Merge pull request #770 from minrk/jsonclean

Ensures all replies from ipkernel are clean for json (not just oinfo), and guesses stdin.encoding before using sys.getdefaultencoding in json_clean.

IPython/config/loader.py

@@ -1,660 +1,661 b''
 """A simple configuration system.
 
 Authors
 -------
 * Brian Granger
 * Fernando Perez
 * Min RK
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2008-2011  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 __builtin__ as builtin_mod
 import re
 import sys
 
 from IPython.external import argparse
 from IPython.utils.path import filefind, get_ipython_dir
-from IPython.utils import py3compat, warn
+from IPython.utils import py3compat, text, warn
 
 #-----------------------------------------------------------------------------
 # Exceptions
 #-----------------------------------------------------------------------------
 
 
 class ConfigError(Exception):
     pass
 
 
 class ConfigLoaderError(ConfigError):
     pass
 
 class ArgumentError(ConfigLoaderError):
     pass
 
 #-----------------------------------------------------------------------------
 # Argparse fix
 #-----------------------------------------------------------------------------
 
 # Unfortunately argparse by default prints help messages to stderr instead of
 # stdout.  This makes it annoying to capture long help screens at the command
 # line, since one must know how to pipe stderr, which many users don't know how
 # to do.  So we override the print_help method with one that defaults to
 # stdout and use our class instead.
 
 class ArgumentParser(argparse.ArgumentParser):
     """Simple argparse subclass that prints help to stdout by default."""
     
     def print_help(self, file=None):
         if file is None:
             file = sys.stdout
         return super(ArgumentParser, self).print_help(file)
     
     print_help.__doc__ = argparse.ArgumentParser.print_help.__doc__
     
 #-----------------------------------------------------------------------------
 # Config class for holding config information
 #-----------------------------------------------------------------------------
 
 
 class Config(dict):
     """An attribute based dict that can do smart merges."""
 
     def __init__(self, *args, **kwds):
         dict.__init__(self, *args, **kwds)
         # This sets self.__dict__ = self, but it has to be done this way
         # because we are also overriding __setattr__.
         dict.__setattr__(self, '__dict__', self)
 
     def _merge(self, other):
         to_update = {}
         for k, v in other.iteritems():
             if not self.has_key(k):
                 to_update[k] = v
             else: # I have this key
                 if isinstance(v, Config):
                     # Recursively merge common sub Configs
                     self[k]._merge(v)
                 else:
                     # Plain updates for non-Configs
                     to_update[k] = v
 
         self.update(to_update)
 
     def _is_section_key(self, key):
         if key[0].upper()==key[0] and not key.startswith('_'):
             return True
         else:
             return False
 
     def __contains__(self, key):
         if self._is_section_key(key):
             return True
         else:
             return super(Config, self).__contains__(key)
     # .has_key is deprecated for dictionaries.
     has_key = __contains__
 
     def _has_section(self, key):
         if self._is_section_key(key):
             if super(Config, self).__contains__(key):
                 return True
         return False
 
     def copy(self):
         return type(self)(dict.copy(self))
 
     def __copy__(self):
         return self.copy()
 
     def __deepcopy__(self, memo):
         import copy
         return type(self)(copy.deepcopy(self.items()))
 
     def __getitem__(self, key):
         # We cannot use directly self._is_section_key, because it triggers
         # infinite recursion on top of PyPy. Instead, we manually fish the
         # bound method.
         is_section_key = self.__class__._is_section_key.__get__(self)
         
         # Because we use this for an exec namespace, we need to delegate
         # the lookup of names in __builtin__ to itself.  This means
         # that you can't have section or attribute names that are 
         # builtins.
         try:
             return getattr(builtin_mod, key)
         except AttributeError:
             pass
         if is_section_key(key):
             try:
                 return dict.__getitem__(self, key)
             except KeyError:
                 c = Config()
                 dict.__setitem__(self, key, c)
                 return c
         else:
             return dict.__getitem__(self, key)
 
     def __setitem__(self, key, value):
         # Don't allow names in __builtin__ to be modified.
         if hasattr(builtin_mod, key):
             raise ConfigError('Config variable names cannot have the same name '
                               'as a Python builtin: %s' % key)
         if self._is_section_key(key):
             if not isinstance(value, Config):
                 raise ValueError('values whose keys begin with an uppercase '
                                  'char must be Config instances: %r, %r' % (key, value))
         else:
             dict.__setitem__(self, key, value)
 
     def __getattr__(self, key):
         try:
             return self.__getitem__(key)
         except KeyError, e:
             raise AttributeError(e)
 
     def __setattr__(self, key, value):
         try:
             self.__setitem__(key, value)
         except KeyError, e:
             raise AttributeError(e)
 
     def __delattr__(self, key):
         try:
             dict.__delitem__(self, key)
         except KeyError, e:
             raise AttributeError(e)
 
 
 #-----------------------------------------------------------------------------
 # Config loading classes
 #-----------------------------------------------------------------------------
 
 
 class ConfigLoader(object):
     """A object for loading configurations from just about anywhere.
     
     The resulting configuration is packaged as a :class:`Struct`.
     
     Notes
     -----
     A :class:`ConfigLoader` does one thing: load a config from a source 
     (file, command line arguments) and returns the data as a :class:`Struct`.
     There are lots of things that :class:`ConfigLoader` does not do.  It does
     not implement complex logic for finding config files.  It does not handle
     default values or merge multiple configs.  These things need to be 
     handled elsewhere.
     """
 
     def __init__(self):
         """A base class for config loaders.
         
         Examples
         --------
         
         >>> cl = ConfigLoader()
         >>> config = cl.load_config()
         >>> config
         {}
         """
         self.clear()
 
     def clear(self):
         self.config = Config()
 
     def load_config(self):
         """Load a config from somewhere, return a :class:`Config` instance.
         
         Usually, this will cause self.config to be set and then returned.
         However, in most cases, :meth:`ConfigLoader.clear` should be called
         to erase any previous state.
         """
         self.clear()
         return self.config
 
 
 class FileConfigLoader(ConfigLoader):
     """A base class for file based configurations.
 
     As we add more file based config loaders, the common logic should go
     here.
     """
     pass
 
 
 class PyFileConfigLoader(FileConfigLoader):
     """A config loader for pure python files.
     
     This calls execfile on a plain python file and looks for attributes
     that are all caps.  These attribute are added to the config Struct.
     """
 
     def __init__(self, filename, path=None):
         """Build a config loader for a filename and path.
 
         Parameters
         ----------
         filename : str
             The file name of the config file.
         path : str, list, tuple
             The path to search for the config file on, or a sequence of
             paths to try in order.
         """
         super(PyFileConfigLoader, self).__init__()
         self.filename = filename
         self.path = path
         self.full_filename = ''
         self.data = None
 
     def load_config(self):
         """Load the config from a file and return it as a Struct."""
         self.clear()
         self._find_file()
         self._read_file_as_dict()
         self._convert_to_config()
         return self.config
 
     def _find_file(self):
         """Try to find the file by searching the paths."""
         self.full_filename = filefind(self.filename, self.path)
 
     def _read_file_as_dict(self):
         """Load the config file into self.config, with recursive loading."""
         # This closure is made available in the namespace that is used
         # to exec the config file.  It allows users to call
         # load_subconfig('myconfig.py') to load config files recursively.
         # It needs to be a closure because it has references to self.path
         # and self.config.  The sub-config is loaded with the same path
         # as the parent, but it uses an empty config which is then merged
         # with the parents.
         
         # If a profile is specified, the config file will be loaded
         # from that profile
         
         def load_subconfig(fname, profile=None):
             # import here to prevent circular imports
             from IPython.core.profiledir import ProfileDir, ProfileDirError
             if profile is not None:
                 try:
                     profile_dir = ProfileDir.find_profile_dir_by_name(
                             get_ipython_dir(),
                             profile,
                     )
                 except ProfileDirError:
                     return
                 path = profile_dir.location
             else:
                 path = self.path
             loader = PyFileConfigLoader(fname, path)
             try:
                 sub_config = loader.load_config()
             except IOError:
                 # Pass silently if the sub config is not there. This happens
                 # when a user s using a profile, but not the default config.
                 pass
             else:
                 self.config._merge(sub_config)
         
         # Again, this needs to be a closure and should be used in config
         # files to get the config being loaded.
         def get_config():
             return self.config
 
         namespace = dict(load_subconfig=load_subconfig, get_config=get_config)
         fs_encoding = sys.getfilesystemencoding() or 'ascii'
         conf_filename = self.full_filename.encode(fs_encoding)
         py3compat.execfile(conf_filename, namespace)
 
     def _convert_to_config(self):
         if self.data is None:
             ConfigLoaderError('self.data does not exist')
 
 
 class CommandLineConfigLoader(ConfigLoader):
     """A config loader for command line arguments.
 
     As we add more command line based loaders, the common logic should go
     here.
     """
 
     def _exec_config_str(self, lhs, rhs):
         exec_str = 'self.config.' + lhs + '=' + rhs
         try:
             # Try to see if regular Python syntax will work. This
             # won't handle strings as the quote marks are removed
             # by the system shell.
             exec exec_str in locals(), globals()
         except (NameError, SyntaxError):
             # This case happens if the rhs is a string but without
             # the quote marks. Use repr, to get quote marks, and
             # 'u' prefix and see if
             # it succeeds. If it still fails, we let it raise.
             exec_str = u'self.config.' + lhs + '=' + repr(rhs)
             exec exec_str in locals(), globals()
     
     def _load_flag(self, cfg):
         """update self.config from a flag, which can be a dict or Config"""
         if isinstance(cfg, (dict, Config)):
             # don't clobber whole config sections, update
             # each section from config:
             for sec,c in cfg.iteritems():
                 self.config[sec].update(c)
         else:
             raise ValueError("Invalid flag: '%s'"%raw)
 
 # raw --identifier=value pattern
 # but *also* accept '-' as wordsep, for aliases
 # accepts:  --foo=a
 #           --Class.trait=value
 #           --alias-name=value
 # rejects:  -foo=value
 #           --foo
 #           --Class.trait
 kv_pattern = re.compile(r'\-\-[A-Za-z][\w\-]*(\.[\w\-]+)*\=.*')
 
 # just flags, no assignments, with two *or one* leading '-'
 # accepts:  --foo
 #           -foo-bar-again
 # rejects:  --anything=anything
 #           --two.word
 
 flag_pattern = re.compile(r'\-\-?\w+[\-\w]*$')
 
 class KeyValueConfigLoader(CommandLineConfigLoader):
     """A config loader that loads key value pairs from the command line.
 
     This allows command line options to be gives in the following form::
     
         ipython --profile="foo" --InteractiveShell.autocall=False
     """
 
     def __init__(self, argv=None, aliases=None, flags=None):
         """Create a key value pair config loader.
 
         Parameters
         ----------
         argv : list
             A list that has the form of sys.argv[1:] which has unicode
             elements of the form u"key=value". If this is None (default),
             then sys.argv[1:] will be used.
         aliases : dict
             A dict of aliases for configurable traits.
             Keys are the short aliases, Values are the resolved trait.
             Of the form: `{'alias' : 'Configurable.trait'}`
         flags : dict
             A dict of flags, keyed by str name. Vaues can be Config objects,
             dicts, or "key=value" strings.  If Config or dict, when the flag
             is triggered, The flag is loaded as `self.config.update(m)`.
 
