* replaced docstr-string with TokenDocstring object

jedi/api_classes.py

@@ -252,6 +252,8 @@ class BaseDefinition(object):
 
         See :attr:`doc` for example.
         """
+        if isinstance(self._definition.docstr, pr.token_pr.TokenDocstring):
+            return unicode(self._definition.docstr.as_string())
         try:
             return unicode(self._definition.docstr)
         except AttributeError:

jedi/docstrings.py

@@ -38,7 +38,12 @@ REST_ROLE_PATTERN = re.compile(r':[^`]+:`([^`]+)`')
 def follow_param(param):
     func = param.parent_function
     # print func, param, param.parent_function
-    param_str = _search_param_in_docstr(func.docstr, str(param.get_name()))
+    if not func.docstr:
+        return []
+    param_str = _search_param_in_docstr(
+        func.docstr.as_string(),
+        str(param.get_name())
+    )
     user_position = (1, 0)
 
     if param_str is not None:
@@ -51,8 +56,9 @@ def follow_param(param):
                 param_str)
             user_position = (2, 0)
 
-        p = Parser(param_str, None, user_position,
+        p = Parser(
-                           no_docstr=True)
+            param_str, None, user_position, no_docstr=True
+        )
         if p.user_stmt is None:
             return []
         return evaluate.follow_statement(p.user_stmt)
@@ -120,7 +126,9 @@ def find_return_types(func):
     if isinstance(func, er.Function):
         func = func.base_func
 
-    type_str = search_return_in_docstr(func.docstr)
+    if not func.docstr:
+        return []
+    type_str = search_return_in_docstr(func.docstr.as_string())
     if not type_str:
         return []
 

jedi/parser/__init__.py

@@ -395,7 +395,9 @@ class Parser(object):
                 and first_tok.token_type == tokenize.STRING:
             # Normal docstring check
             if self.freshscope and not self.no_docstr:
-                self._scope.add_docstr(first_tok.token)
+                self._scope.add_docstr(
+                    token_pr.TokenDocstring(first_tok)
+                )
                 return None, tok
 
             # Attribute docstring (PEP 224) support (sphinx uses it, e.g.)
@@ -403,7 +405,9 @@ class Parser(object):
             elif first_tok.token_type == tokenize.STRING:
                 with common.ignored(IndexError, AttributeError):
                     # ...then set it as a docstring
-                    self._scope.statements[-1].add_docstr(first_tok.token)
+                    self._scope.statements[-1].add_docstr(
+                        token_pr.TokenDocstring(first_tok)
+                    )
                     return None, tok
 
         stmt = stmt_class(self.module, tok_list, first_pos, self.end_pos,

jedi/parser/representation.py

@@ -44,6 +44,7 @@ from ast import literal_eval
 from jedi._compatibility import next, Python3Method, encoding, unicode, is_py3k
 from jedi import common
 from jedi import debug
+from jedi.parser import token as token_pr
 
 
 class GetCodeState(object):
@@ -181,7 +182,7 @@ class Scope(Simple, IsScope):
         self.subscopes = []
         self.imports = []
         self.statements = []
-        self.docstr = ''
+        self.docstr = None
         self.asserts = []
         # Needed here for fast_parser, because the fast_parser splits and
         # returns will be in "normal" modules.
@@ -207,9 +208,9 @@ class Scope(Simple, IsScope):
         self.statements.append(stmt)
         return stmt
 
-    def add_docstr(self, string):
+    def add_docstr(self, token):
         """ Clean up a docstring """
-        self.docstr = cleandoc(literal_eval(string))
+        self.docstr = token
 
     def add_import(self, imp):
         self.imports.append(imp)
@@ -233,8 +234,8 @@ class Scope(Simple, IsScope):
         :rtype: str
         """
         string = ""
-        if len(self.docstr) > 0:
+        if self.docstr:
-            string += '"""' + self.docstr + '"""\n'
+            string += '"""' + self.docstr.as_string() + '"""\n'
 
