Merge lp:~henninge/pocket-lint/bug-776389-pep-0263-encoding into lp:pocket-lint

[Henning Eggers]

Fix bug 776389 by applying the regex from PEP 0263 to the first two lines of each file. Store the encoding instead of the old is_utf8 flag.

Currently, no further use is made of this extended information but I guess check_text could verify the encoding of each line now. It could also check if the encoding is valid in the first place.

[lp:~henninge/pocket-lint/bug-776389-pep-0263-encoding]

383. By Henning Eggers on 2011-05-04

Check text first and let it set the encoding.

=== modified file 'pocketlint/formatcheck.py'
--- pocketlint/formatcheck.py	2011-04-18 19:38:52 +0000
+++ pocketlint/formatcheck.py	2011-05-04 14:48:38 +0000
@@ -448,18 +448,21 @@
 class PythonChecker(BaseChecker, AnyTextMixin):
     """Check python source code."""
 
+    # This regex is taken from PEP 0263.
+    encoding_pattern = re.compile("coding[:=]\s*([-\w.]+)")
+
     def __init__(self, file_path, text, reporter=None):
         super(PythonChecker, self).__init__(
             file_path, text, reporter=reporter)
-        self.is_utf8 = False
+        self.encoding = 'ascii'
 
     def check(self):
         """Check the syntax of the python code."""
         if self.text == '':
             return
+        self.check_text()
         self.check_flakes()
         self.check_pep8()
-        self.check_text()
 
     def check_flakes(self):
         """Check compilation and syntax."""
@@ -500,8 +503,10 @@
         """Call each line_method for each line in text."""
         for line_no, line in enumerate(self.text.splitlines()):
             line_no += 1
-            if line_no in (1, 2) and '# -*- coding: utf-8 -*-' in line:
-                self.is_utf8 = True
+            if line_no in (1, 2):
+                match = self.encoding_pattern.search(line)
+                if match:
+                    self.encoding = match.group(1).lower()
             self.check_pdb(line_no, line)
             self.check_length(line_no, line)
             self.check_trailing_whitespace(line_no, line)
@@ -518,7 +523,7 @@
 
     def check_ascii(self, line_no, line):
         """Check that the line is ascii."""
-        if self.is_utf8:
+        if self.encoding != 'ascii':
             return
         try:
             line.encode('ascii')

=== modified file 'pocketlint/tests/test_python.py'
--- pocketlint/tests/test_python.py	2011-04-19 16:33:22 +0000
+++ pocketlint/tests/test_python.py	2011-05-04 14:48:38 +0000
@@ -201,13 +201,37 @@
         self.assertEqual(
             [(1, 'Line contains a call to pdb.')], self.reporter.messages)
 
-    def test_code_is_utf8(self):
+    def _test_encoding(self, python, expected_encoding='foo-encoding'):
+        checker = PythonChecker(
+           'bogus', python % dict(encoding=expected_encoding), self.reporter)
+        checker.check_text()
+        self.assertEqual(expected_encoding, checker.encoding)
+    
+    def test_pep0263_no_encoding(self):
+        self._test_encoding("# First line\n# Second line\n\n", 'ascii')
+        
+    def test_pep0263_encoding_standard_coding(self):
+        self._test_encoding("# coding=%(encoding)s\n")
+        
+    def test_pep0263_encoding_standard_encoding(self):
+        self._test_encoding("# encoding=%(encoding)s\n")
+
+    def test_pep0263_encoding_emacs(self):
+        self._test_encoding("# -*- coding: %(encoding)s -*-\n")
+        
+    def test_pep0263_encoding_vim(self):
+        self._test_encoding("# vim: set fileencoding=%(encoding)s :\n")
+        
+    def test_pep0263_encoding_2nd_line(self):
+        self._test_encoding("# First line\n# coding=%(encoding)s\n\n")
+        
+    def test_code_utf8(self):
         utf8_python = u"a = 'this is utf-8 [\u272a]'"
         checker = PythonChecker('bogus', utf8_python, self.reporter)
-        checker.is_utf8 = True
+        checker.encoding = 'utf-8'
         checker.check_text()
 
-    def test_code_ascii_is_not_is_utf8(self):
+    def test_code_ascii_is_not_utf8(self):
         utf8_python = u"a = 'this is utf-8 [\u272a]'"
         checker = PythonChecker('bogus', utf8_python, self.reporter)
         checker.check_text()