Issue #18873: IDLE, 2to3, and the findnocoding.py script now detect Python
source code encoding only in comment lines.
diff --git a/Lib/lib2to3/pgen2/tokenize.py b/Lib/lib2to3/pgen2/tokenize.py
index e090aa9..9cf2a69 100644
--- a/Lib/lib2to3/pgen2/tokenize.py
+++ b/Lib/lib2to3/pgen2/tokenize.py
@@ -236,7 +236,7 @@
startline = False
toks_append(tokval)
-cookie_re = re.compile("coding[:=]\s*([-\w.]+)")
+cookie_re = re.compile(r'^[ \t\f]*#.*coding[:=][ \t]*([-\w.]+)', re.ASCII)
def _get_normal_name(orig_enc):
"""Imitates get_normal_name in tokenizer.c."""
@@ -281,11 +281,10 @@
line_string = line.decode('ascii')
except UnicodeDecodeError:
return None
-
- matches = cookie_re.findall(line_string)
- if not matches:
+ match = cookie_re.match(line_string)
+ if not match:
return None
- encoding = _get_normal_name(matches[0])
+ encoding = _get_normal_name(match.group(1))
try:
codec = lookup(encoding)
except LookupError:
diff --git a/Lib/lib2to3/tests/data/false_encoding.py b/Lib/lib2to3/tests/data/false_encoding.py
new file mode 100644
index 0000000..f4e59e7
--- /dev/null
+++ b/Lib/lib2to3/tests/data/false_encoding.py
@@ -0,0 +1,2 @@
+#!/usr/bin/env python
+print '#coding=0'
diff --git a/Lib/lib2to3/tests/test_refactor.py b/Lib/lib2to3/tests/test_refactor.py
index 6020d1f..543fd42 100644
--- a/Lib/lib2to3/tests/test_refactor.py
+++ b/Lib/lib2to3/tests/test_refactor.py
@@ -271,6 +271,10 @@
fn = os.path.join(TEST_DATA_DIR, "different_encoding.py")
self.check_file_refactoring(fn)
+ def test_false_file_encoding(self):
+ fn = os.path.join(TEST_DATA_DIR, "false_encoding.py")
+ data = self.check_file_refactoring(fn)
+
def test_bom(self):
fn = os.path.join(TEST_DATA_DIR, "bom.py")
data = self.check_file_refactoring(fn)