From 3c1910922292876e635f7a7ccf8407909b0e6c3c Mon Sep 17 00:00:00 2001
From: Nose <goyimnose@nothing.blah>
Date: Wed, 21 Jan 2026 14:06:18 -0800
Subject: [PATCH] f

---
 cmdlet/download_file.py | 24 ++++++++++++++++++++----
 tool/ytdlp.py           | 34 ++++++++++++++++++++++++++++++++--
 2 files changed, 52 insertions(+), 6 deletions(-)

diff --git a/cmdlet/download_file.py b/cmdlet/download_file.py
index 23409ef..ff38a84 100644
--- a/cmdlet/download_file.py
+++ b/cmdlet/download_file.py
@@ -1126,12 +1126,20 @@ class Download_File(Cmdlet):
     ) -> bool:
         to_check = []
         if candidate_url:
-            to_check.append(candidate_url)
+            to_check.append(str(candidate_url))
         if extra_urls:
-            to_check.extend(extra_urls)
+            to_check.extend([str(u) for u in extra_urls if u])
+
+        # De-duplicate needles to avoid redundant DB searches.
+        seen = set()
+        unique_to_check = []
+        for u in to_check:
+            if u not in seen:
+                unique_to_check.append(u)
+                seen.add(u)
 
         return sh.check_url_exists_in_storage(
-            urls=to_check,
+            urls=unique_to_check,
             storage=storage,
             hydrus_available=hydrus_available,
             final_output_dir=final_output_dir
@@ -1146,8 +1154,16 @@ class Download_File(Cmdlet):
         final_output_dir: Path,
         **kwargs: Any,
     ) -> bool:
+        if not urls:
+            return True
+        unique_urls = []
+        seen = set()
+        for u in urls:
+            if u and u not in seen:
+                unique_urls.append(u)
+                seen.add(u)
         return sh.check_url_exists_in_storage(
-            urls=urls,
+            urls=unique_urls,
             storage=storage,
             hydrus_available=hydrus_available,
             final_output_dir=final_output_dir
diff --git a/tool/ytdlp.py b/tool/ytdlp.py
index 5ac5605..3da8d47 100644
--- a/tool/ytdlp.py
+++ b/tool/ytdlp.py
@@ -226,6 +226,8 @@ def is_url_supported_by_ytdlp(url: str) -> bool:
     return False
 
 
+_FORMATS_CACHE: Dict[str, tuple[float, List[Dict[str, Any]]]] = {}
+
 def list_formats(
     url: str,
     *,
@@ -242,6 +244,14 @@ def list_formats(
     if not is_url_supported_by_ytdlp(url):
         return None
 
+    # Cache format probes to avoid redundant network hits
+    cache_key = hashlib.md5(f"{url}|{no_playlist}|{playlist_items}|{cookiefile}".encode()).hexdigest()
+    now = time.monotonic()
+    if cache_key in _FORMATS_CACHE:
+        ts, result = _FORMATS_CACHE[cache_key]
+        if now - ts < 300: # 5 minute cache for formats
+            return result
+
     result_container: List[Optional[Any]] = [None, None]  # [result, error]
 
     def _do_list() -> None:
@@ -262,7 +272,7 @@ def list_formats(
                 ydl_opts["cookiefile"] = str(cookiefile)
             else:
                 # Best effort attempt to use browser cookies if no file is explicitly passed
-                ydl_opts["cookiesfrombrowser"] = "chrome"
+                ydl_opts["cookiesfrombrowser"] = ["chrome"]
 
             if no_playlist:
                 ydl_opts["noplaylist"] = True
@@ -302,9 +312,14 @@ def list_formats(
     if result_container[1] is not None:
         return None
 
+    if result_container[0] is not None:
+        _FORMATS_CACHE[cache_key] = (now, cast(List[Dict[str, Any]], result_container[0]))
+
     return cast(Optional[List[Dict[str, Any]]], result_container[0])
 
 
+_PROBE_CACHE: Dict[str, tuple[float, Dict[str, Any]]] = {}
+
 def probe_url(
     url: str,
     no_playlist: bool = False,
@@ -320,6 +335,14 @@ def probe_url(
     if not is_url_supported_by_ytdlp(url):
         return None
 
+    # Simple in-memory cache to avoid duplicate probes for the same URL/options in a short window.
+    cache_key = hashlib.md5(f"{url}|{no_playlist}|{cookiefile}".encode()).hexdigest()
+    now = time.monotonic()
+    if cache_key in _PROBE_CACHE:
+        ts, result = _PROBE_CACHE[cache_key]
+        if now - ts < 60: # 60 second cache
+            return result
+
     result_container: List[Optional[Any]] = [None, None]  # [result, error]
 
     def _do_probe() -> None:
@@ -340,6 +363,10 @@ def probe_url(
 
             if cookiefile:
                 ydl_opts["cookiefile"] = str(cookiefile)
+            else:
+                # Best effort fallback
+                ydl_opts["cookiesfrombrowser"] = ["chrome"]
+
             if no_playlist:
                 ydl_opts["noplaylist"] = True
 
@@ -381,6 +408,9 @@ def probe_url(
     if result_container[1] is not None:
         return None
 
+    if result_container[0] is not None:
+        _PROBE_CACHE[cache_key] = (now, cast(Dict[str, Any], result_container[0]))
+
     return cast(Optional[Dict[str, Any]], result_container[0])
 
 
@@ -678,7 +708,7 @@ class YtDlpTool:
                 # Add browser cookies support "just in case" if no file found (best effort)
                 # This uses yt-dlp's support for extracting from common browsers.
                 # Defaulting to 'chrome' as the most common path.
-                base_options["cookiesfrombrowser"] = "chrome"
+                base_options["cookiesfrombrowser"] = ["chrome"]
 
         # Special handling for format keywords
         if opts.ytdl_format == "audio":