[3.11] gh-101961 fileinput.hookcompressed should not set the encoding value for the...

gh-101961 fileinput.hookcompressed should not set the encoding value for the binary mode (gh-102068)
(cherry picked from commit 6f25657b83d7a680a97849490f6e973b3a695e1a)

Co-authored-by: Gihwan Kim <gihwan.kim@linecorp.com>

diff --git a/Lib/fileinput.py b/Lib/fileinput.py
index e234dc9ea65f15ab389635865733fea9eaccc40a..1b25f28f3d343260038e76f8832718dbe4964e71 100644 (file)
--- a/Lib/fileinput.py
+++ b/Lib/fileinput.py
@@ -399,7 +399,7 @@ class FileInput:
 
 
 def hook_compressed(filename, mode, *, encoding=None, errors=None):
-    if encoding is None:  # EncodingWarning is emitted in FileInput() already.
+    if encoding is None and "b" not in mode:  # EncodingWarning is emitted in FileInput() already.
         encoding = "locale"
     ext = os.path.splitext(filename)[1]
     if ext == '.gz':

diff --git a/Lib/test/test_fileinput.py b/Lib/test/test_fileinput.py
index ac20c74baa09e27df2a0f52ca522e00830aa37c0..786d9186634305b2b8a472c90013e536fed6da06 100644 (file)
--- a/Lib/test/test_fileinput.py
+++ b/Lib/test/test_fileinput.py
@@ -855,29 +855,29 @@ class Test_hook_compressed(unittest.TestCase):
         self.fake_open = InvocationRecorder()
 
     def test_empty_string(self):
-        self.do_test_use_builtin_open("", 1)
+        self.do_test_use_builtin_open_text("", "r")
 
     def test_no_ext(self):
-        self.do_test_use_builtin_open("abcd", 2)
+        self.do_test_use_builtin_open_text("abcd", "r")
 
     @unittest.skipUnless(gzip, "Requires gzip and zlib")
     def test_gz_ext_fake(self):
         original_open = gzip.open
         gzip.open = self.fake_open
         try:
-            result = fileinput.hook_compressed("test.gz", "3")
+            result = fileinput.hook_compressed("test.gz", "r")
         finally:
             gzip.open = original_open
 
         self.assertEqual(self.fake_open.invocation_count, 1)
-        self.assertEqual(self.fake_open.last_invocation, (("test.gz", "3"), {}))
+        self.assertEqual(self.fake_open.last_invocation, (("test.gz", "r"), {}))
 
     @unittest.skipUnless(gzip, "Requires gzip and zlib")
     def test_gz_with_encoding_fake(self):
         original_open = gzip.open
         gzip.open = lambda filename, mode: io.BytesIO(b'Ex-binary string')
         try:
-            result = fileinput.hook_compressed("test.gz", "3", encoding="utf-8")
+            result = fileinput.hook_compressed("test.gz", "r", encoding="utf-8")
         finally:
             gzip.open = original_open
         self.assertEqual(list(result), ['Ex-binary string'])
@@ -887,23 +887,40 @@ class Test_hook_compressed(unittest.TestCase):
         original_open = bz2.BZ2File
         bz2.BZ2File = self.fake_open
         try:
-            result = fileinput.hook_compressed("test.bz2", "4")
+            result = fileinput.hook_compressed("test.bz2", "r")
         finally:
             bz2.BZ2File = original_open
 
         self.assertEqual(self.fake_open.invocation_count, 1)
-        self.assertEqual(self.fake_open.last_invocation, (("test.bz2", "4"), {}))
+        self.assertEqual(self.fake_open.last_invocation, (("test.bz2", "r"), {}))
 
     def test_blah_ext(self):
-        self.do_test_use_builtin_open("abcd.blah", "5")
+        self.do_test_use_builtin_open_binary("abcd.blah", "rb")
 
     def test_gz_ext_builtin(self):
-        self.do_test_use_builtin_open("abcd.Gz", "6")
+        self.do_test_use_builtin_open_binary("abcd.Gz", "rb")
 
     def test_bz2_ext_builtin(self):
-        self.do_test_use_builtin_open("abcd.Bz2", "7")
+        self.do_test_use_builtin_open_binary("abcd.Bz2", "rb")
 
-    def do_test_use_builtin_open(self, filename, mode):
+    def test_binary_mode_encoding(self):
+        self.do_test_use_builtin_open_binary("abcd", "rb")
+
+    def test_text_mode_encoding(self):
+        self.do_test_use_builtin_open_text("abcd", "r")
+
+    def do_test_use_builtin_open_binary(self, filename, mode):
+        original_open = self.replace_builtin_open(self.fake_open)
+        try:
+            result = fileinput.hook_compressed(filename, mode)
+        finally:
+            self.replace_builtin_open(original_open)
+
+        self.assertEqual(self.fake_open.invocation_count, 1)
+        self.assertEqual(self.fake_open.last_invocation,
+                         ((filename, mode), {'encoding': None, 'errors': None}))
+
+    def do_test_use_builtin_open_text(self, filename, mode):
         original_open = self.replace_builtin_open(self.fake_open)
         try:
             result = fileinput.hook_compressed(filename, mode)

diff --git a/Misc/ACKS b/Misc/ACKS
index f674332751c39aec1069765e14edfd32f7cf013b..523b9e8380d5fda5a785ca88a3ed7cb14715529c 100644 (file)
--- a/Misc/ACKS
+++ b/Misc/ACKS
@@ -919,6 +919,7 @@ Tyler Kieft
 Mads Kiilerich
 Jason Killen
 Derek D. Kim
+Gihwan Kim
 Jan Kim
 Taek Joo Kim
 Sam Kimbrel

diff --git a/Misc/NEWS.d/next/Library/2023-02-21-10-05-33.gh-issue-101961.7e56jh.rst b/Misc/NEWS.d/next/Library/2023-02-21-10-05-33.gh-issue-101961.7e56jh.rst
new file mode 100644 (file)
index 0000000..a3d4119
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2023-02-21-10-05-33.gh-issue-101961.7e56jh.rst
@@ -0,0 +1,2 @@
+For the binary mode, :func:`fileinput.hookcompressed` doesn't set the ``encoding`` value
+even if the value is ``None``. Patch by Gihwan Kim.