Merge pull request #1652 from minrk/zmqcompat

add patch_pyzmq() for backporting a few changes from newer pyzmq

* fixes missing constants in super-old pyzmq
* define ioloop.install if not defined
* never allow jsonlib to be used by zmq.utils.jsonapi (#1520).

Closes  #1520.

IPython/frontend/html/notebook/__init__.py

@@ -1,15 +1,20 b''
 """The IPython HTML Notebook"""
 
 # check for tornado 2.1.0
 msg = "The IPython Notebook requires tornado >= 2.1.0"
 try:
     import tornado
 except ImportError:
     raise ImportError(msg)
 try:
     version_info = tornado.version_info
 except AttributeError:
     raise ImportError(msg + ", but you have < 1.1.0")
 if version_info < (2,1,0):
     raise ImportError(msg + ", but you have %s" % tornado.version)
 del msg
+
+# check for pyzmq 2.1.4
+from IPython.zmq import check_for_zmq
+check_for_zmq('2.1.4', 'IPython.frontend.html.notebook')
+del check_for_zmq

IPython/frontend/html/notebook/notebookapp.py

@@ -1,569 +1,563 b''
 # coding: utf-8
 """A tornado based IPython notebook server.
 
 Authors:
 
 * Brian Granger
 """
 #-----------------------------------------------------------------------------
 #  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
 #-----------------------------------------------------------------------------
 
 # stdlib
 import errno
 import logging
 import os
 import re
 import select
 import signal
 import socket
 import sys
 import threading
 import time
 import webbrowser
 
 # Third party
 import zmq
 
 # Install the pyzmq ioloop. This has to be done before anything else from
 # tornado is imported.
 from zmq.eventloop import ioloop
-# FIXME: ioloop.install is new in pyzmq-2.1.7, so remove this conditional
-# when pyzmq dependency is updated beyond that.
-if hasattr(ioloop, 'install'):
-    ioloop.install()
-else:
-    import tornado.ioloop
-    tornado.ioloop.IOLoop = ioloop.IOLoop
+ioloop.install()
 
 from tornado import httpserver
 from tornado import web
 
 # Our own libraries
 from .kernelmanager import MappingKernelManager
 from .handlers import (LoginHandler, LogoutHandler,
     ProjectDashboardHandler, NewHandler, NamedNotebookHandler,
     MainKernelHandler, KernelHandler, KernelActionHandler, IOPubHandler,
     ShellHandler, NotebookRootHandler, NotebookHandler, NotebookCopyHandler,
     RSTHandler, AuthenticatedFileHandler, PrintNotebookHandler,
     MainClusterHandler, ClusterProfileHandler, ClusterActionHandler
 )
 from .notebookmanager import NotebookManager
 from .clustermanager import ClusterManager
 
 from IPython.config.application import catch_config_error, boolean_flag
 from IPython.core.application import BaseIPythonApplication
 from IPython.core.profiledir import ProfileDir
 from IPython.lib.kernel import swallow_argv
 from IPython.zmq.session import Session, default_secure
 from IPython.zmq.zmqshell import ZMQInteractiveShell
 from IPython.zmq.ipkernel import (
     flags as ipkernel_flags,
     aliases as ipkernel_aliases,
     IPKernelApp
 )
 from IPython.utils.traitlets import Dict, Unicode, Integer, List, Enum, Bool
 from IPython.utils import py3compat
 
 #-----------------------------------------------------------------------------
 # Module globals
 #-----------------------------------------------------------------------------
 
 _kernel_id_regex = r"(?P<kernel_id>\w+-\w+-\w+-\w+-\w+)"
 _kernel_action_regex = r"(?P<action>restart|interrupt)"
 _notebook_id_regex = r"(?P<notebook_id>\w+-\w+-\w+-\w+-\w+)"
 _profile_regex = r"(?P<profile>[a-zA-Z0-9]+)"
 _cluster_action_regex = r"(?P<action>start|stop)"
 
 
 LOCALHOST = '127.0.0.1'
 
 _examples = """
 ipython notebook                       # start the notebook
 ipython notebook --profile=sympy       # use the sympy profile
 ipython notebook --pylab=inline        # pylab in inline plotting mode
 ipython notebook --certfile=mycert.pem # use SSL/TLS certificate
 ipython notebook --port=5555 --ip=*    # Listen on port 5555, all interfaces
 """
 
 #-----------------------------------------------------------------------------
 # Helper functions
 #-----------------------------------------------------------------------------
 
 def url_path_join(a,b):
     if a.endswith('/') and b.startswith('/'):
         return a[:-1]+b
     else:
         return a+b
 
 #-----------------------------------------------------------------------------
 # The Tornado web application
 #-----------------------------------------------------------------------------
 
 class NotebookWebApplication(web.Application):
 
     def __init__(self, ipython_app, kernel_manager, notebook_manager, 
                  cluster_manager, log,
                  base_project_url, settings_overrides):
         handlers = [
             (r"/", ProjectDashboardHandler),
             (r"/login", LoginHandler),
             (r"/logout", LogoutHandler),
             (r"/new", NewHandler),
             (r"/%s" % _notebook_id_regex, NamedNotebookHandler),
             (r"/%s/copy" % _notebook_id_regex, NotebookCopyHandler),
             (r"/%s/print" % _notebook_id_regex, PrintNotebookHandler),
             (r"/kernels", MainKernelHandler),
             (r"/kernels/%s" % _kernel_id_regex, KernelHandler),
             (r"/kernels/%s/%s" % (_kernel_id_regex, _kernel_action_regex), KernelActionHandler),
             (r"/kernels/%s/iopub" % _kernel_id_regex, IOPubHandler),
             (r"/kernels/%s/shell" % _kernel_id_regex, ShellHandler),
             (r"/notebooks", NotebookRootHandler),
             (r"/notebooks/%s" % _notebook_id_regex, NotebookHandler),
             (r"/rstservice/render", RSTHandler),
             (r"/files/(.*)", AuthenticatedFileHandler, {'path' : notebook_manager.notebook_dir}),
             (r"/clusters", MainClusterHandler),
             (r"/clusters/%s/%s" % (_profile_regex, _cluster_action_regex), ClusterActionHandler),
             (r"/clusters/%s" % _profile_regex, ClusterProfileHandler),
         ]
         settings = dict(
             template_path=os.path.join(os.path.dirname(__file__), "templates"),
             static_path=os.path.join(os.path.dirname(__file__), "static"),
             cookie_secret=os.urandom(1024),
             login_url="/login",
         )
 
