ISDIR(entry.st_mode):
-            subdirs.append(full_path)
-        elif entry.filename.lower().endswith(".zip"):
-            yield RemoteZip(remote_path=full_path, file_size=entry.st_size or 0)
-            zips_here += 1
+    zips: list[RemoteZip] = []
+    for line in channel.makefile("r", -1):
+        line = line.rstrip("\n")
+        if "\t" not in line:
+            continue
+        size_str, path = line.split("\t", 1)
+        try:
+            zips.append(RemoteZip(remote_path=path, file_size=int(size_str)))
+        except ValueError:
+            continue
 
-    log.info("  %s: %d entries, %d zip(s), %d subdir(s)", remote_dir, len(entries), zips_here, len(subdirs))
-    for subdir in subdirs:
-        yield from _walk_zips_iter(sftp, subdir)
+    stderr_out = channel.makefile_stderr("r", -1).read().strip()
+    if stderr_out:
+        log.warning("find stderr: %s", stderr_out[:500])
+    channel.recv_exit_status()
+    channel.close()
+    return zips