Remove python 2 references from docstrings, where appropriate.

IPython/core/oinspect.py

@@ -1,1032 +1,1031 b''
 # -*- coding: utf-8 -*-
 """Tools for inspecting Python objects.
 
 Uses syntax highlighting for presenting the various information elements.
 
 Similar in spirit to the inspect module, but all calls take a name argument to
 reference the name under which an object is being read.
 """
 
 # Copyright (c) IPython Development Team.
 # Distributed under the terms of the Modified BSD License.
 
 __all__ = ['Inspector','InspectColors']
 
 # stdlib modules
 import ast
 import inspect
 from inspect import signature
 import linecache
 import warnings
 import os
 from textwrap import dedent
 import types
 import io as stdlib_io
 
 from typing import Union
 
 # IPython's own
 from IPython.core import page
 from IPython.lib.pretty import pretty
 from IPython.testing.skipdoctest import skip_doctest
 from IPython.utils import PyColorize
 from IPython.utils import openpy
 from IPython.utils import py3compat
 from IPython.utils.dir2 import safe_hasattr
 from IPython.utils.path import compress_user
 from IPython.utils.text import indent
 from IPython.utils.wildcard import list_namespace
 from IPython.utils.wildcard import typestr2type
 from IPython.utils.coloransi import TermColors, ColorScheme, ColorSchemeTable
 from IPython.utils.py3compat import cast_unicode
 from IPython.utils.colorable import Colorable
 from IPython.utils.decorators import undoc
 
 from pygments import highlight
 from pygments.lexers import PythonLexer
 from pygments.formatters import HtmlFormatter
 
 def pylight(code):
     return highlight(code, PythonLexer(), HtmlFormatter(noclasses=True))
 
 # builtin docstrings to ignore
 _func_call_docstring = types.FunctionType.__call__.__doc__
 _object_init_docstring = object.__init__.__doc__
 _builtin_type_docstrings = {
     inspect.getdoc(t) for t in (types.ModuleType, types.MethodType,
                                 types.FunctionType, property)
 }
 
 _builtin_func_type = type(all)
 _builtin_meth_type = type(str.upper)  # Bound methods have the same type as builtin functions
 #****************************************************************************
 # Builtin color schemes
 
 Colors = TermColors  # just a shorthand
 
 InspectColors = PyColorize.ANSICodeColors
 
 #****************************************************************************
 # Auxiliary functions and objects
 
 # See the messaging spec for the definition of all these fields.  This list
 # effectively defines the order of display
 info_fields = ['type_name', 'base_class', 'string_form', 'namespace',
                'length', 'file', 'definition', 'docstring', 'source',
                'init_definition', 'class_docstring', 'init_docstring',
                'call_def', 'call_docstring',
                # These won't be printed but will be used to determine how to
                # format the object
                'ismagic', 'isalias', 'isclass', 'found', 'name'
                ]
 
 
 def object_info(**kw):
     """Make an object info dict with all fields present."""
     infodict = {k:None for k in info_fields}
     infodict.update(kw)
     return infodict
 
 
 def get_encoding(obj):
     """Get encoding for python source file defining obj
 
     Returns None if obj is not defined in a sourcefile.
     """
     ofile = find_file(obj)
     # run contents of file through pager starting at line where the object
     # is defined, as long as the file isn't binary and is actually on the
     # filesystem.
     if ofile is None:
         return None
     elif ofile.endswith(('.so', '.dll', '.pyd')):
         return None
     elif not os.path.isfile(ofile):
         return None
     else:
         # Print only text files, not extension binaries.  Note that
         # getsourcelines returns lineno with 1-offset and page() uses
         # 0-offset, so we must adjust.
         with stdlib_io.open(ofile, 'rb') as buffer:   # Tweaked to use io.open for Python 2
             encoding, lines = openpy.detect_encoding(buffer.readline)
         return encoding
 
 def getdoc(obj) -> Union[str,None]:
     """Stable wrapper around inspect.getdoc.
 
     This can't crash because of attribute problems.
 
     It also attempts to call a getdoc() method on the given object.  This
     allows objects which provide their docstrings via non-standard mechanisms
     (like Pyro proxies) to still be inspected by ipython's ? system.
     """
     # Allow objects to offer customized documentation via a getdoc method:
     try:
         ds = obj.getdoc()
     except Exception:
         pass
     else:
         if isinstance(ds, str):
             return inspect.cleandoc(ds)
     docstr = inspect.getdoc(obj)
     return docstr
 
 
 def getsource(obj, oname='') -> Union[str,None]:
     """Wrapper around inspect.getsource.
 
     This can be modified by other projects to provide customized source
     extraction.
 
     Parameters
     ----------
     obj : object
         an object whose source code we will attempt to extract
     oname : str
         (optional) a name under which the object is known
 
     Returns
     -------
     src : unicode or None
 
     """
 
     if isinstance(obj, property):
         sources = []
         for attrname in ['fget', 'fset', 'fdel']:
             fn = getattr(obj, attrname)
             if fn is not None:
                 encoding = get_encoding(fn)
                 oname_prefix = ('%s.' % oname) if oname else ''
                 sources.append(''.join(('# ', oname_prefix, attrname)))
                 if inspect.isfunction(fn):
                     sources.append(dedent(getsource(fn)))
                 else:
                     # Default str/repr only prints function name,
                     # pretty.pretty prints module name too.
                     sources.append(
                         '%s%s = %s\n' % (oname_prefix, attrname, pretty(fn))
                     )
         if sources:
             return '\n'.join(sources)
         else:
             return None
 
     else:
         # Get source for non-property objects.
 
         obj = _get_wrapped(obj)
 
         try:
             src = inspect.getsource(obj)
         except TypeError:
             # The object itself provided no meaningful source, try looking for
             # its class definition instead.
             if hasattr(obj, '__class__'):
                 try:
                     src = inspect.getsource(obj.__class__)
                 except TypeError:
                     return None
 
         return src
 
 
 def is_simple_callable(obj):
     """True if obj is a function ()"""
     return (inspect.isfunction(obj) or inspect.ismethod(obj) or \
             isinstance(obj, _builtin_func_type) or isinstance(obj, _builtin_meth_type))
 
 @undoc
 def getargspec(obj):
-    """Wrapper around :func:`inspect.getfullargspec` on Python 3, and
-    :func:inspect.getargspec` on Python 2.
+    """Wrapper around :func:`inspect.getfullargspec`
 
     In addition to functions and methods, this can also handle objects with a
     ``__call__`` attribute.
 
     DEPRECATED: Deprecated since 7.10. Do not use, will be removed.
     """
 
     warnings.warn('`getargspec` function is deprecated as of IPython 7.10'
                   'and will be removed in future versions.', DeprecationWarning, stacklevel=2)
 
     if safe_hasattr(obj, '__call__') and not is_simple_callable(obj):
         obj = obj.__call__
 
     return inspect.getfullargspec(obj)
 
 @undoc
 def format_argspec(argspec):
     """Format argspect, convenience wrapper around inspect's.
 
     This takes a dict instead of ordered arguments and calls
     inspect.format_argspec with the arguments in the necessary order.
 
     DEPRECATED: Do not use; will be removed in future versions.
     """
     
     warnings.warn('`format_argspec` function is deprecated as of IPython 7.10'
                   'and will be removed in future versions.', DeprecationWarning, stacklevel=2)
 
 
     return inspect.formatargspec(argspec['args'], argspec['varargs'],
                                  argspec['varkw'], argspec['defaults'])
 
 @undoc
 def call_tip(oinfo, format_call=True):
     """DEPRECATED. Extract call tip data from an oinfo dict.
     """
     warnings.warn('`call_tip` function is deprecated as of IPython 6.0'
                   'and will be removed in future versions.', DeprecationWarning, stacklevel=2)
     # Get call definition
     argspec = oinfo.get('argspec')
     if argspec is None:
         call_line = None
     else:
         # Callable objects will have 'self' as their first argument, prune
         # it out if it's there for clarity (since users do *not* pass an
         # extra first argument explicitly).
         try:
             has_self = argspec['args'][0] == 'self'
         except (KeyError, IndexError):
             pass
         else:
             if has_self:
                 argspec['args'] = argspec['args'][1:]
 
         call_line = oinfo['name']+format_argspec(argspec)
 
     # Now get docstring.
     # The priority is: call docstring, constructor docstring, main one.
     doc = oinfo.get('call_docstring')
     if doc is None:
         doc = oinfo.get('init_docstring')
     if doc is None:
         doc = oinfo.get('docstring','')
 
     return call_line, doc
 
 
 def _get_wrapped(obj):
     """Get the original object if wrapped in one or more @decorators
 
     Some objects automatically construct similar objects on any unrecognised
     attribute access (e.g. unittest.mock.call). To protect against infinite loops,
     this will arbitrarily cut off after 100 levels of obj.__wrapped__
     attribute access. --TK, Jan 2016
     """
     orig_obj = obj
     i = 0
     while safe_hasattr(obj, '__wrapped__'):
         obj = obj.__wrapped__
         i += 1
         if i > 100:
             # __wrapped__ is probably a lie, so return the thing we started with
             return orig_obj
     return obj
 
 def find_file(obj) -> str:
     """Find the absolute path to the file where an object was defined.
 
     This is essentially a robust wrapper around `inspect.getabsfile`.
 
     Returns None if no file can be found.
 
     Parameters
     ----------
     obj : any Python object
 
     Returns
     -------
     fname : str
       The absolute path to the file where the object was defined.
     """
     obj = _get_wrapped(obj)
 
     fname = None
     try:
         fname = inspect.getabsfile(obj)
     except TypeError:
         # For an instance, the file that matters is where its class was
         # declared.
         if hasattr(obj, '__class__'):
             try:
                 fname = inspect.getabsfile(obj.__class__)
             except TypeError:
                 # Can happen for builtins
                 pass
     except:
         pass
     return cast_unicode(fname)
 
 
 def find_source_lines(obj):
     """Find the line number in a file where an object was defined.
 
     This is essentially a robust wrapper around `inspect.getsourcelines`.
 
     Returns None if no file can be found.
 
     Parameters
     ----------
     obj : any Python object
 
     Returns
     -------
     lineno : int
       The line number where the object definition starts.
     """
     obj = _get_wrapped(obj)
 
     try:
         try:
             lineno = inspect.getsourcelines(obj)[1]
         except TypeError:
             # For instances, try the class object like getsource() does
             if hasattr(obj, '__class__'):
                 lineno = inspect.getsourcelines(obj.__class__)[1]
             else:
                 lineno = None
     except:
         return None
 
     return lineno
 
 class Inspector(Colorable):
 
     def __init__(self, color_table=InspectColors,
                  code_color_table=PyColorize.ANSICodeColors,
                  scheme=None,
                  str_detail_level=0,
                  parent=None, config=None):
         super(Inspector, self).__init__(parent=parent, config=config)
         self.color_table = color_table
         self.parser = PyColorize.Parser(out='str', parent=self, style=scheme)
         self.format = self.parser.format
         self.str_detail_level = str_detail_level
         self.set_active_scheme(scheme)
 
     def _getdef(self,obj,oname='') -> Union[str,None]:
         """Return the call signature for any callable object.
 
