Fix Parquet Reader's Arrow Schema Inference

parquet/src/arrow/array_reader/builder.rs

@@ -151,6 +151,12 @@ fn build_primitive_reader(
 
     let physical_type = primitive_type.get_physical_type();
 
+    // We don't track the column path in ParquetField as it adds a potential source
+    // of bugs when the arrow mapping converts more than one level in the parquet
+    // schema into a single arrow field.
+    //
+    // None of the readers actually use this field, but it is required for this type,
+    // so just stick a placeholder in
     let column_desc = Arc::new(ColumnDescriptor::new(
         primitive_type,
         field.def_level,
@@ -305,7 +311,6 @@ fn build_primitive_reader(
     }
 }
 
-/// Constructs struct array reader without considering repetition.
 fn build_struct_reader(
     field: &ParquetField,
     row_groups: &dyn RowGroupCollection,

parquet/src/arrow/arrow_writer.rs

@@ -1058,7 +1058,7 @@ mod tests {
         let stocks_field = Field::new(
             "stocks",
             DataType::Map(
-                Box::new(Field::new("entries", entries_struct_type, true)),
+                Box::new(Field::new("entries", entries_struct_type, false)),
                 false,
             ),
             true,

parquet/src/arrow/schema.rs

@@ -1128,6 +1128,24 @@ mod tests {
         for i in 0..arrow_fields.len() {
             assert_eq!(arrow_fields[i], converted_fields[i]);
         }
+
+        let err =
+            parquet_to_arrow_schema_by_columns(&parquet_schema, vec![3, 2, 4], None)
+                .unwrap_err()
+                .to_string();
+
+        assert!(
+            err.contains("out of order projection is not supported"),
+            "{}",
+            err
+        );
+
+        let err =
+            parquet_to_arrow_schema_by_columns(&parquet_schema, vec![3, 3, 4], None)
+                .unwrap_err()
+                .to_string();
+
+        assert!(err.contains("repeated column projection is not supported, column 3 appeared multiple times"), "{}", err);
     }
 
     #[test]

parquet/src/arrow/schema/complex.rs

@@ -346,7 +346,7 @@ impl Visitor {
                 let map_field = Field::new(
                     map_key_value.name(),
                     DataType::Struct(vec![key_field, value_field]),
-                    nullable,
+                    false, // The inner map field is always non-nullable (#1697)
                 )
                 .with_metadata(arrow_map.and_then(|f| f.metadata().cloned()));
 
@@ -560,6 +560,10 @@ pub fn convert_schema<T: IntoIterator<Item = usize>>(
         if i < last_idx {
             return Err(general_err!("out of order projection is not supported"));
         }
+        if leaf_mask[i] {
+            return Err(general_err!("repeated column projection is not supported, column {} appeared multiple times", i));
+        }
+
         last_idx = i;
         leaf_mask[i] = true;
     }