Add better line length checks (E501).

parso/python/normalizer.py

@@ -455,15 +455,7 @@ class PEP8Normalizer(Normalizer):
             self._check_spacing(leaf, spacing)
 
         self._analyse_non_prefix(leaf)
-        last_column = leaf.end_pos[1]
+        self._check_line_length(leaf, spacing)
-        if last_column > self._config.max_characters:
-            self.add_issue(
-                501,
-                'Line too long (%s > %s characters)' %
-                    (last_column, self._config.max_characters),
-                leaf
-            )
-
         # -------------------------------
         # Finalizing. Updating the state.
         # -------------------------------
@@ -504,6 +496,29 @@ class PEP8Normalizer(Normalizer):
         self._previous_spacing = spacing
         return value
 
+    def _check_line_length(self, leaf, spacing):
+        if leaf.type == 'backslash':
+            last_column = leaf.start_pos[1] + 1
+        else:
+            last_column = leaf.end_pos[1]
+        if last_column > self._config.max_characters \
+                and spacing.start_pos[1] <= self._config.max_characters :
+            # Special case for long URLs in multi-line docstrings or comments,
+            # but still report the error when the 72 first chars are whitespaces.
+            report = True
+            if leaf.type == 'comment':
+                splitted = leaf.value[1:].split()
+                if len(splitted) == 1 \
+                        and (leaf.end_pos[1] - len(splitted[0])) < 72:
+                    report = False
+            if report:
+                self.add_issue(
+                    501,
+                    'Line too long (%s > %s characters)' %
+                        (last_column, self._config.max_characters),
+                    leaf
+                )
+
     def _check_spacing(self, leaf, spacing):
         def add_if_spaces(*args):
             if spaces:

test/normalizer_issue_files/E50.py

@@ -0,0 +1,126 @@
+#: E501:4
+a = '12345678901234567890123456789012345678901234567890123456789012345678901234567890'
+#: E501:80
+a = '1234567890123456789012345678901234567890123456789012345678901234567890' or \
+    6
+#: E501+1:80
+a = 7 or \
+    '1234567890123456789012345678901234567890123456789012345678901234567890' or \
+    6
+#: E501+1:80 E501+2:80
+a = 7 or \
+    '1234567890123456789012345678901234567890123456789012345678901234567890' or \
+    '1234567890123456789012345678901234567890123456789012345678901234567890' or \
+    6
+#: E501:78
+a = '1234567890123456789012345678901234567890123456789012345678901234567890'  # \
+#: E502:78
+a = ('123456789012345678901234567890123456789012345678901234567890123456789'  \
+     '01234567890')
+#: E502+1:11
+a = ('AAA  \
+      BBB' \
+     'CCC')
+#: E502:38
+if (foo is None and bar is "e000" and \
+        blah == 'yeah'):
+    blah = 'yeahnah'
+#
+# Okay
+a = ('AAA'
+     'BBB')
+
+a = ('AAA  \
+      BBB'
+     'CCC')
+
+a = 'AAA'    \
+    'BBB'    \
+    'CCC'
+
+a = ('AAA\
+BBBBBBBBB\
+CCCCCCCCC\
+DDDDDDDDD')
+#
+# Okay
+if aaa:
+    pass
+elif bbb or \
+        ccc:
+    pass
+
+ddd = \
+    ccc
+
+('\
+    ' + ' \
+')
+('''
+    ''' + ' \
+')
+#: E501:67 E225:21 E225:22
+very_long_identifiers=and_terrible_whitespace_habits(are_no_excuse+for_long_lines)
+#
+#: E501
+'''multiline string
+with a long long long long long long long long long long long long long long long long line
+'''
+#: E501
+'''same thing, but this time without a terminal newline in the string
+long long long long long long long long long long long long long long long long line'''
+#
+# issue 224 (unavoidable long lines in docstrings)
+# Okay
+"""
+I'm some great documentation.  Because I'm some great documentation, I'm
+going to give you a reference to some valuable information about some API
+that I'm calling:
+
+    http://msdn.microsoft.com/en-us/library/windows/desktop/aa363858(v=vs.85).aspx
+"""
+#: E501
+"""
+longnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaces"""
+
+
+# Regression test for #622
+def foo():
+    """Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis pulvinar vitae
+    """
+
+
+# Okay
+"""
+This
+                                                                       almost_empty_line
+"""
+
+"""
+This
+                                                                        almost_empty_line
+"""
+# A basic comment
+#: E501
+# with a long long long long long long long long long long long long long long long long line
+
+#
+# Okay
+# I'm some great comment.  Because I'm so great, I'm going to give you a
+# reference to some valuable information about some API that I'm calling:
+#
+#     http://msdn.microsoft.com/en-us/library/windows/desktop/aa363858(v=vs.85).aspx
+
+x = 3
+
+# longnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaceslongnospaces
+
+#
+# Okay
+# This
+#                                                                      almost_empty_line
+
+#
+#: E501+1
+# This
+#                                                                       almost_empty_line