         Returns
         -------
         config : Config
             The resulting Config object.
 
         Examples
         --------
 
             >>> from IPython.config.loader import KeyValueConfigLoader
             >>> cl = KeyValueConfigLoader()
             >>> cl.load_config(["--A.name='brian'","--B.number=0"])
             {'A': {'name': 'brian'}, 'B': {'number': 0}}
         """
         self.clear()
         if argv is None:
             argv = sys.argv[1:]
         self.argv = argv
         self.aliases = aliases or {}
         self.flags = flags or {}
         
     
     def clear(self):
         super(KeyValueConfigLoader, self).clear()
         self.extra_args = []
         
     
     def _decode_argv(self, argv, enc=None):
         """decode argv if bytes, using stin.encoding, falling back on default enc"""
         uargv = []
         if enc is None:
-            enc = sys.stdin.encoding or sys.getdefaultencoding()
+            enc = text.getdefaultencoding()
         for arg in argv:
             if not isinstance(arg, unicode):
                 # only decode if not already decoded
                 arg = arg.decode(enc)
             uargv.append(arg)
         return uargv
                 
                 
     def load_config(self, argv=None, aliases=None, flags=None):
         """Parse the configuration and generate the Config object.
         
         After loading, any arguments that are not key-value or
         flags will be stored in self.extra_args - a list of
         unparsed command-line arguments.  This is used for
         arguments such as input files or subcommands.
         
         Parameters
         ----------
         argv : list, optional
             A list that has the form of sys.argv[1:] which has unicode
             elements of the form u"key=value". If this is None (default),
             then self.argv will be used.
         aliases : dict
             A dict of aliases for configurable traits.
             Keys are the short aliases, Values are the resolved trait.
             Of the form: `{'alias' : 'Configurable.trait'}`
         flags : dict
             A dict of flags, keyed by str name. Values can be Config objects
             or dicts.  When the flag is triggered, The config is loaded as 
             `self.config.update(cfg)`.
         """
         from IPython.config.configurable import Configurable
 
         self.clear()
         if argv is None:
             argv = self.argv
         if aliases is None:
             aliases = self.aliases
         if flags is None:
             flags = self.flags
         
         # ensure argv is a list of unicode strings:
         uargv = self._decode_argv(argv)
         for idx,raw in enumerate(uargv):
             # strip leading '-'
             item = raw.lstrip('-')
             
             if raw == '--':
                 # don't parse arguments after '--'
                 # this is useful for relaying arguments to scripts, e.g.
                 # ipython -i foo.py --pylab=qt -- args after '--' go-to-foo.py
                 self.extra_args.extend(uargv[idx+1:])
                 break
             
             if kv_pattern.match(raw):
                 lhs,rhs = item.split('=',1)
                 # Substitute longnames for aliases.
                 if lhs in aliases:
                     lhs = aliases[lhs]
                 if '.' not in lhs:
                     # probably a mistyped alias, but not technically illegal
                     warn.warn("Unrecognized alias: '%s', it will probably have no effect."%lhs)
                 self._exec_config_str(lhs, rhs)
                 
             elif flag_pattern.match(raw):
                 if item in flags:
                     cfg,help = flags[item]
                     self._load_flag(cfg)
                 else:
                     raise ArgumentError("Unrecognized flag: '%s'"%raw)
             elif raw.startswith('-'):
                 kv = '--'+item
                 if kv_pattern.match(kv):
                     raise ArgumentError("Invalid argument: '%s', did you mean '%s'?"%(raw, kv))
                 else:
                     raise ArgumentError("Invalid argument: '%s'"%raw)
             else:
                 # keep all args that aren't valid in a list, 
                 # in case our parent knows what to do with them.
                 self.extra_args.append(item)
         return self.config
 
 class ArgParseConfigLoader(CommandLineConfigLoader):
     """A loader that uses the argparse module to load from the command line."""
 
     def __init__(self, argv=None, aliases=None, flags=None, *parser_args, **parser_kw):
         """Create a config loader for use with argparse.
 
         Parameters
         ----------
 
         argv : optional, list
           If given, used to read command-line arguments from, otherwise
           sys.argv[1:] is used.
 
         parser_args : tuple
           A tuple of positional arguments that will be passed to the
           constructor of :class:`argparse.ArgumentParser`.
 
         parser_kw : dict
           A tuple of keyword arguments that will be passed to the
           constructor of :class:`argparse.ArgumentParser`.
 
         Returns
         -------
         config : Config
             The resulting Config object.
         """
         super(CommandLineConfigLoader, self).__init__()
         self.clear()
         if argv is None:
             argv = sys.argv[1:]
         self.argv = argv
         self.aliases = aliases or {}
         self.flags = flags or {}
         
         self.parser_args = parser_args
         self.version = parser_kw.pop("version", None)
         kwargs = dict(argument_default=argparse.SUPPRESS)
         kwargs.update(parser_kw)
         self.parser_kw = kwargs
 
     def load_config(self, argv=None, aliases=None, flags=None):
         """Parse command line arguments and return as a Config object.
 
         Parameters
         ----------
 
         args : optional, list
           If given, a list with the structure of sys.argv[1:] to parse
           arguments from. If not given, the instance's self.argv attribute
           (given at construction time) is used."""
         self.clear()
         if argv is None:
             argv = self.argv
         if aliases is None:
             aliases = self.aliases
         if flags is None:
             flags = self.flags
         self._create_parser(aliases, flags)
         self._parse_args(argv)
         self._convert_to_config()
         return self.config
 
     def get_extra_args(self):
         if hasattr(self, 'extra_args'):
             return self.extra_args
         else:
             return []
 
     def _create_parser(self, aliases=None, flags=None):
         self.parser = ArgumentParser(*self.parser_args, **self.parser_kw)
         self._add_arguments(aliases, flags)
 
     def _add_arguments(self, aliases=None, flags=None):
         raise NotImplementedError("subclasses must implement _add_arguments")
 
     def _parse_args(self, args):
         """self.parser->self.parsed_data"""
         # decode sys.argv to support unicode command-line options
-        uargs = [py3compat.cast_unicode(a) for a in args]
+        enc = text.getdefaultencoding()
+        uargs = [py3compat.cast_unicode(a, enc) for a in args]
         self.parsed_data, self.extra_args = self.parser.parse_known_args(uargs)
 
     def _convert_to_config(self):
         """self.parsed_data->self.config"""
         for k, v in vars(self.parsed_data).iteritems():
             exec "self.config.%s = v"%k in locals(), globals()
 
 class KVArgParseConfigLoader(ArgParseConfigLoader):
     """A config loader that loads aliases and flags with argparse,
     but will use KVLoader for the rest.  This allows better parsing 
     of common args, such as `ipython -c 'print 5'`, but still gets
     arbitrary config with `ipython --InteractiveShell.use_readline=False`"""
     
     def _convert_to_config(self):
         """self.parsed_data->self.config"""
         for k, v in vars(self.parsed_data).iteritems():
             self._exec_config_str(k, v)
 
     def _add_arguments(self, aliases=None, flags=None):
         self.alias_flags = {}
         # print aliases, flags
         if aliases is None:
             aliases = self.aliases
         if flags is None:
             flags = self.flags
         paa = self.parser.add_argument
         for key,value in aliases.iteritems():
             if key in flags:
                 # flags
                 nargs = '?'
             else:
                 nargs = None
             if len(key) is 1:
                 paa('-'+key, '--'+key, type=str, dest=value, nargs=nargs)
             else:
                 paa('--'+key, type=str, dest=value, nargs=nargs)
         for key, (value, help) in flags.iteritems():
             if key in self.aliases:
                 # 
                 self.alias_flags[self.aliases[key]] = value
                 continue
             if len(key) is 1:
                 paa('-'+key, '--'+key, action='append_const', dest='_flags', const=value)
             else:
                 paa('--'+key, action='append_const', dest='_flags', const=value)
     
     def _convert_to_config(self):
         """self.parsed_data->self.config, parse unrecognized extra args via KVLoader."""
         # remove subconfigs list from namespace before transforming the Namespace
         if '_flags' in self.parsed_data:
             subcs = self.parsed_data._flags
             del self.parsed_data._flags
         else:
             subcs = []
         
         for k, v in vars(self.parsed_data).iteritems():
             if v is None:
                 # it was a flag that shares the name of an alias
                 subcs.append(self.alias_flags[k])
             else:
                 # eval the KV assignment
                 self._exec_config_str(k, v)
         
         for subc in subcs:
             self._load_flag(subc)
         
         if self.extra_args:
             sub_parser = KeyValueConfigLoader()
             sub_parser.load_config(self.extra_args)
             self.config._merge(sub_parser.config)
             self.extra_args = sub_parser.extra_args

IPython/utils/_process_win32.py

@@ -1,147 +1,148 b''
 """Windows-specific implementation of process utilities.
 
 This file is only meant to be imported by process.py, not by end-users.
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010  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
 #-----------------------------------------------------------------------------
 from __future__ import print_function
 
 # stdlib
 import os
 import sys
 
 from subprocess import STDOUT
 
 # our own imports
 from ._process_common import read_no_interrupt, process_handler
+from . import text
 
 #-----------------------------------------------------------------------------
 # Function definitions
 #-----------------------------------------------------------------------------
 
 class AvoidUNCPath(object):
     """A context manager to protect command execution from UNC paths.
 
     In the Win32 API, commands can't be invoked with the cwd being a UNC path.
     This context manager temporarily changes directory to the 'C:' drive on
     entering, and restores the original working directory on exit.
 
     The context manager returns the starting working directory *if* it made a
     change and None otherwise, so that users can apply the necessary adjustment
     to their system calls in the event of a change.
 
     Example
     -------
     ::
         cmd = 'dir'
         with AvoidUNCPath() as path:
             if path is not None:
                 cmd = '"pushd %s &&"%s' % (path, cmd)
             os.system(cmd)
     """
     def __enter__(self):
         self.path = os.getcwdu()
         self.is_unc_path = self.path.startswith(r"\\")
         if self.is_unc_path:
             # change to c drive (as cmd.exe cannot handle UNC addresses)
             os.chdir("C:")
             return self.path
         else:
             # We return None to signal that there was no change in the working
             # directory
             return None
 
     def __exit__(self, exc_type, exc_value, traceback):
         if self.is_unc_path:
             os.chdir(self.path)
 
 
 def _find_cmd(cmd):
     """Find the full path to a .bat or .exe using the win32api module."""
     try:
         from win32api import SearchPath
     except ImportError:
         raise ImportError('you need to have pywin32 installed for this to work')
     else:
         PATH = os.environ['PATH']
         extensions = ['.exe', '.com', '.bat', '.py']
         path = None
         for ext in extensions:
             try:
                 path = SearchPath(PATH, cmd + ext)[0]
             except:
                 pass
         if path is None:
             raise OSError("command %r not found" % cmd)
         else:
             return path
 
 
 def _system_body(p):
     """Callback for _system."""
-    enc = sys.stdin.encoding or sys.getdefaultencoding()
+    enc = text.getdefaultencoding()
     for line in read_no_interrupt(p.stdout).splitlines():
         line = line.decode(enc, 'replace')
         print(line, file=sys.stdout)
     for line in read_no_interrupt(p.stderr).splitlines():
         line = line.decode(enc, 'replace')
         print(line, file=sys.stderr)
 
     # Wait to finish for returncode
     return p.wait()
 
 
 def system(cmd):
     """Win32 version of os.system() that works with network shares.
 