         If any exception is generated, None is returned instead and the
         exception is suppressed."""
         try:
             return _render_signature(signature(obj), oname)
         except:
             return None
 
     def __head(self,h) -> str:
         """Return a header string with proper colors."""
         return '%s%s%s' % (self.color_table.active_colors.header,h,
                            self.color_table.active_colors.normal)
 
     def set_active_scheme(self, scheme):
         if scheme is not None:
             self.color_table.set_active_scheme(scheme)
             self.parser.color_table.set_active_scheme(scheme)
 
     def noinfo(self, msg, oname):
         """Generic message when no information is found."""
         print('No %s found' % msg, end=' ')
         if oname:
             print('for %s' % oname)
         else:
             print()
 
     def pdef(self, obj, oname=''):
         """Print the call signature for any callable object.
 
         If the object is a class, print the constructor information."""
 
         if not callable(obj):
             print('Object is not callable.')
             return
 
         header = ''
 
         if inspect.isclass(obj):
             header = self.__head('Class constructor information:\n')
 
 
         output = self._getdef(obj,oname)
         if output is None:
             self.noinfo('definition header',oname)
         else:
             print(header,self.format(output), end=' ')
 
     # In Python 3, all classes are new-style, so they all have __init__.
     @skip_doctest
     def pdoc(self, obj, oname='', formatter=None):
         """Print the docstring for any object.
 
         Optional:
         -formatter: a function to run the docstring through for specially
         formatted docstrings.
 
         Examples
         --------
 
         In [1]: class NoInit:
            ...:     pass
 
         In [2]: class NoDoc:
            ...:     def __init__(self):
            ...:         pass
 
         In [3]: %pdoc NoDoc
         No documentation found for NoDoc
 
         In [4]: %pdoc NoInit
         No documentation found for NoInit
 
         In [5]: obj = NoInit()
 
         In [6]: %pdoc obj
         No documentation found for obj
 
         In [5]: obj2 = NoDoc()
 
         In [6]: %pdoc obj2
         No documentation found for obj2
         """
 
         head = self.__head  # For convenience
         lines = []
         ds = getdoc(obj)
         if formatter:
             ds = formatter(ds).get('plain/text', ds)
         if ds:
             lines.append(head("Class docstring:"))
             lines.append(indent(ds))
         if inspect.isclass(obj) and hasattr(obj, '__init__'):
             init_ds = getdoc(obj.__init__)
             if init_ds is not None:
                 lines.append(head("Init docstring:"))
                 lines.append(indent(init_ds))
         elif hasattr(obj,'__call__'):
             call_ds = getdoc(obj.__call__)
             if call_ds:
                 lines.append(head("Call docstring:"))
                 lines.append(indent(call_ds))
 
         if not lines:
             self.noinfo('documentation',oname)
         else:
             page.page('\n'.join(lines))
 
     def psource(self, obj, oname=''):
         """Print the source code for an object."""
 
         # Flush the source cache because inspect can return out-of-date source
         linecache.checkcache()
         try:
             src = getsource(obj, oname=oname)
         except Exception:
             src = None
 
         if src is None:
             self.noinfo('source', oname)
         else:
             page.page(self.format(src))
 
     def pfile(self, obj, oname=''):
         """Show the whole file where an object was defined."""
 
         lineno = find_source_lines(obj)
         if lineno is None:
             self.noinfo('file', oname)
             return
 
         ofile = find_file(obj)
         # run contents of file through pager starting at line where the object
         # is defined, as long as the file isn't binary and is actually on the
         # filesystem.
         if ofile.endswith(('.so', '.dll', '.pyd')):
             print('File %r is binary, not printing.' % ofile)
         elif not os.path.isfile(ofile):
             print('File %r does not exist, not printing.' % ofile)
         else:
             # Print only text files, not extension binaries.  Note that
             # getsourcelines returns lineno with 1-offset and page() uses
             # 0-offset, so we must adjust.
             page.page(self.format(openpy.read_py_file(ofile, skip_encoding_cookie=False)), lineno - 1)
 
 
     def _mime_format(self, text:str, formatter=None) -> dict:
         """Return a mime bundle representation of the input text.
 
         - if `formatter` is None, the returned mime bundle has
            a `text/plain` field, with the input text.
            a `text/html` field with a `<pre>` tag containing the input text.
 
         - if `formatter` is not None, it must be a callable transforming the
           input text into a mime bundle. Default values for `text/plain` and
           `text/html` representations are the ones described above.
 
         Note:
 
         Formatters returning strings are supported but this behavior is deprecated.
 
         """
         defaults = {
             'text/plain': text,
             'text/html': '<pre>' + text + '</pre>'
         }
 
         if formatter is None:
             return defaults
         else:
             formatted = formatter(text)
 
             if not isinstance(formatted, dict):
                 # Handle the deprecated behavior of a formatter returning
                 # a string instead of a mime bundle.
                 return {
                     'text/plain': formatted,
                     'text/html': '<pre>' + formatted + '</pre>'
                 }
 
             else:
                 return dict(defaults, **formatted)
 
 
     def format_mime(self, bundle):
 
         text_plain = bundle['text/plain']
 
         text = ''
         heads, bodies = list(zip(*text_plain))
         _len = max(len(h) for h in heads)
 
         for head, body in zip(heads, bodies):
             body = body.strip('\n')
             delim = '\n' if '\n' in body else ' '
             text += self.__head(head+':') + (_len - len(head))*' ' +delim + body +'\n'
 
         bundle['text/plain'] = text
         return bundle
 
     def _get_info(self, obj, oname='', formatter=None, info=None, detail_level=0):
         """Retrieve an info dict and format it.
 
         Parameters
         ==========
 
         obj: any
             Object to inspect and return info from
         oname: str (default: ''):
             Name of the variable pointing to `obj`.
         formatter: callable
         info:
             already computed information
         detail_level: integer
             Granularity of detail level, if set to 1, give more information.
         """
 
         info = self._info(obj, oname=oname, info=info, detail_level=detail_level)
 
         _mime = {
             'text/plain': [],
             'text/html': '',
         }
 
         def append_field(bundle, title:str, key:str, formatter=None):
             field = info[key]
             if field is not None:
                 formatted_field = self._mime_format(field, formatter)
                 bundle['text/plain'].append((title, formatted_field['text/plain']))
                 bundle['text/html'] += '<h1>' + title + '</h1>\n' + formatted_field['text/html'] + '\n'
 
         def code_formatter(text):
             return {
                 'text/plain': self.format(text),
                 'text/html': pylight(text)
             }
 
         if info['isalias']:
             append_field(_mime, 'Repr', 'string_form')
 
         elif info['ismagic']:
             if detail_level > 0:
                 append_field(_mime, 'Source', 'source', code_formatter)
             else:
                 append_field(_mime, 'Docstring', 'docstring', formatter)
             append_field(_mime, 'File', 'file')
 
         elif info['isclass'] or is_simple_callable(obj):
             # Functions, methods, classes
             append_field(_mime, 'Signature', 'definition', code_formatter)
             append_field(_mime, 'Init signature', 'init_definition', code_formatter)
             append_field(_mime, 'Docstring', 'docstring', formatter)
             if detail_level > 0 and info['source']:
                 append_field(_mime, 'Source', 'source', code_formatter)
             else:
                 append_field(_mime, 'Init docstring', 'init_docstring', formatter)
 
             append_field(_mime, 'File', 'file')
             append_field(_mime, 'Type', 'type_name')
             append_field(_mime, 'Subclasses', 'subclasses')
 
         else:
             # General Python objects
             append_field(_mime, 'Signature', 'definition', code_formatter)
             append_field(_mime, 'Call signature', 'call_def', code_formatter)
             append_field(_mime, 'Type', 'type_name')
             append_field(_mime, 'String form', 'string_form')
 
             # Namespace
             if info['namespace'] != 'Interactive':
                 append_field(_mime, 'Namespace', 'namespace')
 
             append_field(_mime, 'Length', 'length')
             append_field(_mime, 'File', 'file')
 
             # Source or docstring, depending on detail level and whether
             # source found.
             if detail_level > 0 and info['source']:
                 append_field(_mime, 'Source', 'source', code_formatter)
             else:
                 append_field(_mime, 'Docstring', 'docstring', formatter)
 
             append_field(_mime, 'Class docstring', 'class_docstring', formatter)
             append_field(_mime, 'Init docstring', 'init_docstring', formatter)
             append_field(_mime, 'Call docstring', 'call_docstring', formatter)
 
 
         return self.format_mime(_mime)
 
     def pinfo(self, obj, oname='', formatter=None, info=None, detail_level=0, enable_html_pager=True):
         """Show detailed information about an object.
 
         Optional arguments:
 
         - oname: name of the variable pointing to the object.
 
         - formatter: callable (optional)
               A special formatter for docstrings.
 
               The formatter is a callable that takes a string as an input
               and returns either a formatted string or a mime type bundle
               in the form of a dictionary.
 
               Although the support of custom formatter returning a string
               instead of a mime type bundle is deprecated.
 
         - info: a structure with some information fields which may have been
           precomputed already.
 
         - detail_level: if set to 1, more information is given.
         """
         info = self._get_info(obj, oname, formatter, info, detail_level)
         if not enable_html_pager:
             del info['text/html']
         page.page(info)
 
     def info(self, obj, oname='', formatter=None, info=None, detail_level=0):
         """DEPRECATED. Compute a dict with detailed information about an object.
         """
         if formatter is not None:
             warnings.warn('The `formatter` keyword argument to `Inspector.info`'
                      'is deprecated as of IPython 5.0 and will have no effects.',
                       DeprecationWarning, stacklevel=2)
         return self._info(obj, oname=oname, info=info, detail_level=detail_level)
 
     def _info(self, obj, oname='', info=None, detail_level=0) -> dict:
         """Compute a dict with detailed information about an object.
 
         Parameters
         ==========
 
         obj: any
             An object to find information about
         oname: str (default: ''):
             Name of the variable pointing to `obj`.
         info: (default: None)
             A struct (dict like with attr access) with some information fields
             which may have been precomputed already.
         detail_level: int (default:0)
             If set to 1, more information is given.
 
