ARROW-2218: [Python] PythonFile should infer mode when not given

Author: Antoine Pitrou <antoine@python.org>

Closes apache#1674 from pitrou/ARROW-2218-infer-python-file-mode and squashes the following commits:

a5d704e <Antoine Pitrou> ARROW-2218:  PythonFile should infer mode when not given

Author: pitrou

python/pyarrow/io-hdfs.pxi

@@ -480,7 +480,5 @@ cdef class HdfsFile(NativeFile):
         object mode
         object parent
 
-    cdef object __weakref__
-
     def __dealloc__(self):
         self.parent = None

python/pyarrow/io.pxi

@@ -410,9 +410,21 @@ cdef class PythonFile(NativeFile):
     cdef:
         object handle
 
-    def __cinit__(self, handle, mode='w'):
+    def __cinit__(self, handle, mode=None):
         self.handle = handle
 
+        if mode is None:
+            try:
+                mode = handle.mode
+            except AttributeError:
+                # Not all file-like objects have a mode attribute
+                # (e.g. BytesIO)
+                try:
+                    mode = 'w' if handle.writable() else 'r'
+                except AttributeError:
+                    raise ValueError("could not infer open mode for file-like "
+                                     "object %r, please pass it explicitly"
+                                     % (handle,))
         if mode.startswith('w'):
             self.wr_file.reset(new PyOutputStream(handle))
             self.is_writable = True

python/pyarrow/lib.pxd

@@ -337,6 +337,7 @@ cdef class NativeFile:
         bint is_writable
         readonly bint closed
         bint own_file
+        object __weakref__
 
     # By implementing these "virtual" functions (all functions in Cython
     # extension classes are technically virtual in the C++ sense) we can expose

python/pyarrow/tests/test_io.py

@@ -21,6 +21,7 @@
 import os
 import pytest
 import sys
+import weakref
 
 import numpy as np
 
@@ -124,6 +125,44 @@ def get_buffer():
     assert buf.to_pybytes() == b'sample'
     assert buf.parent is not None
 
+
+def test_python_file_implicit_mode(tmpdir):
+    path = os.path.join(str(tmpdir), 'foo.txt')
+    with open(path, 'wb') as f:
+        pf = pa.PythonFile(f)
+        assert pf.writable()
+        assert not pf.readable()
+        assert not pf.seekable()  # PyOutputStream isn't seekable
+        f.write(b'foobar\n')
+
+    with open(path, 'rb') as f:
+        pf = pa.PythonFile(f)
+        assert pf.readable()
+        assert not pf.writable()
+        assert pf.seekable()
+        assert pf.read() == b'foobar\n'
+
+    bio = BytesIO()
+    pf = pa.PythonFile(bio)
+    assert pf.writable()
+    assert not pf.readable()
+    assert not pf.seekable()
+    pf.write(b'foobar\n')
+    assert bio.getvalue() == b'foobar\n'
+
+
+def test_python_file_closing():
+    bio = BytesIO()
+    pf = pa.PythonFile(bio)
+    wr = weakref.ref(pf)
+    del pf
+    assert wr() is None  # object was destroyed
+    assert not bio.closed
+    pf = pa.PythonFile(bio)
+    pf.close()
+    assert bio.closed
+
+
 # ----------------------------------------------------------------------
 # Buffers