Don't test python 2.6 in tox by default, because the newer pip versions don't support it anymore

f-strings are now parsed as part of the Python grammar and not in separate
steps.

Note that this is not the way that CPython does it. CPython still uses multiple
parse steps in ast.c.

.gitignore

@@ -9,3 +9,4 @@
 /dist/
 parso.egg-info/
 /.cache/
+/.pytest_cache

.travis.yml

@@ -7,6 +7,7 @@ python:
   - 3.4
   - 3.5
   - 3.6
+  - 3.7-dev
   - pypy
 matrix:
   allow_failures:

AUTHORS.txt

@@ -5,6 +5,7 @@ David Halter (@davidhalter) <davidhalter88@gmail.com>
 
 Code Contributors
 =================
+Alisdair Robertson (@robodair)
 
 
 Code Contributors (to Jedi and therefore possibly to this library)

CHANGELOG.rst

@@ -3,8 +3,19 @@
 Changelog
 ---------
 
+0.2.0 (2018-04-15)
++++++++++++++++++++
 
-0.1.0 (2017-05-30)
+- f-strings are now parsed as a part of the normal Python grammar. This makes
+  it way easier to deal with them.
+
+0.1.1 (2017-11-05)
++++++++++++++++++++
+
+- Fixed a few bugs in the caching layer
+- Added support for Python 3.7
+
+0.1.0 (2017-09-04)
 +++++++++++++++++++
 
 - Pulling the library out of Jedi. Some APIs will definitely change.

README.rst

@@ -1,5 +1,5 @@
 ###################################################################
-parso - A Python Parser Written in Python
+parso - A Python Parser
 ###################################################################
 
 .. image:: https://secure.travis-ci.org/davidhalter/parso.png?branch=master
@@ -10,19 +10,52 @@ parso - A Python Parser Written in Python
     :target: https://coveralls.io/r/davidhalter/parso
     :alt: Coverage Status
 
+.. image:: https://raw.githubusercontent.com/davidhalter/parso/master/docs/_static/logo_characters.png
 
-Parso is a Python parser that supports error recovery and round-trip parsing.
+Parso is a Python parser that supports error recovery and round-trip parsing
+for different Python versions (in multiple Python versions). Parso is also able
+to list multiple syntax errors in your python file.
 
 Parso has been battle-tested by jedi_. It was pulled out of jedi to be useful
 for other projects as well.
 
-Parso is very simplistic. It consists of a small API to parse Python and
-analyse the parsing tree.
+Parso consists of a small API to parse Python and analyse the syntax tree.
 
+A simple example:
 
-Ressources
-==========
+.. code-block:: python
 
+    >>> import parso
+    >>> module = parso.parse('hello + 1', version="3.6")
+    >>> expr = module.children[0]
+    >>> expr
+    PythonNode(arith_expr, [<Name: hello@1,0>, <Operator: +>, <Number: 1>])
+    >>> print(expr.get_code())
+    hello + 1
+    >>> name = expr.children[0]
+    >>> name
+    <Name: hello@1,0>
+    >>> name.end_pos
+    (1, 5)
+    >>> expr.end_pos
+    (1, 9)
+
+To list multiple issues:
+
+.. code-block:: python
+
+    >>> grammar = parso.load_grammar()
+    >>> module = grammar.parse('foo +\nbar\ncontinue')
+    >>> error1, error2 = grammar.iter_errors(module)
+    >>> error1.message
+    'SyntaxError: invalid syntax'
+    >>> error2.message
+    "SyntaxError: 'continue' not properly in loop"
+
+Resources
+=========
+
+- `Testing <http://parso.readthedocs.io/en/latest/docs/development.html#testing>`_
 - `PyPI <https://pypi.python.org/pypi/parso>`_
 - `Docs <https://parso.readthedocs.org/en/latest/>`_
 - Uses `semantic versioning <http://semver.org/>`_
@@ -42,32 +75,16 @@ Known Issues
 ============
 
 - `async`/`await` are already used as keywords in Python3.6.
-- `from __future__ import print_function` is not supported,
+- `from __future__ import print_function` is not ignored.
 
-Testing
-=======
-
-The test suite depends on ``tox`` and ``pytest``::
-
-    pip install tox pytest
-
-To run the tests for all supported Python versions::
-
-    tox
-
-If you want to test only a specific Python version (e.g. Python 2.7), it's as
-easy as ::
-
-    tox -e py27
-
-Tests are also run automatically on `Travis CI
-<https://travis-ci.org/davidhalter/parso/>`_.
 
 Acknowledgements
 ================
 
 - Guido van Rossum (@gvanrossum) for creating the parser generator pgen2
   (originally used in lib2to3).
+- `Salome Schneider <https://www.crepes-schnaegg.ch/cr%C3%AApes-schn%C3%A4gg/kunst-f%C3%BCrs-cr%C3%AApes-mobil/>`_
+  for the extremely awesome parso logo.
 
 
 .. _jedi: https://github.com/davidhalter/jedi

conftest.py

@@ -11,9 +11,12 @@ import parso
 from parso import cache
 from parso.utils import parse_version_string
 
-
 collect_ignore = ["setup.py"]
 
+VERSIONS_2 = '2.6', '2.7'
+VERSIONS_3 = '3.3', '3.4', '3.5', '3.6', '3.7'
+
+
 @pytest.fixture(scope='session')
 def clean_parso_cache():
     """
@@ -49,20 +52,11 @@ def pytest_generate_tests(metafunc):
             ids=[c.name for c in cases]
         )
     elif 'each_version' in metafunc.fixturenames:
-        metafunc.parametrize(
-            'each_version',
-            ['2.6', '2.7', '3.3', '3.4', '3.5', '3.6'],
-        )
+        metafunc.parametrize('each_version', VERSIONS_2 + VERSIONS_3)
     elif 'each_py2_version' in metafunc.fixturenames:
-        metafunc.parametrize(
-            'each_py2_version',
-            ['2.6', '2.7'],
-        )
+        metafunc.parametrize('each_py2_version', VERSIONS_2)
     elif 'each_py3_version' in metafunc.fixturenames:
-        metafunc.parametrize(
-            'each_py3_version',
-            ['3.3', '3.4', '3.5', '3.6'],
-        )
+        metafunc.parametrize('each_py3_version', VERSIONS_3)
 
 
 class NormalizerIssueCase(object):
@@ -97,16 +91,6 @@ def pytest_configure(config):
         root.addHandler(ch)
 
 
-@pytest.fixture
-def each_py3_version():
-    return '3.3', '3.4', '3.5', '3.6'
-
-
-@pytest.fixture
-def each_py2_version():
-    return '2.6', '2.7'
-
-
 class Checker():
     def __init__(self, version, is_passing):
         self.version = version

deploy-master.sh

@@ -36,7 +36,7 @@ if [[ $tag_ref ]]; then
         exit 1
     fi
 else
-    git tag $tag
+    git tag -a $tag
     git push --tags
 fi
 
@@ -48,5 +48,5 @@ python setup.py sdist bdist_wheel
 twine upload dist/*
 
 cd $BASE_DIR
-# Back in the development directory fetch tags.
-git fetch --tags
+# The tags have been pushed to this repo. Push the tags to github, now.
+git push --tags

docs/_themes/flask/layout.html

@@ -6,8 +6,8 @@
   {% endif %}
   <link media="only screen and (max-device-width: 480px)" href="{{
     pathto('_static/small_flask.css', 1) }}" type= "text/css" rel="stylesheet" />
-  <a href="https://github.com/davidhalter/jedi">
-    <img style="position: absolute; top: 0; right: 0; border: 0;" src="https://s3.amazonaws.com/github/ribbons/forkme_right_red_aa0000.png" alt="Fork me on GitHub">
+  <a href="https://github.com/davidhalter/parso">
+    <img style="position: absolute; top: 0; right: 0; border: 0;" src="https://s3.amazonaws.com/github/ribbons/forkme_right_red_aa0000.png" alt="Fork me">
   </a>
 {% endblock %}
 {%- block relbar2 %}{% endblock %}
@@ -19,7 +19,6 @@
 {% endblock %}
 {%- block footer %}
   <div class="footer">
-    &copy; Copyright {{ copyright }}.
     Created using <a href="http://sphinx.pocoo.org/">Sphinx</a>.
   </div>
   {% if pagename == 'index' %}

docs/conf.py

@@ -13,7 +13,6 @@
 
 import sys
 import os
-import datetime
 
 # If extensions (or modules to document with autodoc) are in another directory,
 # add these directories to sys.path here. If the directory is relative to the
@@ -45,7 +44,7 @@ master_doc = 'index'
 
 # General information about the project.
 project = u'parso'
-copyright = u'2012 - {today.year}, parso contributors'.format(today=datetime.date.today())
+copyright = u'parso contributors'
 
 import parso
 from parso.utils import version_info
@@ -274,7 +273,7 @@ autodoc_default_flags = []
 # -- Options for intersphinx module --------------------------------------------
 
 intersphinx_mapping = {
-    'http://docs.python.org/': None,
+    'http://docs.python.org/': ('https://docs.python.org/3.6', None),
 }
 
 

docs/docs/development.rst

@@ -4,7 +4,8 @@ Development
 ===========
 
 If you want to contribute anything to |parso|, just open an issue or pull
-request to discuss it. We welcome changes!
+request to discuss it. We welcome changes! Please check the ``CONTRIBUTING.md``
+file in the repository, first.
 
 
 Deprecations Process

docs/docs/installation.rst

@@ -3,8 +3,8 @@
 Installation and Configuration
 ==============================
 
-The preferred way
------------------
+The preferred way (pip)
+-----------------------
 
 On any system you can install |parso| directly from the Python package index
 using pip::

docs/docs/parser-tree.rst

@@ -1,29 +1,42 @@
+.. include:: ../global.rst
+
 .. _parser-tree:
 
 Parser Tree
 ===========
 
-Usage
------
+The parser tree is returned by calling :py:meth:`parso.Grammar.parse`.
 
-.. automodule:: parso.python
+.. note:: Note that parso positions are always 1 based for lines and zero
+   based for columns. This means the first position in a file is (1, 0).
+
+Parser Tree Base Classes
+------------------------
+
+Generally there are two types of classes you will deal with:
+:py:class:`parso.tree.Leaf` and :py:class:`parso.tree.BaseNode`.
+
+.. autoclass:: parso.tree.BaseNode
+    :show-inheritance:
     :members:
-    :undoc-members:
 
-
-Parser Tree Base Class
-----------------------
+.. autoclass:: parso.tree.Leaf
+    :show-inheritance:
+    :members:
 
 All nodes and leaves have these methods/properties:
 
 .. autoclass:: parso.tree.NodeOrLeaf
     :members:
     :undoc-members:
+    :show-inheritance:
 
 
 Python Parser Tree
 ------------------
 
+.. currentmodule:: parso.python.tree
+
 .. automodule:: parso.python.tree
     :members:
     :undoc-members:

docs/docs/usage.rst

@@ -4,12 +4,13 @@ Usage
 =====
 
 |parso| works around grammars. You can simply create Python grammars by calling
-``load_grammar``. Grammars (with a custom tokenizer and custom parser trees)
-can also be created by directly instantiating ``Grammar``. More information
+:py:func:`parso.load_grammar`. Grammars (with a custom tokenizer and custom parser trees)
+can also be created by directly instantiating :py:func:`parso.Grammar`. More information
 about the resulting objects can be found in the :ref:`parser tree documentation
 <parser-tree>`.
 