     Note that this implementation returns None, as meant for use in IPython.
 
     Parameters
     ----------
     cmd : str
       A command to be executed in the system shell.
 
     Returns
     -------
     None : we explicitly do NOT return the subprocess status code, as this
     utility is meant to be used extensively in IPython, where any return value
     would trigger :func:`sys.displayhook` calls.
     """
     with AvoidUNCPath() as path:
         if path is not None:
             cmd = '"pushd %s &&"%s' % (path, cmd)
         return process_handler(cmd, _system_body)
 
 
 def getoutput(cmd):
     """Return standard output of executing cmd in a shell.
 
     Accepts the same arguments as os.system().
 
     Parameters
     ----------
     cmd : str
       A command to be executed in the system shell.
 
     Returns
     -------
     stdout : str
     """
 
     with AvoidUNCPath() as path:
         if path is not None:
             cmd = '"pushd %s &&"%s' % (path, cmd)
         out = process_handler(cmd, lambda p: p.communicate()[0], STDOUT)
 
     if out is None:
         out = ''
     return out

IPython/utils/jsonutil.py

@@ -1,164 +1,165 b''
 """Utilities to manipulate JSON objects.
 """
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING.txt, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 # stdlib
 import re
 import sys
 import types
 from datetime import datetime
 
 from IPython.utils import py3compat
+from IPython.utils import text
 next_attr_name = '__next__' if py3compat.PY3 else 'next'
 
 #-----------------------------------------------------------------------------
 # Globals and constants
 #-----------------------------------------------------------------------------
 
 # timestamp formats
 ISO8601="%Y-%m-%dT%H:%M:%S.%f"
 ISO8601_PAT=re.compile(r"^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d+$")
 
 #-----------------------------------------------------------------------------
 # Classes and functions
 #-----------------------------------------------------------------------------
 
 def rekey(dikt):
     """Rekey a dict that has been forced to use str keys where there should be
     ints by json."""
     for k in dikt.iterkeys():
         if isinstance(k, basestring):
             ik=fk=None
             try:
                 ik = int(k)
             except ValueError:
                 try:
                     fk = float(k)
                 except ValueError:
                     continue
             if ik is not None:
                 nk = ik
             else:
                 nk = fk
             if nk in dikt:
                 raise KeyError("already have key %r"%nk)
             dikt[nk] = dikt.pop(k)
     return dikt
 
 
 def extract_dates(obj):
     """extract ISO8601 dates from unpacked JSON"""
     if isinstance(obj, dict):
         obj = dict(obj) # don't clobber
         for k,v in obj.iteritems():
             obj[k] = extract_dates(v)
     elif isinstance(obj, (list, tuple)):
         obj = [ extract_dates(o) for o in obj ]
     elif isinstance(obj, basestring):
         if ISO8601_PAT.match(obj):
             obj = datetime.strptime(obj, ISO8601)
     return obj
 
 def squash_dates(obj):
     """squash datetime objects into ISO8601 strings"""
     if isinstance(obj, dict):
         obj = dict(obj) # don't clobber
         for k,v in obj.iteritems():
             obj[k] = squash_dates(v)
     elif isinstance(obj, (list, tuple)):
         obj = [ squash_dates(o) for o in obj ]
     elif isinstance(obj, datetime):
         obj = obj.strftime(ISO8601)
     return obj
     
 def date_default(obj):
     """default function for packing datetime objects in JSON."""
     if isinstance(obj, datetime):
         return obj.strftime(ISO8601)
     else:
         raise TypeError("%r is not JSON serializable"%obj)
 
 
 
 def json_clean(obj):
     """Clean an object to ensure it's safe to encode in JSON.
     
     Atomic, immutable objects are returned unmodified.  Sets and tuples are
     converted to lists, lists are copied and dicts are also copied.
 
     Note: dicts whose keys could cause collisions upon encoding (such as a dict
     with both the number 1 and the string '1' as keys) will cause a ValueError
     to be raised.
 
     Parameters
     ----------
     obj : any python object
 
     Returns
     -------
     out : object
     
       A version of the input which will not cause an encoding error when
       encoded as JSON.  Note that this function does not *encode* its inputs,
       it simply sanitizes it so that there will be no encoding errors later.
 
     Examples
     --------
     >>> json_clean(4)
     4
     >>> json_clean(range(10))
     [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
     >>> json_clean(dict(x=1, y=2))
     {'y': 2, 'x': 1}
     >>> json_clean(dict(x=1, y=2, z=[1,2,3]))
     {'y': 2, 'x': 1, 'z': [1, 2, 3]}
     >>> json_clean(True)
     True
     """
     # types that are 'atomic' and ok in json as-is.  bool doesn't need to be
     # listed explicitly because bools pass as int instances
     atomic_ok = (unicode, int, float, types.NoneType)
     
     # containers that we need to convert into lists
     container_to_list = (tuple, set, types.GeneratorType)
     
     if isinstance(obj, atomic_ok):
         return obj
     
     if isinstance(obj, bytes):
-        return obj.decode(sys.getdefaultencoding(), 'replace')
+        return obj.decode(text.getdefaultencoding(), 'replace')
     
     if isinstance(obj, container_to_list) or (
         hasattr(obj, '__iter__') and hasattr(obj, next_attr_name)):
         obj = list(obj)
         
     if isinstance(obj, list):
         return [json_clean(x) for x in obj]
 
     if isinstance(obj, dict):
         # First, validate that the dict won't lose data in conversion due to
         # key collisions after stringification.  This can happen with keys like
         # True and 'true' or 1 and '1', which collide in JSON.
         nkeys = len(obj)
         nkeys_collapsed = len(set(map(str, obj)))
         if nkeys != nkeys_collapsed:
             raise ValueError('dict can not be safely converted to JSON: '
                              'key collision would lead to dropped values')
         # If all OK, proceed by making the new dict that will be json-safe
         out = {}
         for k,v in obj.iteritems():
             out[str(k)] = json_clean(v)
         return out
 
     # If we get here, we don't know how to handle the object, so we just get
     # its repr and return that.  This will catch lambdas, open sockets, class
     # objects, and any other complicated contraption that json can't encode
     return repr(obj)

IPython/utils/text.py

@@ -1,715 +1,739 b''
 # encoding: utf-8
 """
 Utilities for working with strings and text.
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2008-2009  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 __main__
 
+import locale
 import os
 import re
 import shutil
+import sys
 import textwrap
 from string import Formatter
 
 from IPython.external.path import path
 from IPython.utils import py3compat
 from IPython.utils.io import nlprint
 from IPython.utils.data import flatten
 
 #-----------------------------------------------------------------------------
 # Code
 #-----------------------------------------------------------------------------
 
+# 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():
+    """Return IPython's guess for the default encoding for bytes as text.
+    
+    Asks for stdin.encoding first, to match the calling Terminal, but that
+    is often None for subprocesses.  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.
+    """
+    enc = sys.stdin.encoding
+    if not enc:
+        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(False)
+        except Exception:
+            pass
+    return enc or sys.getdefaultencoding()
 
 def unquote_ends(istr):
     """Remove a single pair of quotes from the endpoints of a string."""
 
     if not istr:
         return istr
     if (istr[0]=="'" and istr[-1]=="'") or \
        (istr[0]=='"' and istr[-1]=='"'):
         return istr[1:-1]
     else:
         return istr
 
 
 class LSString(str):
     """String derivative with a special access attributes.
 
     These are normal strings, but with the special attributes:
 
         .l (or .list) : value as list (split on newlines).
         .n (or .nlstr): original value (the string itself).
         .s (or .spstr): value as whitespace-separated string.
         .p (or .paths): list of path objects
 
     Any values which require transformations are computed only once and
     cached.
 
     Such strings are very useful to efficiently interact with the shell, which
     typically only understands whitespace-separated options for commands."""
 
     def get_list(self):
         try:
             return self.__list
         except AttributeError:
             self.__list = self.split('\n')
             return self.__list
 
     l = list = property(get_list)
 
     def get_spstr(self):
         try:
             return self.__spstr
         except AttributeError:
             self.__spstr = self.replace('\n',' ')
             return self.__spstr
 
     s = spstr = property(get_spstr)
 
     def get_nlstr(self):
         return self
 
     n = nlstr = property(get_nlstr)
 
     def get_paths(self):
         try:
             return self.__paths
         except AttributeError:
             self.__paths = [path(p) for p in self.split('\n') if os.path.exists(p)]
             return self.__paths
 
     p = paths = property(get_paths)
 
 # FIXME: We need to reimplement type specific displayhook and then add this
 # back as a custom printer. This should also be moved outside utils into the
 # core.
 
 # def print_lsstring(arg):
 #     """ Prettier (non-repr-like) and more informative printer for LSString """
 #     print "LSString (.p, .n, .l, .s available). Value:"
 #     print arg
 # 
 # 
 # print_lsstring = result_display.when_type(LSString)(print_lsstring)
 
 
 class SList(list):
     """List derivative with a special access attributes.
 
     These are normal lists, but with the special attributes:
 
         .l (or .list) : value as list (the list itself).
         .n (or .nlstr): value as a string, joined on newlines.
         .s (or .spstr): value as a string, joined on spaces.
         .p (or .paths): list of path objects
 
     Any values which require transformations are computed only once and
     cached."""
 
     def get_list(self):
         return self
 
     l = list = property(get_list)
 
     def get_spstr(self):
         try:
             return self.__spstr
         except AttributeError:
             self.__spstr = ' '.join(self)
             return self.__spstr
 
     s = spstr = property(get_spstr)
 
     def get_nlstr(self):
         try:
             return self.__nlstr
         except AttributeError:
             self.__nlstr = '\n'.join(self)
             return self.__nlstr
 
     n = nlstr = property(get_nlstr)
 
     def get_paths(self):
         try:
             return self.__paths
         except AttributeError:
             self.__paths = [path(p) for p in self if os.path.exists(p)]
             return self.__paths
 
     p = paths = property(get_paths)
 
     def grep(self, pattern, prune = False, field = None):
         """ Return all strings matching 'pattern' (a regex or callable)
 
         This is case-insensitive. If prune is true, return all items
         NOT matching the pattern.
 
         If field is specified, the match must occur in the specified
         whitespace-separated field.
 
         Examples::
 
             a.grep( lambda x: x.startswith('C') )
             a.grep('Cha.*log', prune=1)
             a.grep('chm', field=-1)
         """
 
         def match_target(s):
             if field is None:
                 return s
             parts = s.split()
             try:
                 tgt = parts[field]
                 return tgt
             except IndexError:
                 return ""
 
         if isinstance(pattern, basestring):
             pred = lambda x : re.search(pattern, x, re.IGNORECASE)
         else:
             pred = pattern
         if not prune:
             return SList([el for el in self if pred(match_target(el))])
         else:
             return SList([el for el in self if not pred(match_target(el))])
 
     def fields(self, *fields):
         """ Collect whitespace-separated fields from string list
 
         Allows quick awk-like usage of string lists.
 