         # allow custom overrides for the tornado web app.
         settings.update(settings_overrides)
 
         # Python < 2.6.5 doesn't accept unicode keys in f(**kwargs), and
         # base_project_url will always be unicode, which will in turn
         # make the patterns unicode, and ultimately result in unicode
         # keys in kwargs to handler._execute(**kwargs) in tornado.
         # This enforces that base_project_url be ascii in that situation.
         # 
         # Note that the URLs these patterns check against are escaped,
         # and thus guaranteed to be ASCII: 'héllo' is really 'h%C3%A9llo'.
         base_project_url = py3compat.unicode_to_str(base_project_url, 'ascii')
         
         # prepend base_project_url onto the patterns that we match
         new_handlers = []
         for handler in handlers:
             pattern = url_path_join(base_project_url, handler[0])
             new_handler = tuple([pattern]+list(handler[1:]))
             new_handlers.append( new_handler )
 
         super(NotebookWebApplication, self).__init__(new_handlers, **settings)
 
         self.kernel_manager = kernel_manager
         self.notebook_manager = notebook_manager
         self.cluster_manager = cluster_manager
         self.ipython_app = ipython_app
         self.read_only = self.ipython_app.read_only
         self.log = log
 
 
 #-----------------------------------------------------------------------------
 # Aliases and Flags
 #-----------------------------------------------------------------------------
 
 flags = dict(ipkernel_flags)
 flags['no-browser']=(
     {'NotebookApp' : {'open_browser' : False}},
     "Don't open the notebook in a browser after startup."
 )
 flags['no-mathjax']=(
     {'NotebookApp' : {'enable_mathjax' : False}},
     """Disable MathJax
     
     MathJax is the javascript library IPython uses to render math/LaTeX. It is
     very large, so you may want to disable it if you have a slow internet
     connection, or for offline use of the notebook.
     
     When disabled, equations etc. will appear as their untransformed TeX source.
     """
 )
 flags['read-only'] = (
     {'NotebookApp' : {'read_only' : True}},
     """Allow read-only access to notebooks.
     
     When using a password to protect the notebook server, this flag
     allows unauthenticated clients to view the notebook list, and
     individual notebooks, but not edit them, start kernels, or run
     code.
     
     If no password is set, the server will be entirely read-only.
     """
 )
 
 # Add notebook manager flags
 flags.update(boolean_flag('script', 'NotebookManager.save_script',
                'Auto-save a .py script everytime the .ipynb notebook is saved',
                'Do not auto-save .py scripts for every notebook'))
 
 # the flags that are specific to the frontend
 # these must be scrubbed before being passed to the kernel,
 # or it will raise an error on unrecognized flags
 notebook_flags = ['no-browser', 'no-mathjax', 'read-only', 'script', 'no-script']
 
 aliases = dict(ipkernel_aliases)
 
 aliases.update({
     'ip': 'NotebookApp.ip',
     'port': 'NotebookApp.port',
     'keyfile': 'NotebookApp.keyfile',
     'certfile': 'NotebookApp.certfile',
     'notebook-dir': 'NotebookManager.notebook_dir',
     'browser': 'NotebookApp.browser',
 })
 
 # remove ipkernel flags that are singletons, and don't make sense in
 # multi-kernel evironment:
 aliases.pop('f', None)
 
 notebook_aliases = [u'port', u'ip', u'keyfile', u'certfile',
                     u'notebook-dir']
 
 #-----------------------------------------------------------------------------
 # NotebookApp
 #-----------------------------------------------------------------------------
 
 class NotebookApp(BaseIPythonApplication):
 
     name = 'ipython-notebook'
     default_config_file_name='ipython_notebook_config.py'
     
     description = """
         The IPython HTML Notebook.
         
         This launches a Tornado based HTML Notebook Server that serves up an
         HTML5/Javascript Notebook client.
     """
     examples = _examples
     
     classes = [IPKernelApp, ZMQInteractiveShell, ProfileDir, Session,
                MappingKernelManager, NotebookManager]
     flags = Dict(flags)
     aliases = Dict(aliases)
 
     kernel_argv = List(Unicode)
 
     log_level = Enum((0,10,20,30,40,50,'DEBUG','INFO','WARN','ERROR','CRITICAL'),
                     default_value=logging.INFO,
                     config=True,
                     help="Set the log level by value or name.")
 
     # create requested profiles by default, if they don't exist:
     auto_create = Bool(True)
 
     # Network related information.
 
     ip = Unicode(LOCALHOST, config=True,
         help="The IP address the notebook server will listen on."
     )
 
     def _ip_changed(self, name, old, new):
         if new == u'*': self.ip = u''
 
     port = Integer(8888, config=True,
         help="The port the notebook server will listen on."
     )
 
     certfile = Unicode(u'', config=True, 
         help="""The full path to an SSL/TLS certificate file."""
     )
     
     keyfile = Unicode(u'', config=True, 
         help="""The full path to a private key file for usage with SSL/TLS."""
     )
 
     password = Unicode(u'', config=True,
                       help="""Hashed password to use for web authentication.
 
