add extractor lang, #316

tubearchivist/home/config.json

@@ -33,6 +33,7 @@
         "comment_sort": "top",
         "cookie_import": false,
         "throttledratelimit": false,
+        "extractor_lang": false,
         "integrate_ryd": false,
         "integrate_sponsorblock": false
     },

tubearchivist/home/src/frontend/forms.py

@@ -122,6 +122,7 @@ class ApplicationSettingsForm(forms.Form):
     downloads_autodelete_days = forms.IntegerField(required=False)
     downloads_format = forms.CharField(required=False)
     downloads_format_sort = forms.CharField(required=False)
+    downloads_extractor_lang = forms.CharField(required=False)
     downloads_add_metadata = forms.ChoiceField(
         widget=forms.Select, choices=METADATA_CHOICES, required=False
     )

tubearchivist/home/src/index/generic.py

@@ -37,8 +37,14 @@ class YouTubeItem:
     def get_from_youtube(self):
         """use yt-dlp to get meta data from youtube"""
         print(f"{self.youtube_id}: get metadata from youtube")
+        obs_request = self.yt_obs.copy()
+        if self.config["downloads"]["extractor_lang"]:
+            langs = self.config["downloads"]["extractor_lang"]
+            langs_list = [i.strip() for i in langs.split(",")]
+            obs_request["extractor_args"] = {"youtube": {"lang": langs_list}}
+
         url = self.build_yt_url()
-        self.youtube_meta = YtWrap(self.yt_obs, self.config).extract(url)
+        self.youtube_meta = YtWrap(obs_request, self.config).extract(url)
 
     def get_from_es(self):
         """get indexed data from elastic search"""

tubearchivist/home/templates/home/settings.html

@@ -108,6 +108,11 @@
                 {{ app_form.downloads_format_sort }}
                 <br>
             </div>
+            <div class="settings-item">
+                <p>Prefer translated metadata language: <span class="settings-current">{{ config.downloads.extractor_lang }}</span></p>
+                <i>This will change the language this video gets indexed as. That will only be available if the uploader provides translations. Add as two letter ISO language code, check the <a href="https://github.com/yt-dlp/yt-dlp#youtube" target="_blank">documentation</a> which languages are available.</i><br>
+                {{ app_form.downloads_extractor_lang}}
+            </div>
             <div class="settings-item">
                 <p>Current metadata embed setting: <span class="settings-current">{{ config.downloads.add_metadata }}</span></p>
                 <i>Metadata is not embedded into the downloaded files by default.</i><br>