rssVideos: Better sanitization of ytdl info

config/scripts/rssVideos

@@ -51,7 +51,7 @@ class SaveInfoPP(yt_dlp.postprocessor.common.PostProcessor):
     yt_dlp.process_ie_result() doesn't return a completely updated info dict,
     notably the extension is still the one before it realizes the files cannot
     be merged. So we use this PostProcessor to catch the info dict in its final
-    form and save it.
+    form and save what we need from it (it's not serializable in this state).
     """
 
     def __init__(self, rvelement: "RVElement") -> None:
@@ -59,7 +59,7 @@ class SaveInfoPP(yt_dlp.postprocessor.common.PostProcessor):
         super().__init__()
 
     def run(self, info: dict) -> tuple[list, dict]:
-        self.rvelement.ytdl_infos = info
+        self.rvelement.update_post_download(info)
         return [], info
 
 def parse_duration(string: str) -> int:
@@ -109,13 +109,13 @@ def format_duration(duration: int) -> str:
 class RVElement:
     parent: "RVDatabase"
     item: minidom.Element
-    was_downloaded: bool
+    downloaded_filepath: typing.Optional[str]
     watched: bool
 
     def __init__(self, parent: "RVDatabase", item: minidom.Element) -> None:
         self.parent = parent
         self.item = item
-        self.was_downloaded = False
+        self.downloaded_filepath = None
         self.watched = False
 
     def get_tag_data(self, tag_name: str) -> str:
@@ -166,8 +166,8 @@ class RVElement:
         if not self.parent.args.research and cache.is_researched:
             self.__dict__["ytdl_infos"] = cache.__dict__["ytdl_infos"]
             log.debug(f"From cache: {self}")
-        # if cache.was_downloaded:
-        #     self.was_downloaded = True
+        if cache.downloaded_filepath:
+            self.downloaded_filepath = cache.downloaded_filepath
         if cache.watched:
             self.watched = True
 
@@ -191,7 +191,6 @@ class RVElement:
 
     @functools.cached_property
     def ytdl_infos(self) -> typing.Optional[dict]:
-        # TODO Sanitize according to documentation
         log.info(f"Researching: {self}")
         try:
             infos = self.parent.ytdl_dry.extract_info(self.link, download=False)
@@ -201,14 +200,8 @@ class RVElement:
             # TODO Still raise in case of temporary network issue
             log.warning(e)
             infos = None
-        # Apparently that thing is transformed from a LazyList
-        # somewhere in the normal yt_dlp process
-        if (
-            infos
-            and "thumbnails" in infos
-            and isinstance(infos["thumbnails"], yt_dlp.utils.LazyList)
-        ):
-            infos["thumbnails"] = infos["thumbnails"].exhaust()
+        if infos:
+            infos = self.parent.ytdl_dry.sanitize_info(infos)
         # Save database once it's been computed
         self.__dict__["ytdl_infos"] = infos
         self.parent.save()
@@ -228,6 +221,8 @@ class RVElement:
     @property
     def filepath(self) -> str:
         assert self.is_video
+        if self.downloaded_filepath:
+            return self.downloaded_filepath
         return self.parent.ytdl_dry.prepare_filename(self.ytdl_infos)
 
     @property
@@ -242,9 +237,15 @@ class RVElement:
             with yt_dlp.YoutubeDL(self.parent.ytdl_opts) as ydl:
                 ydl.add_post_processor(SaveInfoPP(self))
                 ydl.process_ie_result(self.ytdl_infos, download=True)
-        self.was_downloaded = True
         self.parent.save()
 
+    def update_post_download(self, info: dict) -> None:
+        self.downloaded_filepath = self.parent.ytdl_dry.prepare_filename(info)
+
+    @property
+    def was_downloaded(self) -> bool:
+        return self.downloaded_filepath is not None
+
     def preload(self) -> None:
         assert self.is_video
         if self.downloaded:
@@ -628,7 +629,6 @@ def main() -> None:
     if args.action == "clean":
         database.clean()
     else:
-        database.attempt_clean()
         duration = 0
         for element in database.filter(args):
             if args.action == "download":