reindex comments

tubearchivist/home/src/download/yt_dlp_handler.py

@@ -147,7 +147,7 @@ class DownloadPostProcess:
             return
 
         for video_id in self.download.videos:
-            comment = Comments(video_id)
+            comment = Comments(video_id, config=self.download.config)
             comment.build_json()
             comment.upload_comments()
 

tubearchivist/home/src/index/comments.py

@@ -20,12 +20,17 @@ class Comments:
         self.es_path = f"ta_comment/_doc/{youtube_id}"
         self.json_data = False
         self.config = config
+        self.is_activated = False
         self.comments_format = False
 
     def build_json(self):
         """build json document for es"""
         print(f"{self.youtube_id}: get comments")
         self._check_config()
+
+        if not self.is_activated:
+            return
+
         comments_raw = self.get_yt_comments()
         self.format_comments(comments_raw)
 
@@ -40,6 +45,8 @@ class Comments:
         if not self.config:
             self.config = AppConfig().config
 
+        self.is_activated = bool(self.config["downloads"]["comment_max"])
+
     def build_yt_obs(self):
         """
         get extractor config
@@ -109,6 +116,9 @@ class Comments:
 
     def upload_comments(self):
         """upload comments to es"""
+        if not self.is_activated:
+            return
+
         _, _ = ElasticWrap(self.es_path).put(self.json_data)
 
         vid_path = f"ta_video/_update/{self.youtube_id}"
@@ -117,7 +127,7 @@ class Comments:
 
     def delete_comments(self):
         """delete comments from es"""
-        _, _ = ElasticWrap(self.es_path).delete()
+        _, _ = ElasticWrap(self.es_path).delete(refresh=True)
 
     def get_es_comments(self):
         """get comments from ES"""
@@ -126,4 +136,19 @@ class Comments:
             print(f"comments: not found {self.youtube_id}")
             return False
 
-        return response
+        return response.get("_source")
+
+    def reindex_comments(self):
+        """update comments from youtube"""
+        if not self.is_activated:
+            return
+
+        self.build_json()
+        es_comments = self.get_es_comments()
+
+        if not self.comments_format and es_comments["comment_comments"]:
+            # don't overwrite comments in es
+            return
+
+        self.delete_comments()
+        self.upload_comments()

tubearchivist/home/src/index/reindex.py

@@ -16,6 +16,7 @@ from home.src.download.yt_dlp_base import CookieHandler
 from home.src.download.yt_dlp_handler import VideoDownloader
 from home.src.es.connect import ElasticWrap
 from home.src.index.channel import YoutubeChannel
+from home.src.index.comments import Comments
 from home.src.index.playlist import YoutubePlaylist
 from home.src.index.video import YoutubeVideo
 from home.src.ta.config import AppConfig
@@ -147,8 +148,7 @@ class Reindex:
         if integrate_ryd:
             self._get_unrated_vids()
 
-    @staticmethod
-    def _reindex_single_video(youtube_id):
+    def _reindex_single_video(self, youtube_id):
         """refresh data for single video"""
         video = YoutubeVideo(youtube_id)
 
@@ -182,6 +182,8 @@ class Reindex:
         thumb_handler.delete_video_thumb()
         thumb_handler.download_video_thumb(video.json_data["vid_thumb_url"])
 
+        Comments(youtube_id, config=self.config).reindex_comments()
+
         return
 
     @staticmethod