Replace our implementation of shebang parsing with identify's

pre_commit/parse_shebang.py

@@ -1,13 +1,9 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
-import io
 import os.path
-import shlex
-import string
 
-
+from identify.identify import parse_shebang_from_file
-printable = frozenset(string.printable)
 
 
 class ExecutableNotFoundError(OSError):
@@ -15,34 +11,11 @@ class ExecutableNotFoundError(OSError):
         return (1, self.args[0].encode('UTF-8'), b'')
 
 
-def parse_bytesio(bytesio):
-    """Parse the shebang from a file opened for reading binary."""
-    if bytesio.read(2) != b'#!':
-        return ()
-    first_line = bytesio.readline()
-    try:
-        first_line = first_line.decode('US-ASCII')
-    except UnicodeDecodeError:
-        return ()
-
-    # Require only printable ascii
-    for c in first_line:
-        if c not in printable:
-            return ()
-
-    cmd = tuple(shlex.split(first_line))
-    if cmd[0] == '/usr/bin/env':
-        cmd = cmd[1:]
-    return cmd
-
-
 def parse_filename(filename):
-    """Parse the shebang given a filename."""
+    if not os.path.exists(filename):
-    if not os.path.exists(filename) or not os.access(filename, os.X_OK):
         return ()
-
+    else:
-    with io.open(filename, 'rb') as f:
+        return parse_shebang_from_file(filename)
-        return parse_bytesio(f)
 
 
 def find_executable(exe, _environ=None):

tests/parse_shebang_test.py

@@ -15,36 +15,10 @@ from pre_commit.envcontext import Var
 from pre_commit.util import make_executable
 
 
-@pytest.mark.parametrize(
-    ('s', 'expected'),
-    (
-        (b'', ()),
-        (b'#!/usr/bin/python', ('/usr/bin/python',)),
-        (b'#!/usr/bin/env python', ('python',)),
-        (b'#! /usr/bin/python', ('/usr/bin/python',)),
-        (b'#!/usr/bin/foo  python', ('/usr/bin/foo', 'python')),
-        (b'\xf9\x93\x01\x42\xcd', ()),
-        (b'#!\xf9\x93\x01\x42\xcd', ()),
-        (b'#!\x00\x00\x00\x00', ()),
-    ),
-)
-def test_parse_bytesio(s, expected):
-    assert parse_shebang.parse_bytesio(io.BytesIO(s)) == expected
-
-
 def test_file_doesnt_exist():
     assert parse_shebang.parse_filename('herp derp derp') == ()
 
 
-@pytest.mark.xfail(
-    sys.platform == 'win32', reason='Windows says everything is X_OK',
-)
-def test_file_not_executable(tmpdir):
-    x = tmpdir.join('f')
-    x.write_text('#!/usr/bin/env python', encoding='UTF-8')
-    assert parse_shebang.parse_filename(x.strpath) == ()
-
-
 def test_simple_case(tmpdir):
     x = tmpdir.join('f')
     x.write_text('#!/usr/bin/env python', encoding='UTF-8')