                       To generate, type in a python/IPython shell:
 
                         from IPython.lib import passwd; passwd()
 
                       The string should be of the form type:salt:hashed-password.
                       """
     )
 
     open_browser = Bool(True, config=True,
                         help="""Whether to open in a browser after starting.
                         The specific browser used is platform dependent and
                         determined by the python standard library `webbrowser`
                         module, unless it is overridden using the --browser
                         (NotebookApp.browser) configuration option.
                         """)
 
     browser = Unicode(u'', config=True,
                       help="""Specify what command to use to invoke a web
                       browser when opening the notebook. If not specified, the
                       default browser will be determined by the `webbrowser`
                       standard library module, which allows setting of the
                       BROWSER environment variable to override it.
                       """)
     
     read_only = Bool(False, config=True,
         help="Whether to prevent editing/execution of notebooks."
     )
     
     webapp_settings = Dict(config=True,
             help="Supply overrides for the tornado.web.Application that the "
                  "IPython notebook uses.")
     
     enable_mathjax = Bool(True, config=True,
         help="""Whether to enable MathJax for typesetting math/TeX
 
         MathJax is the javascript library IPython uses to render math/LaTeX. It is
         very large, so you may want to disable it if you have a slow internet
         connection, or for offline use of the notebook.
 
         When disabled, equations etc. will appear as their untransformed TeX source.
         """
     )
     def _enable_mathjax_changed(self, name, old, new):
         """set mathjax url to empty if mathjax is disabled"""
         if not new:
             self.mathjax_url = u''
 
     base_project_url = Unicode('/', config=True,
                                help='''The base URL for the notebook server''')
     base_kernel_url = Unicode('/', config=True,
                                help='''The base URL for the kernel server''')
     websocket_host = Unicode("", config=True,
         help="""The hostname for the websocket server."""
     )
 
     mathjax_url = Unicode("", config=True,
         help="""The url for MathJax.js."""
     )
     def _mathjax_url_default(self):
         if not self.enable_mathjax:
             return u''
         static_path = self.webapp_settings.get("static_path", os.path.join(os.path.dirname(__file__), "static"))
         static_url_prefix = self.webapp_settings.get("static_url_prefix",
                                                      "/static/")
         if os.path.exists(os.path.join(static_path, 'mathjax', "MathJax.js")):
             self.log.info("Using local MathJax")
             return static_url_prefix+u"mathjax/MathJax.js"
         else:
             self.log.info("Using MathJax from CDN")
             hostname = "cdn.mathjax.org"
             try:
                 # resolve mathjax cdn alias to cloudfront, because Amazon's SSL certificate
                 # only works on *.cloudfront.net
                 true_host, aliases, IPs = socket.gethostbyname_ex(hostname)
                 # I've run this on a few machines, and some put the right answer in true_host,
                 # while others gave it in the aliases list, so we check both.
                 aliases.insert(0, true_host)
             except Exception:
                 self.log.warn("Couldn't determine MathJax CDN info")
             else:
                 for alias in aliases:
                     parts = alias.split('.')
                     # want static foo.cloudfront.net, not dynamic foo.lax3.cloudfront.net
                     if len(parts) == 3 and alias.endswith(".cloudfront.net"):
                         hostname = alias
                         break
             
             if not hostname.endswith(".cloudfront.net"):
                 self.log.error("Couldn't resolve CloudFront host, required for HTTPS MathJax.")
                 self.log.error("Loading from https://cdn.mathjax.org will probably fail due to invalid certificate.")
                 self.log.error("For unsecured HTTP access to MathJax use config:")
                 self.log.error("NotebookApp.mathjax_url='http://cdn.mathjax.org/mathjax/latest/MathJax.js'")
             return u"https://%s/mathjax/latest/MathJax.js" % hostname
     
     def _mathjax_url_changed(self, name, old, new):
         if new and not self.enable_mathjax:
             # enable_mathjax=False overrides mathjax_url
             self.mathjax_url = u''
         else:
             self.log.info("Using MathJax: %s", new)
 
     def parse_command_line(self, argv=None):
         super(NotebookApp, self).parse_command_line(argv)
         if argv is None:
             argv = sys.argv[1:]
 
         # Scrub frontend-specific flags
         self.kernel_argv = swallow_argv(argv, notebook_aliases, notebook_flags)
         # Kernel should inherit default config file from frontend
         self.kernel_argv.append("--KernelApp.parent_appname='%s'"%self.name)
 
     def init_configurables(self):
         # force Session default to be secure
         default_secure(self.config)
         # Create a KernelManager and start a kernel.
         self.kernel_manager = MappingKernelManager(
             config=self.config, log=self.log, kernel_argv=self.kernel_argv,
             connection_dir = self.profile_dir.security_dir,
         )
         self.notebook_manager = NotebookManager(config=self.config, log=self.log)
         self.notebook_manager.list_notebooks()
         self.cluster_manager = ClusterManager(config=self.config, log=self.log)
         self.cluster_manager.update_profiles()
 
     def init_logging(self):
         super(NotebookApp, self).init_logging()
         # This prevents double log messages because tornado use a root logger that
         # self.log is a child of. The logging module dipatches log messages to a log
         # and all of its ancenstors until propagate is set to False.
         self.log.propagate = False
     
     def init_webapp(self):
         """initialize tornado webapp and httpserver"""
         self.web_app = NotebookWebApplication(
             self, self.kernel_manager, self.notebook_manager, 
             self.cluster_manager, self.log,
             self.base_project_url, self.webapp_settings
         )
         if self.certfile:
             ssl_options = dict(certfile=self.certfile)
             if self.keyfile:
                 ssl_options['keyfile'] = self.keyfile
         else:
             ssl_options = None
         self.web_app.password = self.password
         self.http_server = httpserver.HTTPServer(self.web_app, ssl_options=ssl_options)
         if ssl_options is None and not self.ip and not (self.read_only and not self.password):
             self.log.critical('WARNING: the notebook server is listening on all IP addresses '
                               'but not using any encryption or authentication. This is highly '
                               'insecure and not recommended.')
 