         Example data (in var a, created by 'a = !ls -l')::
             -rwxrwxrwx  1 ville None      18 Dec 14  2006 ChangeLog
             drwxrwxrwx+ 6 ville None       0 Oct 24 18:05 IPython
 
         a.fields(0) is ['-rwxrwxrwx', 'drwxrwxrwx+']
         a.fields(1,0) is ['1 -rwxrwxrwx', '6 drwxrwxrwx+']
         (note the joining by space).
         a.fields(-1) is ['ChangeLog', 'IPython']
 
         IndexErrors are ignored.
 
         Without args, fields() just split()'s the strings.
         """
         if len(fields) == 0:
             return [el.split() for el in self]
 
         res = SList()
         for el in [f.split() for f in self]:
             lineparts = []
 
             for fd in fields:
                 try:
                     lineparts.append(el[fd])
                 except IndexError:
                     pass
             if lineparts:
                 res.append(" ".join(lineparts))
 
         return res
 
     def sort(self,field= None,  nums = False):
         """ sort by specified fields (see fields())
 
         Example::
             a.sort(1, nums = True)
 
         Sorts a by second field, in numerical order (so that 21 > 3)
 
         """
 
         #decorate, sort, undecorate
         if field is not None:
             dsu = [[SList([line]).fields(field),  line] for line in self]
         else:
             dsu = [[line,  line] for line in self]
         if nums:
             for i in range(len(dsu)):
                 numstr = "".join([ch for ch in dsu[i][0] if ch.isdigit()])
                 try:
                     n = int(numstr)
                 except ValueError:
                     n = 0;
                 dsu[i][0] = n
 
 
         dsu.sort()
         return SList([t[1] for t in dsu])
 
 
 # FIXME: We need to reimplement type specific displayhook and then add this
 # back as a custom printer. This should also be moved outside utils into the
 # core.
 
 # def print_slist(arg):
 #     """ Prettier (non-repr-like) and more informative printer for SList """
 #     print "SList (.p, .n, .l, .s, .grep(), .fields(), sort() available):"
 #     if hasattr(arg,  'hideonce') and arg.hideonce:
 #         arg.hideonce = False
 #         return
 # 
 #     nlprint(arg)
 # 
 # print_slist = result_display.when_type(SList)(print_slist)
 
 
 def esc_quotes(strng):
     """Return the input string with single and double quotes escaped out"""
 
     return strng.replace('"','\\"').replace("'","\\'")
 
 
 def make_quoted_expr(s):
     """Return string s in appropriate quotes, using raw string if possible.
 
     XXX - example removed because it caused encoding errors in documentation
     generation.  We need a new example that doesn't contain invalid chars.
 
     Note the use of raw string and padding at the end to allow trailing
     backslash.
     """
 
     tail = ''
     tailpadding = ''
     raw  = ''
     ucode = '' if py3compat.PY3 else 'u'
     if "\\" in s:
         raw = 'r'
         if s.endswith('\\'):
             tail = '[:-1]'
             tailpadding = '_'
     if '"' not in s:
         quote = '"'
     elif "'" not in s:
         quote = "'"
     elif '"""' not in s and not s.endswith('"'):
         quote = '"""'
     elif "'''" not in s and not s.endswith("'"):
         quote = "'''"
     else:
         # give up, backslash-escaped string will do
         return '"%s"' % esc_quotes(s)
     res = ucode + raw + quote + s + tailpadding + quote + tail
     return res
 
 
 def qw(words,flat=0,sep=None,maxsplit=-1):
     """Similar to Perl's qw() operator, but with some more options.
 
     qw(words,flat=0,sep=' ',maxsplit=-1) -> words.split(sep,maxsplit)
 
     words can also be a list itself, and with flat=1, the output will be
     recursively flattened.
 
     Examples:
 
     >>> qw('1 2')
     ['1', '2']
 
     >>> qw(['a b','1 2',['m n','p q']])
     [['a', 'b'], ['1', '2'], [['m', 'n'], ['p', 'q']]]
 
     >>> qw(['a b','1 2',['m n','p q']],flat=1)
     ['a', 'b', '1', '2', 'm', 'n', 'p', 'q']
     """
 
     if isinstance(words, basestring):
         return [word.strip() for word in words.split(sep,maxsplit)
                 if word and not word.isspace() ]
     if flat:
         return flatten(map(qw,words,[1]*len(words)))
     return map(qw,words)
 
 
 def qwflat(words,sep=None,maxsplit=-1):
     """Calls qw(words) in flat mode. It's just a convenient shorthand."""
     return qw(words,1,sep,maxsplit)
 
 
 def qw_lol(indata):
     """qw_lol('a b') -> [['a','b']],
     otherwise it's just a call to qw().
 
     We need this to make sure the modules_some keys *always* end up as a
     list of lists."""
 
     if isinstance(indata, basestring):
         return [qw(indata)]
     else:
         return qw(indata)
 
 
 def grep(pat,list,case=1):
     """Simple minded grep-like function.
     grep(pat,list) returns occurrences of pat in list, None on failure.
 
     It only does simple string matching, with no support for regexps. Use the
     option case=0 for case-insensitive matching."""
 
     # This is pretty crude. At least it should implement copying only references
     # to the original data in case it's big. Now it copies the data for output.
     out=[]
     if case:
         for term in list:
             if term.find(pat)>-1: out.append(term)
     else:
         lpat=pat.lower()
         for term in list:
             if term.lower().find(lpat)>-1: out.append(term)
 
     if len(out): return out
     else: return None
 
 
 def dgrep(pat,*opts):
     """Return grep() on dir()+dir(__builtins__).
 
     A very common use of grep() when working interactively."""
 
     return grep(pat,dir(__main__)+dir(__main__.__builtins__),*opts)
 
 
 def idgrep(pat):
     """Case-insensitive dgrep()"""
 
     return dgrep(pat,0)
 
 
 def igrep(pat,list):
     """Synonym for case-insensitive grep."""
 
     return grep(pat,list,case=0)
 
 
 def indent(instr,nspaces=4, ntabs=0, flatten=False):
     """Indent a string a given number of spaces or tabstops.
 
     indent(str,nspaces=4,ntabs=0) -> indent str by ntabs+nspaces.
     
     Parameters
     ----------
     
     instr : basestring
         The string to be indented.
     nspaces : int (default: 4)
         The number of spaces to be indented.
     ntabs : int (default: 0)
         The number of tabs to be indented.
     flatten : bool (default: False)
         Whether to scrub existing indentation.  If True, all lines will be
         aligned to the same indentation.  If False, existing indentation will
         be strictly increased.
     
     Returns
     -------
     
     str|unicode : string indented by ntabs and nspaces.
     
     """
     if instr is None:
         return
     ind = '\t'*ntabs+' '*nspaces
     if flatten:
         pat = re.compile(r'^\s*', re.MULTILINE)
     else:
         pat = re.compile(r'^', re.MULTILINE)
     outstr = re.sub(pat, ind, instr)
     if outstr.endswith(os.linesep+ind):
         return outstr[:-len(ind)]
     else:
         return outstr
 
 def native_line_ends(filename,backup=1):
     """Convert (in-place) a file to line-ends native to the current OS.
 
     If the optional backup argument is given as false, no backup of the
     original file is left.  """
 
     backup_suffixes = {'posix':'~','dos':'.bak','nt':'.bak','mac':'.bak'}
 
     bak_filename = filename + backup_suffixes[os.name]
 
     original = open(filename).read()
     shutil.copy2(filename,bak_filename)
     try:
         new = open(filename,'wb')
         new.write(os.linesep.join(original.splitlines()))
         new.write(os.linesep) # ALWAYS put an eol at the end of the file
         new.close()
     except:
         os.rename(bak_filename,filename)
     if not backup:
         try:
             os.remove(bak_filename)
         except:
             pass
 
 
 def list_strings(arg):
     """Always return a list of strings, given a string or list of strings
     as input.
 
     :Examples:
 
         In [7]: list_strings('A single string')
         Out[7]: ['A single string']
 
         In [8]: list_strings(['A single string in a list'])
         Out[8]: ['A single string in a list']
 
         In [9]: list_strings(['A','list','of','strings'])
         Out[9]: ['A', 'list', 'of', 'strings']
     """
 
     if isinstance(arg,basestring): return [arg]
     else: return arg
 
 
 def marquee(txt='',width=78,mark='*'):
     """Return the input string centered in a 'marquee'.
 
     :Examples:
 
         In [16]: marquee('A test',40)
         Out[16]: '**************** A test ****************'
 
         In [17]: marquee('A test',40,'-')
         Out[17]: '---------------- A test ----------------'
 
         In [18]: marquee('A test',40,' ')
         Out[18]: '                 A test                 '
 
     """
     if not txt:
         return (mark*width)[:width]
     nmark = (width-len(txt)-2)//len(mark)//2
     if nmark < 0: nmark =0
     marks = mark*nmark
     return '%s %s %s' % (marks,txt,marks)
 
 
 ini_spaces_re = re.compile(r'^(\s+)')
 
 def num_ini_spaces(strng):
     """Return the number of initial spaces in a string"""
 
     ini_spaces = ini_spaces_re.match(strng)
     if ini_spaces:
         return ini_spaces.end()
     else:
         return 0
 
 
 def format_screen(strng):
     """Format a string for screen printing.
 
     This removes some latex-type format codes."""
     # Paragraph continue
     par_re = re.compile(r'\\$',re.MULTILINE)
     strng = par_re.sub('',strng)
     return strng
 
 def dedent(text):
     """Equivalent of textwrap.dedent that ignores unindented first line.
     
     This means it will still dedent strings like:
     '''foo
     is a bar
     '''
     
     For use in wrap_paragraphs.
     """
     
     if text.startswith('\n'):
         # text starts with blank line, don't ignore the first line
         return textwrap.dedent(text)
     
     # split first line
     splits = text.split('\n',1)
     if len(splits) == 1:
         # only one line
         return textwrap.dedent(text)
     
     first, rest = splits
     # dedent everything but the first line
     rest = textwrap.dedent(rest)
     return '\n'.join([first, rest])
 
 def wrap_paragraphs(text, ncols=80):
     """Wrap multiple paragraphs to fit a specified width.
     
     This is equivalent to textwrap.wrap, but with support for multiple
     paragraphs, as separated by empty lines.
     
     Returns
     -------
     
     list of complete paragraphs, wrapped to fill `ncols` columns.
     """
     paragraph_re = re.compile(r'\n(\s*\n)+', re.MULTILINE)
     text = dedent(text).strip()
     paragraphs = paragraph_re.split(text)[::2] # every other entry is space
     out_ps = []
     indent_re = re.compile(r'\n\s+', re.MULTILINE)
     for p in paragraphs:
         # presume indentation that survives dedent is meaningful formatting, 
         # so don't fill unless text is flush.
         if indent_re.search(p) is None:
             # wrap paragraph
             p = textwrap.fill(p, ncols)
         out_ps.append(p)
     return out_ps
     
 
 
 class EvalFormatter(Formatter):
     """A String Formatter that allows evaluation of simple expressions.
     
