new thumbnails module and new cache layout

tubearchivist/home/apps.py

@@ -6,6 +6,7 @@ from django.apps import AppConfig
 from home.src.config import AppConfig as ArchivistConfig
 from home.src.helper import RedisArchivist
 from home.src.index_management import index_check
+from home.src.thumbnails import validate_thumbnails
 
 
 def sync_redis_state():
@@ -48,3 +49,4 @@ class HomeConfig(AppConfig):
         index_check()
         sync_redis_state()
         make_folders()
+        validate_thumbnails()

tubearchivist/home/src/download.py

@@ -70,6 +70,7 @@ class PendingList:
         all_downloaded = self.get_all_downloaded()
         # loop
         bulk_list = []
+        all_videos_added = []
         for video in missing_videos:
             if isinstance(video, str):
                 youtube_id = video
@@ -79,6 +80,7 @@ class PendingList:
                 # skip already downloaded
                 continue
             video = self.get_youtube_details(youtube_id)
+            thumb_url = video["vid_thumb_url"]
             # skip on download error
             if not video:
                 continue
@@ -91,6 +93,7 @@ class PendingList:
             action = {"create": {"_id": youtube_id, "_index": "ta_download"}}
             bulk_list.append(json.dumps(action))
             bulk_list.append(json.dumps(video))
+            all_videos_added.append((youtube_id, thumb_url))
             # notify
             mess_dict = {
                 "status": "pending",
@@ -108,6 +111,8 @@ class PendingList:
         if not request.ok:
             print(request)
 
+        return all_videos_added
+
     @staticmethod
     def get_youtube_details(youtube_id):
         """get details from youtubedl for single pending video"""
@@ -165,12 +170,11 @@ class PendingList:
             all_hits = json_data["hits"]["hits"]
             if all_hits:
                 for hit in all_hits:
-                    youtube_id = hit["_source"]["youtube_id"]
                     status = hit["_source"]["status"]
                     if status == "pending":
                         all_pending.append(hit["_source"])
                     elif status == "ignore":
-                        all_ignore.append(youtube_id)
+                        all_ignore.append(hit["_source"])
                     search_after = hit["sort"]
                 # update search_after with last hit data
                 data["search_after"] = search_after
@@ -342,9 +346,9 @@ class ChannelSubscription:
         all_channels = self.get_channels()
         pending_handler = PendingList()
         all_pending, all_ignore = pending_handler.get_all_pending()
-        all_pending_ids = [i["youtube_id"] for i in all_pending]
+        all_ids = [i["youtube_id"] for i in all_ignore + all_pending]
         all_downloaded = pending_handler.get_all_downloaded()
-        to_ignore = all_pending_ids + all_ignore + all_downloaded
+        to_ignore = all_ids + all_downloaded
         missing_videos = []
         counter = 1
         for channel in all_channels:

tubearchivist/home/src/searching.py

@@ -14,6 +14,7 @@ from datetime import datetime
 import requests
 from home.src.config import AppConfig
 from home.src.helper import ignore_filelist
+from home.src.thumbnails import ThumbManager
 from PIL import Image
 
 
@@ -63,8 +64,9 @@ class SearchHandler:
                     all_channels.append(channel_dict)
         if self.cache:
             # validate cache
-            self.cache_dl_vids(all_videos)
-            self.cache_dl_chan(all_channels)
+            pass
+            # self.cache_dl_vids(all_videos)
+            # self.cache_dl_chan(all_channels)
 
         return return_value
 
@@ -168,6 +170,11 @@ class SearchHandler:
             date_str = datetime.strftime(date_refresh, "%d %b, %Y")
             hit["source"]["vid_last_refresh"] = date_str
 
