diff --git a/python/pyarrow/parquet.py b/python/pyarrow/parquet.py
index a02dd56c9b1..e75930a1bc4 100644
--- a/python/pyarrow/parquet.py
+++ b/python/pyarrow/parquet.py
@@ -27,6 +27,7 @@
 import os
 import re
 import six
+import operator
 
 import pyarrow as pa
 import pyarrow.lib as lib
@@ -142,15 +143,17 @@ def _build_nested_paths(self):
 
         result = defaultdict(list)
 
-        def _visit_piece(i, key, rest):
-            result[key].append(i)
+        for i, path in enumerate(paths):
+            key = path[0]
+            rest = path[1:]
+            while True:
+                result[key].append(i)
 
-            if len(rest) > 0:
-                nested_key = '.'.join((key, rest[0]))
-                _visit_piece(i, nested_key, rest[1:])
+                if not rest:
+                    break
 
-        for i, path in enumerate(paths):
-            _visit_piece(i, path[0], path[1:])
+                key = '.'.join((key, rest[0]))
+                rest = rest[1:]
 
         return result
 
@@ -922,6 +925,11 @@ def _path_split(path, sep):
 EXCLUDED_PARQUET_PATHS = {'_SUCCESS'}
 
 
+class _ParquetDatasetMetadata:
+    __slots__ = ('fs', 'memory_map', 'read_dictionary', 'common_metadata',
+                 'buffer_size')
+
+
 def _open_dataset_file(dataset, path, meta=None):
     if dataset.fs is not None and not isinstance(dataset.fs, LocalFileSystem):
         path = dataset.fs.open(path, mode='rb')
@@ -999,32 +1007,37 @@ def __init__(self, path_or_paths, filesystem=None, schema=None,
                  metadata=None, split_row_groups=False, validate_schema=True,
                  filters=None, metadata_nthreads=1, read_dictionary=None,
                  memory_map=False, buffer_size=0):
+        self._metadata = _ParquetDatasetMetadata()
         a_path = path_or_paths
         if isinstance(a_path, list):
             a_path = a_path[0]
 
-        self.fs, _ = _get_filesystem_and_path(filesystem, a_path)
+        self._metadata.fs, _ = _get_filesystem_and_path(filesystem, a_path)
         if isinstance(path_or_paths, list):
             self.paths = [_parse_uri(path) for path in path_or_paths]
         else:
             self.paths = _parse_uri(path_or_paths)
 
-        self.read_dictionary = read_dictionary
-        self.memory_map = memory_map
-        self.buffer_size = buffer_size
+        self._metadata.read_dictionary = read_dictionary
+        self._metadata.memory_map = memory_map
+        self._metadata.buffer_size = buffer_size
 
         (self.pieces,
          self.partitions,
          self.common_metadata_path,
          self.metadata_path) = _make_manifest(
              path_or_paths, self.fs, metadata_nthreads=metadata_nthreads,
-             open_file_func=partial(_open_dataset_file, self))
+             open_file_func=partial(_open_dataset_file, self._metadata)
+        )
 
         if self.common_metadata_path is not None:
             with self.fs.open(self.common_metadata_path) as f:
-                self.common_metadata = read_metadata(f, memory_map=memory_map)
+                self._metadata.common_metadata = read_metadata(
+                    f,
+                    memory_map=memory_map
+                )
         else:
-            self.common_metadata = None
+            self._metadata.common_metadata = None
 
         if metadata is None and self.metadata_path is not None:
             with self.fs.open(self.metadata_path) as f:
@@ -1171,6 +1184,16 @@ def all_filters_accept(piece):
 
         self.pieces = [p for p in self.pieces if all_filters_accept(p)]
 
+    fs = property(operator.attrgetter('_metadata.fs'))
+    memory_map = property(operator.attrgetter('_metadata.memory_map'))
+    read_dictionary = property(
+        operator.attrgetter('_metadata.read_dictionary')
+    )
+    common_metadata = property(
+        operator.attrgetter('_metadata.common_metadata')
+    )
+    buffer_size = property(operator.attrgetter('_metadata.buffer_size'))
+
 
 def _make_manifest(path_or_paths, fs, pathsep='/', metadata_nthreads=1,
                    open_file_func=None):