     Any time a format key is not found in the kwargs,
     it will be tried as an expression in the kwargs namespace.
     
     This is to be used in templating cases, such as the parallel batch
     script templates, where simple arithmetic on arguments is useful.
     
     Examples
     --------
     
     In [1]: f = EvalFormatter()
     In [2]: f.format('{n//4}', n=8)
     Out[2]: '2'
     
     In [3]: f.format('{list(range(3))}')
     Out[3]: '[0, 1, 2]'
 
     In [4]: f.format('{3*2}')
     Out[4]: '6'
     """
     
     # should we allow slicing by disabling the format_spec feature?
     allow_slicing = True
     
     # copied from Formatter._vformat with minor changes to allow eval
     # and replace the format_spec code with slicing
     def _vformat(self, format_string, args, kwargs, used_args, recursion_depth):
         if recursion_depth < 0:
             raise ValueError('Max string recursion exceeded')
         result = []
         for literal_text, field_name, format_spec, conversion in \
                 self.parse(format_string):
 
             # output the literal text
             if literal_text:
                 result.append(literal_text)
 
             # if there's a field, output it
             if field_name is not None:
                 # this is some markup, find the object and do
                 #  the formatting
 
                 if self.allow_slicing and format_spec:
                     # override format spec, to allow slicing:
                     field_name = ':'.join([field_name, format_spec])
                     format_spec = ''
 
                 # eval the contents of the field for the object
                 # to be formatted
                 obj = eval(field_name, kwargs)
 
                 # do any conversion on the resulting object
                 obj = self.convert_field(obj, conversion)
 
                 # expand the format spec, if needed
                 format_spec = self._vformat(format_spec, args, kwargs,
                                             used_args, recursion_depth-1)
 
                 # format the object and append to the result
                 result.append(self.format_field(obj, format_spec))
 
         return ''.join(result)
 
 
 def columnize(items, separator='  ', displaywidth=80):
     """ Transform a list of strings into a single string with columns.
 
     Parameters
     ----------
     items : sequence of strings
         The strings to process.
 
     separator : str, optional [default is two spaces]
         The string that separates columns.
 
     displaywidth : int, optional [default is 80]
         Width of the display in number of characters.
     
     Returns
     -------
     The formatted string.
     """
     # Note: this code is adapted from columnize 0.3.2.
     # See http://code.google.com/p/pycolumnize/
 
     # Some degenerate cases.
     size = len(items)
     if size == 0: 
         return '\n'
     elif size == 1:
         return '%s\n' % items[0]
 
     # Special case: if any item is longer than the maximum width, there's no
     # point in triggering the logic below...
     item_len = map(len, items) # save these, we can reuse them below
     longest = max(item_len)
     if longest >= displaywidth:
         return '\n'.join(items+[''])
 
     # Try every row count from 1 upwards
     array_index = lambda nrows, row, col: nrows*col + row
     for nrows in range(1, size):
         ncols = (size + nrows - 1) // nrows
         colwidths = []
         totwidth = -len(separator)
         for col in range(ncols):
             # Get max column width for this column
             colwidth = 0
             for row in range(nrows):
                 i = array_index(nrows, row, col)
                 if i >= size: break
                 x, len_x = items[i], item_len[i]
                 colwidth = max(colwidth, len_x)
             colwidths.append(colwidth)
             totwidth += colwidth + len(separator)
             if totwidth > displaywidth: 
                 break
         if totwidth <= displaywidth: 
             break
 
     # The smallest number of rows computed and the max widths for each
     # column has been obtained. Now we just have to format each of the rows.
     string = ''
     for row in range(nrows):
         texts = []
         for col in range(ncols):
             i = row + nrows*col
             if i >= size:
                 texts.append('')
             else:
                 texts.append(items[i])
         while texts and not texts[-1]:
             del texts[-1]
         for col in range(len(texts)):
             texts[col] = texts[col].ljust(colwidths[col])
         string += '%s\n' % separator.join(texts)
     return string

IPython/zmq/iostream.py

@@ -1,91 +1,91 b''
 import sys
 import time
 from io import StringIO
 
 from session import extract_header, Message
 
-from IPython.utils import io
+from IPython.utils import io, text
 
 #-----------------------------------------------------------------------------
 # Globals
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Stream classes
 #-----------------------------------------------------------------------------
 
 class OutStream(object):
     """A file like object that publishes the stream to a 0MQ PUB socket."""
 
     # The time interval between automatic flushes, in seconds.
     flush_interval = 0.05
     topic=None
 
     def __init__(self, session, pub_socket, name):
         self.session = session
         self.pub_socket = pub_socket
         self.name = name
         self.parent_header = {}
         self._new_buffer()
 
     def set_parent(self, parent):
         self.parent_header = extract_header(parent)
 
     def close(self):
         self.pub_socket = None
 
     def flush(self):
         #io.rprint('>>>flushing output buffer: %s<<<' % self.name)  # dbg
         if self.pub_socket is None:
             raise ValueError(u'I/O operation on closed file')
         else:
             data = self._buffer.getvalue()
             if data:
                 content = {u'name':self.name, u'data':data}
                 msg = self.session.send(self.pub_socket, u'stream', content=content,
                                        parent=self.parent_header, ident=self.topic)
 
                 if hasattr(self.pub_socket, 'flush'):
                     # socket itself has flush (presumably ZMQStream)
                     self.pub_socket.flush()
                 self._buffer.close()
                 self._new_buffer()
 
     def isatty(self):
         return False
 
     def next(self):
         raise IOError('Read not supported on a write only stream.')
 
     def read(self, size=-1):
         raise IOError('Read not supported on a write only stream.')
 
     def readline(self, size=-1):
         raise IOError('Read not supported on a write only stream.')
 
     def write(self, string):
         if self.pub_socket is None:
             raise ValueError('I/O operation on closed file')
         else:
             # Make sure that we're handling unicode
             if not isinstance(string, unicode):
-                enc = sys.stdin.encoding or sys.getdefaultencoding()
+                enc = text.getdefaultencoding()
                 string = string.decode(enc, 'replace')
             
             self._buffer.write(string)
             current_time = time.time()
             if self._start <= 0:
                 self._start = current_time
             elif current_time - self._start > self.flush_interval:
                 self.flush()
 
     def writelines(self, sequence):
         if self.pub_socket is None:
             raise ValueError('I/O operation on closed file')
         else:
             for string in sequence:
                 self.write(string)
 
     def _new_buffer(self):
         self._buffer = StringIO()
         self._start = -1

IPython/zmq/ipkernel.py

@@ -1,699 +1,702 b''
 #!/usr/bin/env python
 """A simple interactive kernel that talks to a frontend over 0MQ.
 
 Things to do:
 
 * Implement `set_parent` logic. Right before doing exec, the Kernel should
   call set_parent on all the PUB objects with the message about to be executed.
 * Implement random port and security key logic.
 * Implement control messages.
 * Implement event loop and poll version.
 """
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 from __future__ import print_function
 
 # Standard library imports.
 import __builtin__
 import atexit
 import sys
 import time
 import traceback
 import logging
 # System library imports.
 import zmq
 
 # Local imports.
 from IPython.config.configurable import Configurable
 from IPython.config.application import boolean_flag
 from IPython.core.application import ProfileDir
 from IPython.core.shellapp import (
     InteractiveShellApp, shell_flags, shell_aliases
 )
 from IPython.utils import io
 from IPython.utils import py3compat
 from IPython.utils.jsonutil import json_clean
 from IPython.lib import pylabtools
 from IPython.utils.traitlets import (
     List, Instance, Float, Dict, Bool, Int, Unicode, CaselessStrEnum
 )
 
 from entry_point import base_launch_kernel
 from kernelapp import KernelApp, kernel_flags, kernel_aliases
 from iostream import OutStream
 from session import Session, Message
 from zmqshell import ZMQInteractiveShell
 
 
 #-----------------------------------------------------------------------------
 # Main kernel class
 #-----------------------------------------------------------------------------
 
 class Kernel(Configurable):
 
     #---------------------------------------------------------------------------
     # Kernel interface
     #---------------------------------------------------------------------------
 
     shell = Instance('IPython.core.interactiveshell.InteractiveShellABC')
     session = Instance(Session)
     shell_socket = Instance('zmq.Socket')
     iopub_socket = Instance('zmq.Socket')
     stdin_socket = Instance('zmq.Socket')
     log = Instance(logging.Logger)
 
     # Private interface
 
     # Time to sleep after flushing the stdout/err buffers in each execute
     # cycle.  While this introduces a hard limit on the minimal latency of the
     # execute cycle, it helps prevent output synchronization problems for
     # clients.
     # Units are in seconds.  The minimum zmq latency on local host is probably
     # ~150 microseconds, set this to 500us for now.  We may need to increase it
     # a little if it's not enough after more interactive testing.
     _execute_sleep = Float(0.0005, config=True)
 
     # Frequency of the kernel's event loop.
     # Units are in seconds, kernel subclasses for GUI toolkits may need to
     # adapt to milliseconds.
     _poll_interval = Float(0.05, config=True)
 
     # If the shutdown was requested over the network, we leave here the
     # necessary reply message so it can be sent by our registered atexit
     # handler.  This ensures that the reply is only sent to clients truly at
     # the end of our shutdown process (which happens after the underlying
     # IPython shell's own shutdown).
     _shutdown_message = None
 
     # This is a dict of port number that the kernel is listening on. It is set
     # by record_ports and used by connect_request.
     _recorded_ports = Dict()
 
 
 
     def __init__(self, **kwargs):
         super(Kernel, self).__init__(**kwargs)
 
         # Before we even start up the shell, register *first* our exit handlers
         # so they come before the shell's
         atexit.register(self._at_shutdown)
 
         # Initialize the InteractiveShell subclass
         self.shell = ZMQInteractiveShell.instance(config=self.config)
         self.shell.displayhook.session = self.session
         self.shell.displayhook.pub_socket = self.iopub_socket
         self.shell.display_pub.session = self.session
         self.shell.display_pub.pub_socket = self.iopub_socket
 
         # TMP - hack while developing
         self.shell._reply_content = None
 
         # Build dict of handlers for message types
         msg_types = [ 'execute_request', 'complete_request', 
                       'object_info_request', 'history_request',
                       'connect_request', 'shutdown_request']
         self.handlers = {}
         for msg_type in msg_types:
             self.handlers[msg_type] = getattr(self, msg_type)
 
     def do_one_iteration(self):
         """Do one iteration of the kernel's evaluation loop.
         """
         try:
             ident,msg = self.session.recv(self.shell_socket, zmq.NOBLOCK)
         except Exception:
             self.log.warn("Invalid Message:", exc_info=True)
             return
         if msg is None:
             return
 
         msg_type = msg['header']['msg_type']
 
         # This assert will raise in versions of zeromq 2.0.7 and lesser.
         # We now require 2.0.8 or above, so we can uncomment for safety.
         # print(ident,msg, file=sys.__stdout__)
         assert ident is not None, "Missing message part."
         