         Returns
         =======
 
         An object info dict with known fields from `info_fields`. Keys are
         strings, values are string or None.
         """
 
         if info is None:
             ismagic = False
             isalias = False
             ospace = ''
         else:
             ismagic = info.ismagic
             isalias = info.isalias
             ospace = info.namespace
 
         # Get docstring, special-casing aliases:
         if isalias:
             if not callable(obj):
                 try:
                     ds = "Alias to the system command:\n  %s" % obj[1]
                 except:
                     ds = "Alias: " + str(obj)
             else:
                 ds = "Alias to " + str(obj)
                 if obj.__doc__:
                     ds += "\nDocstring:\n" + obj.__doc__
         else:
             ds = getdoc(obj)
             if ds is None:
                 ds = '<no docstring>'
 
         # store output in a dict, we initialize it here and fill it as we go
         out = dict(name=oname, found=True, isalias=isalias, ismagic=ismagic, subclasses=None)
 
         string_max = 200 # max size of strings to show (snipped if longer)
         shalf = int((string_max - 5) / 2)
 
         if ismagic:
             out['type_name'] = 'Magic function'
         elif isalias:
             out['type_name'] = 'System alias'
         else:
             out['type_name'] = type(obj).__name__
 
         try:
             bclass = obj.__class__
             out['base_class'] = str(bclass)
         except:
             pass
 
         # String form, but snip if too long in ? form (full in ??)
         if detail_level >= self.str_detail_level:
             try:
                 ostr = str(obj)
                 str_head = 'string_form'
                 if not detail_level and len(ostr)>string_max:
                     ostr = ostr[:shalf] + ' <...> ' + ostr[-shalf:]
                     ostr = ("\n" + " " * len(str_head.expandtabs())).\
                             join(q.strip() for q in ostr.split("\n"))
                 out[str_head] = ostr
             except:
                 pass
 
         if ospace:
             out['namespace'] = ospace
 
         # Length (for strings and lists)
         try:
             out['length'] = str(len(obj))
         except Exception:
             pass
 
         # Filename where object was defined
         binary_file = False
         fname = find_file(obj)
         if fname is None:
             # if anything goes wrong, we don't want to show source, so it's as
             # if the file was binary
             binary_file = True
         else:
             if fname.endswith(('.so', '.dll', '.pyd')):
                 binary_file = True
             elif fname.endswith('<string>'):
                 fname = 'Dynamically generated function. No source code available.'
             out['file'] = compress_user(fname)
 
         # Original source code for a callable, class or property.
         if detail_level:
             # Flush the source cache because inspect can return out-of-date
             # source
             linecache.checkcache()
             try:
                 if isinstance(obj, property) or not binary_file:
                     src = getsource(obj, oname)
                     if src is not None:
                         src = src.rstrip()
                     out['source'] = src
 
             except Exception:
                 pass
 
         # Add docstring only if no source is to be shown (avoid repetitions).
         if ds and not self._source_contains_docstring(out.get('source'), ds):
             out['docstring'] = ds
 
         # Constructor docstring for classes
         if inspect.isclass(obj):
             out['isclass'] = True
 
             # get the init signature:
             try:
                 init_def = self._getdef(obj, oname)
             except AttributeError:
                 init_def = None
 
             # get the __init__ docstring
             try:
                 obj_init = obj.__init__
             except AttributeError:
                 init_ds = None
             else:
                 if init_def is None:
                     # Get signature from init if top-level sig failed.
                     # Can happen for built-in types (list, etc.).
                     try:
                         init_def = self._getdef(obj_init, oname)
                     except AttributeError:
                         pass
                 init_ds = getdoc(obj_init)
                 # Skip Python's auto-generated docstrings
                 if init_ds == _object_init_docstring:
                     init_ds = None
 
             if init_def:
                 out['init_definition'] = init_def
 
             if init_ds:
                 out['init_docstring'] = init_ds
 
             names = [sub.__name__ for sub in type.__subclasses__(obj)]
             if len(names) < 10:
                 all_names = ', '.join(names)
             else:
                 all_names = ', '.join(names[:10]+['...'])
             out['subclasses'] = all_names
         # and class docstring for instances:
         else:
             # reconstruct the function definition and print it:
             defln = self._getdef(obj, oname)
             if defln:
                 out['definition'] = defln
 
             # First, check whether the instance docstring is identical to the
             # class one, and print it separately if they don't coincide.  In
             # most cases they will, but it's nice to print all the info for
             # objects which use instance-customized docstrings.
             if ds:
                 try:
                     cls = getattr(obj,'__class__')
                 except:
                     class_ds = None
                 else:
                     class_ds = getdoc(cls)
                 # Skip Python's auto-generated docstrings
                 if class_ds in _builtin_type_docstrings:
                     class_ds = None
                 if class_ds and ds != class_ds:
                     out['class_docstring'] = class_ds
 
             # Next, try to show constructor docstrings
             try:
                 init_ds = getdoc(obj.__init__)
                 # Skip Python's auto-generated docstrings
                 if init_ds == _object_init_docstring:
                     init_ds = None
             except AttributeError:
                 init_ds = None
             if init_ds:
                 out['init_docstring'] = init_ds
 
             # Call form docstring for callable instances
             if safe_hasattr(obj, '__call__') and not is_simple_callable(obj):
                 call_def = self._getdef(obj.__call__, oname)
                 if call_def and (call_def != out.get('definition')):
                     # it may never be the case that call def and definition differ,
                     # but don't include the same signature twice
                     out['call_def'] = call_def
                 call_ds = getdoc(obj.__call__)
                 # Skip Python's auto-generated docstrings
                 if call_ds == _func_call_docstring:
                     call_ds = None
                 if call_ds:
                     out['call_docstring'] = call_ds
 
         return object_info(**out)
 
     @staticmethod
     def _source_contains_docstring(src, doc):
         """
         Check whether the source *src* contains the docstring *doc*.
 
         This is is helper function to skip displaying the docstring if the
         source already contains it, avoiding repetition of information.
         """
         try:
             def_node, = ast.parse(dedent(src)).body
             return ast.get_docstring(def_node) == doc
         except Exception:
             # The source can become invalid or even non-existent (because it
             # is re-fetched from the source file) so the above code fail in
             # arbitrary ways.
             return False
 
     def psearch(self,pattern,ns_table,ns_search=[],
                 ignore_case=False,show_all=False, *, list_types=False):
         """Search namespaces with wildcards for objects.
 
         Arguments:
 
         - pattern: string containing shell-like wildcards to use in namespace
           searches and optionally a type specification to narrow the search to
           objects of that type.
 
         - ns_table: dict of name->namespaces for search.
 
         Optional arguments:
 
           - ns_search: list of namespace names to include in search.
 
           - ignore_case(False): make the search case-insensitive.
 
           - show_all(False): show all names, including those starting with
             underscores.
             
           - list_types(False): list all available object types for object matching.
         """
         #print 'ps pattern:<%r>' % pattern # dbg
 
         # defaults
         type_pattern = 'all'
         filter = ''
 
         # list all object types
         if list_types:
             page.page('\n'.join(sorted(typestr2type)))
             return
 
         cmds = pattern.split()
         len_cmds  =  len(cmds)
         if len_cmds == 1:
             # Only filter pattern given
             filter = cmds[0]
         elif len_cmds == 2:
             # Both filter and type specified
             filter,type_pattern = cmds
         else:
             raise ValueError('invalid argument string for psearch: <%s>' %
                              pattern)
 
         # filter search namespaces
         for name in ns_search:
             if name not in ns_table:
                 raise ValueError('invalid namespace <%s>. Valid names: %s' %
                                  (name,ns_table.keys()))
 
         #print 'type_pattern:',type_pattern # dbg
         search_result, namespaces_seen = set(), set()
         for ns_name in ns_search:
             ns = ns_table[ns_name]
             # Normally, locals and globals are the same, so we just check one.
             if id(ns) in namespaces_seen:
                 continue
             namespaces_seen.add(id(ns))
             tmp_res = list_namespace(ns, type_pattern, filter,
                                     ignore_case=ignore_case, show_all=show_all)
             search_result.update(tmp_res)
 
         page.page('\n'.join(sorted(search_result)))
 
 
 def _render_signature(obj_signature, obj_name) -> str:
     """
     This was mostly taken from inspect.Signature.__str__.
     Look there for the comments.
     The only change is to add linebreaks when this gets too long.
     """
     result = []
     pos_only = False
     kw_only = True
     for param in obj_signature.parameters.values():
         if param.kind == inspect._POSITIONAL_ONLY:
             pos_only = True
         elif pos_only:
             result.append('/')
             pos_only = False
 
         if param.kind == inspect._VAR_POSITIONAL:
             kw_only = False
         elif param.kind == inspect._KEYWORD_ONLY and kw_only:
             result.append('*')
             kw_only = False
 
         result.append(str(param))
 
     if pos_only:
         result.append('/')
 
     # add up name, parameters, braces (2), and commas
     if len(obj_name) + sum(len(r) + 2 for r in result) > 75:
         # This doesn’t fit behind “Signature: ” in an inspect window.
         rendered = '{}(\n{})'.format(obj_name, ''.join(
             '    {},\n'.format(r) for r in result)
         )
     else:
         rendered = '{}({})'.format(obj_name, ', '.join(result))
 
     if obj_signature.return_annotation is not inspect._empty:
         anno = inspect.formatannotation(obj_signature.return_annotation)
         rendered += ' -> {}'.format(anno)
 
     return rendered

IPython/lib/deepreload.py

@@ -1,347 +1,341 b''
 # -*- coding: utf-8 -*-
 """
 Provides a reload() function that acts recursively.
 
 Python's normal :func:`python:reload` function only reloads the module that it's
 passed. The :func:`reload` function in this module also reloads everything
 imported from that module, which is useful when you're changing files deep
 inside a package.
 
-To use this as your default reload function, type this for Python 2::
-
-    import __builtin__
-    from IPython.lib import deepreload
-    __builtin__.reload = deepreload.reload
-
-Or this for Python 3::
+To use this as your default reload function, type this::
 
     import builtins
     from IPython.lib import deepreload
     builtins.reload = deepreload.reload
 
 A reference to the original :func:`python:reload` is stored in this module as
 :data:`original_reload`, so you can restore it later.
 
 This code is almost entirely based on knee.py, which is a Python
 re-implementation of hierarchical module import.
 """
 #*****************************************************************************
 #       Copyright (C) 2001 Nathaniel Gray <[e-mail]>
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #*****************************************************************************
 
 import builtins as builtin_mod
 from contextlib import contextmanager
 import imp
 import sys
 
 from types import ModuleType
 from warnings import warn
 import types
 
 original_import = builtin_mod.__import__
 
 @contextmanager
 def replace_import_hook(new_import):
     saved_import = builtin_mod.__import__
     builtin_mod.__import__ = new_import
     try:
         yield
     finally:
         builtin_mod.__import__ = saved_import
 
 def get_parent(globals, level):
     """
     parent, name = get_parent(globals, level)
 
     Return the package that an import is being performed in.  If globals comes
     from the module foo.bar.bat (not itself a package), this returns the
     sys.modules entry for foo.bar.  If globals is from a package's __init__.py,
     the package's entry in sys.modules is returned.
 