-The simplest way of using parso is without even loading a grammar:
+The simplest way of using parso is without even loading a grammar
+(:py:func:`parso.parse`):
 
 .. sourcecode:: python
 
@@ -17,7 +18,31 @@ The simplest way of using parso is without even loading a grammar:
    >>> parso.parse('foo + bar')
    <Module: @1-1>
 
-.. automodule:: parso.grammar
+Loading a Grammar
+-----------------
+
+Typically if you want to work with one specific Python version, use:
+
+.. autofunction:: parso.load_grammar
+
+Grammar methods
+---------------
+
+You will get back a grammar object that you can use to parse code and find
+issues in it:
+
+.. autoclass:: parso.Grammar
+    :members:
+    :undoc-members:
+
+
+Error Retrieval
+---------------
+
+|parso| is able to find multiple errors in your source code. Iterating through
+those errors yields the following instances:
+
+.. autoclass:: parso.normalizer.Issue
     :members:
     :undoc-members:
 
@@ -25,17 +50,17 @@ The simplest way of using parso is without even loading a grammar:
 Utility
 -------
 
-.. autofunction:: parso.parse
+|parso| also offers some utility functions that can be really useful:
 
-.. automodule:: parso.utils
-    :members:
-    :undoc-members:
+.. autofunction:: parso.parse
+.. autofunction:: parso.split_lines
+.. autofunction:: parso.python_bytes_to_unicode
 
 
 Used By
 -------
 
-- jedi_ (which is used by IPython and a lot of plugins).
+- jedi_ (which is used by IPython and a lot of editor plugins).
 
 
 .. _jedi: https://github.com/davidhalter/jedi

docs/index.rst

@@ -1,7 +1,7 @@
 .. include global.rst
 
-parso - A Python Parser Written in Python
-=========================================
+parso - A Python Parser
+=======================
 
 Release v\ |release|. (:doc:`Installation <docs/installation>`)
 

parso/__init__.py

@@ -1,10 +1,19 @@
-"""
-parso is a Python parser. It's really easy to use and supports multiple Python
-versions, file caching, round-trips and other stuff:
+r"""
+Parso is a Python parser that supports error recovery and round-trip parsing
+for different Python versions (in multiple Python versions). Parso is also able
+to list multiple syntax errors in your python file.
 
->>> from parso import load_grammar
->>> grammar = load_grammar(version='2.7')
->>> module = grammar.parse('hello + 1')
+Parso has been battle-tested by jedi_. It was pulled out of jedi to be useful
+for other projects as well.
+
+Parso consists of a small API to parse Python and analyse the syntax tree.
+
+.. _jedi: https://github.com/davidhalter/jedi
+
+A simple example:
+
+>>> import parso
+>>> module = parso.parse('hello + 1', version="3.6")
 >>> expr = module.children[0]
 >>> expr
 PythonNode(arith_expr, [<Name: hello@1,0>, <Operator: +>, <Number: 1>])
@@ -17,19 +26,32 @@ hello + 1
 (1, 5)
 >>> expr.end_pos
 (1, 9)
+
+To list multiple issues:
+
+>>> grammar = parso.load_grammar()
+>>> module = grammar.parse('foo +\nbar\ncontinue')
+>>> error1, error2 = grammar.iter_errors(module)
+>>> error1.message
+'SyntaxError: invalid syntax'
+>>> error2.message
+"SyntaxError: 'continue' not properly in loop"
 """
 
 from parso.parser import ParserSyntaxError
 from parso.grammar import Grammar, load_grammar
+from parso.utils import split_lines, python_bytes_to_unicode
 
 
-__version__ = '0.0.4'
+__version__ = '0.2.0'
 
 
 def parse(code=None, **kwargs):
     """
-    A utility function to parse Python with the current Python version. Params
-    are documented in ``Grammar.parse``.
+    A utility function to avoid loading grammars.
+    Params are documented in :py:meth:`parso.Grammar.parse`.
+
+    :param str version: The version used by :py:func:`parso.load_grammar`.
     """
     version = kwargs.pop('version', None)
     grammar = load_grammar(version=version)

parso/_compatibility.py

@@ -36,7 +36,7 @@ except AttributeError:
 def u(string):
     """Cast to unicode DAMMIT!
     Written because Python2 repr always implicitly casts to a string, so we
-    have to cast back to a unicode (and we now that we always deal with valid
+    have to cast back to a unicode (and we know that we always deal with valid
     unicode, because we check that in the beginning).
     """
     if py_version >= 30:

parso/cache.py

@@ -4,13 +4,19 @@ import sys
 import hashlib
 import gc
 import shutil
-import pickle
 import platform
 import errno
 import logging
 
+try:
+    import cPickle as pickle
+except:
+    import pickle
+
 from parso._compatibility import FileNotFoundError
 
+LOG = logging.getLogger(__name__)
+
 
 _PICKLE_VERSION = 30
 """
@@ -111,7 +117,7 @@ def _load_from_file_system(hashed_grammar, path, p_time, cache_path=None):
         return None
     else:
         parser_cache.setdefault(hashed_grammar, {})[path] = module_cache_item
-        logging.debug('pickle loaded: %s', path)
+        LOG.debug('pickle loaded: %s', path)
         return module_cache_item.node
 
 
@@ -125,7 +131,7 @@ def save_module(hashed_grammar, path, module, lines, pickling=True, cache_path=N
     item = _NodeCacheItem(module, lines, p_time)
     parser_cache.setdefault(hashed_grammar, {})[path] = item
     if pickling and path is not None:
-        _save_to_file_system(hashed_grammar, path, item)
+        _save_to_file_system(hashed_grammar, path, item, cache_path=cache_path)
 
 
 def _save_to_file_system(hashed_grammar, path, item, cache_path=None):

parso/grammar.py

@@ -12,17 +12,17 @@ from parso.parser import BaseParser
 from parso.python.parser import Parser as PythonParser
 from parso.python.errors import ErrorFinderConfig
 from parso.python import pep8
-from parso.python import fstring
 
 _loaded_grammars = {}
 
 
 class Grammar(object):
     """
-    Create custom grammars by calling this. It's not really supported, yet.
+    :py:func:`parso.load_grammar` returns instances of this class.
 
-    :param text: A BNF representation of your grammar.
+    Creating custom grammars by calling this is not supported, yet.
     """
+    #:param text: A BNF representation of your grammar.
     _error_normalizer_config = None
     _token_namespace = None
     _default_normalizer_config = pep8.PEP8NormalizerConfig()
@@ -44,33 +44,38 @@ class Grammar(object):
         If you need finer grained control over the parsed instance, there will be
         other ways to access it.
 
-        :param code str: A unicode string that contains Python code.
-        :param path str: The path to the file you want to open. Only needed for caching.
-        :param error_recovery bool: If enabled, any code will be returned. If
+        :param str code: A unicode or bytes string. When it's not possible to
+            decode bytes to a string, returns a
+            :py:class:`UnicodeDecodeError`.
+        :param bool error_recovery: If enabled, any code will be returned. If
             it is invalid, it will be returned as an error node. If disabled,
             you will get a ParseError when encountering syntax errors in your
             code.
-        :param start_symbol str: The grammar symbol that you want to parse. Only
+        :param str start_symbol: The grammar symbol that you want to parse. Only
             allowed to be used when error_recovery is False.
-        :param cache bool: Keeps a copy of the parser tree in RAM and on disk
+        :param str path: The path to the file you want to open. Only needed for caching.
+        :param bool cache: Keeps a copy of the parser tree in RAM and on disk
             if a path is given. Returns the cached trees if the corresponding
             files on disk have not changed.
-        :param diff_cache bool: Diffs the cached python module against the new
+        :param bool diff_cache: Diffs the cached python module against the new
             code and tries to parse only the parts that have changed. Returns
             the same (changed) module that is found in cache. Using this option
-            requires you to not do anything anymore with the old cached module,
-            because the contents of it might have changed.
-        :param cache_path bool: If given saves the parso cache in this
+            requires you to not do anything anymore with the cached modules
+            under that path, because the contents of it might change. This
+            option is still somewhat experimental. If you want stability,
+            please don't use it.
+        :param bool cache_path: If given saves the parso cache in this
             directory. If not given, defaults to the default cache places on
             each platform.
 
-        :return: A syntax tree node. Typically the module.
+        :return: A subclass of :py:class:`parso.tree.NodeOrLeaf`. Typically a
+            :py:class:`parso.python.tree.Module`.
         """
         if 'start_pos' in kwargs:
-            raise TypeError("parse() got an unexpected keyworda argument.")
+            raise TypeError("parse() got an unexpected keyword argument.")
         return self._parse(code=code, **kwargs)
 
-    def _parse(self, code=None, path=None, error_recovery=True,
+    def _parse(self, code=None, error_recovery=True, path=None,
                start_symbol=None, cache=False, diff_cache=False,
                cache_path=None, start_pos=(1, 0)):
         """
@@ -88,10 +93,7 @@ class Grammar(object):
         if error_recovery and start_symbol != 'file_input':
             raise NotImplementedError("This is currently not implemented.")
 
-        if cache and code is None and path is not None:
-            # With the current architecture we cannot load from cache if the
-            # code is given, because we just load from cache if it's not older than
-            # the latest change (file last modified).
+        if cache and path is not None:
             module_node = load_module(self._hashed, path, cache_path=cache_path)
             if module_node is not None:
                 return module_node
@@ -152,6 +154,11 @@ class Grammar(object):
         return ns
 
     def iter_errors(self, node):
+        """
+        Given a :py:class:`parso.tree.NodeOrLeaf` returns a generator of
+        :py:class:`parso.normalizer.Issue` objects. For Python this is
+        a list of syntax/indentation errors.
+        """
         if self._error_normalizer_config is None:
             raise ValueError("No error normalizer specified for this grammar.")
 
@@ -178,7 +185,6 @@ class Grammar(object):
         normalizer.walk(node)
         return normalizer.issues
 
-
     def __repr__(self):
         labels = self._pgen_grammar.number2symbol.values()
         txt = ' '.join(list(labels)[:3]) + ' ...'
@@ -207,46 +213,21 @@ class PythonGrammar(Grammar):
         return tokenize(code, self.version_info)
 
 
-class PythonFStringGrammar(Grammar):
-    _token_namespace = fstring.TokenNamespace
-    _start_symbol = 'fstring'
-
-    def __init__(self):
-        super(PythonFStringGrammar, self).__init__(
-            text=fstring.GRAMMAR,
-            tokenizer=fstring.tokenize,
-            parser=fstring.Parser
-        )
-
-    def parse(self, code, **kwargs):
-        return self._parse(code, **kwargs)
-
-    def _parse(self, code, error_recovery=True, start_pos=(1, 0)):
-        tokens = self._tokenizer(code, start_pos=start_pos)
-        p = self._parser(
-            self._pgen_grammar,
-            error_recovery=error_recovery,
-            start_symbol=self._start_symbol,
-        )
-        return p.parse(tokens=tokens)
-
-    def parse_leaf(self, leaf, error_recovery=True):
-        code = leaf._get_payload()
-        return self.parse(code, error_recovery=True, start_pos=leaf.start_pos)
-
-
 def load_grammar(**kwargs):
     """
-    Loads a Python grammar. The default version is the current Python version.
+    Loads a :py:class:`parso.Grammar`. The default version is the current Python
+    version.
 
-    If you need support for a specific version, please use e.g.
-    `version='3.3'`.
+    :param str version: A python version string, e.g. ``version='3.3'``.
     """
     def load_grammar(language='python', version=None):
         if language == 'python':
             version_info = parse_version_string(version)
 
-            file = 'python/grammar%s%s.txt' % (version_info.major, version_info.minor)
+            file = os.path.join(
+                'python',
+                'grammar%s%s.txt' % (version_info.major, version_info.minor)
+            )
 
             global _loaded_grammars
             path = os.path.join(os.path.dirname(__file__), file)
@@ -262,10 +243,6 @@ def load_grammar(**kwargs):
                 except FileNotFoundError:
                     message = "Python version %s is currently not supported." % version
                     raise NotImplementedError(message)
-        elif language == 'python-f-string':
-            if version is not None:
-                raise NotImplementedError("Currently different versions are not supported.")
-            return PythonFStringGrammar()
         else:
             raise NotImplementedError("No support for language %s." % language)
 

parso/normalizer.py

@@ -121,8 +121,18 @@ class Issue(object):
     def __init__(self, node, code, message):
         self._node = node
         self.code = code
+        """
+        An integer code that stands for the type of error.
+        """
         self.message = message
+        """
+        A message (string) for the issue.
+        """
         self.start_pos = node.start_pos
+        """
+        The start position position of the error as a tuple (line, column). As
+        always in |parso| the first line is 1 and the first column 0.
+        """
 
     def __eq__(self, other):
         return self.start_pos == other.start_pos and self.code == other.code

parso/pgen2/grammar.py

@@ -16,7 +16,10 @@ fallback token code OP, but the parser needs the actual token code.
 
