Sync Module

class py2p.sync.SyncSocket(*args, **kwargs)[source]

This class is used to sync dictionaries between programs. It extends py2p.mesh.MeshSocket

Inheritance diagram of py2p.sync.SyncSocket

Because of this inheritance, this can also be used as an alert network

This also implements and optional leasing system by default. This leasing system means that if node A sets a key, node B cannot overwrite the value at that key for an hour.

This may be turned off by adding leasing=False to the constructor.

Added Events:

Event 'update'(conn, key, data, new_meta)

This event is triggered when a key is updated in your synchronized dictionary. new_meta will be an object containing metadata of this change, including the time of change, and who initiated the change.

Parameters:
Event 'delete'(conn, key)

This event is triggered when a key is deleted from your synchronized dictionary.

Parameters:
__init__(*args, **kwargs)

Initialize a chord socket

Parameters:
  • addr – The address you wish to bind to (ie: “192.168.1.1”)
  • port – The port you wish to bind to (ie: 44565)
  • prot – The Protocol you wish to operate over, defined by a py2p.base.Protocol object
  • out_addr – Your outward facing address. Only needed if you’re connecting over the internet. If you use ‘0.0.0.0’ for the addr argument, this will automatically be set to your LAN address.
  • debug_level – The verbosity you want this socket to use when printing event data
Raises:

SocketException – The address you wanted could not be bound, or is otherwise used

data
metadata
_send_peers(handler)[source]

Shortcut method to send a handshake response. This method is extracted from __handle_handshake() in order to allow cleaner inheritence from py2p.sync.SyncSocket

__setitem__(key, data)[source]

Updates the value at a given key.

Parameters:
  • key – The key that you wish to update. Must be a str or bytes-like object
  • value – The value you wish to put at this key.
Raises:
  • KeyError – If you do not have the lease for this slot. Lease is given automatically for one hour if the slot is open.
  • TypeError – If your key is not bytes -like OR if your value is not serializable. This means your value must be one of the following:
set(key, data)[source]

Updates the value at a given key.

Parameters:
  • key – The key that you wish to update. Must be a str or bytes-like object
  • value – The value you wish to put at this key.
Raises:
  • KeyError – If you do not have the lease for this slot. Lease is given automatically for one hour if the slot is open.
  • TypeError – If your key is not bytes -like OR if your value is not serializable. This means your value must be one of the following:
update(update_dict)[source]

Equivalent to dict.update()

This calls SyncSocket.__setitem__() for each key/value pair in the given dictionary.

Parameters:update_dict – A dict-like object to extract key/value pairs from. Key and value be a str or bytes-like object
Raises:KeyError – If you do not have the lease for this slot. Lease is given automatically for one hour if the slot is open.
__getitem__(key)[source]

Looks up the value at a given key.

Parameters:key – The key that you wish to check. Must be a str or bytes-like object
Returns:The value at said key
Raises:KeyError – If there is no value assigned at that key
get(key, ifError=None)[source]

Retrieves the value at a given key.

Parameters:
  • key – The key that you wish to check. Must be a str or bytes-like object
  • ifError – The value you wish to return on exception (default: None)
Returns:

The value at said key, or the value at ifError if there’s an Exception

keys()[source]
Returns:an iterator of the underlying dict s keys
__iter__()[source]

Returns: an iterator of the underlying dict s keys

values()[source]
Returns:an iterator of the underlying dict s values
items()[source]
Returns:an iterator of the underlying dict s items
pop(key, *args)[source]

Returns a value, with the side effect of deleting that association

Parameters:
  • Key – The key you wish to look up. Must be a str or bytes-like object
  • ifError – The value you wish to return on Exception (default: raise an Exception)
Returns:

The value of the supplied key, or ifError

Raises:

KeyError – If the key does not have an associated value

popitem()[source]

Returns an association, with the side effect of deleting that association

Returns:An arbitrary association
copy()[source]

Returns a dict copy of this synchronized hash table

_BaseSocket__closed
_BaseSocket__handlers
_MeshSocket__clean_waterfalls()

This function cleans the set of recently relayed messages based on the following heuristics:

  • Delete all older than 60 seconds
_MeshSocket__handle_handshake(msg, handler)

This callback is used to deal with handshake signals. Its three primary jobs are:

  • reject connections seeking a different network
  • set connection state
  • deal with connection conflicts
Parameters:
Returns:

Either True or None

_MeshSocket__handle_request(msg, handler)

This callback is used to deal with request signals. Its three primary jobs are:

  • respond with a peers signal if packets[1] is '*'
  • if you know the ID requested, respond to it
  • if you don’t, make a request with your peers
Parameters:
Returns:

Either True or None

_MeshSocket__handle_response(msg, handler)

This callback is used to deal with response signals. Its two primary jobs are:

  • if it was your request, send the deferred message
  • if it was someone else’s request, relay the information
Parameters:
Returns:

Either True or None

_MeshSocket__resolve_connection_conflict(handler, h_id)

Sometimes in trying to recover a network a race condition is created. This function applies a heuristic to try and organize the fallout from that race condition. While it isn’t perfect, it seems to have increased connection recovery rate from ~20% to ~75%. This statistic is from memory on past tests. Much improvement can be made here, but this statistic can likely never be brought to 100%.

In the failure condition, the overall network is unaffected for large networks. In small networks this failure condition causes a fork, usually where an individual node is kicked out.