     If globals doesn't come from a package or a module in a package, or a
     corresponding entry is not found in sys.modules, None is returned.
     """
     orig_level = level
 
     if not level or not isinstance(globals, dict):
         return None, ''
 
     pkgname = globals.get('__package__', None)
 
     if pkgname is not None:
         # __package__ is set, so use it
         if not hasattr(pkgname, 'rindex'):
             raise ValueError('__package__ set to non-string')
         if len(pkgname) == 0:
             if level > 0:
                 raise ValueError('Attempted relative import in non-package')
             return None, ''
         name = pkgname
     else:
         # __package__ not set, so figure it out and set it
         if '__name__' not in globals:
             return None, ''
         modname = globals['__name__']
 
         if '__path__' in globals:
             # __path__ is set, so modname is already the package name
             globals['__package__'] = name = modname
         else:
             # Normal module, so work out the package name if any
             lastdot = modname.rfind('.')
             if lastdot < 0 < level:
                 raise ValueError("Attempted relative import in non-package")
             if lastdot < 0:
                 globals['__package__'] = None
                 return None, ''
             globals['__package__'] = name = modname[:lastdot]
 
     dot = len(name)
     for x in range(level, 1, -1):
         try:
             dot = name.rindex('.', 0, dot)
         except ValueError:
             raise ValueError("attempted relative import beyond top-level "
                              "package")
     name = name[:dot]
 
     try:
         parent = sys.modules[name]
     except:
         if orig_level < 1:
             warn("Parent module '%.200s' not found while handling absolute "
                  "import" % name)
             parent = None
         else:
             raise SystemError("Parent module '%.200s' not loaded, cannot "
                               "perform relative import" % name)
 
     # We expect, but can't guarantee, if parent != None, that:
     # - parent.__name__ == name
     # - parent.__dict__ is globals
     # If this is violated...  Who cares?
     return parent, name
 
 def load_next(mod, altmod, name, buf):
     """
     mod, name, buf = load_next(mod, altmod, name, buf)
 
     altmod is either None or same as mod
     """
 
     if len(name) == 0:
         # completely empty module name should only happen in
         # 'from . import' (or '__import__("")')
         return mod, None, buf
 
     dot = name.find('.')
     if dot == 0:
         raise ValueError('Empty module name')
 
     if dot < 0:
         subname = name
         next = None
     else:
         subname = name[:dot]
         next = name[dot+1:]
 
     if buf != '':
         buf += '.'
     buf += subname
 
     result = import_submodule(mod, subname, buf)
     if result is None and mod != altmod:
         result = import_submodule(altmod, subname, subname)
         if result is not None:
             buf = subname
 
     if result is None:
         raise ImportError("No module named %.200s" % name)
 
     return result, next, buf
 
 
 # Need to keep track of what we've already reloaded to prevent cyclic evil
 found_now = {}
 
 def import_submodule(mod, subname, fullname):
     """m = import_submodule(mod, subname, fullname)"""
     # Require:
     # if mod == None: subname == fullname
     # else: mod.__name__ + "." + subname == fullname
 
     global found_now
     if fullname in found_now and fullname in sys.modules:
         m = sys.modules[fullname]
     else:
         print('Reloading', fullname)
         found_now[fullname] = 1
         oldm = sys.modules.get(fullname, None)
 
         if mod is None:
             path = None
         elif hasattr(mod, '__path__'):
             path = mod.__path__
         else:
             return None
 
         try:
             # This appears to be necessary on Python 3, because imp.find_module()
             # tries to import standard libraries (like io) itself, and we don't
             # want them to be processed by our deep_import_hook.
             with replace_import_hook(original_import):
                 fp, filename, stuff = imp.find_module(subname, path)
         except ImportError:
             return None
 
         try:
             m = imp.load_module(fullname, fp, filename, stuff)
         except:
             # load_module probably removed name from modules because of
             # the error.  Put back the original module object.
             if oldm:
                 sys.modules[fullname] = oldm
             raise
         finally:
             if fp: fp.close()
 
         add_submodule(mod, m, fullname, subname)
 
     return m
 
 def add_submodule(mod, submod, fullname, subname):
     """mod.{subname} = submod"""
     if mod is None:
         return #Nothing to do here.
 
     if submod is None:
         submod = sys.modules[fullname]
 
     setattr(mod, subname, submod)
 
     return
 
 def ensure_fromlist(mod, fromlist, buf, recursive):
     """Handle 'from module import a, b, c' imports."""
     if not hasattr(mod, '__path__'):
         return
     for item in fromlist:
         if not hasattr(item, 'rindex'):
             raise TypeError("Item in ``from list'' not a string")
         if item == '*':
             if recursive:
                 continue # avoid endless recursion
             try:
                 all = mod.__all__
             except AttributeError:
                 pass
             else:
                 ret = ensure_fromlist(mod, all, buf, 1)
                 if not ret:
                     return 0
         elif not hasattr(mod, item):
             import_submodule(mod, item, buf + '.' + item)
 
 def deep_import_hook(name, globals=None, locals=None, fromlist=None, level=-1):
     """Replacement for __import__()"""
     parent, buf = get_parent(globals, level)
 
     head, name, buf = load_next(parent, None if level < 0 else parent, name, buf)
 
     tail = head
     while name:
         tail, name, buf = load_next(tail, tail, name, buf)
 
     # If tail is None, both get_parent and load_next found
     # an empty module name: someone called __import__("") or
     # doctored faulty bytecode
     if tail is None:
         raise ValueError('Empty module name')
 
     if not fromlist:
         return head
 
     ensure_fromlist(tail, fromlist, buf, 0)
     return tail
 
 modules_reloading = {}
 
 def deep_reload_hook(m):
     """Replacement for reload()."""
     # Hardcode this one  as it would raise a NotImplementedError from the
     # bowels of Python and screw up the import machinery after.
     # unlike other imports the `exclude` list already in place is not enough.
 
     if m is types:
         return m
     if not isinstance(m, ModuleType):
         raise TypeError("reload() argument must be module")
 
     name = m.__name__
 
     if name not in sys.modules:
         raise ImportError("reload(): module %.200s not in sys.modules" % name)
 
     global modules_reloading
     try:
         return modules_reloading[name]
     except:
         modules_reloading[name] = m
 
     dot = name.rfind('.')
     if dot < 0:
         subname = name
         path = None
     else:
         try:
             parent = sys.modules[name[:dot]]
         except KeyError:
             modules_reloading.clear()
             raise ImportError("reload(): parent %.200s not in sys.modules" % name[:dot])
         subname = name[dot+1:]
         path = getattr(parent, "__path__", None)
 
     try:
         # This appears to be necessary on Python 3, because imp.find_module()
         # tries to import standard libraries (like io) itself, and we don't
         # want them to be processed by our deep_import_hook.
         with replace_import_hook(original_import):
             fp, filename, stuff  = imp.find_module(subname, path)
     finally:
         modules_reloading.clear()
 
     try:
         newm = imp.load_module(name, fp, filename, stuff)
     except:
          # load_module probably removed name from modules because of
          # the error.  Put back the original module object.
         sys.modules[name] = m
         raise
     finally:
         if fp: fp.close()
 
     modules_reloading.clear()
     return newm
 
 # Save the original hooks
 original_reload = imp.reload
 
 # Replacement for reload()
 def reload(module, exclude=('sys', 'os.path', 'builtins', '__main__',
                             'numpy', 'numpy._globals')):
     """Recursively reload all modules used in the given module.  Optionally
     takes a list of modules to exclude from reloading.  The default exclude
     list contains sys, __main__, and __builtin__, to prevent, e.g., resetting
     display, exception, and io hooks.
     """
     global found_now
     for i in exclude:
         found_now[i] = 1
     try:
         with replace_import_hook(deep_import_hook):
             return deep_reload_hook(module)
     finally:
         found_now = {}

IPython/lib/lexers.py

@@ -1,532 +1,532 b''
 # -*- coding: utf-8 -*-
 """
 Defines a variety of Pygments lexers for highlighting IPython code.
 
 This includes:
 
     IPythonLexer, IPython3Lexer
         Lexers for pure IPython (python + magic/shell commands)
 
     IPythonPartialTracebackLexer, IPythonTracebackLexer
         Supports 2.x and 3.x via keyword `python3`.  The partial traceback
         lexer reads everything but the Python code appearing in a traceback.
         The full lexer combines the partial lexer with an IPython lexer.
 
     IPythonConsoleLexer
         A lexer for IPython console sessions, with support for tracebacks.
 
     IPyLexer
         A friendly lexer which examines the first line of text and from it,
         decides whether to use an IPython lexer or an IPython console lexer.
         This is probably the only lexer that needs to be explicitly added
         to Pygments.
 
 """
 #-----------------------------------------------------------------------------
 # Copyright (c) 2013, the IPython Development Team.
 #
 # Distributed under the terms of the Modified BSD License.
 #
 # The full license is in the file COPYING.txt, distributed with this software.
 #-----------------------------------------------------------------------------
 
 # Standard library
 import re
 
 # Third party
 from pygments.lexers import (
     BashLexer, HtmlLexer, JavascriptLexer, RubyLexer, PerlLexer, PythonLexer,
     Python3Lexer, TexLexer)
 from pygments.lexer import (
     Lexer, DelegatingLexer, RegexLexer, do_insertions, bygroups, using,
 )
 from pygments.token import (
     Generic, Keyword, Literal, Name, Operator, Other, Text, Error,
 )
 from pygments.util import get_bool_opt
 
 # Local
 
 line_re = re.compile('.*?\n')
 
 __all__ = ['build_ipy_lexer', 'IPython3Lexer', 'IPythonLexer',
            'IPythonPartialTracebackLexer', 'IPythonTracebackLexer',
            'IPythonConsoleLexer', 'IPyLexer']
 
 
 def build_ipy_lexer(python3):
     """Builds IPython lexers depending on the value of `python3`.
 
     The lexer inherits from an appropriate Python lexer and then adds
     information about IPython specific keywords (i.e. magic commands,
     shell commands, etc.)
 
     Parameters
     ----------
     python3 : bool
         If `True`, then build an IPython lexer from a Python 3 lexer.
 
     """
     # It would be nice to have a single IPython lexer class which takes
     # a boolean `python3`.  But since there are two Python lexer classes,
     # we will also have two IPython lexer classes.
     if python3:
         PyLexer = Python3Lexer
         name = 'IPython3'
         aliases = ['ipython3']
         doc = """IPython3 Lexer"""
     else:
         PyLexer = PythonLexer
         name = 'IPython'
         aliases = ['ipython2', 'ipython']
         doc = """IPython Lexer"""
 
     ipython_tokens = [
        (r'(?s)(\s*)(%%capture)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%debug)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?is)(\s*)(%%html)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(HtmlLexer))),
         (r'(?s)(\s*)(%%javascript)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(JavascriptLexer))),
         (r'(?s)(\s*)(%%js)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(JavascriptLexer))),
         (r'(?s)(\s*)(%%latex)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(TexLexer))),
         (r'(?s)(\s*)(%%perl)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PerlLexer))),
         (r'(?s)(\s*)(%%prun)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%pypy)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%python)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%python2)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PythonLexer))),
         (r'(?s)(\s*)(%%python3)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(Python3Lexer))),
         (r'(?s)(\s*)(%%ruby)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(RubyLexer))),
         (r'(?s)(\s*)(%%time)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%timeit)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%writefile)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r'(?s)(\s*)(%%file)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(PyLexer))),
         (r"(?s)(\s*)(%%)(\w+)(.*)", bygroups(Text, Operator, Keyword, Text)),
         (r'(?s)(^\s*)(%%!)([^\n]*\n)(.*)', bygroups(Text, Operator, Text, using(BashLexer))),
         (r"(%%?)(\w+)(\?\??)$",  bygroups(Operator, Keyword, Operator)),
         (r"\b(\?\??)(\s*)$",  bygroups(Operator, Text)),
         (r'(%)(sx|sc|system)(.*)(\n)', bygroups(Operator, Keyword,
                                                 using(BashLexer), Text)),
         (r'(%)(\w+)(.*\n)', bygroups(Operator, Keyword, Text)),
         (r'^(!!)(.+)(\n)', bygroups(Operator, using(BashLexer), Text)),
         (r'(!)(?!=)(.+)(\n)', bygroups(Operator, using(BashLexer), Text)),
         (r'^(\s*)(\?\??)(\s*%{0,2}[\w\.\*]*)', bygroups(Text, Operator, Text)),
         (r'(\s*%{0,2}[\w\.\*]*)(\?\??)(\s*)$', bygroups(Text, Operator, Text)),
     ]
 
     tokens = PyLexer.tokens.copy()
     tokens['root'] = ipython_tokens + tokens['root']
 
     attrs = {'name': name, 'aliases': aliases, 'filenames': [],
              '__doc__': doc, 'tokens': tokens}
 
     return type(name, (PyLexer,), attrs)
 
 
 IPython3Lexer = build_ipy_lexer(python3=True)
 IPythonLexer = build_ipy_lexer(python3=False)
 
 
 class IPythonPartialTracebackLexer(RegexLexer):
     """
     Partial lexer for IPython tracebacks.
 