 """
 
-import pickle
+try:
+    import cPickle as pickle
+except:
+    import pickle
 
 
 class Grammar(object):

parso/pgen2/pgen.py

@@ -28,6 +28,7 @@ class ParserGenerator(object):
         c = grammar.Grammar(self._bnf_text)
         names = list(self.dfas.keys())
         names.sort()
+        # TODO do we still need this?
         names.remove(self.startsymbol)
         names.insert(0, self.startsymbol)
         for name in names:
@@ -316,8 +317,8 @@ class ParserGenerator(object):
 
     def _expect(self, type):
         if self.type != type:
-            self._raise_error("expected %s, got %s(%s)",
-                              type, self.type, self.value)
+            self._raise_error("expected %s(%s), got %s(%s)",
+                              type, token.tok_name[type], self.type, self.value)
         value = self.value
         self._gettoken()
         return value

parso/python/diff.py

@@ -16,6 +16,8 @@ from parso.python.tree import EndMarker
 from parso.python.tokenize import (NEWLINE, PythonToken, ERROR_DEDENT,
                                    ENDMARKER, INDENT, DEDENT)
 
+LOG = logging.getLogger(__name__)
+
 
 def _get_last_line(node_or_leaf):
     last_leaf = node_or_leaf.get_last_leaf()
@@ -116,7 +118,7 @@ class DiffParser(object):
 
         Returns the new module node.
         '''
-        logging.debug('diff parser start')
+        LOG.debug('diff parser start')
         # Reset the used names cache so they get regenerated.
         self._module._used_names = None
 
@@ -127,11 +129,11 @@ class DiffParser(object):
         line_length = len(new_lines)
         sm = difflib.SequenceMatcher(None, old_lines, self._parser_lines_new)
         opcodes = sm.get_opcodes()
-        logging.debug('diff parser calculated')
-        logging.debug('diff: line_lengths old: %s, new: %s' % (len(old_lines), line_length))
+        LOG.debug('diff parser calculated')
+        LOG.debug('diff: line_lengths old: %s, new: %s' % (len(old_lines), line_length))
 
         for operation, i1, i2, j1, j2 in opcodes:
-            logging.debug('diff %s old[%s:%s] new[%s:%s]',
+            LOG.debug('diff code[%s] old[%s:%s] new[%s:%s]',
                       operation, i1 + 1, i2, j1 + 1, j2)
 
             if j2 == line_length and new_lines[-1] == '':
@@ -161,12 +163,12 @@ class DiffParser(object):
                 % (last_pos, line_length, ''.join(diff))
             )
 
-        logging.debug('diff parser end')
+        LOG.debug('diff parser end')
         return self._module
 
     def _enabled_debugging(self, old_lines, lines_new):
         if self._module.get_code() != ''.join(lines_new):
-            logging.warning('parser issue:\n%s\n%s', ''.join(old_lines),
+            LOG.warning('parser issue:\n%s\n%s', ''.join(old_lines),
                             ''.join(lines_new))
 
     def _copy_from_old_parser(self, line_offset, until_line_old, until_line_new):
@@ -203,7 +205,7 @@ class DiffParser(object):
                     from_ = copied_nodes[0].get_start_pos_of_prefix()[0] + line_offset
                     to = self._nodes_stack.parsed_until_line
 
-                    logging.debug('diff actually copy %s to %s', from_, to)
+                    LOG.debug('diff actually copy %s to %s', from_, to)
             # Since there are potential bugs that might loop here endlessly, we
             # just stop here.
             assert last_until_line != self._nodes_stack.parsed_until_line \
@@ -248,7 +250,7 @@ class DiffParser(object):
             nodes = node.children
 
             self._nodes_stack.add_parsed_nodes(nodes)
-            logging.debug(
+            LOG.debug(
                 'parse_part from %s to %s (to %s in part parser)',
                 nodes[0].get_start_pos_of_prefix()[0],
                 self._nodes_stack.parsed_until_line,
@@ -452,7 +454,7 @@ class _NodesStack(object):
         self._last_prefix = ''
         if is_endmarker:
             try:
-                separation = last_leaf.prefix.rindex('\n')
+                separation = last_leaf.prefix.rindex('\n') + 1
             except ValueError:
                 pass
             else:
@@ -460,7 +462,7 @@ class _NodesStack(object):
                 # That is not relevant if parentheses were opened. Always parse
                 # until the end of a line.
                 last_leaf.prefix, self._last_prefix = \
-                    last_leaf.prefix[:separation + 1], last_leaf.prefix[separation + 1:]
+                    last_leaf.prefix[:separation], last_leaf.prefix[separation:]
 
         first_leaf = tree_nodes[0].get_first_leaf()
         first_leaf.prefix = self.prefix + first_leaf.prefix
@@ -470,7 +472,6 @@ class _NodesStack(object):
             self.prefix = last_leaf.prefix
 
             tree_nodes = tree_nodes[:-1]
-
         return tree_nodes
 
     def copy_nodes(self, tree_nodes, until_line, line_offset):
@@ -490,6 +491,13 @@ class _NodesStack(object):
         new_tos = tos
         for node in nodes:
             if node.type == 'endmarker':
+                # We basically removed the endmarker, but we are not allowed to
+                # remove the newline at the end of the line, otherwise it's
+                # going to be missing.
+                try:
+                    self.prefix = node.prefix[:node.prefix.rindex('\n') + 1]
+                except ValueError:
+                    pass
                 # Endmarkers just distort all the checks below. Remove them.
                 break
 

parso/python/errors.py

@@ -563,7 +563,8 @@ class _ReturnAndYieldChecks(SyntaxRule):
                     and self._normalizer.version == (3, 5):
                 self.add_issue(self.get_node(leaf), message=self.message_async_yield)
 
-@ErrorFinder.register_rule(type='atom')
+
+@ErrorFinder.register_rule(type='strings')
 class _BytesAndStringMix(SyntaxRule):
     # e.g. 's' b''
     message = "cannot mix bytes and nonbytes literals"
@@ -744,7 +745,12 @@ class _NonlocalModuleLevelRule(SyntaxRule):
 
 @ErrorFinder.register_rule(type='arglist')
 class _ArglistRule(SyntaxRule):
-    message = "Generator expression must be parenthesized if not sole argument"
+    @property
+    def message(self):
+        if self._normalizer.version < (3, 7):
+            return "Generator expression must be parenthesized if not sole argument"
+        else:
+            return "Generator expression must be parenthesized"
 
     def is_issue(self, node):
         first_arg = node.children[0]
@@ -837,101 +843,36 @@ class _TryStmtRule(SyntaxRule):
                 self.add_issue(default_except, message=self.message)
 
 
-@ErrorFinder.register_rule(type='string')
+@ErrorFinder.register_rule(type='fstring')
 class _FStringRule(SyntaxRule):
     _fstring_grammar = None
-    message_empty = "f-string: empty expression not allowed"  # f'{}'
-    message_single_closing = "f-string: single '}' is not allowed"  # f'}'
     message_nested = "f-string: expressions nested too deeply"
-    message_backslash = "f-string expression part cannot include a backslash"  # f'{"\"}' or f'{"\\"}'
-    message_comment = "f-string expression part cannot include '#'"  # f'{#}'
-    message_unterminated_string = "f-string: unterminated string"  # f'{"}'
     message_conversion = "f-string: invalid conversion character: expected 's', 'r', or 'a'"
-    message_incomplete = "f-string: expecting '}'"  # f'{'
-    message_syntax = "invalid syntax"
 
-    @classmethod
-    def _load_grammar(cls):
-        import parso
+    def _check_format_spec(self, format_spec, depth):
+        self._check_fstring_contents(format_spec.children[1:], depth)
 
-        if cls._fstring_grammar is None:
-            cls._fstring_grammar = parso.load_grammar(language='python-f-string')
-        return cls._fstring_grammar
+    def _check_fstring_expr(self, fstring_expr, depth):
+        if depth >= 2:
+            self.add_issue(fstring_expr, message=self.message_nested)
+
+        conversion = fstring_expr.children[2]
+        if conversion.type == 'fstring_conversion':
+            name = conversion.children[1]
+            if name.value not in ('s', 'r', 'a'):
+                self.add_issue(name, message=self.message_conversion)
+
+        format_spec = fstring_expr.children[-2]
+        if format_spec.type == 'fstring_format_spec':
+            self._check_format_spec(format_spec, depth + 1)
 
     def is_issue(self, fstring):
-        if 'f' not in fstring.string_prefix.lower():
-            return
+        self._check_fstring_contents(fstring.children[1:-1])
 
-        parsed = self._load_grammar().parse_leaf(fstring)
-        for child in parsed.children:
-            if child.type == 'expression':
-                self._check_expression(child)
-            elif child.type == 'error_node':
-                next_ = child.get_next_leaf()
-                if next_.type == 'error_leaf' and next_.original_type == 'unterminated_string':
-                    self.add_issue(next_, message=self.message_unterminated_string)
-                    # At this point nothing more is comming except the error
-                    # leaf that we've already checked here.
-                    break
-                self.add_issue(child, message=self.message_incomplete)
-            elif child.type == 'error_leaf':
-                self.add_issue(child, message=self.message_single_closing)
-
-    def _check_python_expr(self, python_expr):
-        value = python_expr.value
-        if '\\' in value:
-            self.add_issue(python_expr, message=self.message_backslash)
-            return
-        if '#' in value:
-            self.add_issue(python_expr, message=self.message_comment)
-            return
-        if re.match('\s*$', value) is not None:
-            self.add_issue(python_expr, message=self.message_empty)
-            return
-
-        # This is now nested parsing. We parsed the fstring and now
-        # we're parsing Python again.
-        try:
-            # CPython has a bit of a special ways to parse Python code within
-            # f-strings. It wraps the code in brackets to make sure that
-            # whitespace doesn't make problems (indentation/newlines).
-            # Just use that algorithm as well here and adapt start positions.
-            start_pos = python_expr.start_pos
-            start_pos = start_pos[0], start_pos[1] - 1
-            eval_input = self._normalizer.grammar._parse(
-                '(%s)' % value,
-                start_symbol='eval_input',
-                start_pos=start_pos,
-                error_recovery=False
-            )
-        except ParserSyntaxError as e:
-            self.add_issue(e.error_leaf, message=self.message_syntax)
-            return
-
-        issues = self._normalizer.grammar.iter_errors(eval_input)
-        self._normalizer.issues += issues
-
-    def _check_format_spec(self, format_spec):
-        for expression in format_spec.children[1:]:
-            nested_format_spec = expression.children[-2]
-            if nested_format_spec.type == 'format_spec':
-                if len(nested_format_spec.children) > 1:
-                    self.add_issue(
-                        nested_format_spec.children[1],
-                        message=self.message_nested
-                    )
-
-            self._check_expression(expression)
-
-    def _check_expression(self, expression):
-        for c in expression.children:
-            if c.type == 'python_expr':
-                self._check_python_expr(c)
-            elif c.type == 'conversion':
-                if c.value not in ('s', 'r', 'a'):
-                    self.add_issue(c, message=self.message_conversion)
-            elif c.type == 'format_spec':
-                self._check_format_spec(c)
+    def _check_fstring_contents(self, children, depth=0):
+        for fstring_content in children:
+            if fstring_content.type == 'fstring_expr':
+                self._check_fstring_expr(fstring_content, depth)
 
 
 class _CheckAssignmentRule(SyntaxRule):
@@ -944,7 +885,7 @@ class _CheckAssignmentRule(SyntaxRule):
             first, second = node.children[:2]
             error = _get_comprehension_type(node)
             if error is None:
-                if second.type in ('dictorsetmaker', 'string'):
+                if second.type == 'dictorsetmaker':
                     error = 'literal'
                 elif first in ('(', '['):
                     if second.type == 'yield_expr':
@@ -963,7 +904,7 @@ class _CheckAssignmentRule(SyntaxRule):
                 error = 'Ellipsis'
         elif type_ == 'comparison':
             error = 'comparison'
-        elif type_ in ('string', 'number'):
+        elif type_ in ('string', 'number', 'strings'):
             error = 'literal'
         elif type_ == 'yield_expr':
             # This one seems to be a slightly different warning in Python.

