Use the proper python2 pygments lexer

Since pygments 2.5 `PythonLexer` refers to python3 [1]. We have to be
explicit if we want to use python2.

[1] https://pygments.org/docs/changelog/

IPython/core/oinspect.py

@@ -1,1011 +1,1015 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.
 
 from __future__ import print_function
 
 __all__ = ['Inspector','InspectColors']
 
 # stdlib modules
 import inspect
 import linecache
 import warnings
 import os
 from textwrap import dedent
 import types
 import io as stdlib_io
 
 try:
     from itertools import izip_longest
 except ImportError:
     from itertools import zip_longest as izip_longest
 
 # IPython's own
 from IPython.core import page
 from IPython.lib.pretty import pretty
 from IPython.testing.skipdoctest import skip_doctest_py3
 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.coloransi import TermColors, ColorScheme, ColorSchemeTable
 from IPython.utils.py3compat import cast_unicode, string_types, PY3
 from IPython.utils.signatures import signature
 from IPython.utils.colorable import Colorable
 
 from pygments import highlight
-from pygments.lexers import PythonLexer
+try:
+    # PythonLexer was renamed to Python2Lexer in pygments 2.5
+    from pygments.lexers import Python2Lexer
+except ImportError:
+    from pygments.lexers import PythonLexer as Python2Lexer
 from pygments.formatters import HtmlFormatter
 
 def pylight(code):
-    return highlight(code, PythonLexer(), HtmlFormatter(noclasses=True))
+    return highlight(code, Python2Lexer(), 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', 'argspec', 'found', 'name'
                ]
 
 
 def object_info(**kw):
     """Make an object info dict with all fields present."""
     infodict = dict(izip_longest(info_fields, [None]))
     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):
     """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 we get extra info, we add it to the normal docstring.
         if isinstance(ds, string_types):
             return inspect.cleandoc(ds)
     try:
         docstr = inspect.getdoc(obj)
         encoding = get_encoding(obj)
         return py3compat.cast_unicode(docstr, encoding=encoding)
     except Exception:
         # Harden against an inspect failure, which can occur with
         # extensions modules.
         raise
         return None
 
 
 def getsource(obj, oname=''):
     """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(cast_unicode(
                     ''.join(('# ', oname_prefix, attrname)),
                     encoding=encoding))
                 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(cast_unicode(
                         '%s%s = %s\n' % (
                             oname_prefix, attrname, pretty(fn)),
                         encoding=encoding))
         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
 
         encoding = get_encoding(obj)
         return cast_unicode(src, encoding=encoding)
 
 
 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))
 
 
 def getargspec(obj):
     """Wrapper around :func:`inspect.getfullargspec` on Python 3, and
     :func:inspect.getargspec` on Python 2.
     
     In addition to functions and methods, this can also handle objects with a
     ``__call__`` attribute.
     """
     if safe_hasattr(obj, '__call__') and not is_simple_callable(obj):
         obj = obj.__call__
 
     return inspect.getfullargspec(obj) if PY3 else inspect.getargspec(obj)
 
 
 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.
     """
     return inspect.formatargspec(argspec['args'], argspec['varargs'],
                                  argspec['varkw'], argspec['defaults'])
 
 
 def call_tip(oinfo, format_call=True):
     """Extract call tip data from an oinfo dict.
 
     Parameters
     ----------
     oinfo : dict
 
     format_call : bool, optional
       If True, the call line is formatted and returned as a string.  If not, a
       tuple of (name, argspec) is returned.
 
     Returns
     -------
     call_info : None, str or (str, dict) tuple.
       When format_call is True, the whole call information is formattted as a
       single string.  Otherwise, the object's name and its argspec dict are
       returned.  If no call information is available, None is returned.
 
     docstring : str or None
       The most relevant docstring for calling purposes is returned, if
       available.  The priority is: call docstring for callable instances, then
       constructor docstring for classes, then main object's docstring otherwise
       (regular functions).
     """
     # 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):
     """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='NoColor',
                  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=''):
         """Return the call signature for any callable object.
 
         If any exception is generated, None is returned instead and the
         exception is suppressed."""
         try:
             hdef = oname + str(signature(obj))
             return cast_unicode(hdef)
         except:
             return None
 
     def __head(self,h):
         """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):
         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')
         elif (not py3compat.PY3) and type(obj) is types.InstanceType:
             obj = obj.__call__
 
         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_py3
     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 _format_fields(self, fields, title_width=0):
         """Formats a list of fields for display.
 