         # Try random ports centered around the default.
         from random import randint
         n = 50  # Max number of attempts, keep reasonably large.
         for port in range(self.port, self.port+5) + [self.port + randint(-2*n, 2*n) for i in range(n-5)]:
             try:
                 self.http_server.listen(port, self.ip)
             except socket.error, e:
                 if e.errno != errno.EADDRINUSE:
                     raise
                 self.log.info('The port %i is already in use, trying another random port.' % port)
             else:
                 self.port = port
                 break
     
     def init_signal(self):
         # FIXME: remove this check when pyzmq dependency is >= 2.1.11
         # safely extract zmq version info:
         try:
             zmq_v = zmq.pyzmq_version_info()
         except AttributeError:
             zmq_v = [ int(n) for n in re.findall(r'\d+', zmq.__version__) ]
             if 'dev' in zmq.__version__:
                 zmq_v.append(999)
             zmq_v = tuple(zmq_v)
         if zmq_v >= (2,1,9):
             # This won't work with 2.1.7 and
             # 2.1.9-10 will log ugly 'Interrupted system call' messages,
             # but it will work
             signal.signal(signal.SIGINT, self._handle_sigint)
         signal.signal(signal.SIGTERM, self._signal_stop)
     
     def _handle_sigint(self, sig, frame):
         """SIGINT handler spawns confirmation dialog"""
         # register more forceful signal handler for ^C^C case
         signal.signal(signal.SIGINT, self._signal_stop)
         # request confirmation dialog in bg thread, to avoid
         # blocking the App
         thread = threading.Thread(target=self._confirm_exit)
         thread.daemon = True
         thread.start()
     
     def _restore_sigint_handler(self):
         """callback for restoring original SIGINT handler"""
         signal.signal(signal.SIGINT, self._handle_sigint)
     
     def _confirm_exit(self):
         """confirm shutdown on ^C
         
         A second ^C, or answering 'y' within 5s will cause shutdown,
         otherwise original SIGINT handler will be restored.
         """
         # FIXME: remove this delay when pyzmq dependency is >= 2.1.11
         time.sleep(0.1)
         sys.stdout.write("Shutdown Notebook Server (y/[n])? ")
         sys.stdout.flush()
         r,w,x = select.select([sys.stdin], [], [], 5)
         if r:
             line = sys.stdin.readline()
             if line.lower().startswith('y'):
                 self.log.critical("Shutdown confirmed")
                 ioloop.IOLoop.instance().stop()
                 return
         else:
             print "No answer for 5s:",
         print "resuming operation..."
         # no answer, or answer is no:
         # set it back to original SIGINT handler
         # use IOLoop.add_callback because signal.signal must be called
         # from main thread
         ioloop.IOLoop.instance().add_callback(self._restore_sigint_handler)
     
     def _signal_stop(self, sig, frame):
         self.log.critical("received signal %s, stopping", sig)
         ioloop.IOLoop.instance().stop()
     
     @catch_config_error
     def initialize(self, argv=None):
         super(NotebookApp, self).initialize(argv)
         self.init_configurables()
         self.init_webapp()
         self.init_signal()
 
     def cleanup_kernels(self):
         """shutdown all kernels
         
         The kernels will shutdown themselves when this process no longer exists,
         but explicit shutdown allows the KernelManagers to cleanup the connection files.
         """
         self.log.info('Shutting down kernels')
         km = self.kernel_manager
         # copy list, since kill_kernel deletes keys
         for kid in list(km.kernel_ids):
             km.kill_kernel(kid)
 
     def start(self):
         ip = self.ip if self.ip else '[all ip addresses on your system]'
         proto = 'https' if self.certfile else 'http'
         info = self.log.info
         info("The IPython Notebook is running at: %s://%s:%i%s" %
              (proto, ip, self.port,self.base_project_url) )
         info("Use Control-C to stop this server and shut down all kernels.")
 
         if self.open_browser:
             ip = self.ip or '127.0.0.1'
             if self.browser:
                 browser = webbrowser.get(self.browser)
             else:
                 browser = webbrowser.get()
             b = lambda : browser.open("%s://%s:%i%s" % (proto, ip, self.port,
                                                            self.base_project_url),
                                          new=2)
             threading.Thread(target=b).start()
         try:
             ioloop.IOLoop.instance().start()
         except KeyboardInterrupt:
             info("Interrupted...")
         finally:
             self.cleanup_kernels()
     
 
 #-----------------------------------------------------------------------------
 # Main entry point
 #-----------------------------------------------------------------------------
 
 def launch_new_instance():
     app = NotebookApp.instance()
     app.initialize()
     app.start()
 

IPython/zmq/__init__.py

@@ -1,39 +1,68 b''
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010-2011  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.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Verify zmq version dependency >= 2.1.4
 #-----------------------------------------------------------------------------
 
 import warnings
 from distutils.version import LooseVersion as V
 
+
+def patch_pyzmq():
+    """backport a few patches from newer pyzmq
+    
+    These can be removed as we bump our minimum pyzmq version
+    """
+    
+    import zmq
+    
+    # ioloop.install, introduced in pyzmq 2.1.7
+    from zmq.eventloop import ioloop
+    
+    def install():
+        import tornado.ioloop
+        tornado.ioloop.IOLoop = ioloop.IOLoop
+    
+    if not hasattr(ioloop, 'install'):
+        ioloop.install = install
+    
+    # fix missing DEALER/ROUTER aliases in pyzmq < 2.1.9
+    if not hasattr(zmq, 'DEALER'):
+        zmq.DEALER = zmq.XREQ
+    if not hasattr(zmq, 'ROUTER'):
+        zmq.ROUTER = zmq.XREP
+    
+    # fallback on stdlib json if jsonlib is selected, because jsonlib breaks things.
+    # jsonlib support is removed from pyzmq >= 2.2.0
+
+    from zmq.utils import jsonapi
+    if jsonapi.jsonmod.__name__ == 'jsonlib':
+        import json
+        jsonapi.jsonmod = json
+
+
 def check_for_zmq(minimum_version, module='IPython.zmq'):
     try:
         import zmq
     except ImportError:
         raise ImportError("%s requires pyzmq >= %s"%(module, minimum_version))
 
     pyzmq_version = zmq.__version__
     
     if 'dev' not in pyzmq_version and V(pyzmq_version) < V(minimum_version):
         raise ImportError("%s requires pyzmq >= %s, but you have %s"%(
                         module, minimum_version, pyzmq_version))
 
-    # fix missing DEALER/ROUTER aliases in pyzmq < 2.1.9
-    if not hasattr(zmq, 'DEALER'):
-        zmq.DEALER = zmq.XREQ
-    if not hasattr(zmq, 'ROUTER'):
-        zmq.ROUTER = zmq.XREP
-
     if V(zmq.zmq_version()) >= V('4.0.0'):
         warnings.warn("""libzmq 4 detected.
         It is unlikely that IPython's zmq code will work properly.
         Please install libzmq stable, which is 2.1.x or 2.2.x""",
         RuntimeWarning)
 
 check_for_zmq('2.1.4')
+patch_pyzmq()

IPython/zmq/session.py

@@ -1,768 +1,756 b''
 """Session object for building, serializing, sending, and receiving messages in
 IPython. The Session object supports serialization, HMAC signatures, and
 metadata on messages.
 
