feat(ipc): Read and write DictionaryEncoding.custom_metadata for dictionary values

src/nanoarrow/ipc/decoder.c

@@ -1275,9 +1275,11 @@ static int ArrowIpcDecoderSetType(struct ArrowSchema* schema, ns(Field_table_t)
 }
 
 // When decoding dictionaries, we move the value type to the schema->dictionary
-// member, but we need to move the field metadata back because in IPC there
-// is no such thing as dictionary metadata (even extension metadata)
-// https://github.com/apache/arrow/issues/49704
+// member, but we need to move the field metadata back because Field.custom_metadata
+// is the metadata of the field (i.e., the dictionary-encoded ArrowSchema). The
+// metadata of the dictionary value type is DictionaryEncoding.custom_metadata
+// (format version 1.6, https://github.com/apache/arrow/issues/49704), which is
+// set on schema->dictionary separately.
 static int ArrowIpcMoveDictionaryMetadataBackToField(struct ArrowSchema* schema) {
   NANOARROW_DCHECK(schema->dictionary != NULL);
   NANOARROW_RETURN_NOT_OK(ArrowSchemaSetMetadata(schema, schema->dictionary->metadata));
@@ -1338,6 +1340,13 @@ static int ArrowIpcSetDictionaryEncoding(
   NANOARROW_RETURN_NOT_OK_WITH_ERROR(ArrowIpcMoveDictionaryMetadataBackToField(schema),
                                      error);
 
+  // The metadata of the dictionary value type (e.g., the extension type of the values),
+  // which is absent if the writer predates format version 1.6 or the value type has no
+  // metadata.
+  NANOARROW_RETURN_NOT_OK(ArrowIpcDecoderSetMetadata(
+      schema->dictionary, ns(DictionaryEncoding_custom_metadata(dictionary_encoding)),
+      error));
+
   // Track the identifier if we have a dictionaries object in which to track it
   if (dictionaries != NULL) {
     int64_t id = ns(DictionaryEncoding_id(dictionary_encoding));

src/nanoarrow/ipc/encoder.c

@@ -608,20 +608,37 @@ static ArrowErrorCode ArrowIpcEncodeField(
         return EINVAL;
     }
 
-    // Create the Int type for the index type
-    ns(Int_ref_t) index_type_ref =
-        ns(Int_create(builder, index_bitwidth, index_is_signed));
-    FLATCC_RETURN_IF_NULL(index_type_ref, error);
-
-    // Create the DictionaryEncoding with id, indexType, isOrdered, and dictionaryKind
+    // Add the DictionaryEncoding with id, indexType, isOrdered, dictionaryKind, and
+    // custom_metadata to the field
     flatbuffers_bool_t is_ordered = (schema->flags & ARROW_FLAG_DICTIONARY_ORDERED) != 0;
-    ns(DictionaryEncoding_ref_t) dict_encoding_ref =
-        ns(DictionaryEncoding_create(builder, encoding->id, index_type_ref, is_ordered,
-                                     ns(DictionaryKind_DenseArray)));
-    FLATCC_RETURN_IF_NULL(dict_encoding_ref, error);
+    FLATCC_RETURN_UNLESS_0(Field_dictionary_start(builder), error);
+    FLATCC_RETURN_UNLESS_0(DictionaryEncoding_id_add(builder, encoding->id), error);
+    FLATCC_RETURN_UNLESS_0(
+        DictionaryEncoding_indexType_create(builder, index_bitwidth, index_is_signed),
+        error);
+    FLATCC_RETURN_UNLESS_0(DictionaryEncoding_isOrdered_add(builder, is_ordered), error);
+    FLATCC_RETURN_UNLESS_0(
+        DictionaryEncoding_dictionaryKind_add(builder, ns(DictionaryKind_DenseArray)),
+        error);
+
+    // The metadata of the dictionary (value) schema is the metadata of the dictionary
+    // value type, whose IPC counterpart is DictionaryEncoding.custom_metadata (format
+    // version 1.6, https://github.com/apache/arrow/issues/49704). The metadata of the
+    // field itself (i.e., schema->metadata) is written to Field.custom_metadata below.
+    // Writers omit DictionaryEncoding.custom_metadata if there is no value metadata.
+    struct ArrowMetadataReader value_metadata;
+    NANOARROW_RETURN_NOT_OK_WITH_ERROR(
+        ArrowMetadataReaderInit(&value_metadata, schema->dictionary->metadata), error);
+    if (value_metadata.remaining_keys > 0) {
+      FLATCC_RETURN_UNLESS_0(DictionaryEncoding_custom_metadata_start(builder), error);
+      NANOARROW_RETURN_NOT_OK(ArrowIpcEncodeMetadata(
+          builder, schema->dictionary->metadata,
+          &ns(DictionaryEncoding_custom_metadata_push_start),
+          &ns(DictionaryEncoding_custom_metadata_push_end), error));
+      FLATCC_RETURN_UNLESS_0(DictionaryEncoding_custom_metadata_end(builder), error);
+    }
 
-    // Add the dictionary encoding to the field
-    FLATCC_RETURN_UNLESS_0(Field_dictionary_add(builder, dict_encoding_ref), error);
+    FLATCC_RETURN_UNLESS_0(Field_dictionary_end(builder), error);
 
     // Support dictionary values with children by encoding children from
     // schema->dictionary (and add a roundtrip test for a nested value type).