-    Handles all the non-python output. This works for both Python 2.x and 3.x.
+    Handles all the non-python output.
 
     """
     name = 'IPython Partial Traceback'
 
     tokens = {
         'root': [
             # Tracebacks for syntax errors have a different style.
             # For both types of tracebacks, we mark the first line with
             # Generic.Traceback.  For syntax errors, we mark the filename
             # as we mark the filenames for non-syntax tracebacks.
             #
             # These two regexps define how IPythonConsoleLexer finds a
             # traceback.
             #
             ## Non-syntax traceback
             (r'^(\^C)?(-+\n)', bygroups(Error, Generic.Traceback)),
             ## Syntax traceback
             (r'^(  File)(.*)(, line )(\d+\n)',
              bygroups(Generic.Traceback, Name.Namespace,
                       Generic.Traceback, Literal.Number.Integer)),
 
             # (Exception Identifier)(Whitespace)(Traceback Message)
             (r'(?u)(^[^\d\W]\w*)(\s*)(Traceback.*?\n)',
              bygroups(Name.Exception, Generic.Whitespace, Text)),
             # (Module/Filename)(Text)(Callee)(Function Signature)
             # Better options for callee and function signature?
             (r'(.*)( in )(.*)(\(.*\)\n)',
              bygroups(Name.Namespace, Text, Name.Entity, Name.Tag)),
             # Regular line: (Whitespace)(Line Number)(Python Code)
             (r'(\s*?)(\d+)(.*?\n)',
              bygroups(Generic.Whitespace, Literal.Number.Integer, Other)),
             # Emphasized line: (Arrow)(Line Number)(Python Code)
             # Using Exception token so arrow color matches the Exception.
             (r'(-*>?\s?)(\d+)(.*?\n)',
              bygroups(Name.Exception, Literal.Number.Integer, Other)),
             # (Exception Identifier)(Message)
             (r'(?u)(^[^\d\W]\w*)(:.*?\n)',
              bygroups(Name.Exception, Text)),
             # Tag everything else as Other, will be handled later.
             (r'.*\n', Other),
         ],
     }
 
 
 class IPythonTracebackLexer(DelegatingLexer):
     """
     IPython traceback lexer.
 
     For doctests, the tracebacks can be snipped as much as desired with the
     exception to the lines that designate a traceback. For non-syntax error
     tracebacks, this is the line of hyphens. For syntax error tracebacks,
     this is the line which lists the File and line number.
 
     """
     # The lexer inherits from DelegatingLexer.  The "root" lexer is an
     # appropriate IPython lexer, which depends on the value of the boolean
     # `python3`.  First, we parse with the partial IPython traceback lexer.
     # Then, any code marked with the "Other" token is delegated to the root
     # lexer.
     #
     name = 'IPython Traceback'
     aliases = ['ipythontb']
 
     def __init__(self, **options):
         self.python3 = get_bool_opt(options, 'python3', False)
         if self.python3:
             self.aliases = ['ipython3tb']
         else:
             self.aliases = ['ipython2tb', 'ipythontb']
 
         if self.python3:
             IPyLexer = IPython3Lexer
         else:
             IPyLexer = IPythonLexer
 
         DelegatingLexer.__init__(self, IPyLexer,
                                  IPythonPartialTracebackLexer, **options)
 
 class IPythonConsoleLexer(Lexer):
     """
     An IPython console lexer for IPython code-blocks and doctests, such as:
 
     .. code-block:: rst
 
         .. code-block:: ipythonconsole
 
             In [1]: a = 'foo'
 
             In [2]: a
             Out[2]: 'foo'
 
             In [3]: print a
             foo
 
             In [4]: 1 / 0
 
 
     Support is also provided for IPython exceptions:
 
     .. code-block:: rst
 
         .. code-block:: ipythonconsole
 
             In [1]: raise Exception
 
             ---------------------------------------------------------------------------
             Exception                                 Traceback (most recent call last)
             <ipython-input-1-fca2ab0ca76b> in <module>
             ----> 1 raise Exception
 
             Exception:
 
     """
     name = 'IPython console session'
     aliases = ['ipythonconsole']
     mimetypes = ['text/x-ipython-console']
 
     # The regexps used to determine what is input and what is output.
     # The default prompts for IPython are:
     #
     #    in           = 'In [#]: '
     #    continuation = '   .D.: '
     #    template     = 'Out[#]: '
     #
     # Where '#' is the 'prompt number' or 'execution count' and 'D' 
     # D is a number of dots  matching the width of the execution count 
     #
     in1_regex = r'In \[[0-9]+\]: '
     in2_regex = r'   \.\.+\.: '
     out_regex = r'Out\[[0-9]+\]: '
 
     #: The regex to determine when a traceback starts.
     ipytb_start = re.compile(r'^(\^C)?(-+\n)|^(  File)(.*)(, line )(\d+\n)')
 
     def __init__(self, **options):
         """Initialize the IPython console lexer.
 
         Parameters
         ----------
         python3 : bool
             If `True`, then the console inputs are parsed using a Python 3
             lexer. Otherwise, they are parsed using a Python 2 lexer.
         in1_regex : RegexObject
             The compiled regular expression used to detect the start
             of inputs. Although the IPython configuration setting may have a
             trailing whitespace, do not include it in the regex. If `None`,
             then the default input prompt is assumed.
         in2_regex : RegexObject
             The compiled regular expression used to detect the continuation
             of inputs. Although the IPython configuration setting may have a
             trailing whitespace, do not include it in the regex. If `None`,
             then the default input prompt is assumed.
         out_regex : RegexObject
             The compiled regular expression used to detect outputs. If `None`,
             then the default output prompt is assumed.
 
         """
         self.python3 = get_bool_opt(options, 'python3', False)
         if self.python3:
             self.aliases = ['ipython3console']
         else:
             self.aliases = ['ipython2console', 'ipythonconsole']
 
         in1_regex = options.get('in1_regex', self.in1_regex)
         in2_regex = options.get('in2_regex', self.in2_regex)
         out_regex = options.get('out_regex', self.out_regex)
 
         # So that we can work with input and output prompts which have been
         # rstrip'd (possibly by editors) we also need rstrip'd variants. If
         # we do not do this, then such prompts will be tagged as 'output'.
         # The reason can't just use the rstrip'd variants instead is because
         # we want any whitespace associated with the prompt to be inserted
         # with the token. This allows formatted code to be modified so as hide
         # the appearance of prompts, with the whitespace included. One example
         # use of this is in copybutton.js from the standard lib Python docs.
         in1_regex_rstrip = in1_regex.rstrip() + '\n'
         in2_regex_rstrip = in2_regex.rstrip() + '\n'
         out_regex_rstrip = out_regex.rstrip() + '\n'
 
         # Compile and save them all.
         attrs = ['in1_regex', 'in2_regex', 'out_regex',
                  'in1_regex_rstrip', 'in2_regex_rstrip', 'out_regex_rstrip']
         for attr in attrs:
             self.__setattr__(attr, re.compile(locals()[attr]))
 
         Lexer.__init__(self, **options)
 
         if self.python3:
             pylexer = IPython3Lexer
             tblexer = IPythonTracebackLexer
         else:
             pylexer = IPythonLexer
             tblexer = IPythonTracebackLexer
 
         self.pylexer = pylexer(**options)
         self.tblexer = tblexer(**options)
 
         self.reset()
 
     def reset(self):
         self.mode = 'output'
         self.index = 0
         self.buffer = u''
         self.insertions = []
 
     def buffered_tokens(self):
         """
         Generator of unprocessed tokens after doing insertions and before
         changing to a new state.
 
         """
         if self.mode == 'output':
             tokens = [(0, Generic.Output, self.buffer)]
         elif self.mode == 'input':
             tokens = self.pylexer.get_tokens_unprocessed(self.buffer)
         else: # traceback
             tokens = self.tblexer.get_tokens_unprocessed(self.buffer)
 
         for i, t, v in do_insertions(self.insertions, tokens):
             # All token indexes are relative to the buffer.
             yield self.index + i, t, v
 
         # Clear it all
         self.index += len(self.buffer)
         self.buffer = u''
         self.insertions = []
 
     def get_mci(self, line):
         """
         Parses the line and returns a 3-tuple: (mode, code, insertion).
 
         `mode` is the next mode (or state) of the lexer, and is always equal
         to 'input', 'output', or 'tb'.
 
         `code` is a portion of the line that should be added to the buffer
         corresponding to the next mode and eventually lexed by another lexer.
         For example, `code` could be Python code if `mode` were 'input'.
 
         `insertion` is a 3-tuple (index, token, text) representing an
         unprocessed "token" that will be inserted into the stream of tokens
         that are created from the buffer once we change modes. This is usually
         the input or output prompt.
 
         In general, the next mode depends on current mode and on the contents
         of `line`.
 
         """
         # To reduce the number of regex match checks, we have multiple
         # 'if' blocks instead of 'if-elif' blocks.
 
         # Check for possible end of input
         in2_match = self.in2_regex.match(line)
         in2_match_rstrip = self.in2_regex_rstrip.match(line)
         if (in2_match and in2_match.group().rstrip() == line.rstrip()) or \
            in2_match_rstrip:
             end_input = True
         else:
             end_input = False
         if end_input and self.mode != 'tb':
             # Only look for an end of input when not in tb mode.
             # An ellipsis could appear within the traceback.
             mode = 'output'
             code = u''
             insertion = (0, Generic.Prompt, line)
             return mode, code, insertion
 
         # Check for output prompt
         out_match = self.out_regex.match(line)
         out_match_rstrip = self.out_regex_rstrip.match(line)
         if out_match or out_match_rstrip:
             mode = 'output'
             if out_match:
                 idx = out_match.end()
             else:
                 idx = out_match_rstrip.end()
             code = line[idx:]
             # Use the 'heading' token for output.  We cannot use Generic.Error
             # since it would conflict with exceptions.
             insertion = (0, Generic.Heading, line[:idx])
             return mode, code, insertion
 
 
         # Check for input or continuation prompt (non stripped version)
         in1_match = self.in1_regex.match(line)
         if in1_match or (in2_match and self.mode != 'tb'):
             # New input or when not in tb, continued input.
             # We do not check for continued input when in tb since it is
             # allowable to replace a long stack with an ellipsis.
             mode = 'input'
             if in1_match:
                 idx = in1_match.end()
             else: # in2_match
                 idx = in2_match.end()
             code = line[idx:]
             insertion = (0, Generic.Prompt, line[:idx])
             return mode, code, insertion
 
         # Check for input or continuation prompt (stripped version)
         in1_match_rstrip = self.in1_regex_rstrip.match(line)
         if in1_match_rstrip or (in2_match_rstrip and self.mode != 'tb'):
             # New input or when not in tb, continued input.
             # We do not check for continued input when in tb since it is
             # allowable to replace a long stack with an ellipsis.
             mode = 'input'
             if in1_match_rstrip:
                 idx = in1_match_rstrip.end()
             else: # in2_match
                 idx = in2_match_rstrip.end()
             code = line[idx:]
             insertion = (0, Generic.Prompt, line[:idx])
             return mode, code, insertion
 
         # Check for traceback
         if self.ipytb_start.match(line):
             mode = 'tb'
             code = line
             insertion = None
             return mode, code, insertion
 
         # All other stuff...
         if self.mode in ('input', 'output'):
             # We assume all other text is output. Multiline input that
             # does not use the continuation marker cannot be detected.
             # For example, the 3 in the following is clearly output:
             #
             #    In [1]: print 3
             #    3
             #
             # But the following second line is part of the input:
             #
             #    In [2]: while True:
             #        print True
             #
             # In both cases, the 2nd line will be 'output'.
             #
             mode = 'output'
         else:
             mode = 'tb'
 
         code = line
         insertion = None
 
         return mode, code, insertion
 
     def get_tokens_unprocessed(self, text):
         self.reset()
         for match in line_re.finditer(text):
             line = match.group()
             mode, code, insertion = self.get_mci(line)
 
             if mode != self.mode:
                 # Yield buffered tokens before transitioning to new mode.
                 for token in self.buffered_tokens():
                     yield token
                 self.mode = mode
 
             if insertion:
                 self.insertions.append((len(self.buffer), [insertion]))
             self.buffer += code
 
         for token in self.buffered_tokens():
             yield token
 
 class IPyLexer(Lexer):
     r"""
     Primary lexer for all IPython-like code.
 