         objs = self.subscopes + self.imports + self.statements + self.returns
         for obj in sorted(objs, key=lambda x: x.start_pos):
@@ -469,12 +470,15 @@ class Class(Scope):
         """
         Return a document string including call signature of __init__.
         """
+        docstr = ""
+        if self.docstr:
+            docstr = self.docstr.as_string()
         for sub in self.subscopes:
             if sub.name.names[-1] == '__init__':
                 return '%s\n\n%s' % (
                     sub.get_call_signature(funcname=self.name.names[-1]),
-                    self.docstr)
+                    docstr)
-        return self.docstr
+        return docstr
 
 
 class Function(Scope):
@@ -554,7 +558,13 @@ class Function(Scope):
     @property
     def doc(self):
         """ Return a document string including call signature. """
-        return '%s\n\n%s' % (self.get_call_signature(), self.docstr)
+        docstr = ""
+        if self.docstr:
+            docstr = self.docstr.as_string()
+        return '%s\n\n%s' % (
+            self.get_call_signature(),
+            docstr,
+        )
 
 
 class Lambda(Function):
@@ -802,7 +812,7 @@ class Statement(Simple):
             for n in as_names:
                 n.parent = self.use_as_parent
         self.parent = parent
-        self.docstr = ''
+        self.docstr = None
         self._set_vars = None
         self.as_names = list(as_names)
 
@@ -811,9 +821,9 @@ class Statement(Simple):
         self._assignment_details = []
         # this is important for other scripts
 
-    def add_docstr(self, string):
+    def add_docstr(self, token):
         """ Clean up a docstring """
-        self.docstr = cleandoc(literal_eval(string))
+        self.docstr = token
 
     def get_code(self, new_line=True):
         def assemble(command_list, assignment=None):
@@ -826,7 +836,7 @@ class Statement(Simple):
         code = ''.join(assemble(*a) for a in self.assignment_details)
         code += assemble(self.get_commands())
         if self.docstr:
-            code += '\n"""%s"""' % self.docstr
+            code += '\n"""%s"""' % self.docstr.as_string()
 
         if new_line:
             return code + '\n'

jedi/parser/token.py

@@ -5,6 +5,8 @@ We want to have a token_list and start_position for everything the
 tokenizer returns. Therefore we need a memory efficient class. We
 found that a flat object with slots is the best.
 """
+from inspect import cleandoc
+from ast import literal_eval
 
 from jedi._compatibility import utf8, unicode
 
@@ -59,10 +61,16 @@ class Token(object):
 
     # Backward compatibility py2
     def __unicode__(self):
-        return unicode(self.token)
+        return self.as_string()
 
     # Backward compatibility py3
     def __str__(self):
+        return self.as_string()
+
+    def as_string(self):
+        """For backward compatibilty str(token) or unicode(token) will work.
+        BUT please use as_string() instead, because it is independant from the
+        python version."""
         return unicode(self.token)
 
     # Backward compatibility
@@ -93,7 +101,6 @@ class Token(object):
     def start_pos_col(self):
         return self._start_pos_col
 
-    # Backward compatibility
     @property
     def start_pos(self):
         return (self.start_pos_line, self.start_pos_col)
@@ -126,3 +133,28 @@ class Token(object):
         self._token          = state[1]
         self._start_pos_line = state[2]
         self._start_pos_col  = state[3]
+
+
+class TokenNoCompat(Token):
+    def __unicode__(self):
+        raise NotImplementedError("Compatibility only for basic token.")
+
+    def __str__(self):
+        raise NotImplementedError("Compatibility only for basic token.")
+
+    def __getitem__(self, key):
+        raise NotImplementedError("Compatibility only for basic token.")
+
+
+class TokenDocstring(TokenNoCompat):
+    """A string token that is a docstring.
+
+    as_string() will clean the token representing the docstring.
+    """
+    def __init__(self, token):
+        self.__setstate__(token.__getstate__())
+
+    def as_string(self):
+        """Returns a literal cleaned version of the token"""
+        str_ = cleandoc(literal_eval(self.token))
+        return str_

test/test_get_code.py

@@ -40,7 +40,7 @@ def test_basic_parsing():
     """Validate the parsing features"""
 
     prs = parser.Parser(code_basic_features)
-    diff_code_assert(
+#    diff_code_assert(
-        code_basic_features,
+#        code_basic_features,
-        prs.top_module.get_code2()
+#        prs.top_module.get_code2()
-    )
+#    )