parso/python/fstring.py

@@ -1,211 +0,0 @@
-import re
-
-from itertools import count
-from parso.utils import PythonVersionInfo
-from parso.utils import split_lines
-from parso.python.tokenize import Token
-from parso import parser
-from parso.tree import TypedLeaf, ErrorNode, ErrorLeaf
-
-version36 = PythonVersionInfo(3, 6)
-
-
-class TokenNamespace:
-    _c = count()
-    LBRACE = next(_c)
-    RBRACE = next(_c)
-    ENDMARKER = next(_c)
-    COLON = next(_c)
-    CONVERSION = next(_c)
-    PYTHON_EXPR = next(_c)
-    EXCLAMATION_MARK = next(_c)
-    UNTERMINATED_STRING = next(_c)
-
-    token_map = dict((v, k) for k, v in locals().items() if not k.startswith('_'))
-
-    @classmethod
-    def generate_token_id(cls, string):
-        if string == '{':
-            return cls.LBRACE
-        elif string == '}':
-            return cls.RBRACE
-        elif string == '!':
-            return cls.EXCLAMATION_MARK
-        elif string == ':':
-            return cls.COLON
-        return getattr(cls, string)
-
-
-GRAMMAR = """
-fstring: expression* ENDMARKER
-format_spec: ':' expression*
-expression: '{' PYTHON_EXPR [ '!' CONVERSION ] [ format_spec ] '}'
-"""
-
-_prefix = r'((?:[^{}]+)*)'
-_expr = _prefix + r'(\{|\}|$)'
-_in_expr = r'([^{}\[\]:"\'!]*)(.?)'
-# There's only one conversion character allowed. But the rules have to be
-# checked later anyway, so allow more here. This makes error recovery nicer.
-_conversion = r'([^={}:]*)(.?)'
-
-_compiled_expr = re.compile(_expr)
-_compiled_in_expr = re.compile(_in_expr)
-_compiled_conversion = re.compile(_conversion)
-
-
-def tokenize(code, start_pos=(1, 0)):
-    def add_to_pos(string):
-        lines = split_lines(string)
-        l = len(lines[-1])
-        if len(lines) > 1:
-            start_pos[0] += len(lines) - 1
-            start_pos[1] = l
-        else:
-            start_pos[1] += l
-
-    def tok(value, type=None, prefix=''):
-        if type is None:
-            type = TokenNamespace.generate_token_id(value)
-
-        add_to_pos(prefix)
-        token = Token(type, value, tuple(start_pos), prefix)
-        add_to_pos(value)
-        return token
-
-    start = 0
-    recursion_level = 0
-    added_prefix = ''
-    start_pos = list(start_pos)
-    while True:
-        match = _compiled_expr.match(code, start)
-        prefix = added_prefix + match.group(1)
-        found = match.group(2)
-        start = match.end()
-        if not found:
-            # We're at the end.
-            break
-
-        if found == '}':
-            if recursion_level == 0 and len(code) > start  and code[start] == '}':
-                # This is a }} escape.
-                added_prefix = prefix + '}}'
-                start += 1
-                continue
-
-            recursion_level = max(0, recursion_level - 1)
-            yield tok(found, prefix=prefix)
-            added_prefix = ''
-        else:
-            assert found == '{'
-            if recursion_level == 0 and len(code) > start and code[start] == '{':
-                # This is a {{ escape.
-                added_prefix = prefix + '{{'
-                start += 1
-                continue
-
-            recursion_level += 1
-            yield tok(found, prefix=prefix)
-            added_prefix = ''
-
-            expression = ''
-            squared_count = 0
-            curly_count = 0
-            while True:
-                expr_match = _compiled_in_expr.match(code, start)
-                expression += expr_match.group(1)
-                found = expr_match.group(2)
-                start = expr_match.end()
-
-                if found == '{':
-                    curly_count += 1
-                    expression += found
-                elif found == '}' and curly_count > 0:
-                    curly_count -= 1
-                    expression += found
-                elif found == '[':
-                    squared_count += 1
-                    expression += found
-                elif found == ']':
-                    # Use a max function here, because the Python code might
-                    # just have syntax errors.
-                    squared_count = max(0, squared_count - 1)
-                    expression += found
-                elif found == ':' and (squared_count or curly_count):
-                    expression += found
-                elif found in ('"', "'"):
-                    search = found
-                    if len(code) > start + 1 and  \
-                            code[start] == found == code[start+1]:
-                        search *= 3
-                        start += 2
-
-                    index = code.find(search, start)
-                    if index == -1:
-                        yield tok(expression, type=TokenNamespace.PYTHON_EXPR)
-                        yield tok(
-                            found + code[start:],
-                            type=TokenNamespace.UNTERMINATED_STRING,
-                        )
-                        start = len(code)
-                        break
-                    expression += found + code[start:index+1]
-                    start = index + 1
-                elif found == '!' and len(code) > start and code[start] == '=':
-                    # This is a python `!=` and not a conversion.
-                    expression += found
-                else:
-                    yield tok(expression, type=TokenNamespace.PYTHON_EXPR)
-                    if found:
-                        yield tok(found)
-                    break
-
-            if found == '!':
-                conversion_match = _compiled_conversion.match(code, start)
-                found = conversion_match.group(2)
-                start = conversion_match.end()
-                yield tok(conversion_match.group(1), type=TokenNamespace.CONVERSION)
-                if found:
-                    yield tok(found)
-            if found == '}':
-                recursion_level -= 1
-
-            # We don't need to handle everything after ':', because that is
-            # basically new tokens.
-
-    yield tok('', type=TokenNamespace.ENDMARKER, prefix=prefix)
-
-
-class Parser(parser.BaseParser):
-    def parse(self, tokens):
-        node = super(Parser, self).parse(tokens)
-        if isinstance(node, self.default_leaf):  # Is an endmarker.
-            # If there's no curly braces we get back a non-module. We always
-            # want an fstring.
-            node = self.default_node('fstring', [node])
-
-        return node
-
-    def convert_leaf(self, pgen_grammar, type, value, prefix, start_pos):
-        # TODO this is so ugly.
-        leaf_type = TokenNamespace.token_map[type].lower()
-        return TypedLeaf(leaf_type, value, start_pos, prefix)
-
-    def error_recovery(self, pgen_grammar, stack, arcs, typ, value, start_pos, prefix,
-                       add_token_callback):
-        if not self._error_recovery:
-            return super(Parser, self).error_recovery(
-                pgen_grammar, stack, arcs, typ, value, start_pos, prefix,
-                add_token_callback
-            )
-
-        token_type = TokenNamespace.token_map[typ].lower()
-        if len(stack) == 1:
-            error_leaf = ErrorLeaf(token_type, value, start_pos, prefix)
-            stack[0][2][1].append(error_leaf)
-        else:
-            dfa, state, (type_, nodes) = stack[1]
-            stack[0][2][1].append(ErrorNode(nodes))
-            stack[1:] = []
-
-            add_token_callback(typ, value, start_pos, prefix)

parso/python/grammar26.txt

@@ -119,7 +119,8 @@ atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [listmaker] ']' |
        '{' [dictorsetmaker] '}' |
        '`' testlist1 '`' |
-       NAME | NUMBER | STRING+)
+       NAME | NUMBER | strings)
+strings: STRING+
 listmaker: test ( list_for | (',' test)* [','] )
 # Dave: Renamed testlist_gexpr to testlist_comp, because in 2.7+ this is the
 #       default. It's more consistent like this.

parso/python/grammar27.txt

@@ -104,7 +104,8 @@ atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [listmaker] ']' |
        '{' [dictorsetmaker] '}' |
        '`' testlist1 '`' |
-       NAME | NUMBER | STRING+)
+       NAME | NUMBER | strings)
+strings: STRING+
 listmaker: test ( list_for | (',' test)* [','] )
 testlist_comp: test ( comp_for | (',' test)* [','] )
 lambdef: 'lambda' [varargslist] ':' test

parso/python/grammar33.txt

@@ -103,7 +103,8 @@ power: atom trailer* ['**' factor]
 atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [testlist_comp] ']' |
        '{' [dictorsetmaker] '}' |
-       NAME | NUMBER | STRING+ | '...' | 'None' | 'True' | 'False')
+       NAME | NUMBER | strings | '...' | 'None' | 'True' | 'False')
+strings: STRING+
 testlist_comp: (test|star_expr) ( comp_for | (',' (test|star_expr))* [','] )
 trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 subscriptlist: subscript (',' subscript)* [',']

parso/python/grammar34.txt

@@ -103,7 +103,8 @@ power: atom trailer* ['**' factor]
 atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [testlist_comp] ']' |
        '{' [dictorsetmaker] '}' |
-       NAME | NUMBER | STRING+ | '...' | 'None' | 'True' | 'False')
+       NAME | NUMBER | strings | '...' | 'None' | 'True' | 'False')
+strings: STRING+
 testlist_comp: (test|star_expr) ( comp_for | (',' (test|star_expr))* [','] )
 trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 subscriptlist: subscript (',' subscript)* [',']

parso/python/grammar35.txt

@@ -110,7 +110,8 @@ atom_expr: ['await'] atom trailer*
 atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [testlist_comp] ']' |
        '{' [dictorsetmaker] '}' |
-       NAME | NUMBER | STRING+ | '...' | 'None' | 'True' | 'False')
+       NAME | NUMBER | strings | '...' | 'None' | 'True' | 'False')
+strings: STRING+
 testlist_comp: (test|star_expr) ( comp_for | (',' (test|star_expr))* [','] )
 trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 subscriptlist: subscript (',' subscript)* [',']

parso/python/grammar36.txt

@@ -108,7 +108,7 @@ atom_expr: ['await'] atom trailer*
 atom: ('(' [yield_expr|testlist_comp] ')' |
        '[' [testlist_comp] ']' |
        '{' [dictorsetmaker] '}' |
-       NAME | NUMBER | STRING+ | '...' | 'None' | 'True' | 'False')
+       NAME | NUMBER | strings | '...' | 'None' | 'True' | 'False')
 testlist_comp: (test|star_expr) ( comp_for | (',' (test|star_expr))* [','] )
 trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
 subscriptlist: subscript (',' subscript)* [',']
@@ -148,3 +148,10 @@ encoding_decl: NAME
 
 yield_expr: 'yield' [yield_arg]
 yield_arg: 'from' test | testlist
+
+strings: (STRING | fstring)+
+fstring: FSTRING_START fstring_content* FSTRING_END
+fstring_content: FSTRING_STRING | fstring_expr
+fstring_conversion: '!' NAME
+fstring_expr: '{' testlist_comp [ fstring_conversion ] [ fstring_format_spec ] '}'
+fstring_format_spec: ':' fstring_content*