     This is a simple helper lexer.  If the first line of the text begins with
     "In \[[0-9]+\]:", then the entire text is parsed with an IPython console
     lexer. If not, then the entire text is parsed with an IPython lexer.
 
     The goal is to reduce the number of lexers that are registered
     with Pygments.
 
     """
     name = 'IPy session'
     aliases = ['ipy']
 
     def __init__(self, **options):
         self.python3 = get_bool_opt(options, 'python3', False)
         if self.python3:
             self.aliases = ['ipy3']
         else:
             self.aliases = ['ipy2', 'ipy']
 
         Lexer.__init__(self, **options)
 
         self.IPythonLexer = IPythonLexer(**options)
         self.IPythonConsoleLexer = IPythonConsoleLexer(**options)
 
     def get_tokens_unprocessed(self, text):
         # Search for the input prompt anywhere...this allows code blocks to
         # begin with comments as well.
         if re.match(r'.*(In \[[0-9]+\]:)', text.strip(), re.DOTALL):
             lex = self.IPythonConsoleLexer
         else:
             lex = self.IPythonLexer
         for token in lex.get_tokens_unprocessed(text):
             yield token
 

IPython/lib/pretty.py

@@ -1,871 +1,860 b''
 # -*- coding: utf-8 -*-
 """
 Python advanced pretty printer.  This pretty printer is intended to
 replace the old `pprint` python module which does not allow developers
 to provide their own pretty print callbacks.
 
 This module is based on ruby's `prettyprint.rb` library by `Tanaka Akira`.
 
 
 Example Usage
 -------------
 
 To directly print the representation of an object use `pprint`::
 
     from pretty import pprint
     pprint(complex_object)
 
 To get a string of the output use `pretty`::
 
     from pretty import pretty
     string = pretty(complex_object)
 
 
 Extending
 ---------
 
 The pretty library allows developers to add pretty printing rules for their
 own objects.  This process is straightforward.  All you have to do is to
 add a `_repr_pretty_` method to your object and call the methods on the
 pretty printer passed::
 
     class MyObject(object):
 
         def _repr_pretty_(self, p, cycle):
             ...
 
 Here is an example implementation of a `_repr_pretty_` method for a list
 subclass::
 
     class MyList(list):
 
         def _repr_pretty_(self, p, cycle):
             if cycle:
                 p.text('MyList(...)')
             else:
                 with p.group(8, 'MyList([', '])'):
                     for idx, item in enumerate(self):
                         if idx:
                             p.text(',')
                             p.breakable()
                         p.pretty(item)
 
 The `cycle` parameter is `True` if pretty detected a cycle.  You *have* to
 react to that or the result is an infinite loop.  `p.text()` just adds
 non breaking text to the output, `p.breakable()` either adds a whitespace
 or breaks here.  If you pass it an argument it's used instead of the
 default space.  `p.pretty` prettyprints another object using the pretty print
 method.
 
 The first parameter to the `group` function specifies the extra indentation
 of the next line.  In this example the next item will either be on the same
 line (if the items are short enough) or aligned with the right edge of the
 opening bracket of `MyList`.
 
 If you just want to indent something you can use the group function
 without open / close parameters.  You can also use this code::
 
     with p.indent(2):
         ...
 
 Inheritance diagram:
 
 .. inheritance-diagram:: IPython.lib.pretty
    :parts: 3
 
 :copyright: 2007 by Armin Ronacher.
             Portions (c) 2009 by Robert Kern.
 :license: BSD License.
 """
 
 from contextlib import contextmanager
 import datetime
 import os
 import re
 import sys
 import types
 from collections import deque
 from inspect import signature
 from io import StringIO
 from warnings import warn
 
 from IPython.utils.decorators import undoc
 from IPython.utils.py3compat import PYPY
 
 __all__ = ['pretty', 'pprint', 'PrettyPrinter', 'RepresentationPrinter',
     'for_type', 'for_type_by_name']
 
 
 MAX_SEQ_LENGTH = 1000
 _re_pattern_type = type(re.compile(''))
 
 def _safe_getattr(obj, attr, default=None):
     """Safe version of getattr.
 
     Same as getattr, but will return ``default`` on any Exception,
     rather than raising.
     """
     try:
         return getattr(obj, attr, default)
     except Exception:
         return default
 
 @undoc
 class CUnicodeIO(StringIO):
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         warn(("CUnicodeIO is deprecated since IPython 6.0. "
               "Please use io.StringIO instead."),
              DeprecationWarning, stacklevel=2)
 
 def _sorted_for_pprint(items):
     """
     Sort the given items for pretty printing. Since some predictable
     sorting is better than no sorting at all, we sort on the string
     representation if normal sorting fails.
     """
     items = list(items)
     try:
         return sorted(items)
     except Exception:
         try:
             return sorted(items, key=str)
         except Exception:
             return items
 
 def pretty(obj, verbose=False, max_width=79, newline='\n', max_seq_length=MAX_SEQ_LENGTH):
     """
     Pretty print the object's representation.
     """
     stream = StringIO()
     printer = RepresentationPrinter(stream, verbose, max_width, newline, max_seq_length=max_seq_length)
     printer.pretty(obj)
     printer.flush()
     return stream.getvalue()
 
 
 def pprint(obj, verbose=False, max_width=79, newline='\n', max_seq_length=MAX_SEQ_LENGTH):
     """
     Like `pretty` but print to stdout.
     """
     printer = RepresentationPrinter(sys.stdout, verbose, max_width, newline, max_seq_length=max_seq_length)
     printer.pretty(obj)
     printer.flush()
     sys.stdout.write(newline)
     sys.stdout.flush()
 
 class _PrettyPrinterBase(object):
 
     @contextmanager
     def indent(self, indent):
         """with statement support for indenting/dedenting."""
         self.indentation += indent
         try:
             yield
         finally:
             self.indentation -= indent
 
     @contextmanager
     def group(self, indent=0, open='', close=''):
         """like begin_group / end_group but for the with statement."""
         self.begin_group(indent, open)
         try:
             yield
         finally:
             self.end_group(indent, close)
 
 class PrettyPrinter(_PrettyPrinterBase):
     """
     Baseclass for the `RepresentationPrinter` prettyprinter that is used to
     generate pretty reprs of objects.  Contrary to the `RepresentationPrinter`
     this printer knows nothing about the default pprinters or the `_repr_pretty_`
     callback method.
     """
 
     def __init__(self, output, max_width=79, newline='\n', max_seq_length=MAX_SEQ_LENGTH):
         self.output = output
         self.max_width = max_width
         self.newline = newline
         self.max_seq_length = max_seq_length
         self.output_width = 0
         self.buffer_width = 0
         self.buffer = deque()
 
         root_group = Group(0)
         self.group_stack = [root_group]
         self.group_queue = GroupQueue(root_group)
         self.indentation = 0
 
     def _break_one_group(self, group):
         while group.breakables:
             x = self.buffer.popleft()
             self.output_width = x.output(self.output, self.output_width)
             self.buffer_width -= x.width
         while self.buffer and isinstance(self.buffer[0], Text):
             x = self.buffer.popleft()
             self.output_width = x.output(self.output, self.output_width)
             self.buffer_width -= x.width
 
     def _break_outer_groups(self):
         while self.max_width < self.output_width + self.buffer_width:
             group = self.group_queue.deq()
             if not group:
                 return
             self._break_one_group(group)
 
     def text(self, obj):
         """Add literal text to the output."""
         width = len(obj)
         if self.buffer:
             text = self.buffer[-1]
             if not isinstance(text, Text):
                 text = Text()
                 self.buffer.append(text)
             text.add(obj, width)
             self.buffer_width += width
             self._break_outer_groups()
         else:
             self.output.write(obj)
             self.output_width += width
 
     def breakable(self, sep=' '):
         """
         Add a breakable separator to the output.  This does not mean that it
         will automatically break here.  If no breaking on this position takes
         place the `sep` is inserted which default to one space.
         """
         width = len(sep)
         group = self.group_stack[-1]
         if group.want_break:
             self.flush()
             self.output.write(self.newline)
             self.output.write(' ' * self.indentation)
             self.output_width = self.indentation
             self.buffer_width = 0
         else:
             self.buffer.append(Breakable(sep, width, self))
             self.buffer_width += width
             self._break_outer_groups()
 
     def break_(self):
         """
         Explicitly insert a newline into the output, maintaining correct indentation.
         """
         group = self.group_queue.deq()
         if group:
             self._break_one_group(group)
         self.flush()
         self.output.write(self.newline)
         self.output.write(' ' * self.indentation)
         self.output_width = self.indentation
         self.buffer_width = 0
 
 
     def begin_group(self, indent=0, open=''):
         """
-        Begin a group.  If you want support for python < 2.5 which doesn't has
-        the with statement this is the preferred way:
-
-            p.begin_group(1, '{')
-            ...
-            p.end_group(1, '}')
-
-        The python 2.5 expression would be this:
-
-            with p.group(1, '{', '}'):
-                ...
-
+        Begin a group.
         The first parameter specifies the indentation for the next line (usually
         the width of the opening text), the second the opening text.  All
         parameters are optional.
         """
         if open:
             self.text(open)
         group = Group(self.group_stack[-1].depth + 1)
         self.group_stack.append(group)
         self.group_queue.enq(group)
         self.indentation += indent
 
     def _enumerate(self, seq):
         """like enumerate, but with an upper limit on the number of items"""
         for idx, x in enumerate(seq):
             if self.max_seq_length and idx >= self.max_seq_length:
                 self.text(',')
                 self.breakable()
                 self.text('...')
                 return
             yield idx, x
 
     def end_group(self, dedent=0, close=''):
         """End a group. See `begin_group` for more details."""
         self.indentation -= dedent
         group = self.group_stack.pop()
         if not group.breakables:
             self.group_queue.remove(group)
         if close:
             self.text(close)
 
     def flush(self):
         """Flush data that is left in the buffer."""
         for data in self.buffer:
             self.output_width += data.output(self.output, self.output_width)
         self.buffer.clear()
         self.buffer_width = 0
 
 
 def _get_mro(obj_class):
     """ Get a reasonable method resolution order of a class and its superclasses
     for both old-style and new-style classes.
     """
     if not hasattr(obj_class, '__mro__'):
         # Old-style class. Mix in object to make a fake new-style class.
         try:
             obj_class = type(obj_class.__name__, (obj_class, object), {})
         except TypeError:
             # Old-style extension type that does not descend from object.
             # FIXME: try to construct a more thorough MRO.
             mro = [obj_class]
         else:
             mro = obj_class.__mro__[1:-1]
     else:
         mro = obj_class.__mro__
     return mro
 
 
 class RepresentationPrinter(PrettyPrinter):
     """
     Special pretty printer that has a `pretty` method that calls the pretty
     printer for a python object.
 
