From 8f5311ff04f37ae0ae4701854fbda46ee2e0fa70 Mon Sep 17 00:00:00 2001
From: simon <simobilleter@gmail.com>
Date: Fri, 18 Nov 2022 11:33:06 +0700
Subject: [PATCH] delete comments when deleting channel

---
 tubearchivist/home/src/es/index_mapping.json |  3 +++
 tubearchivist/home/src/index/channel.py      | 12 +++++++++++-
 tubearchivist/home/src/index/comments.py     |  7 +++++--
 3 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/tubearchivist/home/src/es/index_mapping.json b/tubearchivist/home/src/es/index_mapping.json
index 79f1d81..7270563 100644
--- a/tubearchivist/home/src/es/index_mapping.json
+++ b/tubearchivist/home/src/es/index_mapping.json
@@ -473,6 +473,9 @@
                 "comment_last_refresh": {
                     "type": "date"
                 },
+                "comment_channel_id": {
+                    "type": "keyword"
+                },
                 "comment_comments": {
                     "properties": {
                         "comment_id": {
diff --git a/tubearchivist/home/src/index/channel.py b/tubearchivist/home/src/index/channel.py
index 7a554a9..7108d43 100644
--- a/tubearchivist/home/src/index/channel.py
+++ b/tubearchivist/home/src/index/channel.py
@@ -50,7 +50,7 @@ class ChannelScraper:
         url = f"https://www.youtube.com/channel/{self.channel_id}/about?hl=en"
         cookies = {"CONSENT": "YES+xxxxxxxxxxxxxxxxxxxxxxxxxxx"}
         response = requests.get(
-            url, cookies=cookies, headers=requests_headers()
+            url, cookies=cookies, headers=requests_headers(), timeout=10
         )
         if response.ok:
             channel_page = response.text
@@ -275,6 +275,15 @@ class YoutubeChannel(YouTubeItem):
         }
         _, _ = ElasticWrap("ta_video/_delete_by_query").post(data)
 
+    def delete_es_comments(self):
+        """delete all comments from this channel"""
+        data = {
+            "query": {
+                "term": {"comment_channel_id": {"value": self.youtube_id}}
+            }
+        }
+        _, _ = ElasticWrap("ta_comment/_delete_by_query").post(data)
+
     def delete_playlists(self):
         """delete all indexed playlist from es"""
         all_playlists = self.get_indexed_playlists()
@@ -301,6 +310,7 @@ class YoutubeChannel(YouTubeItem):
         self.delete_playlists()
         print(f"{self.youtube_id}: delete indexed videos")
         self.delete_es_videos()
+        self.delete_es_comments()
         self.del_in_es()
 
     def index_channel_playlists(self):
diff --git a/tubearchivist/home/src/index/comments.py b/tubearchivist/home/src/index/comments.py
index 3ad7c9e..0ac2436 100644
--- a/tubearchivist/home/src/index/comments.py
+++ b/tubearchivist/home/src/index/comments.py
@@ -31,12 +31,13 @@ class Comments:
         if not self.is_activated:
             return
 
-        comments_raw = self.get_yt_comments()
+        comments_raw, channel_id = self.get_yt_comments()
         self.format_comments(comments_raw)
 
         self.json_data = {
             "youtube_id": self.youtube_id,
             "comment_last_refresh": int(datetime.now().strftime("%s")),
+            "comment_channel_id": channel_id,
             "comment_comments": self.comments_format,
         }
 
@@ -75,7 +76,8 @@ class Comments:
         yt_obs = self.build_yt_obs()
         info_json = YtWrap(yt_obs).extract(self.youtube_id)
         comments_raw = info_json.get("comments")
-        return comments_raw
+        channel_id = info_json.get("channel_id")
+        return comments_raw, channel_id
 
     def format_comments(self, comments_raw):
         """process comments to match format"""
@@ -141,6 +143,7 @@ class Comments:
 
     def reindex_comments(self):
         """update comments from youtube"""
+        self.check_config()
         if not self.is_activated:
             return