(fix) DVC_SITE_CACHE_DIR environment variable should override core.site_cache_dir (#10778)

* Prioritize DVC_SITE_CACHE_DIR over core.site_cache_dir

* Add tests for proper prioritization of env var and config

* Tests require python-3.9 compatible typing, use Optional

* Remove comment

* Fixup

Author: rgoya

dvc/dirs.py

@@ -1,4 +1,5 @@
 import os
+from typing import Optional
 
 import platformdirs
 
@@ -20,13 +21,16 @@ def global_config_dir():
     )
 
 
-def site_cache_dir():
+def site_cache_dir(config_site_cache_dir: Optional[str] = None):
     from platformdirs import PlatformDirs
     from platformdirs.unix import Unix
 
     if dvc_site_cache_dir := os.getenv(env.DVC_SITE_CACHE_DIR):
         return dvc_site_cache_dir
 
+    if config_site_cache_dir is not None:
+        return config_site_cache_dir
+
     if issubclass(Unix, PlatformDirs):
         # Return the cache directory shared by users, e.g. `/var/tmp/$appname`
         # NOTE: platformdirs>=5 changed `site_cache_dir` to return /var/cache/$appname.

dvc/repo/__init__.py

@@ -609,7 +609,7 @@ def site_cache_dir(self) -> str:
         from dvc.fs import GitFileSystem
         from dvc.version import version_tuple
 
-        cache_dir = self.config["core"].get("site_cache_dir") or site_cache_dir()
+        cache_dir = site_cache_dir(self.config["core"].get("site_cache_dir"))
 
         subdir = None
         if isinstance(self.fs, GitFileSystem):

tests/unit/test_dirs.py

@@ -21,3 +21,13 @@ def test_site_cache_dir_on_unix(monkeypatch):
 def test_site_cache_dir_env_var(monkeypatch):
     monkeypatch.setenv(DVC_SITE_CACHE_DIR, "foo_bar")
     assert site_cache_dir() == "foo_bar"
+
+
+def test_site_cache_dir_with_config_parameter(monkeypatch):
+    monkeypatch.delenv(DVC_SITE_CACHE_DIR)
+    assert site_cache_dir(config_site_cache_dir="foo_bar") == "foo_bar"
+
+
+def test_site_cache_dir_env_var_precedence(monkeypatch):
+    monkeypatch.setenv(DVC_SITE_CACHE_DIR, "foo")
+    assert site_cache_dir(config_site_cache_dir="bar") == "foo"