         # Print some info about this message and leave a '--->' marker, so it's
         # easier to trace visually the message chain when debugging.  Each
         # handler prints its message at the end.
         self.log.debug('\n*** MESSAGE TYPE:'+str(msg_type)+'***')
         self.log.debug('   Content: '+str(msg['content'])+'\n   --->\n   ')
 
         # Find and call actual handler for message
         handler = self.handlers.get(msg_type, None)
         if handler is None:
             self.log.error("UNKNOWN MESSAGE TYPE:" +str(msg))
         else:
             handler(ident, msg)
             
         # Check whether we should exit, in case the incoming message set the
         # exit flag on
         if self.shell.exit_now:
             self.log.debug('\nExiting IPython kernel...')
             # We do a normal, clean exit, which allows any actions registered
             # via atexit (such as history saving) to take place.
             sys.exit(0)
 
 
     def start(self):
         """ Start the kernel main loop.
         """
         poller = zmq.Poller()
         poller.register(self.shell_socket, zmq.POLLIN)
         while True:
             try:
                 # scale by extra factor of 10, because there is no
                 # reason for this to be anything less than ~ 0.1s
                 # since it is a real poller and will respond
                 # to events immediately
                 
                 # double nested try/except, to properly catch KeyboardInterrupt
                 # due to pyzmq Issue #130
                 try:
                     poller.poll(10*1000*self._poll_interval)
                     self.do_one_iteration()
                 except:
                     raise
             except KeyboardInterrupt:
                 # Ctrl-C shouldn't crash the kernel
                 io.raw_print("KeyboardInterrupt caught in kernel")
 
     def record_ports(self, ports):
         """Record the ports that this kernel is using.
 
         The creator of the Kernel instance must call this methods if they
         want the :meth:`connect_request` method to return the port numbers.
         """
         self._recorded_ports = ports
 
     #---------------------------------------------------------------------------
     # Kernel request handlers
     #---------------------------------------------------------------------------
 
     def _publish_pyin(self, code, parent):
         """Publish the code request on the pyin stream."""
 
         pyin_msg = self.session.send(self.iopub_socket, u'pyin',{u'code':code}, parent=parent)
 
     def execute_request(self, ident, parent):
         
         status_msg = self.session.send(self.iopub_socket,
             u'status',
             {u'execution_state':u'busy'},
             parent=parent
         )
         
         try:
             content = parent[u'content']
             code = content[u'code']
             silent = content[u'silent'] 
         except:
             self.log.error("Got bad msg: ")
             self.log.error(str(Message(parent)))
             return
 
         shell = self.shell # we'll need this a lot here
 
         # Replace raw_input. Note that is not sufficient to replace 
         # raw_input in the user namespace.
         raw_input = lambda prompt='': self._raw_input(prompt, ident, parent)
         if py3compat.PY3:
             __builtin__.input = raw_input
         else:
             __builtin__.raw_input = raw_input
 
         # Set the parent message of the display hook and out streams.
         shell.displayhook.set_parent(parent)
         shell.display_pub.set_parent(parent)
         sys.stdout.set_parent(parent)
         sys.stderr.set_parent(parent)
 
         # Re-broadcast our input for the benefit of listening clients, and
         # start computing output
         if not silent:
             self._publish_pyin(code, parent)
 
         reply_content = {}
         try:
             if silent:
                 # run_code uses 'exec' mode, so no displayhook will fire, and it
                 # doesn't call logging or history manipulations.  Print
                 # statements in that code will obviously still execute.
                 shell.run_code(code)
             else:
                 # FIXME: the shell calls the exception handler itself.
                 shell.run_cell(code)
         except:
             status = u'error'
             # FIXME: this code right now isn't being used yet by default,
             # because the run_cell() call above directly fires off exception
             # reporting.  This code, therefore, is only active in the scenario
             # where runlines itself has an unhandled exception.  We need to
             # uniformize this, for all exception construction to come from a
             # single location in the codbase.
             etype, evalue, tb = sys.exc_info()
             tb_list = traceback.format_exception(etype, evalue, tb)
             reply_content.update(shell._showtraceback(etype, evalue, tb_list))
         else:
             status = u'ok'
 
         reply_content[u'status'] = status
         
         # Return the execution counter so clients can display prompts
         reply_content['execution_count'] = shell.execution_count -1
 
         # FIXME - fish exception info out of shell, possibly left there by
         # runlines.  We'll need to clean up this logic later.
         if shell._reply_content is not None:
             reply_content.update(shell._reply_content)
             # reset after use
             shell._reply_content = None
 
         # At this point, we can tell whether the main code execution succeeded
         # or not.  If it did, we proceed to evaluate user_variables/expressions
         if reply_content['status'] == 'ok':
             reply_content[u'user_variables'] = \
                          shell.user_variables(content[u'user_variables'])
             reply_content[u'user_expressions'] = \
                          shell.user_expressions(content[u'user_expressions'])
         else:
             # If there was an error, don't even try to compute variables or
             # expressions
             reply_content[u'user_variables'] = {}
             reply_content[u'user_expressions'] = {}
 
         # Payloads should be retrieved regardless of outcome, so we can both
         # recover partial output (that could have been generated early in a
         # block, before an error) and clear the payload system always.
         reply_content[u'payload'] = shell.payload_manager.read_payload()
         # Be agressive about clearing the payload because we don't want
         # it to sit in memory until the next execute_request comes in.
         shell.payload_manager.clear_payload()
 
         # Flush output before sending the reply.
         sys.stdout.flush()
         sys.stderr.flush()
         # FIXME: on rare occasions, the flush doesn't seem to make it to the
         # clients... This seems to mitigate the problem, but we definitely need
         # to better understand what's going on.
         if self._execute_sleep:
             time.sleep(self._execute_sleep)
         
         # Send the reply.
+        reply_content = json_clean(reply_content)
         reply_msg = self.session.send(self.shell_socket, u'execute_reply',
                                       reply_content, parent, ident=ident)
         self.log.debug(str(reply_msg))
 
         if reply_msg['content']['status'] == u'error':
             self._abort_queue()
 
         status_msg = self.session.send(self.iopub_socket,
             u'status',
             {u'execution_state':u'idle'},
             parent=parent
         )
 
     def complete_request(self, ident, parent):
         txt, matches = self._complete(parent)
         matches = {'matches' : matches,
                    'matched_text' : txt,
                    'status' : 'ok'}
+        matches = json_clean(matches)
         completion_msg = self.session.send(self.shell_socket, 'complete_reply',
                                            matches, parent, ident)
         self.log.debug(str(completion_msg))
 
     def object_info_request(self, ident, parent):
         object_info = self.shell.object_inspect(parent['content']['oname'])
         # Before we send this object over, we scrub it for JSON usage
         oinfo = json_clean(object_info)
         msg = self.session.send(self.shell_socket, 'object_info_reply',
                                 oinfo, parent, ident)
         self.log.debug(msg)
 
     def history_request(self, ident, parent):
         # We need to pull these out, as passing **kwargs doesn't work with
         # unicode keys before Python 2.6.5.
         hist_access_type = parent['content']['hist_access_type']
         raw = parent['content']['raw']
         output = parent['content']['output']
         if hist_access_type == 'tail':
             n = parent['content']['n']
             hist = self.shell.history_manager.get_tail(n, raw=raw, output=output,
                                                             include_latest=True)
         
         elif hist_access_type == 'range':
             session = parent['content']['session']
             start = parent['content']['start']
             stop = parent['content']['stop']
             hist = self.shell.history_manager.get_range(session, start, stop,
                                                         raw=raw, output=output)
         
         elif hist_access_type == 'search':
             pattern = parent['content']['pattern']
             hist = self.shell.history_manager.search(pattern, raw=raw, output=output)
         
         else:
             hist = []
         content = {'history' : list(hist)}
+        content = json_clean(content)
         msg = self.session.send(self.shell_socket, 'history_reply',
                                 content, parent, ident)
         self.log.debug(str(msg))
 
     def connect_request(self, ident, parent):
         if self._recorded_ports is not None:
             content = self._recorded_ports.copy()
         else:
             content = {}
         msg = self.session.send(self.shell_socket, 'connect_reply',
                                 content, parent, ident)
         self.log.debug(msg)
 
     def shutdown_request(self, ident, parent):
         self.shell.exit_now = True
         self._shutdown_message = self.session.msg(u'shutdown_reply', parent['content'], parent)
         sys.exit(0)
 
     #---------------------------------------------------------------------------
     # Protected interface
     #---------------------------------------------------------------------------
 
     def _abort_queue(self):
         while True:
             try:
                 ident,msg = self.session.recv(self.shell_socket, zmq.NOBLOCK)
             except Exception:
                 self.log.warn("Invalid Message:", exc_info=True)
                 continue
             if msg is None:
                 break
             else:
                 assert ident is not None, \
                        "Unexpected missing message part."
 
             self.log.debug("Aborting:\n"+str(Message(msg)))
             msg_type = msg['header']['msg_type']
             reply_type = msg_type.split('_')[0] + '_reply'
             reply_msg = self.session.send(self.shell_socket, reply_type,
                     {'status' : 'aborted'}, msg, ident=ident)
             self.log.debug(reply_msg)
             # We need to wait a bit for requests to come in. This can probably
             # be set shorter for true asynchronous clients.
             time.sleep(0.1)
 
     def _raw_input(self, prompt, ident, parent):
         # Flush output before making the request.
         sys.stderr.flush()
         sys.stdout.flush()
 
         # Send the input request.
-        content = dict(prompt=prompt)
+        content = json_clean(dict(prompt=prompt))
         msg = self.session.send(self.stdin_socket, u'input_request', content, parent)
 
         # Await a response.
         while True:
             try:
                 ident, reply = self.session.recv(self.stdin_socket, 0)
             except Exception:
                 self.log.warn("Invalid Message:", exc_info=True)
             else:
                 break
         try:
             value = reply['content']['value']
         except:
             self.log.error("Got bad raw_input reply: ")
             self.log.error(str(Message(parent)))
             value = ''
         return value
     
     def _complete(self, msg):
         c = msg['content']
         try:
             cpos = int(c['cursor_pos'])
         except:
             # If we don't get something that we can convert to an integer, at
             # least attempt the completion guessing the cursor is at the end of
             # the text, if there's any, and otherwise of the line
             cpos = len(c['text'])
             if cpos==0:
                 cpos = len(c['line'])
         return self.shell.complete(c['text'], c['line'], cpos)
 
     def _object_info(self, context):
         symbol, leftover = self._symbol_from_context(context)
         if symbol is not None and not leftover:
             doc = getattr(symbol, '__doc__', '')
         else:
             doc = ''
         object_info = dict(docstring = doc)
         return object_info
 
     def _symbol_from_context(self, context):
         if not context:
             return None, context
 
         base_symbol_string = context[0]
         symbol = self.shell.user_ns.get(base_symbol_string, None)
         if symbol is None:
             symbol = __builtin__.__dict__.get(base_symbol_string, None)
         if symbol is None:
             return None, context
 
         context = context[1:]
         for i, name in enumerate(context):
             new_symbol = getattr(symbol, name, None)
             if new_symbol is None:
                 return symbol, context[i:]
             else:
                 symbol = new_symbol
 
         return symbol, []
 
     def _at_shutdown(self):
         """Actions taken at shutdown by the kernel, called by python's atexit.
         """
         # io.rprint("Kernel at_shutdown") # dbg
         if self._shutdown_message is not None:
             self.session.send(self.shell_socket, self._shutdown_message)
             self.session.send(self.iopub_socket, self._shutdown_message)
             self.log.debug(str(self._shutdown_message))
             # A very short sleep to give zmq time to flush its message buffers
             # before Python truly shuts down.
             time.sleep(0.01)
 
 
 class QtKernel(Kernel):
     """A Kernel subclass with Qt support."""
 
