Drop download_archive for generic extractors when using default presets. Fixes #506

app/library/Download.py

@@ -20,6 +20,9 @@ from .ItemDTO import ItemDTO
 from .Utils import create_cookies_file, delete_dir, extract_info, extract_ytdlp_logs
 from .ytdlp import YTDLP
 
+GENERIC_EXTRACTORS = ("HTML5MediaEmbed", "generic")
+"Generic extractors identifiers."
+
 
 class Terminator:
     pass
@@ -246,13 +249,19 @@ class Download:
                     self.logger.error(err_msg)
                     raise ValueError(err_msg) from e
 
-            # Safe-guard in-case downloading take too long and the info expires.
-            if self.info_dict and isinstance(self.info_dict, dict) and self.download_info_expires > 0:
-                _ts: int | None = self.info_dict.get("epoch", self.info_dict.get("timestamp", None))
-                _ts = datetime.fromtimestamp(_ts, tz=UTC) if _ts else None
-                if not _ts or (datetime.now(tz=UTC) - _ts).total_seconds() > self.download_info_expires:
-                    self.info_dict = None
-                    self.logger.warning(f"Info for '{self.info.url}' has expired, re-extracting info.")
+            if self.info_dict and isinstance(self.info_dict, dict):
+                # If info extractor_key is generic, we need to remove download_archive param from default preset.
+                if self.info_dict.get("extractor_key") in GENERIC_EXTRACTORS and self.info.get_preset().default:
+                    self.logger.debug(f"Removing 'download_archive' for generic extractor. {self.info.url=}")
+                    params.pop("download_archive", None)
+
+                # Safe-guard in-case downloading take too long and the info expires.
+                if self.download_info_expires > 0:
+                    _ts: int | None = self.info_dict.get("epoch", self.info_dict.get("timestamp", None))
+                    _ts = datetime.fromtimestamp(_ts, tz=UTC) if _ts else None
+                    if not _ts or (datetime.now(tz=UTC) - _ts).total_seconds() > self.download_info_expires:
+                        self.info_dict = None
+                        self.logger.warning(f"Info for '{self.info.url}' has expired, re-extracting info.")
 
             if not self.info_dict or not isinstance(self.info_dict, dict):
                 self.logger.info(f"Extracting info for '{self.info.url}'.")

app/library/ItemDTO.py

@@ -538,6 +538,18 @@ class ItemDTO:
 
         return params
 
+    def get_preset(self) -> "Preset | None":
+        """
+        Get the preset for the item.
+
+        Returns:
+            Preset | None: The preset for the item. If not found, None.
+
+        """
+        from .Presets import Presets
+
+        return Presets.get_instance().get(self.preset if self.preset else "default")
+
     def archive_status(self, force: bool = False) -> None:
         """
         Recompute the archive status of the item.

app/tests/test_itemdto.py

@@ -279,3 +279,51 @@ class TestItemDTO:
         mock_utils_sidecar.assert_not_called()
         assert result is existing
         assert dto.sidecar is existing
+
+    def test_get_preset_returns_preset_instance(self):
+        """Test ItemDTO.get_preset returns the Preset instance."""
+        from app.library.Presets import Preset
+
+        mock_preset = Preset(id="test-id", name="test-preset", cli="--test")
+
+        with patch.object(ItemDTO, "__post_init__", lambda _: None):
+            dto = ItemDTO(id="vid", title="t", url="u", folder="f", preset="test-preset")
+
+        with patch("app.library.Presets.Presets.get_instance") as mock_presets:
+            mock_presets.return_value.get.return_value = mock_preset
+
+            result = dto.get_preset()
+
+            mock_presets.return_value.get.assert_called_once_with("test-preset")
+            assert result is mock_preset
+            assert result.name == "test-preset"
+
+    def test_get_preset_uses_default_when_no_preset_set(self):
+        """Test ItemDTO.get_preset uses 'default' when preset is empty."""
+        from app.library.Presets import Preset
+
+        mock_preset = Preset(id="default-id", name="default", cli="--default")
+
+        with patch.object(ItemDTO, "__post_init__", lambda _: None):
+            dto = ItemDTO(id="vid", title="t", url="u", folder="f", preset="")
+
+        with patch("app.library.Presets.Presets.get_instance") as mock_presets:
+            mock_presets.return_value.get.return_value = mock_preset
+
+            result = dto.get_preset()
+
+            mock_presets.return_value.get.assert_called_once_with("default")
+            assert result is mock_preset
+
+    def test_get_preset_returns_none_when_not_found(self):
+        """Test ItemDTO.get_preset returns None when preset not found."""
+        with patch.object(ItemDTO, "__post_init__", lambda _: None):
+            dto = ItemDTO(id="vid", title="t", url="u", folder="f", preset="nonexistent")
+
+        with patch("app.library.Presets.Presets.get_instance") as mock_presets:
+            mock_presets.return_value.get.return_value = None
+
+            result = dto.get_preset()
+
+            mock_presets.return_value.get.assert_called_once_with("nonexistent")
+            assert result is None