Parameters:
  • handler – The handler with whom you have a connection conflict
  • h_id – The id of this handler
_SyncSocket__check_lease(key, new_data, new_meta)
_SyncSocket__handle_store(msg, handler)

This callback is used to deal with data storage signals. Its two primary jobs are:

  • store data in a given key
  • delete data in a given key
Args:
msg: A Message handler: A MeshConnection
Returns:
Either True or None
_SyncSocket__leasing
_SyncSocket__store(key, new_data, new_meta, error=True)

Private API method for storing data. You have permission to store something if:

  • The network is not enforcing leases, or
  • There is no value at that key, or
  • The lease on that key has lapsed (not been set in the last hour), or
  • You are the owner of that key
Parameters:
  • key – The key you wish to store data at
  • new_data – The data you wish to store in said key
  • new_meta – The metadata associated with this storage
  • error – A boolean which says whether to raise a KeyError if you can’t store there
Raises:

KeyError – If someone else has a lease at this value, and error is True

_get_peer_list()

This function is used to generate a list-formatted group of your peers. It goes in format [ ((addr, port), ID), ...]

_handle_peers(msg, handler)

This callback is used to deal with peer signals. Its primary jobs is to connect to the given peers, if this does not exceed py2p.mesh.max_outgoing

Parameters:
Returns:

Either True or None

_logger
_send_handshake(handler)

Shortcut method for sending a handshake to a given handler

Parameters:handler – A MeshConnection
awaiting_ids
close()

If the socket is not closed, close the socket

Raises:RuntimeError – The socket was already closed
connect(addr, port, id=None, conn_type=<class 'py2p.mesh.MeshConnection'>)

This function connects you to a specific node in the overall network. Connecting to one node should connect you to the rest of the network, however if you connect to the wrong subnet, the handshake failure involved is silent. You can check this by looking at the truthiness of this objects routing table. Example:

>>> conn = mesh.MeshSocket('localhost', 4444)
>>> conn.connect('localhost', 5555)
>>> # do some other setup for your program
>>> if not conn.routing_table:
...     conn.connect('localhost', 6666)  # any fallback address
Parameters:
  • addr – A string address
  • port – A positive, integral port
  • id – A string-like object which represents the expected ID of this node
daemon
debug_level
disconnect(handler)

Closes a given connection, and removes it from your routing tables

Parameters:handler – the connection you would like to close
emit(event, *args, **kwargs)

Emit event, passing *args and **kwargs to each attached function. Returns True if any functions are attached to event; otherwise returns False.

Example:

ee.emit('data', '00101001')

Assuming data is an attached function, this will call data('00101001')'.

For coroutine event handlers, calling emit is non-blocking. In other words, you do not have to await any results from emit, and the coroutine is scheduled in a fire-and-forget fashion.

handle_msg(msg, conn)

Decides how to handle various message types, allowing some to be handled automatically

Parameters:
Returns:

True if an action was taken, None if not.

id
incoming

IDs of incoming connections

listeners(event)

Returns the list of all listeners registered to the event.

on(event, f=None)

Registers the function (or optionally an asyncio coroutine function) f to the event name event.

If f isn’t provided, this method returns a function that takes f as a callback; in other words, you can use this method as a decorator, like so:

@ee.on('data')
def data_handler(data):
    print(data)

As mentioned, this method can also take an asyncio coroutine function:

@ee.on('data')
async def data_handler(data)
    await do_async_thing(data)

This will automatically schedule the coroutine using the configured scheduling function (defaults to asyncio.ensure_future) and the configured event loop (defaults to asyncio.get_event_loop()).

once(event, f=None)

The same as ee.on, except that the listener is automatically removed after being called.

out_addr
outgoing

IDs of outgoing connections

protocol
queue
recv(quantity=1)

This function has two behaviors depending on whether quantity is left as default.

If quantity is given, it will return a list of Message objects up to length quantity.

If quantity is left alone, it will return either a single Message object, or None

Parameters:quantity – The maximum number of Message s you would like to pull (default: 1)
Returns:A list of Message s, an empty list, a single Message , or None
register_handler(method)

Register a handler for incoming method.

Parameters:method – A function with two given arguments. Its signature should be of the form handler(msg, handler), where msg is a py2p.base.Message object, and handler is a py2p.base.BaseConnection object. It should return True if it performed an action, to reduce the number of handlers checked.
Raises:ValueError – If the method signature doesn’t parse correctly
remove_all_listeners(event=None)

Remove all listeners attached to event. If event is None, remove all listeners on all events.

remove_listener(event, f)

Removes the function f from event.

request_peers()

Requests your peers’ routing tables

requests
routing_table
send(*args, **kargs)

This sends a message to all of your peers. If you use default values it will send it to everyone on the network

Parameters:
  • *args – A list of objects you want your peers to receive
  • **kargs – There are two keywords available:
  • flag – A string or bytes-like object which defines your flag. In other words, this defines packet 0.
  • type – A string or bytes-like object which defines your message type. Changing this from default can have adverse effects.
Raises:

TypeError – If any of the arguments are not serializable. This means your objects must be one of the following:

Warning

If you change the type attribute from default values, bad things could happen. It MUST be a value from py2p.base.flags, and more specifically, it MUST be either broadcast or whisper. The only other valid flags are waterfall and renegotiate, but these are RESERVED and must NOT be used.

status

The status of the socket.

Returns:"Nominal" if all is going well, or a list of unexpected (Exception, traceback) tuples if not
waterfall(msg)

This function handles message relays. Its return value is based on whether it took an action or not.

Parameters:msg – The Message in question
Returns:True if the message was then forwarded. False if not.
waterfalls
class py2p.sync.metatuple[source]

This class is used to store metadata for a particular key