add json indentation to config (#1952)

Author: d-v-b

src/zarr/config.py

@@ -11,6 +11,7 @@
             "array": {"order": "C"},
             "async": {"concurrency": None, "timeout": None},
             "codec_pipeline": {"batch_size": 1},
+            "json_indent": 2,
         }
     ],
 )

src/zarr/group.py

@@ -25,6 +25,7 @@
     ChunkCoords,
     ZarrFormat,
 )
+from zarr.config import config
 from zarr.store import StoreLike, StorePath, make_store_path
 from zarr.sync import SyncMixin, sync
 
@@ -79,14 +80,21 @@ class GroupMetadata(Metadata):
     node_type: Literal["group"] = field(default="group", init=False)
 
     def to_buffer_dict(self) -> dict[str, Buffer]:
+        json_indent = config.get("json_indent")
         if self.zarr_format == 3:
-            return {ZARR_JSON: Buffer.from_bytes(json.dumps(self.to_dict()).encode())}
+            return {
+                ZARR_JSON: Buffer.from_bytes(
+                    json.dumps(self.to_dict(), indent=json_indent).encode()
+                )
+            }
         else:
             return {
                 ZGROUP_JSON: Buffer.from_bytes(
-                    json.dumps({"zarr_format": self.zarr_format}).encode()
+                    json.dumps({"zarr_format": self.zarr_format}, indent=json_indent).encode()
+                ),
+                ZATTRS_JSON: Buffer.from_bytes(
+                    json.dumps(self.attributes, indent=json_indent).encode()
                 ),
-                ZATTRS_JSON: Buffer.from_bytes(json.dumps(self.attributes).encode()),
             }
 
     def __init__(self, attributes: dict[str, Any] | None = None, zarr_format: ZarrFormat = 3):

src/zarr/metadata.py

@@ -16,6 +16,7 @@
 from zarr.chunk_grids import ChunkGrid, RegularChunkGrid
 from zarr.chunk_key_encodings import ChunkKeyEncoding, parse_separator
 from zarr.codecs._v2 import V2Compressor, V2Filters
+from zarr.config import config
 
 if TYPE_CHECKING:
     from typing_extensions import Self
@@ -272,8 +273,11 @@ def _json_convert(o: np.dtype[Any] | Enum | Codec) -> str | dict[str, Any]:
                 return config
             raise TypeError
 
+        json_indent = config.get("json_indent")
         return {
-            ZARR_JSON: Buffer.from_bytes(json.dumps(self.to_dict(), default=_json_convert).encode())
+            ZARR_JSON: Buffer.from_bytes(
+                json.dumps(self.to_dict(), default=_json_convert, indent=json_indent).encode()
+            )
         }
 
     @classmethod
@@ -394,9 +398,12 @@ def _json_convert(
         assert isinstance(zarray_dict, dict)
         zattrs_dict = zarray_dict.pop("attributes", {})
         assert isinstance(zattrs_dict, dict)
+        json_indent = config.get("json_indent")
         return {
-            ZARRAY_JSON: Buffer.from_bytes(json.dumps(zarray_dict, default=_json_convert).encode()),
-            ZATTRS_JSON: Buffer.from_bytes(json.dumps(zattrs_dict).encode()),
+            ZARRAY_JSON: Buffer.from_bytes(
+                json.dumps(zarray_dict, default=_json_convert, indent=json_indent).encode()
+            ),
+            ZATTRS_JSON: Buffer.from_bytes(json.dumps(zattrs_dict, indent=json_indent).encode()),
         }
 
     @classmethod

tests/v3/test_config.py

@@ -1,19 +1,32 @@
+from typing import Any
+
+import pytest
+
 from zarr.config import config
 
 
-def test_config_defaults_set():
+def test_config_defaults_set() -> None:
     # regression test for available defaults
     assert config.defaults == [
         {
             "array": {"order": "C"},
             "async": {"concurrency": None, "timeout": None},
             "codec_pipeline": {"batch_size": 1},
+            "json_indent": 2,
         }
     ]
     assert config.get("array.order") == "C"
+    assert config.get("async.concurrency") is None
+    assert config.get("async.timeout") is None
+    assert config.get("codec_pipeline.batch_size") == 1
+    assert config.get("json_indent") == 2
 
 
-def test_config_defaults_can_be_overridden():
-    assert config.get("array.order") == "C"
-    with config.set({"array.order": "F"}):
-        assert config.get("array.order") == "F"
+@pytest.mark.parametrize(
+    "key, old_val, new_val",
+    [("array.order", "C", "F"), ("async.concurrency", None, 10), ("json_indent", 2, 0)],
+)
+def test_config_defaults_can_be_overridden(key: str, old_val: Any, new_val: Any) -> None:
+    assert config.get(key) == old_val
+    with config.set({key: new_val}):
+        assert config.get(key) == new_val