parso/python/grammar37.txt

@@ -0,0 +1,157 @@
+# Grammar for Python
+
+# NOTE WELL: You should also follow all the steps listed at
+# https://docs.python.org/devguide/grammar.html
+
+# Start symbols for the grammar:
+#       single_input is a single interactive statement;
+#       file_input is a module or sequence of commands read from an input file;
+#       eval_input is the input for the eval() functions.
+# NB: compound_stmt in single_input is followed by extra NEWLINE!
+single_input: NEWLINE | simple_stmt | compound_stmt NEWLINE
+file_input: (NEWLINE | stmt)* ENDMARKER
+eval_input: testlist NEWLINE* ENDMARKER
+decorator: '@' dotted_name [ '(' [arglist] ')' ] NEWLINE
+decorators: decorator+
+decorated: decorators (classdef | funcdef | async_funcdef)
+
+# NOTE: Francisco Souza/Reinoud Elhorst, using ASYNC/'await' keywords instead of
+# skipping python3.5+ compatibility, in favour of 3.7 solution
+async_funcdef: 'async' funcdef
+funcdef: 'def' NAME parameters ['->' test] ':' suite
+
+parameters: '(' [typedargslist] ')'
+typedargslist: (tfpdef ['=' test] (',' tfpdef ['=' test])* [',' [
+        '*' [tfpdef] (',' tfpdef ['=' test])* [',' ['**' tfpdef [',']]]
+      | '**' tfpdef [',']]]
+  | '*' [tfpdef] (',' tfpdef ['=' test])* [',' ['**' tfpdef [',']]]
+  | '**' tfpdef [','])
+tfpdef: NAME [':' test]
+varargslist: (vfpdef ['=' test] (',' vfpdef ['=' test])* [',' [
+        '*' [vfpdef] (',' vfpdef ['=' test])* [',' ['**' vfpdef [',']]]
+      | '**' vfpdef [',']]]
+  | '*' [vfpdef] (',' vfpdef ['=' test])* [',' ['**' vfpdef [',']]]
+  | '**' vfpdef [',']
+)
+vfpdef: NAME
+
+stmt: simple_stmt | compound_stmt
+simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
+small_stmt: (expr_stmt | del_stmt | pass_stmt | flow_stmt |
+             import_stmt | global_stmt | nonlocal_stmt | assert_stmt)
+expr_stmt: testlist_star_expr (annassign | augassign (yield_expr|testlist) |
+                     ('=' (yield_expr|testlist_star_expr))*)
+annassign: ':' test ['=' test]
+testlist_star_expr: (test|star_expr) (',' (test|star_expr))* [',']
+augassign: ('+=' | '-=' | '*=' | '@=' | '/=' | '%=' | '&=' | '|=' | '^=' |
+            '<<=' | '>>=' | '**=' | '//=')
+# For normal and annotated assignments, additional restrictions enforced by the interpreter
+del_stmt: 'del' exprlist
+pass_stmt: 'pass'
+flow_stmt: break_stmt | continue_stmt | return_stmt | raise_stmt | yield_stmt
+break_stmt: 'break'
+continue_stmt: 'continue'
+return_stmt: 'return' [testlist]
+yield_stmt: yield_expr
+raise_stmt: 'raise' [test ['from' test]]
+import_stmt: import_name | import_from
+import_name: 'import' dotted_as_names
+# note below: the ('.' | '...') is necessary because '...' is tokenized as ELLIPSIS
+import_from: ('from' (('.' | '...')* dotted_name | ('.' | '...')+)
+              'import' ('*' | '(' import_as_names ')' | import_as_names))
+import_as_name: NAME ['as' NAME]
+dotted_as_name: dotted_name ['as' NAME]
+import_as_names: import_as_name (',' import_as_name)* [',']
+dotted_as_names: dotted_as_name (',' dotted_as_name)*
+dotted_name: NAME ('.' NAME)*
+global_stmt: 'global' NAME (',' NAME)*
+nonlocal_stmt: 'nonlocal' NAME (',' NAME)*
+assert_stmt: 'assert' test [',' test]
+
+compound_stmt: if_stmt | while_stmt | for_stmt | try_stmt | with_stmt | funcdef | classdef | decorated | async_stmt
+async_stmt: 'async' (funcdef | with_stmt | for_stmt)
+if_stmt: 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite]
+while_stmt: 'while' test ':' suite ['else' ':' suite]
+for_stmt: 'for' exprlist 'in' testlist ':' suite ['else' ':' suite]
+try_stmt: ('try' ':' suite
+           ((except_clause ':' suite)+
+            ['else' ':' suite]
+            ['finally' ':' suite] |
+           'finally' ':' suite))
+with_stmt: 'with' with_item (',' with_item)*  ':' suite
+with_item: test ['as' expr]
+# NB compile.c makes sure that the default except clause is last
+except_clause: 'except' [test ['as' NAME]]
+suite: simple_stmt | NEWLINE INDENT stmt+ DEDENT
+
+test: or_test ['if' or_test 'else' test] | lambdef
+test_nocond: or_test | lambdef_nocond
+lambdef: 'lambda' [varargslist] ':' test
+lambdef_nocond: 'lambda' [varargslist] ':' test_nocond
+or_test: and_test ('or' and_test)*
+and_test: not_test ('and' not_test)*
+not_test: 'not' not_test | comparison
+comparison: expr (comp_op expr)*
+# <> isn't actually a valid comparison operator in Python. It's here for the
+# sake of a __future__ import described in PEP 401 (which really works :-)
+comp_op: '<'|'>'|'=='|'>='|'<='|'<>'|'!='|'in'|'not' 'in'|'is'|'is' 'not'
+star_expr: '*' expr
+expr: xor_expr ('|' xor_expr)*
+xor_expr: and_expr ('^' and_expr)*
+and_expr: shift_expr ('&' shift_expr)*
+shift_expr: arith_expr (('<<'|'>>') arith_expr)*
+arith_expr: term (('+'|'-') term)*
+term: factor (('*'|'@'|'/'|'%'|'//') factor)*
+factor: ('+'|'-'|'~') factor | power
+power: atom_expr ['**' factor]
+atom_expr: ['await'] atom trailer*
+atom: ('(' [yield_expr|testlist_comp] ')' |
+       '[' [testlist_comp] ']' |
+       '{' [dictorsetmaker] '}' |
+       NAME | NUMBER | strings | '...' | 'None' | 'True' | 'False')
+testlist_comp: (test|star_expr) ( comp_for | (',' (test|star_expr))* [','] )
+trailer: '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
+subscriptlist: subscript (',' subscript)* [',']
+subscript: test | [test] ':' [test] [sliceop]
+sliceop: ':' [test]
+exprlist: (expr|star_expr) (',' (expr|star_expr))* [',']
+testlist: test (',' test)* [',']
+dictorsetmaker: ( ((test ':' test | '**' expr)
+                   (comp_for | (',' (test ':' test | '**' expr))* [','])) |
+                  ((test | star_expr)
+                   (comp_for | (',' (test | star_expr))* [','])) )
+
+classdef: 'class' NAME ['(' [arglist] ')'] ':' suite
+
+arglist: argument (',' argument)*  [',']
+
+# The reason that keywords are test nodes instead of NAME is that using NAME
+# results in an ambiguity. ast.c makes sure it's a NAME.
+# "test '=' test" is really "keyword '=' test", but we have no such token.
+# These need to be in a single rule to avoid grammar that is ambiguous
+# to our LL(1) parser. Even though 'test' includes '*expr' in star_expr,
+# we explicitly match '*' here, too, to give it proper precedence.
+# Illegal combinations and orderings are blocked in ast.c:
+# multiple (test comp_for) arguments are blocked; keyword unpackings
+# that precede iterable unpackings are blocked; etc.
+argument: ( test [comp_for] |
+            test '=' test |
+            '**' test |
+            '*' test )
+
+comp_iter: comp_for | comp_if
+comp_for: ['async'] 'for' exprlist 'in' or_test [comp_iter]
+comp_if: 'if' test_nocond [comp_iter]
+
+# not used in grammar, but may appear in "node" passed from Parser to Compiler
+encoding_decl: NAME
+
+yield_expr: 'yield' [yield_arg]
+yield_arg: 'from' test | testlist
+
+strings: (STRING | fstring)+
+fstring: FSTRING_START fstring_content* FSTRING_END
+fstring_content: FSTRING_STRING | fstring_expr
+fstring_conversion: '!' NAME
+fstring_expr: '{' testlist [ fstring_conversion ] [ fstring_format_spec ] '}'
+fstring_format_spec: ':' fstring_content*

parso/python/parser.py

@@ -1,6 +1,7 @@
 from parso.python import tree
 from parso.python.token import (DEDENT, INDENT, ENDMARKER, NEWLINE, NUMBER,
-                                STRING, tok_name, NAME)
+                                STRING, tok_name, NAME, FSTRING_STRING,
+                                FSTRING_START, FSTRING_END)
 from parso.parser import BaseParser
 from parso.pgen2.parse import token_to_ilabel
 
@@ -50,6 +51,17 @@ class Parser(BaseParser):
     }
     default_node = tree.PythonNode
 
+    # Names/Keywords are handled separately
+    _leaf_map = {
+        STRING: tree.String,
+        NUMBER: tree.Number,
+        NEWLINE: tree.Newline,
+        ENDMARKER: tree.EndMarker,
+        FSTRING_STRING: tree.FStringString,
+        FSTRING_START: tree.FStringStart,
+        FSTRING_END: tree.FStringEnd,
+    }
+
     def __init__(self, pgen_grammar, error_recovery=True, start_symbol='file_input'):
         super(Parser, self).__init__(pgen_grammar, start_symbol, error_recovery=error_recovery)
 
@@ -121,16 +133,8 @@ class Parser(BaseParser):
                 return tree.Keyword(value, start_pos, prefix)
             else:
                 return tree.Name(value, start_pos, prefix)
-        elif type == STRING:
-            return tree.String(value, start_pos, prefix)
-        elif type == NUMBER:
-            return tree.Number(value, start_pos, prefix)
-        elif type == NEWLINE:
-            return tree.Newline(value, start_pos, prefix)
-        elif type == ENDMARKER:
-            return tree.EndMarker(value, start_pos, prefix)
-        else:
-            return tree.Operator(value, start_pos, prefix)
+
+        return self._leaf_map.get(type, tree.Operator)(value, start_pos, prefix)
 
     def error_recovery(self, pgen_grammar, stack, arcs, typ, value, start_pos, prefix,
                        add_token_callback):

parso/python/pep8.py

@@ -145,7 +145,7 @@ class BackslashNode(IndentationNode):
 
 
 def _is_magic_name(name):
-    return name.value.startswith('__') and name.value.startswith('__')
+    return name.value.startswith('__') and name.value.endswith('__')
 
 
 class PEP8Normalizer(ErrorFinder):

parso/python/token.py

@@ -32,6 +32,14 @@ if py_version < 35:
 ERROR_DEDENT = next(_counter)
 tok_name[ERROR_DEDENT] = 'ERROR_DEDENT'
 
+FSTRING_START = next(_counter)
+tok_name[FSTRING_START] = 'FSTRING_START'
+FSTRING_END = next(_counter)
+tok_name[FSTRING_END] = 'FSTRING_END'
+FSTRING_STRING = next(_counter)
+tok_name[FSTRING_STRING] = 'FSTRING_STRING'
+EXCLAMATION = next(_counter)
+tok_name[EXCLAMATION] = 'EXCLAMATION'
 
 # Map from operator to number (since tokenize doesn't do this)
 
@@ -84,6 +92,7 @@ opmap_raw = """\
 //= DOUBLESLASHEQUAL
 -> RARROW
 ... ELLIPSIS
+! EXCLAMATION
 """
 
 opmap = {}

parso/python/tokenize.py

@@ -20,14 +20,15 @@ from codecs import BOM_UTF8
 
 from parso.python.token import (tok_name, ENDMARKER, STRING, NUMBER, opmap,
                                 NAME, ERRORTOKEN, NEWLINE, INDENT, DEDENT,
-                                ERROR_DEDENT)
+                                ERROR_DEDENT, FSTRING_STRING, FSTRING_START,
+                                FSTRING_END)
 from parso._compatibility import py_version
 from parso.utils import split_lines
 
 
 TokenCollection = namedtuple(
     'TokenCollection',
-    'pseudo_token single_quoted triple_quoted endpats always_break_tokens',
+    'pseudo_token single_quoted triple_quoted endpats fstring_pattern_map always_break_tokens',
 )
 
 BOM_UTF8_STRING = BOM_UTF8.decode('utf-8')
@@ -52,32 +53,35 @@ def group(*choices, **kwargs):
     return start + '|'.join(choices) + ')'
 
 
-def any(*choices):
-    return group(*choices) + '*'
-
-
 def maybe(*choices):
     return group(*choices) + '?'
 
