Fix for --float-trunc-prec to propagate to nested columns too

Author: FrancescAlted

src/blosc2/cli/parquet_to_blosc2.py

@@ -1101,6 +1101,8 @@ def import_unnamed_root_separate_cols(  # noqa: C901
     """
 
     inner_schema = blosc2.CTable._inner_schema_for_unnamed_root(pa, parquet_schema)
+    flat_inner_schema = blosc2.CTable._flatten_arrow_struct_schema(pa, inner_schema)
+    float_trunc_column_cparams = build_float_trunc_column_cparams(pa, flat_inner_schema, args)
     total_parquet_rows = pf.metadata.num_rows if pf.metadata is not None else None
 
     # ------------------------------------------------------------------
@@ -1187,6 +1189,13 @@ def import_unnamed_root_separate_cols(  # noqa: C901
         print(f"Chunks:                {args.chunks:,}")
     if args.blocks is not None:
         print(f"Blocks:                {args.blocks:,}")
+    trunc_global = getattr(args, "float_trunc_prec_global", None)
+    trunc_columns = getattr(args, "float_trunc_prec_columns", {})
+    if trunc_global is not None:
+        print(f"Float trunc precision: {trunc_global} bits (all float columns)")
+    if trunc_columns:
+        formatted = ", ".join(f"{name}={bits}" for name, bits in sorted(trunc_columns.items()))
+        print(f"Float trunc columns:   {formatted}")
     print()
 
     cparams = blosc2.CParams(codec=blosc2.Codec[args.codec], clevel=args.clevel, use_dict=args.use_dict)
@@ -1204,6 +1213,7 @@ def import_unnamed_root_separate_cols(  # noqa: C901
         blosc2_batch_size=args.blosc2_batch_size,
         blosc2_items_per_block=args.blosc2_items_per_block,
         list_serializer=args.list_serializer,
+        column_cparams=float_trunc_column_cparams or None,
         create_summary_index=args.create_summary_index,
         chunks=args.chunks,
         blocks=args.blocks,

tests/ctable/test_arrow_interop.py

@@ -189,6 +189,46 @@ def test_from_arrow_column_cparams(tmp_path):
     assert reopened._cols["x"].cparams.filters[:2] == [blosc2.Filter.TRUNC_PREC, blosc2.Filter.SHUFFLE]
 
 
+def test_from_arrow_column_cparams_nested_struct(tmp_path):
+    # Regression: column_cparams with TRUNC_PREC must be applied to leaf columns
+    # produced by struct flattening, not silently dropped.
+    struct_type = pa.struct(
+        [
+            pa.field("lon", pa.float64()),
+            pa.field("lat", pa.float64()),
+        ]
+    )
+    at = pa.table(
+        {
+            "pos": pa.array(
+                [{"lon": -87.6, "lat": 41.8}, {"lon": -87.7, "lat": 41.9}],
+                type=struct_type,
+            ),
+            "fare": pa.array([10.0, 20.0], type=pa.float32()),
+        }
+    )
+    trunc_cparams = {
+        "codec": blosc2.Codec.ZSTD.value,
+        "clevel": 5,
+        "typesize": 8,
+        "filters": [blosc2.Filter.TRUNC_PREC.value, blosc2.Filter.SHUFFLE.value],
+        "filters_meta": [22, 0],
+    }
+    t = CTable.from_arrow(
+        at.schema,
+        at.to_batches(),
+        urlpath=str(tmp_path / "trunc_nested.b2d"),
+        column_cparams={"pos.lon": trunc_cparams, "pos.lat": trunc_cparams},
+    )
+    assert t.col_names == ["pos.lon", "pos.lat", "fare"]
+    assert t._cols["pos.lon"].cparams.filters[:2] == [blosc2.Filter.TRUNC_PREC, blosc2.Filter.SHUFFLE]
+    assert t._cols["pos.lon"].cparams.filters_meta[:2] == [22, 0]
+    assert t._cols["pos.lat"].cparams.filters[:2] == [blosc2.Filter.TRUNC_PREC, blosc2.Filter.SHUFFLE]
+    # fare is float32, no TRUNC_PREC requested
+    assert blosc2.Filter.TRUNC_PREC not in t._cols["fare"].cparams.filters
+    t.close()
+
+
 def test_from_arrow_string_values():
     # Without string_max_length, scalar strings become vlstring columns.
     # Accessing [:] on a vlstring column returns a Python list, not an ndarray.