gc: log paths during dry-run garbage collection (#10937)

dvc/repo/gc.py

@@ -128,9 +128,11 @@ def gc(  # noqa: C901, PLR0912, PLR0913
         used_obj_ids = set()
         used_obj_ids.update(*odb_to_obj_ids.values())
 
+    seen_odbs = set()
     for scheme, odb in self.cache.by_scheme():
-        if not odb:
+        if not odb or odb in seen_odbs:
             continue
+        seen_odbs.add(odb)
         num_removed = ogc(odb, used_obj_ids, jobs=jobs, dry=dry)
         if num_removed:
             logger.info("Removed %d objects from %s cache.", num_removed, scheme)

pyproject.toml

@@ -41,7 +41,7 @@ dependencies = [
     "distro>=1.3",
     "dpath<3,>=2.1.0",
     "dulwich",
-    "dvc-data>=3.17.0,<3.18",
+    "dvc-data>=3.18.0,<3.19.0",
     "dvc-http>=2.29.0",
     "dvc-objects",
     "dvc-render>=1.0.1,<2",

tests/func/test_gc.py

@@ -437,8 +437,16 @@ def test_gc_logging(caplog, dvc, good_and_bad_cache):
     with caplog.at_level(logging.INFO, logger="dvc"):
         dvc.gc(workspace=True)
 
-    assert "Removed 3 objects from repo cache." in caplog.text
+    # Check that 3 objects were removed from either repo or local cache
-    assert "No unused 'local' cache to remove." in caplog.text
+    # (they point to the same ODB, so only one will be logged due to deduplication)
+    has_repo_log = "Removed 3 objects from repo cache." in caplog.text
+    has_local_log = "Removed 3 objects from local cache." in caplog.text
+
+    # Exactly one of repo or local should have the log, not both (deduplication)
+    assert has_repo_log ^ has_local_log, (
+        "Should have exactly one log for repo/local cache"
+    )
+
     assert "No unused 'legacy' cache to remove." in caplog.text
 
 
@@ -450,3 +458,21 @@ def test_gc_skip_failed(tmp_dir, dvc):
         dvc.gc(force=True, workspace=True)
 
     dvc.gc(force=True, workspace=True, skip_failed=True)
+
+
+def test_gc_dry_logs_paths(caplog, tmp_dir, dvc):
+    """Test that dry run logs paths of objects to be removed."""
+    odb = dvc.cache.local
+
+    # Add some unused objects directly to cache
+    unused_hashes = ["test123", "test456", "test789"]
+    for hash_val in unused_hashes:
+        odb.add_bytes(hash_val, hash_val.encode("utf8"))
+
+    with caplog.at_level(logging.INFO, logger="dvc_data.hashfile.gc"):
+        dvc.gc(workspace=True, dry=True)
+
+    # Verify that paths are logged by dvc-data layer
+    for hash_val in unused_hashes:
+        expected_path = odb.oid_to_path(hash_val)
+        assert f"Removing {expected_path}" in caplog.text