undo warnings.warn

Author: rambleraptor

pyiceberg/table/__init__.py

@@ -641,7 +641,7 @@ def delete(
             self.table_metadata.properties.get(TableProperties.DELETE_MODE, TableProperties.DELETE_MODE_DEFAULT)
             == TableProperties.DELETE_MODE_MERGE_ON_READ
         ):
-            raise NotImplementedError("Merge on read is not yet supported")
+            warnings.warn("Merge on read is not yet supported, falling back to copy-on-write", stacklevel=2)
 
         if isinstance(delete_filter, str):
             delete_filter = _parse_row_filter(delete_filter)
@@ -1829,9 +1829,6 @@ def from_rest_response(
 
         Returns:
             A FileScanTask with the converted data and delete files.
-
-        Raises:
-            NotImplementedError: If equality delete files are encountered.
         """
         data_file = _rest_file_to_data_file(rest_task.data_file)
 
@@ -1850,18 +1847,28 @@ def from_rest_response(
 
 def _rest_file_to_data_file(rest_file: RESTContentFile) -> DataFile:
     """Convert a REST content file to a manifest DataFile."""
-    from pyiceberg.catalog.rest.scan_planning import RESTDataFile
+    from pyiceberg.catalog.rest.scan_planning import RESTDataFile, RESTEqualityDeleteFile, RESTPositionDeleteFile
+
+    column_sizes = None
+    value_counts = None
+    null_value_counts = None
+    nan_value_counts = None
+    equality_ids = None
+    referenced_data_file = None
+    content_offset = None
+    content_size_in_bytes = None
 
     if isinstance(rest_file, RESTDataFile):
         column_sizes = rest_file.column_sizes.to_dict() if rest_file.column_sizes else None
         value_counts = rest_file.value_counts.to_dict() if rest_file.value_counts else None
         null_value_counts = rest_file.null_value_counts.to_dict() if rest_file.null_value_counts else None
         nan_value_counts = rest_file.nan_value_counts.to_dict() if rest_file.nan_value_counts else None
-    else:
-        column_sizes = None
-        value_counts = None
-        null_value_counts = None
-        nan_value_counts = None
+    elif isinstance(rest_file, RESTEqualityDeleteFile):
+        equality_ids = rest_file.equality_ids
+    elif isinstance(rest_file, RESTPositionDeleteFile):
+        referenced_data_file = rest_file.referenced_data_file
+        content_offset = rest_file.content_offset
+        content_size_in_bytes = rest_file.content_size_in_bytes
 
     data_file = DataFile.from_args(
         content=DataFileContent.from_rest_type(rest_file.content),
@@ -1875,6 +1882,10 @@ def _rest_file_to_data_file(rest_file: RESTContentFile) -> DataFile:
         null_value_counts=null_value_counts,
         nan_value_counts=nan_value_counts,
         split_offsets=rest_file.split_offsets,
+        equality_ids=equality_ids,
+        referenced_data_file=referenced_data_file,
+        content_offset=content_offset,
+        content_size_in_bytes=content_size_in_bytes,
         sort_order_id=rest_file.sort_order_id,
     )
     data_file.spec_id = rest_file.spec_id

tests/catalog/test_scan_planning_models.py

@@ -38,7 +38,7 @@
     ValueMap,
 )
 from pyiceberg.expressions import AlwaysTrue, EqualTo, Reference
-from pyiceberg.manifest import FileFormat
+from pyiceberg.manifest import DataFileContent, FileFormat
 
 TEST_URI = "https://iceberg-test-catalog/"
 
@@ -545,7 +545,7 @@ def test_plan_scan_cancelled(rest_scan_catalog: RestCatalog, requests_mock: Mock
         list(rest_scan_catalog.plan_scan(("db", "tbl"), request))
 
 
-def test_plan_scan_equality_deletes_not_supported(rest_scan_catalog: RestCatalog, requests_mock: Mocker) -> None:
+def test_plan_scan_equality_deletes_supported(rest_scan_catalog: RestCatalog, requests_mock: Mocker) -> None:
     file_one = _rest_data_file(file_path="s3://bucket/tbl/data/file1.parquet")
     equality_delete = _rest_equality_delete_file(equality_ids=[1, 2])
     requests_mock.post(
@@ -566,5 +566,11 @@ def test_plan_scan_equality_deletes_not_supported(rest_scan_catalog: RestCatalog
     )
 
     request = PlanTableScanRequest()
-    with pytest.raises(NotImplementedError, match="PyIceberg does not yet support equality deletes"):
-        list(rest_scan_catalog.plan_scan(("db", "tbl"), request))
+    tasks = list(rest_scan_catalog.plan_scan(("db", "tbl"), request))
+    assert len(tasks) == 1
+    task = tasks[0]
+    assert task.file.file_path == "s3://bucket/tbl/data/file1.parquet"
+    assert len(task.delete_files) == 1
+    delete_file = list(task.delete_files)[0]
+    assert delete_file.content == DataFileContent.EQUALITY_DELETES
+    assert delete_file.equality_ids == [1, 2]

tests/table/test_snapshots.py

@@ -587,8 +587,8 @@ def test_block_writing_equality_deletes(table_schema_simple: Schema, catalog: Ca
     assert "PyIceberg does not support writing DataFileContent.EQUALITY_DELETES" in str(exc.value)
 
 
-def test_delete_mode_mor_raises_error(table_schema_simple: Schema, catalog: Catalog) -> None:
-    identifier = "default.test_delete_mode_mor_raises_error"
+def test_delete_mode_mor_warns(table_schema_simple: Schema, catalog: Catalog) -> None:
+    identifier = "default.test_delete_mode_mor_warns"
     catalog.create_namespace("default")
 
     catalog.create_table(
@@ -599,7 +599,7 @@ def test_delete_mode_mor_raises_error(table_schema_simple: Schema, catalog: Cata
 
     tbl = catalog.load_table(identifier)
 
-    with pytest.raises(NotImplementedError) as exc:
-        tbl.delete("id = 1")
+    with pytest.warns() as record:
+        tbl.delete("foo = 'a'")
 
-    assert "Merge on read is not yet supported" in str(exc.value)
+    assert any("Merge on read is not yet supported, falling back to copy-on-write" in str(w.message) for w in record)