     This class stores processing data on `self` so you must *never* use
     this class in a threaded environment.  Always lock it or reinstanciate
     it.
 
     Instances also have a verbose flag callbacks can access to control their
     output.  For example the default instance repr prints all attributes and
     methods that are not prefixed by an underscore if the printer is in
     verbose mode.
     """
 
     def __init__(self, output, verbose=False, max_width=79, newline='\n',
         singleton_pprinters=None, type_pprinters=None, deferred_pprinters=None,
         max_seq_length=MAX_SEQ_LENGTH):
 
         PrettyPrinter.__init__(self, output, max_width, newline, max_seq_length=max_seq_length)
         self.verbose = verbose
         self.stack = []
         if singleton_pprinters is None:
             singleton_pprinters = _singleton_pprinters.copy()
         self.singleton_pprinters = singleton_pprinters
         if type_pprinters is None:
             type_pprinters = _type_pprinters.copy()
         self.type_pprinters = type_pprinters
         if deferred_pprinters is None:
             deferred_pprinters = _deferred_type_pprinters.copy()
         self.deferred_pprinters = deferred_pprinters
 
     def pretty(self, obj):
         """Pretty print the given object."""
         obj_id = id(obj)
         cycle = obj_id in self.stack
         self.stack.append(obj_id)
         self.begin_group()
         try:
             obj_class = _safe_getattr(obj, '__class__', None) or type(obj)
             # First try to find registered singleton printers for the type.
             try:
                 printer = self.singleton_pprinters[obj_id]
             except (TypeError, KeyError):
                 pass
             else:
                 return printer(obj, self, cycle)
             # Next walk the mro and check for either:
             #   1) a registered printer
             #   2) a _repr_pretty_ method
             for cls in _get_mro(obj_class):
                 if cls in self.type_pprinters:
                     # printer registered in self.type_pprinters
                     return self.type_pprinters[cls](obj, self, cycle)
                 else:
                     # deferred printer
                     printer = self._in_deferred_types(cls)
                     if printer is not None:
                         return printer(obj, self, cycle)
                     else:
                         # Finally look for special method names.
                         # Some objects automatically create any requested
                         # attribute. Try to ignore most of them by checking for
                         # callability.
                         if '_repr_pretty_' in cls.__dict__:
                             meth = cls._repr_pretty_
                             if callable(meth):
                                 return meth(obj, self, cycle)
                         if cls is not object \
                                 and callable(cls.__dict__.get('__repr__')):
                             return _repr_pprint(obj, self, cycle)
 
             return _default_pprint(obj, self, cycle)
         finally:
             self.end_group()
             self.stack.pop()
 
     def _in_deferred_types(self, cls):
         """
         Check if the given class is specified in the deferred type registry.
 
         Returns the printer from the registry if it exists, and None if the
         class is not in the registry. Successful matches will be moved to the
         regular type registry for future use.
         """
         mod = _safe_getattr(cls, '__module__', None)
         name = _safe_getattr(cls, '__name__', None)
         key = (mod, name)
         printer = None
         if key in self.deferred_pprinters:
             # Move the printer over to the regular registry.
             printer = self.deferred_pprinters.pop(key)
             self.type_pprinters[cls] = printer
         return printer
 
 
 class Printable(object):
 
     def output(self, stream, output_width):
         return output_width
 
 
 class Text(Printable):
 
     def __init__(self):
         self.objs = []
         self.width = 0
 
     def output(self, stream, output_width):
         for obj in self.objs:
             stream.write(obj)
         return output_width + self.width
 
     def add(self, obj, width):
         self.objs.append(obj)
         self.width += width
 
 
 class Breakable(Printable):
 
     def __init__(self, seq, width, pretty):
         self.obj = seq
         self.width = width
         self.pretty = pretty
         self.indentation = pretty.indentation
         self.group = pretty.group_stack[-1]
         self.group.breakables.append(self)
 
     def output(self, stream, output_width):
         self.group.breakables.popleft()
         if self.group.want_break:
             stream.write(self.pretty.newline)
             stream.write(' ' * self.indentation)
             return self.indentation
         if not self.group.breakables:
             self.pretty.group_queue.remove(self.group)
         stream.write(self.obj)
         return output_width + self.width
 
 
 class Group(Printable):
 
     def __init__(self, depth):
         self.depth = depth
         self.breakables = deque()
         self.want_break = False
 
 
 class GroupQueue(object):
 
     def __init__(self, *groups):
         self.queue = []
         for group in groups:
             self.enq(group)
 
     def enq(self, group):
         depth = group.depth
         while depth > len(self.queue) - 1:
             self.queue.append([])
         self.queue[depth].append(group)
 
     def deq(self):
         for stack in self.queue:
             for idx, group in enumerate(reversed(stack)):
                 if group.breakables:
                     del stack[idx]
                     group.want_break = True
                     return group
             for group in stack:
                 group.want_break = True
             del stack[:]
 
     def remove(self, group):
         try:
             self.queue[group.depth].remove(group)
         except ValueError:
             pass
 
 
 def _default_pprint(obj, p, cycle):
     """
     The default print function.  Used if an object does not provide one and
     it's none of the builtin objects.
     """
     klass = _safe_getattr(obj, '__class__', None) or type(obj)
     if _safe_getattr(klass, '__repr__', None) is not object.__repr__:
         # A user-provided repr. Find newlines and replace them with p.break_()
         _repr_pprint(obj, p, cycle)
         return
     p.begin_group(1, '<')
     p.pretty(klass)
     p.text(' at 0x%x' % id(obj))
     if cycle:
         p.text(' ...')
     elif p.verbose:
         first = True
         for key in dir(obj):
             if not key.startswith('_'):
                 try:
                     value = getattr(obj, key)
                 except AttributeError:
                     continue
                 if isinstance(value, types.MethodType):
                     continue
                 if not first:
                     p.text(',')
                 p.breakable()
                 p.text(key)
                 p.text('=')
                 step = len(key) + 1
                 p.indentation += step
                 p.pretty(value)
                 p.indentation -= step
                 first = False
     p.end_group(1, '>')
 
 
 def _seq_pprinter_factory(start, end):
     """
     Factory that returns a pprint function useful for sequences.  Used by
     the default pprint for tuples, dicts, and lists.
     """
     def inner(obj, p, cycle):
         if cycle:
             return p.text(start + '...' + end)
         step = len(start)
         p.begin_group(step, start)
         for idx, x in p._enumerate(obj):
             if idx:
                 p.text(',')
                 p.breakable()
             p.pretty(x)
         if len(obj) == 1 and type(obj) is tuple:
             # Special case for 1-item tuples.
             p.text(',')
         p.end_group(step, end)
     return inner
 
 
 def _set_pprinter_factory(start, end):
     """
     Factory that returns a pprint function useful for sets and frozensets.
     """
     def inner(obj, p, cycle):
         if cycle:
             return p.text(start + '...' + end)
         if len(obj) == 0:
             # Special case.
             p.text(type(obj).__name__ + '()')
         else:
             step = len(start)
             p.begin_group(step, start)
             # Like dictionary keys, we will try to sort the items if there aren't too many
             if not (p.max_seq_length and len(obj) >= p.max_seq_length):
                 items = _sorted_for_pprint(obj)
             else:
                 items = obj
             for idx, x in p._enumerate(items):
                 if idx:
                     p.text(',')
                     p.breakable()
                 p.pretty(x)
             p.end_group(step, end)
     return inner
 
 
 def _dict_pprinter_factory(start, end):
     """
     Factory that returns a pprint function used by the default pprint of
     dicts and dict proxies.
     """
     def inner(obj, p, cycle):
         if cycle:
             return p.text('{...}')
         step = len(start)
         p.begin_group(step, start)
         keys = obj.keys()
         for idx, key in p._enumerate(keys):
             if idx:
                 p.text(',')
                 p.breakable()
             p.pretty(key)
             p.text(': ')
             p.pretty(obj[key])
         p.end_group(step, end)
     return inner
 
 
 def _super_pprint(obj, p, cycle):
     """The pprint for the super type."""
     p.begin_group(8, '<super: ')
     p.pretty(obj.__thisclass__)
     p.text(',')
     p.breakable()
     if PYPY: # In PyPy, super() objects don't have __self__ attributes
         dself = obj.__repr__.__self__
         p.pretty(None if dself is obj else dself)
     else:
         p.pretty(obj.__self__)
     p.end_group(8, '>')
 
 
 def _re_pattern_pprint(obj, p, cycle):
     """The pprint function for regular expression patterns."""
     p.text('re.compile(')
     pattern = repr(obj.pattern)
     if pattern[:1] in 'uU':
         pattern = pattern[1:]
         prefix = 'ur'
     else:
         prefix = 'r'
     pattern = prefix + pattern.replace('\\\\', '\\')
     p.text(pattern)
     if obj.flags:
         p.text(',')
         p.breakable()
         done_one = False
         for flag in ('TEMPLATE', 'IGNORECASE', 'LOCALE', 'MULTILINE', 'DOTALL',
             'UNICODE', 'VERBOSE', 'DEBUG'):
             if obj.flags & getattr(re, flag):
                 if done_one:
                     p.text('|')
                 p.text('re.' + flag)
                 done_one = True
     p.text(')')
 
 
 def _type_pprint(obj, p, cycle):
     """The pprint for classes and types."""
     # Heap allocated types might not have the module attribute,
     # and others may set it to None.
 