 Also defined here are utilities for working with Sessions:
 * A SessionFactory to be used as a base class for configurables that work with
 Sessions.
 * A Message object for convenience that allows attribute-access to the msg dict.
 
 Authors:
 
 * Min RK
 * Brian Granger
 * Fernando Perez
 """
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010-2011  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 hmac
 import logging
 import os
 import pprint
 import uuid
 from datetime import datetime
 
 try:
     import cPickle
     pickle = cPickle
 except:
     cPickle = None
     import pickle
 
 import zmq
 from zmq.utils import jsonapi
 from zmq.eventloop.ioloop import IOLoop
 from zmq.eventloop.zmqstream import ZMQStream
 
 from IPython.config.application import Application, boolean_flag
 from IPython.config.configurable import Configurable, LoggingConfigurable
 from IPython.utils.importstring import import_item
 from IPython.utils.jsonutil import extract_dates, squash_dates, date_default
 from IPython.utils.py3compat import str_to_bytes
 from IPython.utils.traitlets import (CBytes, Unicode, Bool, Any, Instance, Set,
                                         DottedObjectName, CUnicode)
 
 #-----------------------------------------------------------------------------
 # utility functions
 #-----------------------------------------------------------------------------
 
 def squash_unicode(obj):
     """coerce unicode back to bytestrings."""
     if isinstance(obj,dict):
         for key in obj.keys():
             obj[key] = squash_unicode(obj[key])
             if isinstance(key, unicode):
                 obj[squash_unicode(key)] = obj.pop(key)
     elif isinstance(obj, list):
         for i,v in enumerate(obj):
             obj[i] = squash_unicode(v)
     elif isinstance(obj, unicode):
         obj = obj.encode('utf8')
     return obj
 
 #-----------------------------------------------------------------------------
 # globals and defaults
 #-----------------------------------------------------------------------------
 
 
-# jsonlib behaves a bit differently, so handle that where it affects us
-if jsonapi.jsonmod.__name__ == 'jsonlib':
-    # kwarg for serializing unknown types (datetime) is different
-    dumps_kwargs = dict(on_unknown=date_default)
-    # By default, jsonlib unpacks floats as Decimal instead of float,
-    # which can foul things up
-    loads_kwargs = dict(use_float=True)
-else:
-    # ISO8601-ify datetime objects
-    dumps_kwargs = dict(default=date_default)
-    # nothing to specify for loads
-    loads_kwargs = dict()
-
-json_packer = lambda obj: jsonapi.dumps(obj, **dumps_kwargs)
-json_unpacker = lambda s: extract_dates(jsonapi.loads(s, **loads_kwargs))
+# ISO8601-ify datetime objects
+json_packer = lambda obj: jsonapi.dumps(obj, default=date_default)
+json_unpacker = lambda s: extract_dates(jsonapi.loads(s))
 
 pickle_packer = lambda o: pickle.dumps(o,-1)
 pickle_unpacker = pickle.loads
 
 default_packer = json_packer
 default_unpacker = json_unpacker
 
 DELIM=b"<IDS|MSG>"
 
 
 #-----------------------------------------------------------------------------
 # Mixin tools for apps that use Sessions
 #-----------------------------------------------------------------------------
 
 session_aliases = dict(
     ident = 'Session.session',
     user = 'Session.username',
     keyfile = 'Session.keyfile',
 )
 
 session_flags  = {
     'secure' : ({'Session' : { 'key' : str_to_bytes(str(uuid.uuid4())),
                             'keyfile' : '' }},
         """Use HMAC digests for authentication of messages.
         Setting this flag will generate a new UUID to use as the HMAC key.
         """),
     'no-secure' : ({'Session' : { 'key' : b'', 'keyfile' : '' }},
         """Don't authenticate messages."""),
 }
 
 def default_secure(cfg):
     """Set the default behavior for a config environment to be secure.
     
     If Session.key/keyfile have not been set, set Session.key to
     a new random UUID.
     """
     
     if 'Session' in cfg:
         if 'key' in cfg.Session or 'keyfile' in cfg.Session:
             return
     # key/keyfile not specified, generate new UUID:
     cfg.Session.key = str_to_bytes(str(uuid.uuid4()))
 
 
 #-----------------------------------------------------------------------------
 # Classes
 #-----------------------------------------------------------------------------
 
 class SessionFactory(LoggingConfigurable):
     """The Base class for configurables that have a Session, Context, logger,
     and IOLoop.
     """
 
     logname = Unicode('')
     def _logname_changed(self, name, old, new):
         self.log = logging.getLogger(new)
 
     # not configurable:
     context = Instance('zmq.Context')
     def _context_default(self):
         return zmq.Context.instance()
 
     session = Instance('IPython.zmq.session.Session')
 
     loop = Instance('zmq.eventloop.ioloop.IOLoop', allow_none=False)
     def _loop_default(self):
         return IOLoop.instance()
 
     def __init__(self, **kwargs):
         super(SessionFactory, self).__init__(**kwargs)
 
         if self.session is None:
             # construct the session
             self.session = Session(**kwargs)
 
 
 class Message(object):
     """A simple message object that maps dict keys to attributes.
 