     def start(self):
         """Start a kernel with QtPy4 event loop integration."""
 
         from IPython.external.qt_for_kernel import QtCore
         from IPython.lib.guisupport import get_app_qt4, start_event_loop_qt4
 
         self.app = get_app_qt4([" "])
         self.app.setQuitOnLastWindowClosed(False)
         self.timer = QtCore.QTimer()
         self.timer.timeout.connect(self.do_one_iteration)
         # Units for the timer are in milliseconds
         self.timer.start(1000*self._poll_interval)
         start_event_loop_qt4(self.app)
 
 
 class WxKernel(Kernel):
     """A Kernel subclass with Wx support."""
 
     def start(self):
         """Start a kernel with wx event loop support."""
 
         import wx
         from IPython.lib.guisupport import start_event_loop_wx
 
         doi = self.do_one_iteration
          # Wx uses milliseconds
         poll_interval = int(1000*self._poll_interval)
 
         # We have to put the wx.Timer in a wx.Frame for it to fire properly.
         # We make the Frame hidden when we create it in the main app below.
         class TimerFrame(wx.Frame):
             def __init__(self, func):
                 wx.Frame.__init__(self, None, -1)
                 self.timer = wx.Timer(self)
                 # Units for the timer are in milliseconds
                 self.timer.Start(poll_interval)
                 self.Bind(wx.EVT_TIMER, self.on_timer)
                 self.func = func
 
             def on_timer(self, event):
                 self.func()
 
         # We need a custom wx.App to create our Frame subclass that has the
         # wx.Timer to drive the ZMQ event loop.
         class IPWxApp(wx.App):
             def OnInit(self):
                 self.frame = TimerFrame(doi)
                 self.frame.Show(False)
                 return True
 
         # The redirect=False here makes sure that wx doesn't replace
         # sys.stdout/stderr with its own classes.
         self.app = IPWxApp(redirect=False)
         start_event_loop_wx(self.app)
 
 
 class TkKernel(Kernel):
     """A Kernel subclass with Tk support."""
 
     def start(self):
         """Start a Tk enabled event loop."""
 
         import Tkinter
         doi = self.do_one_iteration
         # Tk uses milliseconds
         poll_interval = int(1000*self._poll_interval)
         # For Tkinter, we create a Tk object and call its withdraw method.
         class Timer(object):
             def __init__(self, func):
                 self.app = Tkinter.Tk()
                 self.app.withdraw()
                 self.func = func
                 
             def on_timer(self):
                 self.func()
                 self.app.after(poll_interval, self.on_timer)
                 
             def start(self):
                 self.on_timer()  # Call it once to get things going.
                 self.app.mainloop()
 
         self.timer = Timer(doi)
         self.timer.start()
 
 
 class GTKKernel(Kernel):
     """A Kernel subclass with GTK support."""
     
     def start(self):
         """Start the kernel, coordinating with the GTK event loop"""
         from .gui.gtkembed import GTKEmbed
         
         gtk_kernel = GTKEmbed(self)
         gtk_kernel.start()
 
 
 #-----------------------------------------------------------------------------
 # Aliases and Flags for the IPKernelApp
 #-----------------------------------------------------------------------------
 
 flags = dict(kernel_flags)
 flags.update(shell_flags)
 
 addflag = lambda *args: flags.update(boolean_flag(*args))
 
 flags['pylab'] = (
     {'IPKernelApp' : {'pylab' : 'auto'}},
     """Pre-load matplotlib and numpy for interactive use with
     the default matplotlib backend."""
 )
 
 aliases = dict(kernel_aliases)
 aliases.update(shell_aliases)
 
 # it's possible we don't want short aliases for *all* of these:
 aliases.update(dict(
     pylab='IPKernelApp.pylab',
 ))
 
 #-----------------------------------------------------------------------------
 # The IPKernelApp class
 #-----------------------------------------------------------------------------
 
 class IPKernelApp(KernelApp, InteractiveShellApp):
     name = 'ipkernel'
 
     aliases = Dict(aliases)
     flags = Dict(flags)
     classes = [Kernel, ZMQInteractiveShell, ProfileDir, Session]
     # configurables
     pylab = CaselessStrEnum(['tk', 'qt', 'wx', 'gtk', 'osx', 'inline', 'auto'],
         config=True,
         help="""Pre-load matplotlib and numpy for interactive use,
         selecting a particular matplotlib backend and loop integration.
         """
     )
     def initialize(self, argv=None):
         super(IPKernelApp, self).initialize(argv)
         self.init_shell()
         self.init_extensions()
         self.init_code()
 
     def init_kernel(self):
         kernel_factory = Kernel
 
         kernel_map = {
             'qt' : QtKernel,
             'qt4': QtKernel,
             'inline': Kernel,
             'osx': TkKernel,
             'wx' : WxKernel,
             'tk' : TkKernel,
             'gtk': GTKKernel,
         }
 
         if self.pylab:
             key = None if self.pylab == 'auto' else self.pylab
             gui, backend = pylabtools.find_gui_and_backend(key)
             kernel_factory = kernel_map.get(gui)
             if kernel_factory is None:
                 raise ValueError('GUI is not supported: %r' % gui)
             pylabtools.activate_matplotlib(backend)
 
         kernel = kernel_factory(config=self.config, session=self.session,
                                 shell_socket=self.shell_socket,
                                 iopub_socket=self.iopub_socket,
                                 stdin_socket=self.stdin_socket,
                                 log=self.log
         )
         self.kernel = kernel
         kernel.record_ports(self.ports)
 
         if self.pylab:
             import_all = self.pylab_import_all
             pylabtools.import_pylab(kernel.shell.user_ns, backend, import_all,
                                     shell=kernel.shell)
     
     def init_shell(self):
         self.shell = self.kernel.shell
 
 
 #-----------------------------------------------------------------------------
 # Kernel main and launch functions
 #-----------------------------------------------------------------------------
 
 def launch_kernel(*args, **kwargs):
     """Launches a localhost IPython kernel, binding to the specified ports.
 
     This function simply calls entry_point.base_launch_kernel with the right first
     command to start an ipkernel.  See base_launch_kernel for arguments.
 
     Returns
     -------
     A tuple of form:
         (kernel_process, shell_port, iopub_port, stdin_port, hb_port)
     where kernel_process is a Popen object and the ports are integers.
     """
     return base_launch_kernel('from IPython.zmq.ipkernel import main; main()',
                               *args, **kwargs)
 
 
 def main():
     """Run an IPKernel as an application"""
     app = IPKernelApp.instance()
     app.initialize()
     app.start()
 
 
 if __name__ == '__main__':
     main()

IPython/zmq/zmqshell.py

@@ -1,441 +1,442 b''
 """A ZMQ-based subclass of InteractiveShell.
 
 This code is meant to ease the refactoring of the base InteractiveShell into
 something with a cleaner architecture for 2-process use, without actually
 breaking InteractiveShell itself.  So we're doing something a bit ugly, where
 we subclass and override what we want to fix.  Once this is working well, we
 can go back to the base class and refactor the code for a cleaner inheritance
 implementation that doesn't rely on so much monkeypatching.
 
 But this lets us maintain a fully working IPython as we develop the new
 machinery.  This should thus be thought of as scaffolding.
 """
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 from __future__ import print_function
 
 # Stdlib
 import inspect
 import os
 
 # Our own
 from IPython.core.interactiveshell import (
     InteractiveShell, InteractiveShellABC
 )
 from IPython.core import page
 from IPython.core.autocall import ZMQExitAutocall
 from IPython.core.displaypub import DisplayPublisher
 from IPython.core.macro import Macro
 from IPython.core.magic import MacroToEdit
 from IPython.core.payloadpage import install_payload_page
 from IPython.utils import io
+from IPython.utils.jsonutil import json_clean
 from IPython.utils.path import get_py_filename
 from IPython.utils.traitlets import Instance, Type, Dict, CBool
 from IPython.utils.warn import warn
 from IPython.zmq.displayhook import ZMQShellDisplayHook, _encode_binary
 from IPython.zmq.session import extract_header
 from session import Session
 
 #-----------------------------------------------------------------------------
 # Globals and side-effects
 #-----------------------------------------------------------------------------
 
 # Install the payload version of page.
 install_payload_page()
 
 #-----------------------------------------------------------------------------
 # Functions and classes
 #-----------------------------------------------------------------------------
 
 class ZMQDisplayPublisher(DisplayPublisher):
     """A display publisher that publishes data using a ZeroMQ PUB socket."""
 
     session = Instance(Session)
     pub_socket = Instance('zmq.Socket')
     parent_header = Dict({})
 
     def set_parent(self, parent):
         """Set the parent for outbound messages."""
         self.parent_header = extract_header(parent)
 
     def publish(self, source, data, metadata=None):
         if metadata is None:
             metadata = {}
         self._validate_data(source, data, metadata)
         content = {}
         content['source'] = source
         _encode_binary(data)
         content['data'] = data
         content['metadata'] = metadata
         self.session.send(
-            self.pub_socket, u'display_data', content,
+            self.pub_socket, u'display_data', json_clean(content),
             parent=self.parent_header
         )
 
 
 class ZMQInteractiveShell(InteractiveShell):
     """A subclass of InteractiveShell for ZMQ."""
 
     displayhook_class = Type(ZMQShellDisplayHook)
     display_pub_class = Type(ZMQDisplayPublisher)
     
     # Override the traitlet in the parent class, because there's no point using
     # readline for the kernel. Can be removed when the readline code is moved
     # to the terminal frontend.
 
     # FIXME.  This is disabled for now, even though it may cause problems under
     # Windows, because it breaks %run in the Qt console.  See gh-617 for more
     # details.  Re-enable once we've fully tested that %run works in the Qt
     # console with syntax highlighting in tracebacks.
     # readline_use = CBool(False)
     # /FIXME
     
     exiter = Instance(ZMQExitAutocall)
     def _exiter_default(self):
         return ZMQExitAutocall(self)
 
     keepkernel_on_exit = None
 
     def init_environment(self):
         """Configure the user's environment.
 
         """
         env = os.environ
         # These two ensure 'ls' produces nice coloring on BSD-derived systems
         env['TERM'] = 'xterm-color'
         env['CLICOLOR'] = '1'
         # Since normal pagers don't work at all (over pexpect we don't have
         # single-key control of the subprocess), try to disable paging in
         # subprocesses as much as possible.
         env['PAGER'] = 'cat'
         env['GIT_PAGER'] = 'cat'
 
     def auto_rewrite_input(self, cmd):
         """Called to show the auto-rewritten input for autocall and friends.
 