 
 # Return the empty string, plus all of the valid string prefixes.
-def _all_string_prefixes(version_info):
+def _all_string_prefixes(version_info, include_fstring=False, only_fstring=False):
     def different_case_versions(prefix):
         for s in _itertools.product(*[(c, c.upper()) for c in prefix]):
             yield ''.join(s)
     # The valid string prefixes. Only contain the lower case versions,
     #  and don't contain any permuations (include 'fr', but not
     #  'rf'). The various permutations will be generated.
-    _valid_string_prefixes = ['b', 'r', 'u']
+    valid_string_prefixes = ['b', 'r', 'u']
     if version_info >= (3, 0):
-        _valid_string_prefixes.append('br')
+        valid_string_prefixes.append('br')
 
-    if version_info >= (3, 6):
-        _valid_string_prefixes += ['f', 'fr']
+    result = set([''])
+    if version_info >= (3, 6) and include_fstring:
+        f = ['f', 'fr']
+        if only_fstring:
+            valid_string_prefixes = f
+            result = set()
+        else:
+            valid_string_prefixes += f
+    elif only_fstring:
+        return set()
 
     # if we add binary f-strings, add: ['fb', 'fbr']
-    result = set([''])
-    for prefix in _valid_string_prefixes:
+    for prefix in valid_string_prefixes:
         for t in _itertools.permutations(prefix):
             # create a list with upper and lower versions of each
             #  character
@@ -102,6 +106,10 @@ def _get_token_collection(version_info):
         return result
 
 
+fstring_string_single_line = _compile(r'(?:[^{}\r\n]+|\{\{|\}\})+')
+fstring_string_multi_line = _compile(r'(?:[^{}]+|\{\{|\}\})+')
+
+
 def _create_token_collection(version_info):
     # Note: we use unicode matching for names ("\w") but ascii matching for
     # number literals.
@@ -141,6 +149,9 @@ def _create_token_collection(version_info):
     #  StringPrefix can be the empty string (making it optional).
     possible_prefixes = _all_string_prefixes(version_info)
     StringPrefix = group(*possible_prefixes)
+    StringPrefixWithF = group(*_all_string_prefixes(version_info, include_fstring=True))
+    fstring_prefixes = _all_string_prefixes(version_info, include_fstring=True, only_fstring=True)
+    FStringStart = group(*fstring_prefixes)
 
     # Tail end of ' string.
     Single = r"[^'\\]*(?:\\.[^'\\]*)*'"
@@ -150,14 +161,14 @@ def _create_token_collection(version_info):
     Single3 = r"[^'\\]*(?:(?:\\.|'(?!''))[^'\\]*)*'''"
     # Tail end of """ string.
     Double3 = r'[^"\\]*(?:(?:\\.|"(?!""))[^"\\]*)*"""'
-    Triple = group(StringPrefix + "'''", StringPrefix + '"""')
+    Triple = group(StringPrefixWithF + "'''", StringPrefixWithF + '"""')
 
     # Because of leftmost-then-longest match semantics, be sure to put the
     # longest operators first (e.g., if = came before ==, == would get
     # recognized as two instances of =).
-    Operator = group(r"\*\*=?", r">>=?", r"<<=?", r"!=",
+    Operator = group(r"\*\*=?", r">>=?", r"<<=?",
                      r"//=?", r"->",
-                     r"[+\-*/%&@`|^=<>]=?",
+                     r"[+\-*/%&@`|^!=<>]=?",
                      r"~")
 
     Bracket = '[][(){}]'
@@ -174,7 +185,12 @@ def _create_token_collection(version_info):
                     group("'", r'\\\r?\n'),
                     StringPrefix + r'"[^\n"\\]*(?:\\.[^\n"\\]*)*' +
                     group('"', r'\\\r?\n'))
-    PseudoExtras = group(r'\\\r?\n|\Z', Comment, Triple)
+    pseudo_extra_pool = [Comment, Triple]
+    all_quotes = '"', "'", '"""', "'''"
+    if fstring_prefixes:
+        pseudo_extra_pool.append(FStringStart + group(*all_quotes))
+
+    PseudoExtras = group(r'\\\r?\n|\Z', *pseudo_extra_pool)
     PseudoToken = group(Whitespace, capture=True) + \
         group(PseudoExtras, Number, Funny, ContStr, Name, capture=True)
 
@@ -192,18 +208,24 @@ def _create_token_collection(version_info):
     #  including the opening quotes.
     single_quoted = set()
     triple_quoted = set()
+    fstring_pattern_map = {}
     for t in possible_prefixes:
-        for p in (t + '"', t + "'"):
-            single_quoted.add(p)
-        for p in (t + '"""', t + "'''"):
-            triple_quoted.add(p)
+        for quote in '"', "'":
+            single_quoted.add(t + quote)
+
+        for quote in '"""', "'''":
+            triple_quoted.add(t + quote)
+
+    for t in fstring_prefixes:
+        for quote in all_quotes:
+            fstring_pattern_map[t + quote] = quote
 
     ALWAYS_BREAK_TOKENS = (';', 'import', 'class', 'def', 'try', 'except',
                            'finally', 'while', 'with', 'return')
     pseudo_token_compiled = _compile(PseudoToken)
     return TokenCollection(
         pseudo_token_compiled, single_quoted, triple_quoted, endpats,
-        ALWAYS_BREAK_TOKENS
+        fstring_pattern_map, ALWAYS_BREAK_TOKENS
     )
 
 
@@ -226,12 +248,104 @@ class PythonToken(Token):
                 self._replace(type=self._get_type_name()))
 
 
+class FStringNode(object):
+    def __init__(self, quote):
+        self.quote = quote
+        self.parentheses_count = 0
+        self.previous_lines = ''
+        self.last_string_start_pos = None
+        # In the syntax there can be multiple format_spec's nested:
+        # {x:{y:3}}
+        self.format_spec_count = 0
+
+    def open_parentheses(self, character):
+        self.parentheses_count += 1
+
+    def close_parentheses(self, character):
+        self.parentheses_count -= 1
+
+    def allow_multiline(self):
+        return len(self.quote) == 3
+
+    def is_in_expr(self):
+        return (self.parentheses_count - self.format_spec_count) > 0
+
+
+def _check_fstring_ending(fstring_stack, token, from_start=False):
+    fstring_end = float('inf')
+    fstring_index = None
+    for i, node in enumerate(fstring_stack):
+        if from_start:
+            if token.startswith(node.quote):
+                fstring_index = i
+                fstring_end = len(node.quote)
+            else:
+                continue
+        else:
+            try:
+                end = token.index(node.quote)
+            except ValueError:
+                pass
+            else:
+                if fstring_index is None or end < fstring_end:
+                    fstring_index = i
+                    fstring_end = end
+    return fstring_index, fstring_end
+
+
+def _find_fstring_string(fstring_stack, line, lnum, pos):
+    tos = fstring_stack[-1]
+    if tos.is_in_expr():
+        return '', pos
+    else:
+        new_pos = pos
+        allow_multiline = tos.allow_multiline()
+        if allow_multiline:
+            match = fstring_string_multi_line.match(line, pos)
+        else:
+            match = fstring_string_single_line.match(line, pos)
+        if match is None:
+            string = tos.previous_lines
+        else:
+            if not tos.previous_lines:
+                tos.last_string_start_pos = (lnum, pos)
+
+            string = match.group(0)
+            for fstring_stack_node in fstring_stack:
+                try:
+                    string = string[:string.index(fstring_stack_node.quote)]
+                except ValueError:
+                    pass  # The string was not found.
+
+            new_pos += len(string)
+            if allow_multiline and string.endswith('\n'):
+                tos.previous_lines += string
+                string = ''
+            else:
+                string = tos.previous_lines + string
+
+        return string, new_pos
+
+
 def tokenize(code, version_info, start_pos=(1, 0)):
     """Generate tokens from a the source code (string)."""
     lines = split_lines(code, keepends=True)
     return tokenize_lines(lines, version_info, start_pos=start_pos)
 
 
+def _print_tokens(func):
+    """
+    A small helper function to help debug the tokenize_lines function.
+    """
+    def wrapper(*args, **kwargs):
+        for token in func(*args, **kwargs):
+            print(token)
+            yield token
+
+    return wrapper
+
+
+# @_print_tokens
 def tokenize_lines(lines, version_info, start_pos=(1, 0)):
     """
     A heavily modified Python standard library tokenizer.
@@ -240,7 +354,7 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
     token. This idea comes from lib2to3. The prefix contains all information
     that is irrelevant for the parser like newlines in parentheses or comments.
     """
-    pseudo_token, single_quoted, triple_quoted, endpats, always_break_tokens, = \
+    pseudo_token, single_quoted, triple_quoted, endpats, fstring_pattern_map, always_break_tokens, = \
         _get_token_collection(version_info)
     paren_level = 0  # count parentheses
     indents = [0]
@@ -257,6 +371,7 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
     additional_prefix = ''
     first = True
     lnum = start_pos[0] - 1
+    fstring_stack = []
     for line in lines:  # loop over lines in stream
         lnum += 1
         pos = 0
@@ -287,6 +402,37 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
                 continue
 
         while pos < max:
+            if fstring_stack:
+                string, pos = _find_fstring_string(fstring_stack, line, lnum, pos)
+                if string:
+                    yield PythonToken(
+                        FSTRING_STRING, string,
+                        fstring_stack[-1].last_string_start_pos,
+                        # Never has a prefix because it can start anywhere and
+                        # include whitespace.
+                        prefix=''
+                    )
+                    fstring_stack[-1].previous_lines = ''
+                    continue
+
+                if pos == max:
+                    break
+
+                rest = line[pos:]
+                fstring_index, end = _check_fstring_ending(fstring_stack, rest, from_start=True)
+
+                if fstring_index is not None:
+                    yield PythonToken(
+                        FSTRING_END,
+                        fstring_stack[fstring_index].quote,
+                        (lnum, pos),
+                        prefix=additional_prefix,
+                    )
+                    additional_prefix = ''
+                    del fstring_stack[fstring_index:]
+                    pos += end
+                    continue
+
             pseudomatch = pseudo_token.match(line, pos)
             if not pseudomatch:                             # scan for tokens
                 txt = line[pos:]
@@ -311,10 +457,11 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
 
             if new_line and initial not in '\r\n#':
                 new_line = False
-                if paren_level == 0:
+                if paren_level == 0 and not fstring_stack:
                     i = 0
                     while line[i] == '\f':
                         i += 1
+                        # TODO don't we need to change spos as well?
                         start -= 1
                     if start > indents[-1]:
                         yield PythonToken(INDENT, '', spos, '')
@@ -326,11 +473,33 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
                         yield PythonToken(DEDENT, '', spos, '')
                         indents.pop()
 
+            if fstring_stack:
+                fstring_index, end = _check_fstring_ending(fstring_stack, token)
+                if fstring_index is not None:
+                    if end != 0:
+                        yield PythonToken(ERRORTOKEN, token[:end], spos, prefix)
+                        prefix = ''
+
+                    yield PythonToken(
+                        FSTRING_END,
+                        fstring_stack[fstring_index].quote,
+                        (lnum, spos[1] + 1),
+                        prefix=prefix
+                    )
+                    del fstring_stack[fstring_index:]
+                    pos -= len(token) - end
+                    continue
+
             if (initial in numchars or                      # ordinary number
                     (initial == '.' and token != '.' and token != '...')):
                 yield PythonToken(NUMBER, token, spos, prefix)
             elif initial in '\r\n':
-                if not new_line and paren_level == 0:
+                if any(not f.allow_multiline() for f in fstring_stack):
+                    # Would use fstring_stack.clear, but that's not available
+                    # in Python 2.
+                    fstring_stack[:] = []
+
+                if not new_line and paren_level == 0 and not fstring_stack:
                     yield PythonToken(NEWLINE, token, spos, prefix)
                 else:
                     additional_prefix = prefix + token
@@ -362,8 +531,12 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
                     break
                 else:                                       # ordinary string
                     yield PythonToken(STRING, token, spos, prefix)
+            elif token in fstring_pattern_map:  # The start of an fstring.
+                fstring_stack.append(FStringNode(fstring_pattern_map[token]))
+                yield PythonToken(FSTRING_START, token, spos, prefix)
             elif is_identifier(initial):                      # ordinary name
                 if token in always_break_tokens:
+                    fstring_stack[:] = []
                     paren_level = 0
                     while True:
                         indent = indents.pop()
@@ -378,9 +551,18 @@ def tokenize_lines(lines, version_info, start_pos=(1, 0)):
                 break
             else:
                 if token in '([{':
-                    paren_level += 1
+                    if fstring_stack:
+                        fstring_stack[-1].open_parentheses(token)
+                    else:
+                        paren_level += 1
                 elif token in ')]}':
-                    paren_level -= 1
+                    if fstring_stack:
+                        fstring_stack[-1].close_parentheses(token)
+                    else:
+                        paren_level -= 1
+                elif token == ':' and fstring_stack \
+                        and fstring_stack[-1].parentheses_count == 1:
+                    fstring_stack[-1].format_spec_count += 1
 
                 try:
                     # This check is needed in any case to check if it's a valid

parso/python/tree.py

@@ -1,16 +1,18 @@
 """
-If you know what an syntax tree is, you'll see that this module is pretty much
-that. The classes represent syntax elements like functions and imports.
+This is the syntax tree for Python syntaxes (2 & 3).  The classes represent
+syntax elements like functions and imports.
 
