def get_comments(video_id, youtube, pageToken=None): """fetch all comments and replies for a given youtube video ID""" video_id = video_id parser = HTMLParser() to_replace = ['\ufeff', '\n'] all_comments = [] page_token = pageToken comment_thread_id = 0 count = 0 while page_token is not 0: comments = youtube.commentThreads().list( part="snippet", videoId=video_id, textFormat="plainText", pageToken=page_token ).execute() next_page_token = "" next_page_tokens = [] print(count) count += 1 for comment in comments['items']: comment_id = comment['id'] replies = comment['snippet']['totalReplyCount'] comment_text = comment['snippet']['topLevelComment']['snippet']['textDisplay'] comment_text = parser.unescape(comment_text) for symbol in to_replace: comment_text = comment_text.replace(symbol, '') all_comments.append(comment_text) published_at = comment['snippet']['topLevelComment']['snippet']['publishedAt'] like_count = comment['snippet']['topLevelComment']['snippet']['likeCount'] author_display_name = comment['snippet']['topLevelComment']['snippet']['authorDisplayName'] new_comment = Comment(place_in_thread=0, comment_thread_id=0, video_id=video_id, username=author_display_name, url="www.example.com", comment_text=comment_text, not_sure=0, is_a_troll=0, not_a_troll=0) if replies > 0: new_comment.place_in_thread = 0 comment_replies = youtube.comments().list(part="snippet", parentId=comment_id, textFormat="plainText").execute() #reply data starts with last comment in thread thread_location = len(comment_replies['items']) for reply in comment_replies['items']: reply_text = reply['snippet']['textDisplay'] reply_author_display_name = reply['snippet']['authorDisplayName'] reply_text = parser.unescape(reply_text) for symbol in to_replace: reply_text = reply_text.replace(symbol, '') all_comments.append(reply_text) reply_comment = Comment(place_in_thread=thread_location, comment_thread_id=comment_thread_id, video_id=video_id, not_sure=0, username=reply_author_display_name, url="www.example.com", comment_text=reply_text, is_a_troll=0, not_a_troll=0) #reply_comment.place_in_thread = thread_location #reply_comment.comment_thread_id = comment_thread_id reply_comment.video_id = video_id db.session.add(reply_comment) db.session.commit() thread_location -= 1 new_comment.comment_thread_id = comment_thread_id comment_thread_id += 1 new_comment.video_id = video_id db.session.add(new_comment) db.session.commit() if 'nextPageToken' in comments: next_page_token = comments['nextPageToken'] next_page_tokens.append(next_page_token) page_token = next_page_token else: page_token = 0 else: return all_comments