     A Message can be created from a dict and a dict from a Message instance
     simply by calling dict(msg_obj)."""
 
     def __init__(self, msg_dict):
         dct = self.__dict__
         for k, v in dict(msg_dict).iteritems():
             if isinstance(v, dict):
                 v = Message(v)
             dct[k] = v
 
     # Having this iterator lets dict(msg_obj) work out of the box.
     def __iter__(self):
         return iter(self.__dict__.iteritems())
 
     def __repr__(self):
         return repr(self.__dict__)
 
     def __str__(self):
         return pprint.pformat(self.__dict__)
 
     def __contains__(self, k):
         return k in self.__dict__
 
     def __getitem__(self, k):
         return self.__dict__[k]
 
 
 def msg_header(msg_id, msg_type, username, session):
     date = datetime.now()
     return locals()
 
 def extract_header(msg_or_header):
     """Given a message or header, return the header."""
     if not msg_or_header:
         return {}
     try:
         # See if msg_or_header is the entire message.
         h = msg_or_header['header']
     except KeyError:
         try:
             # See if msg_or_header is just the header
             h = msg_or_header['msg_id']
         except KeyError:
             raise
         else:
             h = msg_or_header
     if not isinstance(h, dict):
         h = dict(h)
     return h
 
 class Session(Configurable):
     """Object for handling serialization and sending of messages.
 
     The Session object handles building messages and sending them
     with ZMQ sockets or ZMQStream objects.  Objects can communicate with each
     other over the network via Session objects, and only need to work with the
     dict-based IPython message spec. The Session will handle
     serialization/deserialization, security, and metadata.
 
     Sessions support configurable serialiization via packer/unpacker traits,
     and signing with HMAC digests via the key/keyfile traits.
 
     Parameters
     ----------
 
     debug : bool
         whether to trigger extra debugging statements
     packer/unpacker : str : 'json', 'pickle' or import_string
         importstrings for methods to serialize message parts.  If just
         'json' or 'pickle', predefined JSON and pickle packers will be used.
         Otherwise, the entire importstring must be used.
 
         The functions must accept at least valid JSON input, and output *bytes*.
 
         For example, to use msgpack:
         packer = 'msgpack.packb', unpacker='msgpack.unpackb'
     pack/unpack : callables
         You can also set the pack/unpack callables for serialization directly.
     session : bytes
         the ID of this Session object.  The default is to generate a new UUID.
     username : unicode
         username added to message headers.  The default is to ask the OS.
     key : bytes
         The key used to initialize an HMAC signature.  If unset, messages
         will not be signed or checked.
     keyfile : filepath
         The file containing a key.  If this is set, `key` will be initialized
         to the contents of the file.
 
     """
 
     debug=Bool(False, config=True, help="""Debug output in the Session""")
 
     packer = DottedObjectName('json',config=True,
             help="""The name of the packer for serializing messages.
             Should be one of 'json', 'pickle', or an import name
             for a custom callable serializer.""")
     def _packer_changed(self, name, old, new):
         if new.lower() == 'json':
             self.pack = json_packer
             self.unpack = json_unpacker
         elif new.lower() == 'pickle':
             self.pack = pickle_packer
             self.unpack = pickle_unpacker
         else:
             self.pack = import_item(str(new))
 
     unpacker = DottedObjectName('json', config=True,
         help="""The name of the unpacker for unserializing messages.
         Only used with custom functions for `packer`.""")
     def _unpacker_changed(self, name, old, new):
         if new.lower() == 'json':
             self.pack = json_packer
             self.unpack = json_unpacker
         elif new.lower() == 'pickle':
             self.pack = pickle_packer
             self.unpack = pickle_unpacker
         else:
             self.unpack = import_item(str(new))
 
     session = CUnicode(u'', config=True,
         help="""The UUID identifying this session.""")
     def _session_default(self):
         u = unicode(uuid.uuid4())
         self.bsession = u.encode('ascii')
         return u
 
     def _session_changed(self, name, old, new):
         self.bsession = self.session.encode('ascii')
 
     # bsession is the session as bytes
     bsession = CBytes(b'')
 
     username = Unicode(os.environ.get('USER',u'username'), config=True,
         help="""Username for the Session. Default is your system username.""")
 
     # message signature related traits:
     
     key = CBytes(b'', config=True,
         help="""execution key, for extra authentication.""")
     def _key_changed(self, name, old, new):
         if new:
             self.auth = hmac.HMAC(new)
         else:
             self.auth = None
     auth = Instance(hmac.HMAC)
     digest_history = Set()
 
     keyfile = Unicode('', config=True,
         help="""path to file containing execution key.""")
     def _keyfile_changed(self, name, old, new):
         with open(new, 'rb') as f:
             self.key = f.read().strip()
 
     # serialization traits:
     
     pack = Any(default_packer) # the actual packer function
     def _pack_changed(self, name, old, new):
         if not callable(new):
             raise TypeError("packer must be callable, not %s"%type(new))
 
     unpack = Any(default_unpacker) # the actual packer function
     def _unpack_changed(self, name, old, new):
         # unpacker is not checked - it is assumed to be
         if not callable(new):
             raise TypeError("unpacker must be callable, not %s"%type(new))
 
     def __init__(self, **kwargs):
         """create a Session object
 
         Parameters
         ----------
 
         debug : bool
             whether to trigger extra debugging statements
         packer/unpacker : str : 'json', 'pickle' or import_string
             importstrings for methods to serialize message parts.  If just
             'json' or 'pickle', predefined JSON and pickle packers will be used.
             Otherwise, the entire importstring must be used.
 
             The functions must accept at least valid JSON input, and output
             *bytes*.
 