     # Checks for a __repr__ override in the metaclass. Can't compare the
     # type(obj).__repr__ directly because in PyPy the representation function
     # inherited from type isn't the same type.__repr__
     if [m for m in _get_mro(type(obj)) if "__repr__" in vars(m)][:1] != [type]:
         _repr_pprint(obj, p, cycle)
         return
 
     mod = _safe_getattr(obj, '__module__', None)
     try:
         name = obj.__qualname__
         if not isinstance(name, str):
             # This can happen if the type implements __qualname__ as a property
             # or other descriptor in Python 2.
             raise Exception("Try __name__")
     except Exception:
         name = obj.__name__
         if not isinstance(name, str):
             name = '<unknown type>'
 
     if mod in (None, '__builtin__', 'builtins', 'exceptions'):
         p.text(name)
     else:
         p.text(mod + '.' + name)
 
 
 def _repr_pprint(obj, p, cycle):
     """A pprint that just redirects to the normal repr function."""
     # Find newlines and replace them with p.break_()
     output = repr(obj)
     lines = output.splitlines()
     with p.group():
         for idx, output_line in enumerate(lines):
             if idx:
                 p.break_()
             p.text(output_line)
 
 
 def _function_pprint(obj, p, cycle):
     """Base pprint for all functions and builtin functions."""
     name = _safe_getattr(obj, '__qualname__', obj.__name__)
     mod = obj.__module__
     if mod and mod not in ('__builtin__', 'builtins', 'exceptions'):
         name = mod + '.' + name
     try:
        func_def = name + str(signature(obj))
     except ValueError:
        func_def = name
     p.text('<function %s>' % func_def)
 
 
 def _exception_pprint(obj, p, cycle):
     """Base pprint for all exceptions."""
     name = getattr(obj.__class__, '__qualname__', obj.__class__.__name__)
     if obj.__class__.__module__ not in ('exceptions', 'builtins'):
         name = '%s.%s' % (obj.__class__.__module__, name)
     step = len(name) + 1
     p.begin_group(step, name + '(')
     for idx, arg in enumerate(getattr(obj, 'args', ())):
         if idx:
             p.text(',')
             p.breakable()
         p.pretty(arg)
     p.end_group(step, ')')
 
 
 #: the exception base
 try:
     _exception_base = BaseException
 except NameError:
     _exception_base = Exception
 
 
 #: printers for builtin types
 _type_pprinters = {
     int:                        _repr_pprint,
     float:                      _repr_pprint,
     str:                        _repr_pprint,
     tuple:                      _seq_pprinter_factory('(', ')'),
     list:                       _seq_pprinter_factory('[', ']'),
     dict:                       _dict_pprinter_factory('{', '}'),
     set:                        _set_pprinter_factory('{', '}'),
     frozenset:                  _set_pprinter_factory('frozenset({', '})'),
     super:                      _super_pprint,
     _re_pattern_type:           _re_pattern_pprint,
     type:                       _type_pprint,
     types.FunctionType:         _function_pprint,
     types.BuiltinFunctionType:  _function_pprint,
     types.MethodType:           _repr_pprint,
     datetime.datetime:          _repr_pprint,
     datetime.timedelta:         _repr_pprint,
     _exception_base:            _exception_pprint
 }
 
 # render os.environ like a dict
 _env_type = type(os.environ)
 # future-proof in case os.environ becomes a plain dict?
 if _env_type is not dict:
     _type_pprinters[_env_type] = _dict_pprinter_factory('environ{', '}')
 
 try:
     # In PyPy, types.DictProxyType is dict, setting the dictproxy printer
     # using dict.setdefault avoids overwriting the dict printer
     _type_pprinters.setdefault(types.DictProxyType,
                                _dict_pprinter_factory('dict_proxy({', '})'))
     _type_pprinters[types.ClassType] = _type_pprint
     _type_pprinters[types.SliceType] = _repr_pprint
 except AttributeError: # Python 3
     _type_pprinters[types.MappingProxyType] = \
         _dict_pprinter_factory('mappingproxy({', '})')
     _type_pprinters[slice] = _repr_pprint
 
 try:
     _type_pprinters[long] = _repr_pprint
     _type_pprinters[unicode] = _repr_pprint
 except NameError:
     _type_pprinters[range] = _repr_pprint
     _type_pprinters[bytes] = _repr_pprint
 
 #: printers for types specified by name
 _deferred_type_pprinters = {
 }
 
 def for_type(typ, func):
     """
     Add a pretty printer for a given type.
     """
     oldfunc = _type_pprinters.get(typ, None)
     if func is not None:
         # To support easy restoration of old pprinters, we need to ignore Nones.
         _type_pprinters[typ] = func
     return oldfunc
 
 def for_type_by_name(type_module, type_name, func):
     """
     Add a pretty printer for a type specified by the module and name of a type
     rather than the type object itself.
     """
     key = (type_module, type_name)
     oldfunc = _deferred_type_pprinters.get(key, None)
     if func is not None:
         # To support easy restoration of old pprinters, we need to ignore Nones.
         _deferred_type_pprinters[key] = func
     return oldfunc
 
 
 #: printers for the default singletons
 _singleton_pprinters = dict.fromkeys(map(id, [None, True, False, Ellipsis,
                                       NotImplemented]), _repr_pprint)
 
 
 def _defaultdict_pprint(obj, p, cycle):
     name = obj.__class__.__name__
     with p.group(len(name) + 1, name + '(', ')'):
         if cycle:
             p.text('...')
         else:
             p.pretty(obj.default_factory)
             p.text(',')
             p.breakable()
             p.pretty(dict(obj))
 
 def _ordereddict_pprint(obj, p, cycle):
     name = obj.__class__.__name__
     with p.group(len(name) + 1, name + '(', ')'):
         if cycle:
             p.text('...')
         elif len(obj):
             p.pretty(list(obj.items()))
 
 def _deque_pprint(obj, p, cycle):
     name = obj.__class__.__name__
     with p.group(len(name) + 1, name + '(', ')'):
         if cycle:
             p.text('...')
         else:
             p.pretty(list(obj))
 
 
 def _counter_pprint(obj, p, cycle):
     name = obj.__class__.__name__
     with p.group(len(name) + 1, name + '(', ')'):
         if cycle:
             p.text('...')
         elif len(obj):
             p.pretty(dict(obj))
 
 for_type_by_name('collections', 'defaultdict', _defaultdict_pprint)
 for_type_by_name('collections', 'OrderedDict', _ordereddict_pprint)
 for_type_by_name('collections', 'deque', _deque_pprint)
 for_type_by_name('collections', 'Counter', _counter_pprint)
 
 if __name__ == '__main__':
     from random import randrange
     class Foo(object):
         def __init__(self):
             self.foo = 1
             self.bar = re.compile(r'\s+')
             self.blub = dict.fromkeys(range(30), randrange(1, 40))
             self.hehe = 23424.234234
             self.list = ["blub", "blah", self]
 
         def get_foo(self):
             print("foo")
 
     pprint(Foo(), verbose=True)

IPython/utils/encoding.py

@@ -1,71 +1,71 b''
 # coding: utf-8
 """
 Utilities for dealing with text encodings
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2008-2012  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 import sys
 import locale
 import warnings
 
 # to deal with the possibility of sys.std* not being a stream at all
 def get_stream_enc(stream, default=None):
     """Return the given stream's encoding or a default.
 
     There are cases where ``sys.std*`` might not actually be a stream, so
     check for the encoding attribute prior to returning it, and return
     a default if it doesn't exist or evaluates as False. ``default``
     is None if not provided.
     """
     if not hasattr(stream, 'encoding') or not stream.encoding:
         return default
     else:
         return stream.encoding
 
 # Less conservative replacement for sys.getdefaultencoding, that will try
 # to match the environment.
 # Defined here as central function, so if we find better choices, we
 # won't need to make changes all over IPython.
 def getdefaultencoding(prefer_stream=True):
     """Return IPython's guess for the default encoding for bytes as text.
     
     If prefer_stream is True (default), asks for stdin.encoding first,
     to match the calling Terminal, but that is often None for subprocesses.
     
     Then fall back on locale.getpreferredencoding(),
     which should be a sensible platform default (that respects LANG environment),
     and finally to sys.getdefaultencoding() which is the most conservative option,
-    and usually ASCII on Python 2 or UTF8 on Python 3.
+    and usually UTF8 as of Python 3.
     """
     enc = None
     if prefer_stream:
         enc = get_stream_enc(sys.stdin)
     if not enc or enc=='ascii':
         try:
             # There are reports of getpreferredencoding raising errors
             # in some cases, which may well be fixed, but let's be conservative here.
             enc = locale.getpreferredencoding()
         except Exception:
             pass
     enc = enc or sys.getdefaultencoding()
     # On windows `cp0` can be returned to indicate that there is no code page.
     # Since cp0 is an invalid encoding return instead cp1252 which is the
     # Western European default.
     if enc == 'cp0':
         warnings.warn(
             "Invalid code page cp0 detected - using cp1252 instead."
             "If cp1252 is incorrect please ensure a valid code page "
             "is defined for the process.", RuntimeWarning)
         return 'cp1252'
     return enc
 
 DEFAULT_ENCODING = getdefaultencoding()

IPython/utils/openpy.py

@@ -1,105 +1,103 b''
 """
 Tools to open .py files as Unicode, using the encoding specified within the file,
 as per PEP 263.
 
 Much of the code is taken from the tokenize module in Python 3.2.
 """
 
 import io
 from io import TextIOWrapper, BytesIO
 import re
 from tokenize import open, detect_encoding
 
 cookie_re = re.compile(r"coding[:=]\s*([-\w.]+)", re.UNICODE)
 cookie_comment_re = re.compile(r"^\s*#.*coding[:=]\s*([-\w.]+)", re.UNICODE)
 
 def source_to_unicode(txt, errors='replace', skip_encoding_cookie=True):
     """Converts a bytes string with python source code to unicode.
 
     Unicode strings are passed through unchanged. Byte strings are checked
     for the python source file encoding cookie to determine encoding.
     txt can be either a bytes buffer or a string containing the source
     code.
     """
     if isinstance(txt, str):
         return txt
     if isinstance(txt, bytes):
         buffer = BytesIO(txt)
     else:
         buffer = txt
     try:
         encoding, _ = detect_encoding(buffer.readline)
     except SyntaxError:
         encoding = "ascii"
     buffer.seek(0)
     with TextIOWrapper(buffer, encoding, errors=errors, line_buffering=True) as text:
         text.mode = 'r'
         if skip_encoding_cookie:
             return u"".join(strip_encoding_cookie(text))
         else:
             return text.read()
 
 def strip_encoding_cookie(filelike):
     """Generator to pull lines from a text-mode file, skipping the encoding
     cookie if it is found in the first two lines.
     """
     it = iter(filelike)
     try:
         first = next(it)
         if not cookie_comment_re.match(first):
             yield first
         second = next(it)
         if not cookie_comment_re.match(second):
             yield second
     except StopIteration:
         return
     
     for line in it:
         yield line
 
 def read_py_file(filename, skip_encoding_cookie=True):
     """Read a Python file, using the encoding declared inside the file.
     
     Parameters
     ----------
     filename : str
       The path to the file to read.
     skip_encoding_cookie : bool
       If True (the default), and the encoding declaration is found in the first
-      two lines, that line will be excluded from the output - compiling a
-      unicode string with an encoding declaration is a SyntaxError in Python 2.
+      two lines, that line will be excluded from the output.
     
     Returns
     -------
     A unicode string containing the contents of the file.
     """
     with open(filename) as f:   # the open function defined in this module.
         if skip_encoding_cookie:
             return "".join(strip_encoding_cookie(f))
         else:
             return f.read()
 
 def read_py_url(url, errors='replace', skip_encoding_cookie=True):
     """Read a Python file from a URL, using the encoding declared inside the file.
     
     Parameters
     ----------
     url : str
       The URL from which to fetch the file.
     errors : str
       How to handle decoding errors in the file. Options are the same as for
       bytes.decode(), but here 'replace' is the default.
     skip_encoding_cookie : bool
       If True (the default), and the encoding declaration is found in the first
-      two lines, that line will be excluded from the output - compiling a
-      unicode string with an encoding declaration is a SyntaxError in Python 2.
+      two lines, that line will be excluded from the output.
     
     Returns
     -------
     A unicode string containing the contents of the file.
     """
     # Deferred import for faster start
     from urllib.request import urlopen 
     response = urlopen(url)
     buffer = io.BytesIO(response.read())
     return source_to_unicode(buffer, errors, skip_encoding_cookie)