+        if "vid_thumb_url" in hit_keys:
+            youtube_id = hit["source"]["youtube_id"]
+            thumb_path = ThumbManager().vid_thumb_path(youtube_id)
+            hit["source"]["vid_thumb_url"] = thumb_path
+
         if "channel_last_refresh" in hit_keys:
             refreshed = hit["source"]["channel_last_refresh"]
             date_refresh = datetime.fromtimestamp(refreshed)

tubearchivist/home/src/thumbnails.py

@@ -0,0 +1,114 @@
+"""
+functionality:
+- handle download and caching for thumbnails
+"""
+
+import os
+
+import requests
+from home.src.config import AppConfig
+from home.src.download import PendingList
+from home.src.helper import RedisArchivist, ignore_filelist
+from PIL import Image
+
+
+class ThumbManager:
+    """handle thumbnails related functions"""
+
+    CONFIG = AppConfig().config
+    CACHE_DIR = CONFIG["application"]["cache_dir"]
+    VIDEO_DIR = os.path.join(CACHE_DIR, "videos")
+
+    def get_all_thumbs(self):
+        """raise exception if cache not clean"""
+        all_thumb_folders = ignore_filelist(os.listdir(self.VIDEO_DIR))
+        all_thumbs = []
+        for folder in all_thumb_folders:
+            folder_path = os.path.join(self.VIDEO_DIR, folder)
+            if os.path.isfile(folder_path):
+                self.update_path(folder)
+                all_thumbs.append(folder_path)
+                continue
+                # raise exemption here in a future version
+                # raise FileExistsError("video cache dir has files inside")
+
+            all_folder_thumbs = ignore_filelist(os.listdir(folder_path))
+            all_thumbs.extend(all_folder_thumbs)
+
+        return all_thumbs
+
+    def update_path(self, file_name):
+        """reorganize thumbnails into folders as update path from v0.0.5"""
+        folder_name = file_name[0].lower()
+        folder_path = os.path.join(self.VIDEO_DIR, folder_name)
+        old_file = os.path.join(self.VIDEO_DIR, file_name)
+        new_file = os.path.join(folder_path, file_name)
+        os.makedirs(folder_path, exist_ok=True)
+        os.rename(old_file, new_file)
+
+    def get_missing_thumbs(self):
+        """get a list of all missing thumbnails"""
+        all_thumbs = self.get_all_thumbs()
+        all_indexed = PendingList().get_all_indexed()
+        all_in_queue, all_ignored = PendingList().get_all_pending()
+
+        missing_thumbs = []
+        for video in all_indexed:
+            youtube_id = video["_source"]["youtube_id"]
+            if youtube_id + ".jpg" not in all_thumbs:
+                thumb_url = video["_source"]["vid_thumb_url"]
+                missing_thumbs.append((youtube_id, thumb_url))
+
+        for video in all_in_queue + all_ignored:
+            youtube_id = video["youtube_id"]
+            if youtube_id + ".jpg" not in all_thumbs:
+                thumb_url = video["vid_thumb_url"]
+                missing_thumbs.append((youtube_id, thumb_url))
+
+        return missing_thumbs
+
+    def download_missing(self, missing_thumbs):
+        """download all missing thumbnails from list"""
+        print(f"downloading {len(missing_thumbs)} thumbnails")
+        vid_cache = os.path.join(self.CACHE_DIR, "videos")
+        # videos
+        for youtube_id, thumb_url in missing_thumbs:
+            folder_name = youtube_id[0].lower()
+            folder_path = os.path.join(vid_cache, folder_name)
+            thumb_path_part = self.vid_thumb_path(youtube_id)
+            thumb_path = os.path.join(self.CACHE_DIR, thumb_path_part)
+
+            os.makedirs(folder_path, exist_ok=True)
+            img_raw = requests.get(thumb_url, stream=True).raw
+            img = Image.open(img_raw)
+
+            width, height = img.size
+            if not width / height == 16 / 9:
+                new_height = width / 16 * 9
+                offset = (height - new_height) / 2
+                img = img.crop((0, offset, width, height - offset))
+
+            img.convert("RGB").save(thumb_path)
+
+            mess_dict = {
+                "status": "pending",
+                "level": "info",
+                "title": "Adding to download queue.",
+                "message": "Downloading Thumbnails...",
+            }
+            RedisArchivist().set_message("progress:download", mess_dict)
+
+    @staticmethod
+    def vid_thumb_path(youtube_id):
+        """build expected path for video thumbnail from youtube_id"""
+        folder_name = youtube_id[0].lower()
+        folder_path = os.path.join("videos", folder_name)
+        thumb_path = os.path.join(folder_path, youtube_id + ".jpg")
+        return thumb_path
+
+
+def validate_thumbnails():
+    """check if all thumbnails are there and organized correctly"""
+    handler = ThumbManager()
+    thumbs_to_download = handler.get_missing_thumbs()
+    handler.download_missing(thumbs_to_download)

tubearchivist/home/tasks.py

@@ -16,6 +16,7 @@ from home.src.reindex import (
     reindex_old_documents,
     scan_filesystem,
 )
+from home.src.thumbnails import ThumbManager
 
