Support binary read when encoding=None (python#29)

Closes python#28

importlib_resources/_py2.py

@@ -73,7 +73,10 @@ def read(package, file_name, encoding='utf-8', errors='strict'):
     package = _get_package(package)
     # Note this is **not** builtins.open()!
     with open(package, file_name) as binary_file:
-        return binary_file.read().decode(encoding=encoding, errors=errors)
+        contents = binary_file.read()
+        if encoding is None:
+            return contents
+        return contents.decode(encoding=encoding, errors=errors)
 
 
 @contextmanager

importlib_resources/_py3.py

@@ -77,8 +77,10 @@ def open(package: Package, file_name: FileName) -> BinaryIO:
                 return BytesIO(data)
 
 
-def read(package: Package, file_name: FileName, encoding: str = 'utf-8',
-         errors: str = 'strict') -> str:
+def read(package: Package,
+         file_name: FileName,
+         encoding: str = 'utf-8',
+         errors: str = 'strict') -> Union[str, bytes]:
     """Return the decoded string of the resource.
 
     The decoding-related arguments have the same semantics as those of
@@ -88,6 +90,8 @@ def read(package: Package, file_name: FileName, encoding: str = 'utf-8',
     package = _get_package(package)
     # Note this is **not** builtins.open()!
     with open(package, file_name) as binary_file:
+        if encoding is None:
+            return binary_file.read()
         # Decoding from io.TextIOWrapper() instead of str.decode() in hopes
         # that the former will be smarter about memory usage.
         text_file = TextIOWrapper(

importlib_resources/tests/test_read.py

@@ -15,17 +15,21 @@ class ReadTests:
 
     def test_default_encoding(self):
         result = resources.read(self.data, 'utf-8.file')
-        self.assertEqual("Hello, UTF-8 world!\n", result)
+        self.assertEqual('Hello, UTF-8 world!\n', result)
 
     def test_encoding(self):
         result = resources.read(self.data, 'utf-16.file', encoding='utf-16')
-        self.assertEqual("Hello, UTF-16 world!\n", result)
+        self.assertEqual('Hello, UTF-16 world!\n', result)
 
     def test_errors(self):
         # Raises UnicodeError without the 'errors' argument.
         resources.read(
             self.data, 'utf-16.file', encoding='utf-8', errors='ignore')
 
+    def test_no_encoding(self):
+        result = resources.read(self.data, 'binary.file', encoding=None)
+        self.assertEqual(b'\0\1\2\3', result)
+
 
 class ReadDiskTests(ReadTests, unittest.TestCase):
 

update-tests.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python3
+
+"""Remake the ziptestdata.zip file.
+
+Run this to rebuild the importlib_resources/tests/data/ziptestdata.zip file,
+e.g. if you want to add a new file to the zip.
+
+This will replace the file with the new build, but it won't commit anything to
+git.
+"""
+
+import os
+from zipfile import ZipFile
+
+RELPATH = 'importlib_resources/tests/data'
+CONTENTS = [
+    # filenames - the source will always be prepended by
+    # importlib_resources/tests/data/ziptestdata.zip and the destination will
+    # always be prepended by ziptestdata/
+    '__init__.py',
+    'binary.file',
+    'utf-16.file',
+    'utf-8.file',
+    ]
+
+
+zip_file_path = os.path.join(RELPATH, 'ziptestdata.zip')
+
+with ZipFile(zip_file_path, 'w') as zf:
+    for filename in CONTENTS:
+        src = os.path.join(RELPATH, filename)
+        dst = os.path.join('ziptestdata', filename)
+        zf.write(src, dst)