         Parameters
         ----------
         fields : list
           A list of 2-tuples: (field_title, field_content)
         title_width : int
           How many characters to pad titles to. Default to longest title.
         """
         out = []
         header = self.__head
         if title_width == 0:
             title_width = max(len(title) + 2 for title, _ in fields)
         for title, content in fields:
             if len(content.splitlines()) > 1:
                 title = header(title + ':') + '\n'
             else:
                 title = header((title + ':').ljust(title_width))
             out.append(cast_unicode(title) + cast_unicode(content))
         return "\n".join(out)
 
     def _mime_format(self, text, formatter=None):
         """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.
 
         """
         text = cast_unicode(text)
         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."""
 
         info = self._info(obj, oname=oname, info=info, detail_level=detail_level)
 
         _mime = {
             'text/plain': [],
             'text/html': '',
         }
 
         def append_field(bundle, title, key, 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)
             if detail_level > 0 and info['source']:
                 append_field(_mime, 'Source', 'source', code_formatter)
             else:
                 append_field(_mime, 'Docstring', 'docstring', formatter)
                 append_field(_mime, 'Init docstring', 'init_docstring', formatter)
 
             append_field(_mime, 'File', 'file')
             append_field(_mime, 'Type', 'type_name')
 
         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')
 
             # Base class for old-style instances
             if (not py3compat.PY3) and isinstance(obj, types.InstanceType) and info['base_class']:
                 append_field(_mime, 'Base Class', 'base_class')
 
             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:
                 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 dictionnary.
 
               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):
         """Compute a dict with detailed information about an object.
 
         Optional arguments:
 
         - oname: name of the variable pointing to the object.
 
         - info: a structure with some information fields which may have been
           precomputed already.
 
         - detail_level: if set to 1, more information is given.
         """
 
         obj_type = type(obj)
 
         if info is None:
             ismagic = 0
             isalias = 0
             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)
 
         string_max = 200 # max size of strings to show (snipped if longer)
         shalf = int((string_max - 5) / 2)
 
         if ismagic:
             obj_type_name = 'Magic function'
         elif isalias:
             obj_type_name = 'System alias'
         else:
             obj_type_name = obj_type.__name__
         out['type_name'] = obj_type_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: 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 out.get('source', None) is None:
             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
 
         # 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
 
         # Compute the object's argspec as a callable.  The key is to decide
         # whether to pull it from the object itself, from its __init__ or
         # from its __call__ method.
 
         if inspect.isclass(obj):
             # Old-style classes need not have an __init__
             callable_obj = getattr(obj, "__init__", None)
         elif callable(obj):
             callable_obj = obj
         else:
             callable_obj = None
 
         if callable_obj is not None:
             try:
                 argspec = getargspec(callable_obj)
             except (TypeError, AttributeError):
                 # For extensions/builtins we can't retrieve the argspec
                 pass
             else:
                 # named tuples' _asdict() method returns an OrderedDict, but we
                 # we want a normal
                 out['argspec'] = argspec_dict = dict(argspec._asdict())
                 # We called this varkw before argspec became a named tuple.
                 # With getfullargspec it's also called varkw.
                 if 'varkw' not in argspec_dict:
                     argspec_dict['varkw'] = argspec_dict.pop('keywords')
 
         return object_info(**out)
 
     def psearch(self,pattern,ns_table,ns_search=[],
                 ignore_case=False,show_all=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.
         """
         #print 'ps pattern:<%r>' % pattern # dbg
 
         # defaults
         type_pattern = 'all'
         filter = ''
 
         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)))

IPython/lib/lexers.py

@@ -1,512 +1,517 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, PythonLexer, Python3Lexer
+from pygments.lexers import BashLexer, Python3Lexer
+try:
+    # PythonLexer was renamed to Python2Lexer in pygments 2.5
+    from pygments.lexers import Python2Lexer
+except ImportError:
+    from pygments.lexers import PythonLexer as Python2Lexer
 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']
 
 ipython_tokens = [
   (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)),
 ]
 
 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
+        PyLexer = Python2Lexer
         name = 'IPython'
         aliases = ['ipython2', 'ipython']
         doc = """IPython Lexer"""
 
     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.
 
     """
     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):
     """
     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