             For example, to use msgpack:
             packer = 'msgpack.packb', unpacker='msgpack.unpackb'
         pack/unpack : callables
             You can also set the pack/unpack callables for serialization
             directly.
         session : unicode (must be ascii)
             the ID of this Session object.  The default is to generate a new
             UUID.
         bsession : bytes
             The session as bytes
         username : unicode
             username added to message headers.  The default is to ask the OS.
         key : bytes
             The key used to initialize an HMAC signature.  If unset, messages
             will not be signed or checked.
         keyfile : filepath
             The file containing a key.  If this is set, `key` will be
             initialized to the contents of the file.
         """
         super(Session, self).__init__(**kwargs)
         self._check_packers()
         self.none = self.pack({})
         # ensure self._session_default() if necessary, so bsession is defined:
         self.session
 
     @property
     def msg_id(self):
         """always return new uuid"""
         return str(uuid.uuid4())
 
     def _check_packers(self):
         """check packers for binary data and datetime support."""
         pack = self.pack
         unpack = self.unpack
 
         # check simple serialization
         msg = dict(a=[1,'hi'])
         try:
             packed = pack(msg)
         except Exception:
             raise ValueError("packer could not serialize a simple message")
 
         # ensure packed message is bytes
         if not isinstance(packed, bytes):
             raise ValueError("message packed to %r, but bytes are required"%type(packed))
 
         # check that unpack is pack's inverse
         try:
             unpacked = unpack(packed)
         except Exception:
             raise ValueError("unpacker could not handle the packer's output")
 
         # check datetime support
         msg = dict(t=datetime.now())
         try:
             unpacked = unpack(pack(msg))
         except Exception:
             self.pack = lambda o: pack(squash_dates(o))
             self.unpack = lambda s: extract_dates(unpack(s))
 
     def msg_header(self, msg_type):
         return msg_header(self.msg_id, msg_type, self.username, self.session)
 
     def msg(self, msg_type, content=None, parent=None, subheader=None, header=None):
         """Return the nested message dict.
 
         This format is different from what is sent over the wire. The
         serialize/unserialize methods converts this nested message dict to the wire
         format, which is a list of message parts.
         """
         msg = {}
         header = self.msg_header(msg_type) if header is None else header
         msg['header'] = header
         msg['msg_id'] = header['msg_id']
         msg['msg_type'] = header['msg_type']
         msg['parent_header'] = {} if parent is None else extract_header(parent)
         msg['content'] = {} if content is None else content
         sub = {} if subheader is None else subheader
         msg['header'].update(sub)
         return msg
 
     def sign(self, msg_list):
         """Sign a message with HMAC digest. If no auth, return b''.
 
         Parameters
         ----------
         msg_list : list
             The [p_header,p_parent,p_content] part of the message list.
         """
         if self.auth is None:
             return b''
         h = self.auth.copy()
         for m in msg_list:
             h.update(m)
         return str_to_bytes(h.hexdigest())
 
     def serialize(self, msg, ident=None):
         """Serialize the message components to bytes.
 
         This is roughly the inverse of unserialize. The serialize/unserialize
         methods work with full message lists, whereas pack/unpack work with
         the individual message parts in the message list.
 
         Parameters
         ----------
         msg : dict or Message
             The nexted message dict as returned by the self.msg method.
 
         Returns
         -------
         msg_list : list
             The list of bytes objects to be sent with the format:
             [ident1,ident2,...,DELIM,HMAC,p_header,p_parent,p_content,
              buffer1,buffer2,...]. In this list, the p_* entities are
             the packed or serialized versions, so if JSON is used, these
             are utf8 encoded JSON strings.
         """
         content = msg.get('content', {})
         if content is None:
             content = self.none
         elif isinstance(content, dict):
             content = self.pack(content)
         elif isinstance(content, bytes):
             # content is already packed, as in a relayed message
             pass
         elif isinstance(content, unicode):
             # should be bytes, but JSON often spits out unicode
             content = content.encode('utf8')
         else:
             raise TypeError("Content incorrect type: %s"%type(content))
 
         real_message = [self.pack(msg['header']),
                         self.pack(msg['parent_header']),
                         content
         ]
 
         to_send = []
 
         if isinstance(ident, list):
             # accept list of idents
             to_send.extend(ident)
         elif ident is not None:
             to_send.append(ident)
         to_send.append(DELIM)
 
         signature = self.sign(real_message)
         to_send.append(signature)
 
         to_send.extend(real_message)
 
         return to_send
 
     def send(self, stream, msg_or_type, content=None, parent=None, ident=None,
              buffers=None, subheader=None, track=False, header=None):
         """Build and send a message via stream or socket.
 
         The message format used by this function internally is as follows:
 
         [ident1,ident2,...,DELIM,HMAC,p_header,p_parent,p_content,
          buffer1,buffer2,...]
 
         The serialize/unserialize methods convert the nested message dict into this
         format.
 
         Parameters
         ----------
 
         stream : zmq.Socket or ZMQStream
             The socket-like object used to send the data.
         msg_or_type : str or Message/dict
             Normally, msg_or_type will be a msg_type unless a message is being
             sent more than once. If a header is supplied, this can be set to
             None and the msg_type will be pulled from the header.
 
         content : dict or None
             The content of the message (ignored if msg_or_type is a message).
         header : dict or None
             The header dict for the message (ignores if msg_to_type is a message).
         parent : Message or dict or None
             The parent or parent header describing the parent of this message
             (ignored if msg_or_type is a message).
         ident : bytes or list of bytes
             The zmq.IDENTITY routing path.
         subheader : dict or None
             Extra header keys for this message's header (ignored if msg_or_type
             is a message).
         buffers : list or None
             The already-serialized buffers to be appended to the message.
         track : bool
             Whether to track.  Only for use with Sockets, because ZMQStream
             objects cannot track messages.
 