         FIXME: this payload is currently not correctly processed by the
         frontend.
         """
         new = self.displayhook.prompt1.auto_rewrite() + cmd
         payload = dict(
             source='IPython.zmq.zmqshell.ZMQInteractiveShell.auto_rewrite_input',
             transformed_input=new,
             )
         self.payload_manager.write_payload(payload)
         
     def ask_exit(self):
         """Engage the exit actions."""
         payload = dict(
             source='IPython.zmq.zmqshell.ZMQInteractiveShell.ask_exit',
             exit=True,
             keepkernel=self.keepkernel_on_exit,
             )
         self.payload_manager.write_payload(payload)
 
     def _showtraceback(self, etype, evalue, stb):
 
         exc_content = {
             u'traceback' : stb,
             u'ename' : unicode(etype.__name__),
             u'evalue' : unicode(evalue)
         }
 
         dh = self.displayhook
         # Send exception info over pub socket for other clients than the caller
         # to pick up
-        exc_msg = dh.session.send(dh.pub_socket, u'pyerr', exc_content, dh.parent_header)
+        exc_msg = dh.session.send(dh.pub_socket, u'pyerr', json_clean(exc_content), dh.parent_header)
 
         # FIXME - Hack: store exception info in shell object.  Right now, the
         # caller is reading this info after the fact, we need to fix this logic
         # to remove this hack.  Even uglier, we need to store the error status
         # here, because in the main loop, the logic that sets it is being
         # skipped because runlines swallows the exceptions.
         exc_content[u'status'] = u'error'
         self._reply_content = exc_content
         # /FIXME
         
         return exc_content
 
     #------------------------------------------------------------------------
     # Magic overrides
     #------------------------------------------------------------------------
     # Once the base class stops inheriting from magic, this code needs to be
     # moved into a separate machinery as well.  For now, at least isolate here
     # the magics which this class needs to implement differently from the base
     # class, or that are unique to it.
 
     def magic_doctest_mode(self,parameter_s=''):
         """Toggle doctest mode on and off.
 
         This mode is intended to make IPython behave as much as possible like a
         plain Python shell, from the perspective of how its prompts, exceptions
         and output look.  This makes it easy to copy and paste parts of a
         session into doctests.  It does so by:
 
         - Changing the prompts to the classic ``>>>`` ones.
         - Changing the exception reporting mode to 'Plain'.
         - Disabling pretty-printing of output.
 
         Note that IPython also supports the pasting of code snippets that have
         leading '>>>' and '...' prompts in them.  This means that you can paste
         doctests from files or docstrings (even if they have leading
         whitespace), and the code will execute correctly.  You can then use
         '%history -t' to see the translated history; this will give you the
         input after removal of all the leading prompts and whitespace, which
         can be pasted back into an editor.
 
         With these features, you can switch into this mode easily whenever you
         need to do testing and changes to doctests, without having to leave
         your existing IPython session.
         """
 
         from IPython.utils.ipstruct import Struct
 
         # Shorthands
         shell = self.shell
         disp_formatter = self.shell.display_formatter
         ptformatter = disp_formatter.formatters['text/plain']
         # dstore is a data store kept in the instance metadata bag to track any
         # changes we make, so we can undo them later.
         dstore = shell.meta.setdefault('doctest_mode', Struct())
         save_dstore = dstore.setdefault
 
         # save a few values we'll need to recover later
         mode = save_dstore('mode', False)
         save_dstore('rc_pprint', ptformatter.pprint)
         save_dstore('rc_plain_text_only',disp_formatter.plain_text_only)
         save_dstore('xmode', shell.InteractiveTB.mode)
         
         if mode == False:
             # turn on
             ptformatter.pprint = False
             disp_formatter.plain_text_only = True
             shell.magic_xmode('Plain')
         else:
             # turn off
             ptformatter.pprint = dstore.rc_pprint
             disp_formatter.plain_text_only = dstore.rc_plain_text_only
             shell.magic_xmode(dstore.xmode)
 
         # Store new mode and inform on console
         dstore.mode = bool(1-int(mode))
         mode_label = ['OFF','ON'][dstore.mode]
         print('Doctest mode is:', mode_label)
         
         # Send the payload back so that clients can modify their prompt display
         payload = dict(
             source='IPython.zmq.zmqshell.ZMQInteractiveShell.magic_doctest_mode',
             mode=dstore.mode)
         self.payload_manager.write_payload(payload)
 
     def magic_edit(self,parameter_s='',last_call=['','']):
         """Bring up an editor and execute the resulting code.
 
         Usage:
           %edit [options] [args]
 
         %edit runs an external text editor. You will need to set the command for
         this editor via the ``TerminalInteractiveShell.editor`` option in your
         configuration file before it will work.
 
         This command allows you to conveniently edit multi-line code right in
         your IPython session.
         
         If called without arguments, %edit opens up an empty editor with a
         temporary file and will execute the contents of this file when you
         close it (don't forget to save it!).
 
 
         Options:
 
         -n <number>: open the editor at a specified line number.  By default,
         the IPython editor hook uses the unix syntax 'editor +N filename', but
         you can configure this by providing your own modified hook if your
         favorite editor supports line-number specifications with a different
         syntax.
         
         -p: this will call the editor with the same data as the previous time
         it was used, regardless of how long ago (in your current session) it
         was.
 
         -r: use 'raw' input.  This option only applies to input taken from the
         user's history.  By default, the 'processed' history is used, so that
         magics are loaded in their transformed version to valid Python.  If
         this option is given, the raw input as typed as the command line is
         used instead.  When you exit the editor, it will be executed by
         IPython's own processor.
         
         -x: do not execute the edited code immediately upon exit. This is
         mainly useful if you are editing programs which need to be called with
         command line arguments, which you can then do using %run.
 
 
         Arguments:
 
         If arguments are given, the following possibilites exist:
 
         - The arguments are numbers or pairs of colon-separated numbers (like
         1 4:8 9). These are interpreted as lines of previous input to be
         loaded into the editor. The syntax is the same of the %macro command.
 
         - If the argument doesn't start with a number, it is evaluated as a
         variable and its contents loaded into the editor. You can thus edit
         any string which contains python code (including the result of
         previous edits).
 
         - If the argument is the name of an object (other than a string),
         IPython will try to locate the file where it was defined and open the
         editor at the point where it is defined. You can use `%edit function`
         to load an editor exactly at the point where 'function' is defined,
         edit it and have the file be executed automatically.
 
         If the object is a macro (see %macro for details), this opens up your
         specified editor with a temporary file containing the macro's data.
         Upon exit, the macro is reloaded with the contents of the file.
 
         Note: opening at an exact line is only supported under Unix, and some
         editors (like kedit and gedit up to Gnome 2.8) do not understand the
         '+NUMBER' parameter necessary for this feature. Good editors like
         (X)Emacs, vi, jed, pico and joe all do.
 
         - If the argument is not found as a variable, IPython will look for a
         file with that name (adding .py if necessary) and load it into the
         editor. It will execute its contents with execfile() when you exit,
         loading any code in the file into your interactive namespace.
 
         After executing your code, %edit will return as output the code you
         typed in the editor (except when it was an existing file). This way
         you can reload the code in further invocations of %edit as a variable,
         via _<NUMBER> or Out[<NUMBER>], where <NUMBER> is the prompt number of
         the output.
 
         Note that %edit is also available through the alias %ed.
 
         This is an example of creating a simple function inside the editor and
         then modifying it. First, start up the editor:
 
         In [1]: ed
         Editing... done. Executing edited code...
         Out[1]: 'def foo():n    print "foo() was defined in an editing session"n'
 
         We can then call the function foo():
         
         In [2]: foo()
         foo() was defined in an editing session
 
         Now we edit foo.  IPython automatically loads the editor with the
         (temporary) file where foo() was previously defined:
         
         In [3]: ed foo
         Editing... done. Executing edited code...
 
         And if we call foo() again we get the modified version:
         
         In [4]: foo()
         foo() has now been changed!
 
         Here is an example of how to edit a code snippet successive
         times. First we call the editor:
 
         In [5]: ed
         Editing... done. Executing edited code...
         hello
         Out[5]: "print 'hello'n"
 
         Now we call it again with the previous output (stored in _):
 
         In [6]: ed _
         Editing... done. Executing edited code...
         hello world
         Out[6]: "print 'hello world'n"
 
         Now we call it with the output #8 (stored in _8, also as Out[8]):
 
         In [7]: ed _8
         Editing... done. Executing edited code...
         hello again
         Out[7]: "print 'hello again'n"
         """
         
         opts,args = self.parse_options(parameter_s,'prn:')
         
         try:
             filename, lineno, _ = self._find_edit_target(args, opts, last_call)
         except MacroToEdit as e:
             # TODO: Implement macro editing over 2 processes.
             print("Macro editing not yet implemented in 2-process model.")
             return
 
         # Make sure we send to the client an absolute path, in case the working
         # directory of client and kernel don't match
         filename = os.path.abspath(filename)
 
         payload = {
             'source' : 'IPython.zmq.zmqshell.ZMQInteractiveShell.edit_magic',
             'filename' : filename,
             'line_number' : lineno
         }
         self.payload_manager.write_payload(payload)
 
     def magic_gui(self, *args, **kwargs):
         raise NotImplementedError(
             'Kernel GUI support is not implemented yet, except for --pylab.')
 
     def magic_pylab(self, *args, **kwargs):
         raise NotImplementedError(
             'pylab support must be enabled in command line options.')
 
     # A few magics that are adapted to the specifics of using pexpect and a
     # remote terminal
 
     def magic_clear(self, arg_s):
         """Clear the terminal."""
         if os.name == 'posix':
             self.shell.system("clear")
         else:
             self.shell.system("cls")
 
     if os.name == 'nt':
         # This is the usual name in windows
         magic_cls = magic_clear
 
     # Terminal pagers won't work over pexpect, but we do have our own pager
     
     def magic_less(self, arg_s):
         """Show a file through the pager.
 
         Files ending in .py are syntax-highlighted."""
         cont = open(arg_s).read()
         if arg_s.endswith('.py'):
             cont = self.shell.pycolorize(cont)
         page.page(cont)
 
     magic_more = magic_less
 
     # Man calls a pager, so we also need to redefine it
     if os.name == 'posix':
         def magic_man(self, arg_s):
             """Find the man page for the given command and display in pager."""
             page.page(self.shell.getoutput('man %s | col -b' % arg_s,
                                            split=False))
 
     # FIXME: this is specific to the GUI, so we should let the gui app load
     # magics at startup that are only for the gui.  Once the gui app has proper
     # profile and configuration management, we can have it initialize a kernel
     # with a special config file that provides these.
     def magic_guiref(self, arg_s):
         """Show a basic reference about the GUI console."""
         from IPython.core.usage import gui_reference
         page.page(gui_reference, auto_html=True)
     
     def set_next_input(self, text):
         """Send the specified text to the frontend to be presented at the next
         input cell."""
         payload = dict(
             source='IPython.zmq.zmqshell.ZMQInteractiveShell.set_next_input', 
             text=text
         )
         self.payload_manager.write_payload(payload)
 
 InteractiveShellABC.register(ZMQInteractiveShell)