Improve directory checks

src/fairseq2/assets/download_manager.py

@@ -19,7 +19,6 @@
 from urllib.request import Request, urlopen
 from zipfile import BadZipFile, ZipFile
 
-import torch
 from tqdm import tqdm  # type: ignore[import]
 
 from fairseq2.assets.error import AssetError
@@ -283,24 +282,35 @@ def _try_uri_as_path(self) -> Optional[Path]:
     def _init_download_path(self) -> None:
         assert self.download_filename is not None
 
-        if pathname := os.getenv("FAIRSEQ2_CACHE_DIR"):
-            try:
-                cache_root_path = Path(pathname)
-            except ValueError as ex:
-                raise RuntimeError(
-                    f"`FAIRSEQ2_CACHE_DIR` environment variable must contain a valid pathname, but contains '{pathname}' instead."
-                ) from ex
-        else:
-            cache_root_path = Path(torch.hub.get_dir()).joinpath("fairseq2")
+        cache_dir = self._get_path_from_env("FAIRSEQ2_CACHE_DIR")
+        if cache_dir is None:
+            cache_dir = self._get_path_from_env("XDG_CACHE_HOME")
+            if cache_dir is None:
+                cache_dir = Path("~/.cache")
+
+            cache_dir = cache_dir.joinpath("fairseq2")
 
         hash_ = sha1(self.uri.encode()).hexdigest()
 
         hash_ = hash_[:24]
 
-        self.download_path = cache_root_path.expanduser().joinpath(
+        self.download_path = cache_dir.expanduser().joinpath(
             "assets", hash_, self.download_filename
         )
 
+    @staticmethod
+    def _get_path_from_env(var_name: str) -> Optional[Path]:
+        pathname = os.getenv(var_name)
+        if not pathname:
+            return None
+
+        try:
+            return Path(pathname)
+        except ValueError as ex:
+            raise RuntimeError(
+                f"`{var_name}` environment variable must contain a valid pathname, but contains '{pathname}' instead."
+            ) from ex
+
     def _download_asset(self) -> None:
         assert self.download_path is not None
 

src/fairseq2/assets/store.py

@@ -161,49 +161,57 @@ def _create_asset_store() -> ProviderBackedAssetStore:
 asset_store = _create_asset_store()
 
 
-def _get_asset_dir_from_env(var_name: str) -> Optional[Path]:
+def _get_path_from_env(var_name: str) -> Optional[Path]:
     pathname = os.getenv(var_name)
     if not pathname:
         return None
 
     try:
-        asset_dir = Path(pathname)
+        path = Path(pathname)
     except ValueError as ex:
         raise RuntimeError(
             f"`{var_name}` environment variable must contain a valid pathname, but contains '{pathname}' instead."
         ) from ex
 
-    if not asset_dir.exists():
+    if not path.exists():
         logger = logging.getLogger("fairseq2.assets")
 
         logger.warning(
-            f"The path '{asset_dir}' pointed to by the `{var_name}` environment variable does not exist."
+            f"The path '{path}' pointed to by the `{var_name}` environment variable does not exist."
         )
 
         return None
 
-    return asset_dir
+    return path
 
 
 def _load_asset_directory() -> None:
-    asset_dir = _get_asset_dir_from_env("FAIRSEQ2_ASSET_DIR")
+    asset_dir = _get_path_from_env("FAIRSEQ2_ASSET_DIR")
     if asset_dir is None:
         asset_dir = Path("/etc/fairseq2/assets")
         if not asset_dir.exists():
             return
 
+    asset_dir = asset_dir.expanduser()
+
     asset_store.metadata_providers.append(FileAssetMetadataProvider(asset_dir))
 
 
 _load_asset_directory()
 
 
 def _load_user_asset_directory() -> None:
-    asset_dir = _get_asset_dir_from_env("FAIRSEQ2_USER_ASSET_DIR")
+    asset_dir = _get_path_from_env("FAIRSEQ2_USER_ASSET_DIR")
     if asset_dir is None:
-        asset_dir = Path("~/.config/fairseq2/assets").expanduser()
-        if not asset_dir.exists():
-            return
+        asset_dir = _get_path_from_env("XDG_CONFIG_HOME")
+        if asset_dir is None:
+            asset_dir = Path("~/.config").expanduser()
+            if not asset_dir.exists():
+                return
+
+        asset_dir = asset_dir.joinpath("fairseq2/assets")
+
+    asset_dir = asset_dir.expanduser()
 
     asset_store.user_metadata_providers.append(FileAssetMetadataProvider(asset_dir))