         Returns
         -------
         msg : dict
             The constructed message.
         (msg,tracker) : (dict, MessageTracker)
             if track=True, then a 2-tuple will be returned,
             the first element being the constructed
             message, and the second being the MessageTracker
 
         """
 
         if not isinstance(stream, (zmq.Socket, ZMQStream)):
             raise TypeError("stream must be Socket or ZMQStream, not %r"%type(stream))
         elif track and isinstance(stream, ZMQStream):
             raise TypeError("ZMQStream cannot track messages")
 
         if isinstance(msg_or_type, (Message, dict)):
             # We got a Message or message dict, not a msg_type so don't
             # build a new Message.
             msg = msg_or_type
         else:
             msg = self.msg(msg_or_type, content=content, parent=parent,
                            subheader=subheader, header=header)
 
         buffers = [] if buffers is None else buffers
         to_send = self.serialize(msg, ident)
         flag = 0
         if buffers:
             flag = zmq.SNDMORE
             _track = False
         else:
             _track=track
         if track:
             tracker = stream.send_multipart(to_send, flag, copy=False, track=_track)
         else:
             tracker = stream.send_multipart(to_send, flag, copy=False)
         for b in buffers[:-1]:
             stream.send(b, flag, copy=False)
         if buffers:
             if track:
                 tracker = stream.send(buffers[-1], copy=False, track=track)
             else:
                 tracker = stream.send(buffers[-1], copy=False)
 
         # omsg = Message(msg)
         if self.debug:
             pprint.pprint(msg)
             pprint.pprint(to_send)
             pprint.pprint(buffers)
 
         msg['tracker'] = tracker
 
         return msg
 
     def send_raw(self, stream, msg_list, flags=0, copy=True, ident=None):
         """Send a raw message via ident path.
 
         This method is used to send a already serialized message.
 
         Parameters
         ----------
         stream : ZMQStream or Socket
             The ZMQ stream or socket to use for sending the message.
         msg_list : list
             The serialized list of messages to send. This only includes the
             [p_header,p_parent,p_content,buffer1,buffer2,...] portion of
             the message.
         ident : ident or list
             A single ident or a list of idents to use in sending.
         """
         to_send = []
         if isinstance(ident, bytes):
             ident = [ident]
         if ident is not None:
             to_send.extend(ident)
 
         to_send.append(DELIM)
         to_send.append(self.sign(msg_list))
         to_send.extend(msg_list)
         stream.send_multipart(msg_list, flags, copy=copy)
 
     def recv(self, socket, mode=zmq.NOBLOCK, content=True, copy=True):
         """Receive and unpack a message.
 
         Parameters
         ----------
         socket : ZMQStream or Socket
             The socket or stream to use in receiving.
 
         Returns
         -------
         [idents], msg
             [idents] is a list of idents and msg is a nested message dict of
             same format as self.msg returns.
         """
         if isinstance(socket, ZMQStream):
             socket = socket.socket
         try:
             msg_list = socket.recv_multipart(mode)
         except zmq.ZMQError as e:
             if e.errno == zmq.EAGAIN:
                 # We can convert EAGAIN to None as we know in this case
                 # recv_multipart won't return None.
                 return None,None
             else:
                 raise
         # split multipart message into identity list and message dict
         # invalid large messages can cause very expensive string comparisons
         idents, msg_list = self.feed_identities(msg_list, copy)
         try:
             return idents, self.unserialize(msg_list, content=content, copy=copy)
         except Exception as e:
             # TODO: handle it
             raise e
 
     def feed_identities(self, msg_list, copy=True):
         """Split the identities from the rest of the message.
 
         Feed until DELIM is reached, then return the prefix as idents and
         remainder as msg_list. This is easily broken by setting an IDENT to DELIM,
         but that would be silly.
 
         Parameters
         ----------
         msg_list : a list of Message or bytes objects
             The message to be split.
         copy : bool
             flag determining whether the arguments are bytes or Messages
 
         Returns
         -------
         (idents, msg_list) : two lists
             idents will always be a list of bytes, each of which is a ZMQ
             identity. msg_list will be a list of bytes or zmq.Messages of the
             form [HMAC,p_header,p_parent,p_content,buffer1,buffer2,...] and
             should be unpackable/unserializable via self.unserialize at this
             point.
         """
         if copy:
             idx = msg_list.index(DELIM)
             return msg_list[:idx], msg_list[idx+1:]
         else:
             failed = True
             for idx,m in enumerate(msg_list):
                 if m.bytes == DELIM:
                     failed = False
                     break
             if failed:
                 raise ValueError("DELIM not in msg_list")
             idents, msg_list = msg_list[:idx], msg_list[idx+1:]
             return [m.bytes for m in idents], msg_list
 
     def unserialize(self, msg_list, content=True, copy=True):
         """Unserialize a msg_list to a nested message dict.
 
         This is roughly the inverse of serialize. The serialize/unserialize
         methods work with full message lists, whereas pack/unpack work with
         the individual message parts in the message list.
 
         Parameters:
         -----------
         msg_list : list of bytes or Message objects
             The list of message parts of the form [HMAC,p_header,p_parent,
             p_content,buffer1,buffer2,...].
         content : bool (True)
             Whether to unpack the content dict (True), or leave it packed
             (False).
         copy : bool (True)
             Whether to return the bytes (True), or the non-copying Message
             object in each place (False).
 
         Returns
         -------
         msg : dict
             The nested message dict with top-level keys [header, parent_header,
             content, buffers].
         """
         minlen = 4
         message = {}
         if not copy:
             for i in range(minlen):
                 msg_list[i] = msg_list[i].bytes
         if self.auth is not None:
             signature = msg_list[0]
             if not signature:
                 raise ValueError("Unsigned Message")
             if signature in self.digest_history:
                 raise ValueError("Duplicate Signature: %r"%signature)
             self.digest_history.add(signature)
             check = self.sign(msg_list[1:4])
             if not signature == check:
                 raise ValueError("Invalid Signature: %r"%signature)
         if not len(msg_list) >= minlen:
             raise TypeError("malformed message, must have at least %i elements"%minlen)
         header = self.unpack(msg_list[1])
         message['header'] = header
         message['msg_id'] = header['msg_id']
         message['msg_type'] = header['msg_type']
         message['parent_header'] = self.unpack(msg_list[2])
         if content:
             message['content'] = self.unpack(msg_list[3])
         else:
             message['content'] = msg_list[3]
 
         message['buffers'] = msg_list[4:]
         return message
 
 def test_msg2obj():
     am = dict(x=1)
     ao = Message(am)
     assert ao.x == am['x']
 
     am['y'] = dict(z=1)
     ao = Message(am)
     assert ao.y.z == am['y']['z']
 
     k1, k2 = 'y', 'z'
     assert ao[k1][k2] == am[k1][k2]
 
     am2 = dict(ao)
     assert am['x'] == am2['x']
     assert am['y']['z'] == am2['y']['z']