-This is the "business logic" part of the parser. There's a lot of logic here
-that makes it easier for Jedi (and other libraries) to deal with a Python syntax
-tree.
+All of the nodes can be traced back to the `Python grammar file
+<https://docs.python.org/3/reference/grammar.html>`_. If you want to know how
+a tree is structured, just analyse that file (for each Python version it's a
+bit different).
 
-By using `get_code` on a module, you can get back the 1-to-1 representation of
-the input given to the parser. This is important if you are using refactoring.
+There's a lot of logic here that makes it easier for Jedi (and other libraries)
+to deal with a Python syntax tree.
 
-The easiest way to play with this module is to use :class:`parsing.Parser`.
-:attr:`parsing.Parser.module` holds an instance of :class:`Module`:
+By using :py:meth:`parso.tree.NodeOrLeaf.get_code` on a module, you can get
+back the 1-to-1 representation of the input given to the parser. This is
+important if you want to refactor a parser tree.
 
 >>> from parso import parse
 >>> parser = parse('import os')
@@ -23,6 +25,21 @@ Any subclasses of :class:`Scope`, including :class:`Module` has an attribute
 
 >>> list(module.iter_imports())
 [<ImportName: import os@1,0>]
+
+Changes to the Python Grammar
+-----------------------------
+
+A few things have changed when looking at Python grammar files:
+
+- :class:`Param` does not exist in Python grammar files. It is essentially a
+  part of a ``parameters`` node.  |parso| splits it up to make it easier to
+  analyse parameters. However this just makes it easier to deal with the syntax
+  tree, it doesn't actually change the valid syntax.
+- A few nodes like `lambdef` and `lambdef_nocond` have been merged in the
+  syntax tree to make it easier to do deal with them.
+
+Parser Tree Classes
+-------------------
 """
 
 import re
@@ -79,6 +96,10 @@ class PythonMixin(object):
     __slots__ = ()
 
     def get_name_of_position(self, position):
+        """
+        Given a (line, column) tuple, returns a :py:class:`Name` or ``None`` if
+        there is no name at that position.
+        """
         for c in self.children:
             if isinstance(c, Leaf):
                 if c.type == 'name' and c.start_pos <= position <= c.end_pos:
@@ -97,6 +118,9 @@ class PythonLeaf(PythonMixin, Leaf):
         return split_prefix(self, self.get_start_pos_of_prefix())
 
     def get_start_pos_of_prefix(self):
+        """
+        Basically calls :py:meth:`parso.tree.NodeOrLeaf.get_start_pos_of_prefix`.
+        """
         # TODO it is really ugly that we have to override it. Maybe change
         #   indent error leafs somehow? No idea how, though.
         previous_leaf = self.get_previous_leaf()
@@ -190,11 +214,6 @@ class Name(_LeafWithoutNewlines):
                 return node
             return None
 
-        if type_ in ():
-            if self in node.get_defined_names():
-                return node
-            return None
-
         if type_ == 'except_clause':
             # TODO in Python 2 this doesn't work correctly. See grammar file.
             #      I think we'll just let it be. Python 2 will be gone in a few
@@ -243,6 +262,33 @@ class String(Literal):
         return match.group(2)[:-len(match.group(1))]
 
 
+class FStringString(Leaf):
+    """
+    f-strings contain f-string expressions and normal python strings. These are
+    the string parts of f-strings.
+    """
+    type = 'fstring_string'
+    __slots__ = ()
+
+
+class FStringStart(Leaf):
+    """
+    f-strings contain f-string expressions and normal python strings. These are
+    the string parts of f-strings.
+    """
+    type = 'fstring_start'
+    __slots__ = ()
+
+
+class FStringEnd(Leaf):
+    """
+    f-strings contain f-string expressions and normal python strings. These are
+    the string parts of f-strings.
+    """
+    type = 'fstring_end'
+    __slots__ = ()
+
+
 class _StringComparisonMixin(object):
     def __eq__(self, other):
         """
@@ -343,7 +389,8 @@ class Module(Scope):
 
     def _iter_future_import_names(self):
         """
-        :return list of str: A list of future import names.
+        :return: A list of future import names.
+        :rtype: list of str
         """
         # In Python it's not allowed to use future imports after the first
         # actual (non-future) statement. However this is not a linter here,
@@ -370,8 +417,8 @@ class Module(Scope):
 
     def get_used_names(self):
         """
-        Returns all the `Name` leafs that exist in this module. Tihs includes
-        both definitions and references of names.
+        Returns all the :class:`Name` leafs that exist in this module. This
+        includes both definitions and references of names.
         """
         if self._used_names is None:
             # Don't directly use self._used_names to eliminate a lookup.
@@ -410,7 +457,7 @@ class ClassOrFunc(Scope):
 
     def get_decorators(self):
         """
-        :return list of Decorator:
+        :rtype: list of :class:`Decorator`
         """
         decorated = self.parent
         if decorated.type == 'decorated':
@@ -425,13 +472,6 @@ class ClassOrFunc(Scope):
 class Class(ClassOrFunc):
     """
     Used to store the parsed contents of a python class.
-
-    :param name: The Class name.
-    :type name: str
-    :param supers: The super classes of a Class.
-    :type supers: list
-    :param start_pos: The start position (line, column) of the class.
-    :type start_pos: tuple(int, int)
     """
     type = 'classdef'
     __slots__ = ()
@@ -579,6 +619,21 @@ class Function(ClassOrFunc):
 
         return scan(self.children)
 
+    def iter_raise_stmts(self):
+        """
+        Returns a generator of `raise_stmt`. Includes raise statements inside try-except blocks
+        """
+        def scan(children):
+            for element in children:
+                if element.type == 'raise_stmt' \
+                        or element.type == 'keyword' and element.value == 'raise':
+                    yield element
+                if element.type in _RETURN_STMT_CONTAINERS:
+                    for e in scan(element.children):
+                        yield e
+
+        return scan(self.children)
+
     def is_generator(self):
         """
         :return bool: Checks if a function is a generator or not.
@@ -1049,7 +1104,7 @@ class Param(PythonBaseNode):
     @property
     def annotation(self):
         """
-        The default is the test node that appears after `->`. Is `None` in case
+        The default is the test node that appears after `:`. Is `None` in case
         no annotation is present.
         """
         tfpdef = self._tfpdef()
@@ -1133,4 +1188,5 @@ class CompFor(PythonBaseNode):
         """
         Returns the a list of `Name` that the comprehension defines.
         """
-        return _defined_names(self.children[1])
+        # allow async for
+        return _defined_names(self.children[self.children.index('for') + 1])

parso/tree.py

@@ -4,12 +4,12 @@ from parso._compatibility import utf8_repr, encoding, py_version
 
 def search_ancestor(node, *node_types):
     """
-    Recursively looks at the parents of a node and checks if the type names
-    match.
+    Recursively looks at the parents of a node and returns the first found node
+    that matches node_types. Returns ``None`` if no matching node is found.
 
-    :param node: The node that is looked at.
-    :param node_types: A tuple or a string of type names that are
-        searched for.
+    :param node: The ancestors of this node will be checked.
+    :param node_types: type names that are searched for.
+    :type node_types: tuple of str
     """
     while True:
         node = node.parent
@@ -22,6 +22,10 @@ class NodeOrLeaf(object):
     The base class for nodes and leaves.
     """
     __slots__ = ()
+    type = None
+    '''
+    The type is a string that typically matches the types of the grammar file.
+    '''
 
     def get_root_node(self):
         """
@@ -35,8 +39,8 @@ class NodeOrLeaf(object):
 
     def get_next_sibling(self):
         """
-        The node immediately following the invocant in their parent's children
-        list. If the invocant does not have a next sibling, it is None
+        Returns the node immediately following this node in this parent's
+        children list. If this node does not have a next sibling, it is None
         """
         # Can't use index(); we need to test by identity
         for i, child in enumerate(self.parent.children):
@@ -48,8 +52,8 @@ class NodeOrLeaf(object):
 
     def get_previous_sibling(self):
         """
-        The node/leaf immediately preceding the invocant in their parent's
-        children list. If the invocant does not have a previous sibling, it is
+        Returns the node immediately preceding this node in this parent's
+        children list. If this node does not have a previous sibling, it is
         None.
         """
         # Can't use index(); we need to test by identity
@@ -62,7 +66,7 @@ class NodeOrLeaf(object):
     def get_previous_leaf(self):
         """
         Returns the previous leaf in the parser tree.
-        Raises an IndexError if it's the first element in the parser tree.
+        Returns `None` if this is the first element in the parser tree.
         """
         node = self
         while True:
@@ -85,7 +89,7 @@ class NodeOrLeaf(object):
     def get_next_leaf(self):
         """
         Returns the next leaf in the parser tree.
-        Returns `None` if it's the last element in the parser tree.
+        Returns None if this is the last element in the parser tree.
         """
         node = self
         while True:
@@ -135,19 +139,19 @@ class NodeOrLeaf(object):
     @abstractmethod
     def get_first_leaf(self):
         """
-        Returns the first leaf of a node or itself it's a leaf.
+        Returns the first leaf of a node or itself if this is a leaf.
         """
 
     @abstractmethod
     def get_last_leaf(self):
         """
-        Returns the last leaf of a node or itself it's a leaf.
+        Returns the last leaf of a node or itself if this is a leaf.
         """
 
     @abstractmethod
     def get_code(self, include_prefix=True):
         """
-        Returns the code that was the input of the parser.
+        Returns the code that was input the input for the parser for this node.
 
         :param include_prefix: Removes the prefix (whitespace and comments) of
             e.g. a statement.
@@ -155,13 +159,27 @@ class NodeOrLeaf(object):
 
 
 class Leaf(NodeOrLeaf):
+    '''
+    Leafs are basically tokens with a better API. Leafs exactly know where they
+    were defined and what text preceeds them.
+    '''
     __slots__ = ('value', 'parent', 'line', 'column', 'prefix')
 
     def __init__(self, value, start_pos, prefix=''):
         self.value = value
+        '''
+        :py:func:`str` The value of the current token.
+        '''
         self.start_pos = start_pos
         self.prefix = prefix
+        '''
+        :py:func:`str` Typically a mixture of whitespace and comments. Stuff
+        that is syntactically irrelevant for the syntax tree.
+        '''
         self.parent = None
+        '''
+        The parent :class:`BaseNode` of this leaf.
+        '''
 
     @property
     def start_pos(self):
@@ -219,9 +237,7 @@ class TypedLeaf(Leaf):
 class BaseNode(NodeOrLeaf):
     """
     The super class for all nodes.
-
-    If you create custom nodes, you will probably want to inherit from this
-    ``BaseNode``.
+    A node has children, a type and possibly a parent node.
     """
     __slots__ = ('children', 'parent')
     type = None
@@ -230,7 +246,14 @@ class BaseNode(NodeOrLeaf):
         for c in children:
             c.parent = self
         self.children = children
+        """
+        A list of :class:`NodeOrLeaf` child nodes.
+        """
         self.parent = None
+        '''
+        The parent :class:`BaseNode` of this leaf.
+        None if this is the root node.
+        '''
 
     @property
     def start_pos(self):
@@ -254,6 +277,14 @@ class BaseNode(NodeOrLeaf):
         return self._get_code_for_children(self.children, include_prefix)
 
     def get_leaf_for_position(self, position, include_prefixes=False):
+        """
+        Get the :py:class:`parso.tree.Leaf` at ``position``
+
+        :param tuple position: A position tuple, row, column. Rows start from 1
+        :param bool include_prefixes: If ``False``, ``None`` will be returned if ``position`` falls
+            on whitespace or comments before a leaf
+        :return: :py:class:`parso.tree.Leaf` at ``position``, or ``None``
+        """
         def binary_search(lower, upper):
             if lower == upper:
                 element = self.children[lower]
@@ -307,7 +338,7 @@ class Node(BaseNode):
 
 class ErrorNode(BaseNode):
     """
-    A node that containes valid nodes/leaves that we're follow by a token that
+    A node that contains valid nodes/leaves that we're follow by a token that
     was invalid. This basically means that the leaf after this node is where
     Python would mark a syntax error.
     """

parso/utils.py

@@ -11,10 +11,10 @@ Version = namedtuple('Version', 'major, minor, micro')
 
 def split_lines(string, keepends=False):
     r"""
-    A str.splitlines for Python code. In contrast to Python's ``str.splitlines``,
+    Intended for Python code. In contrast to Python's :py:meth:`str.splitlines`,
     looks at form feeds and other special characters as normal text. Just
     splits ``\n`` and ``\r\n``.
-    Also different: Returns ``['']`` for an empty string input.
+    Also different: Returns ``[""]`` for an empty string input.
 
