add text.getdefaultencoding() for central default encoding guess

This is a central location for the many places we call sys.stdin.encoding or sys.getdefaultencoding(), which
now adds locale.getpreferredencoding(False) after stdin.encoding,
which should be a better guess when stdin.encoding is None.

IPython/config/loader.py

@@ -1,660 +1,661
 """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
 """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
 """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
 # 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
 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