 CONFIG = AppConfig().config
 REDIS_HOST = os.environ.get("REDIS_HOST")
@@ -90,7 +91,8 @@ def extrac_dl(youtube_ids):
     """parse list passed and add to pending"""
     pending_handler = PendingList()
     missing_videos = pending_handler.parse_url_list(youtube_ids)
-    pending_handler.add_to_pending(missing_videos)
+    all_videos_added = pending_handler.add_to_pending(missing_videos)
+    ThumbManager().download_missing(all_videos_added)
 
 
 @shared_task

tubearchivist/home/templates/home/channel_id.html

@@ -75,10 +75,10 @@
     {% if videos %}
         {% for video in videos %}
             <div class="video-item {{ view_style }}">
-                <a href="#player" data-src="/media/{{ video.source.media_url }}" data-thumb="/cache/videos/{{ video.source.youtube_id }}.jpg" data-title="{{ video.source.title }}" data-channel="{{ video.source.channel.channel_name }}" data-id="{{ video.source.youtube_id }}" onclick="createPlayer(this)">
+                <a href="#player" data-src="/media/{{ video.source.media_url }}" data-thumb="/cache/{{ video.source.vid_thumb_url }}" data-title="{{ video.source.title }}" data-channel="{{ video.source.channel.channel_name }}" data-id="{{ video.source.youtube_id }}" onclick="createPlayer(this)">
                     <div class="video-thumb-wrap {{ view_style }}">
                         <div class="video-thumb">
-                            <img src="/cache/videos/{{ video.source.youtube_id }}.jpg" alt="video-thumb">
+                            <img src="/cache/{{ video.source.vid_thumb_url }}" alt="video-thumb">
                         </div>
                         <div class="video-play">
                             <img src="{% static 'img/icon-play.svg' %}" alt="play-icon">

tubearchivist/home/templates/home/downloads.html

@@ -57,7 +57,7 @@
         {% for video in all_video_hits %}
             <div class="dl-item {{ view_style }}" id="dl-{{ video.youtube_id }}">
                 <div class="dl-thumb {{ view_style }}">
-                    <img src="{{ video.vid_thumb_url }}" alt="video_thumb">
+                    <img src="/cache/{{ video.vid_thumb_url }}" alt="video_thumb">
                 </div>
                 <div class="dl-desc {{ view_style }}">
                     {% if show_ignored_only %}

tubearchivist/home/templates/home/home.html

@@ -48,10 +48,10 @@
     {% if videos %}
         {% for video in videos %}
             <div class="video-item {{ view_style }}">
-                <a href="#player" data-src="/media/{{ video.source.media_url }}" data-thumb="/cache/videos/{{ video.source.youtube_id }}.jpg" data-title="{{ video.source.title }}" data-channel="{{ video.source.channel.channel_name }}" data-id="{{ video.source.youtube_id }}" onclick="createPlayer(this)">
+                <a href="#player" data-src="/media/{{ video.source.media_url }}" data-thumb="/cache/{{ video.source.vid_thumb_url }}" data-title="{{ video.source.title }}" data-channel="{{ video.source.channel.channel_name }}" data-id="{{ video.source.youtube_id }}" onclick="createPlayer(this)">
                     <div class="video-thumb-wrap {{ view_style }}">
                         <div class="video-thumb">
-                            <img src="/cache/videos/{{ video.source.youtube_id }}.jpg" alt="video-thumb">
+                            <img src="/cache/{{ video.source.vid_thumb_url }}" alt="video-thumb">
                         </div>
                         <div class="video-play">
                             <img src="{% static 'img/icon-play.svg' %}" alt="play-icon">

tubearchivist/home/templates/home/video.html

@@ -8,7 +8,7 @@
 <div class="video-main">
     <video 
         src="/media/{{ video.media_url }}" 
-        poster="/cache/videos/{{ video.youtube_id }}.jpg" controls preload="false" 
+        poster="/cache/{{ video.vid_thumb_url }}" controls preload="false" 
         type='video/mp4' width="100%">
     </video>
 </div>