     In Python 2.7 form feeds are used as normal characters when using
     str.splitlines. However in Python 3 somewhere there was a decision to split
@@ -48,9 +48,14 @@ def split_lines(string, keepends=False):
         return re.split('\n|\r\n', string)
 
 
-def python_bytes_to_unicode(source, default_encoding='utf-8', errors='strict'):
+def python_bytes_to_unicode(source, encoding='utf-8', errors='strict'):
     """
-    `errors` can be 'strict', 'replace' or 'ignore'.
+    Checks for unicode BOMs and PEP 263 encoding declarations. Then returns a
+    unicode object like in :py:meth:`bytes.decode`.
+
+    :param encoding: See :py:meth:`bytes.decode` documentation.
+    :param errors: See :py:meth:`bytes.decode` documentation. ``errors`` can be
+        ``'strict'``, ``'replace'`` or ``'ignore'``.
     """
     def detect_encoding():
         """
@@ -70,7 +75,7 @@ def python_bytes_to_unicode(source, default_encoding='utf-8', errors='strict'):
             return possible_encoding.group(1)
         else:
             # the default if nothing else has been set -> PEP 263
-            return default_encoding
+            return encoding
 
     if isinstance(source, unicode):
         # only cast str/bytes

setup.py

@@ -14,7 +14,7 @@ readme = open('README.rst').read() + '\n\n' + open('CHANGELOG.rst').read()
 
 setup(name='parso',
       version=parso.__version__,
-      description='A Python parser written in Python.',
+      description='A Python Parser',
       author=__AUTHOR__,
       author_email=__AUTHOR_EMAIL__,
       include_package_data=True,

test/failing_examples.py

@@ -141,7 +141,7 @@ FAILING_EXAMPLES = [
 
     # f-strings
     'f"{}"',
-    'f"{\\}"',
+    r'f"{\}"',
     'f"{\'\\\'}"',
     'f"{#}"',
     "f'{1!b}'",

test/test_diff_parser.py

@@ -484,3 +484,21 @@ def test_indentation_issue(differ):
 
     differ.initialize(code1)
     differ.parse(code2, parsers=2)
+
+
+def test_endmarker_newline(differ):
+    code1 = dedent('''\
+        docu = None
+        # some comment
+        result = codet
+        incomplete_dctassign = {
+            "module"
+
+        if "a":
+            x = 3 # asdf
+    ''')
+
+    code2 = code1.replace('codet', 'coded')
+
+    differ.initialize(code1)
+    differ.parse(code2, parsers=2, copies=2, expect_error_leaves=True)

test/test_fstring.py

@@ -1,17 +1,19 @@
 import pytest
+from textwrap import dedent
 
 from parso import load_grammar, ParserSyntaxError
-from parso.python.fstring import tokenize
+from parso.python.tokenize import tokenize
 
 
 @pytest.fixture
 def grammar():
-    return load_grammar(language="python-f-string")
+    return load_grammar(version='3.6')
 
 
 @pytest.mark.parametrize(
     'code', [
         '{1}',
+        '{1:}',
         '',
         '{1!a}',
         '{1!a:1}',
@@ -26,22 +28,12 @@ def grammar():
         '{{{1}',
         '1{{2{{3',
         '}}',
-        '{:}}}',
-
-        # Invalid, but will be checked, later.
-        '{}',
-        '{1:}',
-        '{:}',
-        '{:1}',
-        '{!:}',
-        '{!}',
-        '{!a}',
-        '{1:{}}',
-        '{1:{:}}',
     ]
 )
 def test_valid(code, grammar):
-    fstring = grammar.parse(code, error_recovery=False)
+    code = 'f"""%s"""' % code
+    module = grammar.parse(code, error_recovery=False)
+    fstring = module.children[0]
     assert fstring.type == 'fstring'
     assert fstring.get_code() == code
 
@@ -52,24 +44,46 @@ def test_valid(code, grammar):
         '{',
         '{1!{a}}',
         '{!{a}}',
+        '{}',
+        '{:}',
+        '{:}}}',
+        '{:1}',
+        '{!:}',
+        '{!}',
+        '{!a}',
+        '{1:{}}',
+        '{1:{:}}',
     ]
 )
 def test_invalid(code, grammar):
+    code = 'f"""%s"""' % code
     with pytest.raises(ParserSyntaxError):
         grammar.parse(code, error_recovery=False)
 
     # It should work with error recovery.
-    #grammar.parse(code, error_recovery=True)
+    grammar.parse(code, error_recovery=True)
 
 
 @pytest.mark.parametrize(
-    ('code', 'start_pos', 'positions'), [
+    ('code', 'positions'), [
         # 2 times 2, 5 because python expr and endmarker.
-        ('}{', (2, 3), [(2, 3), (2, 4), (2, 5), (2, 5)]),
-        (' :{ 1 : } ', (1, 0), [(1, 2), (1, 3), (1, 6), (1, 8), (1, 10)]),
-        ('\n{\nfoo\n }', (2, 1), [(3, 0), (3, 1), (5, 1), (5, 2)]),
+        ('f"}{"', [(1, 0), (1, 2), (1, 3), (1, 4), (1, 5)]),
+        ('f" :{ 1 : } "', [(1, 0), (1, 2), (1, 4), (1, 6), (1, 8), (1, 9),
+                           (1, 10), (1, 11), (1, 12), (1, 13)]),
+        ('f"""\n {\nfoo\n }"""', [(1, 0), (1, 4), (2, 1), (3, 0), (4, 1),
+                                  (4, 2), (4, 5)]),
     ]
 )
-def test_tokenize_start_pos(code, start_pos, positions):
-    tokens = tokenize(code, start_pos)
+def test_tokenize_start_pos(code, positions):
+    tokens = list(tokenize(code, version_info=(3, 6)))
     assert positions == [p.start_pos for p in tokens]
+
+
+def test_roundtrip(grammar):
+    code = dedent("""\
+        f'''s{
+           str.uppe
+        '''
+        """)
+    tree = grammar.parse(code)
+    assert tree.get_code() == code

test/test_parser_tree.py

@@ -125,6 +125,10 @@ def get_return_stmts(code):
     return list(parse(code).children[0].iter_return_stmts())
 
 
+def get_raise_stmts(code, child):
+    return list(parse(code).children[child].iter_raise_stmts())
+
+
 def test_yields(each_version):
     y, = get_yield_exprs('def x(): yield', each_version)
     assert y.value == 'yield'
@@ -149,3 +153,30 @@ def test_returns():
 
     r, = get_return_stmts('def x(): return 1')
     assert r.type == 'return_stmt'
+
+
+def test_raises():
+    code = """
+def single_function():
+    raise Exception
+def top_function():
+    def inner_function():
+        raise NotImplementedError()
+    inner_function()
+    raise Exception
+def top_function_three():
+    try:
+        raise NotImplementedError()
+    except NotImplementedError:
+        pass
+    raise Exception
+    """
+
+    r = get_raise_stmts(code, 0) #  Lists in a simple Function
+    assert len(list(r)) == 1
+
+    r = get_raise_stmts(code, 1) #  Doesn't Exceptions list in closures
+    assert len(list(r)) == 1
+
+    r = get_raise_stmts(code, 2) #  Lists inside try-catch
+    assert len(list(r)) == 2

test/test_python_errors.py

@@ -114,6 +114,22 @@ def _get_actual_exception(code):
         # Python 3.4/3.4 have a bit of a different warning than 3.5/3.6 in
         # certain places. But in others this error makes sense.
         return [wanted, "SyntaxError: can't use starred expression here"], line_nr
+    elif wanted == 'SyntaxError: f-string: unterminated string':
+        wanted = 'SyntaxError: EOL while scanning string literal'
+    elif wanted == 'SyntaxError: f-string expression part cannot include a backslash':
+        return [
+            wanted,
+            "SyntaxError: EOL while scanning string literal",
+            "SyntaxError: unexpected character after line continuation character",
+        ], line_nr
+    elif wanted == "SyntaxError: f-string: expecting '}'":
+        wanted = 'SyntaxError: EOL while scanning string literal'
+    elif wanted == 'SyntaxError: f-string: empty expression not allowed':
+        wanted = 'SyntaxError: invalid syntax'
+    elif wanted == "SyntaxError: f-string expression part cannot include '#'":
+        wanted = 'SyntaxError: invalid syntax'
+    elif wanted == "SyntaxError: f-string: single '}' is not allowed":
+        wanted = 'SyntaxError: invalid syntax'
     return [wanted], line_nr
 
 

test/test_tokenize.py

@@ -7,7 +7,8 @@ import pytest
 from parso._compatibility import py_version
 from parso.utils import split_lines, parse_version_string
 from parso.python.token import (
-    NAME, NEWLINE, STRING, INDENT, DEDENT, ERRORTOKEN, ENDMARKER, ERROR_DEDENT)
+    NAME, NEWLINE, STRING, INDENT, DEDENT, ERRORTOKEN, ENDMARKER, ERROR_DEDENT,
+    FSTRING_START)
 from parso.python import tokenize
 from parso import parse
 from parso.python.tokenize import PythonToken
@@ -162,8 +163,9 @@ def test_ur_literals():
         token_list = _get_token_list(literal)
         typ, result_literal, _, _ = token_list[0]
         if is_literal:
-            assert typ == STRING
-            assert result_literal == literal
+            if typ != FSTRING_START:
+                assert typ == STRING
+                assert result_literal == literal
         else:
             assert typ == NAME
 
@@ -175,6 +177,7 @@ def test_ur_literals():
     # Starting with Python 3.3 this ordering is also possible.
     if py_version >= 33:
         check('Rb""')
+
     # Starting with Python 3.6 format strings where introduced.
     check('fr""', is_literal=py_version >= 36)
     check('rF""', is_literal=py_version >= 36)

tox.ini

@@ -1,20 +1,21 @@
 [tox]
-envlist = py26, py27, py33, py34, py35, py36
+envlist = py27, py33, py34, py35, py36, py37
 [testenv]
 deps =
-    pytest>=3.0.7
+    {env:_PARSO_TEST_PYTEST_DEP:pytest>=3.0.7}
 # For --lf and --ff.
     pytest-cache
 setenv =
 # https://github.com/tomchristie/django-rest-framework/issues/1957
 # tox corrupts __pycache__, solution from here:
     PYTHONDONTWRITEBYTECODE=1
+    py26,py33: _PARSO_TEST_PYTEST_DEP=pytest>=3.0.7,<3.3
 commands =
-    py.test {posargs:parso test}
+    pytest {posargs:parso test}
 [testenv:cov]
 deps =
     coverage
     {[testenv]deps}
 commands =
-    coverage run --source parso -m py.test
+    coverage run --source parso -m pytest
     coverage report