backup: Add support for IPv6 addresses in socket backend

Support the bracketed `socket:[::]:1234` syntax for IPv6 addresses.

Also, add factor out the socket URI parsing to a function and add
tests for it. As a forward compatibility measure, reject query strings
(`?a=b`) because I intend to use these for parameters such as SOCKS5
proxy (for Tor) in a future patch.

backup/remote.md

@@ -15,7 +15,8 @@ The remote backup system consists of two parts:
 - A server daemon that receives changes from the backup backend and communicates with a local backup backend
   to store them. The server side does not need to be running c-lightning, nor have it installed.
 
-The backend URL format is `socket:<host>:<port>`. For example `socket:127.0.0.1:1234`.
+The backend URL format is `socket:<host>:<port>`. For example `socket:127.0.0.1:1234`. To supply a IPv6
+address use the bracketed syntax `socket:[::1]:1234`.
 
 To run the server against a local backend use `backup-cli server file://.../ 127.0.0.1:1234`.
 

backup/socketbackend.py

@@ -1,21 +1,71 @@
-import json
-import logging, socket, struct
+from collections import namedtuple
+import json, logging, socket, re, struct
 from typing import Tuple, Iterator
-from urllib.parse import urlparse
+from urllib.parse import urlparse, parse_qs
 
 from backend import Backend, Change
 from protocol import PacketType, recvall, PKT_CHANGE_TYPES, change_from_packet, packet_from_change, send_packet, recv_packet
 
+SocketURLInfo = namedtuple('SocketURLInfo', ['host', 'port', 'addrtype'])
+
+class AddrType:
+    IPv4 = 0
+    IPv6 = 1
+    NAME = 2
+
+def parse_socket_url(destination: str) -> SocketURLInfo:
+    '''Parse a socket: URL to extract the information contained in it.'''
+    url = urlparse(destination)
+    if url.scheme != 'socket':
+        raise ValueError('Scheme for socket backend must be socket:...')
+
+    if url.path.startswith('['): # bracketed IPv6 address
+        eidx = url.path.find(']')
+        if eidx == -1:
+            raise ValueError('Unterminated bracketed host address.')
+        host = url.path[1:eidx]
+        addrtype = AddrType.IPv6
+        eidx += 1
+        if eidx >= len(url.path) or url.path[eidx] != ':':
+            raise ValueError('Port number missing.')
+        eidx += 1
+    else:
+        eidx = url.path.find(':')
+        if eidx == -1:
+            raise ValueError('Port number missing.')
+        host = url.path[0:eidx]
+        if re.match('\d+\.\d+\.\d+\.\d+$', host): # matches IPv4 address format
+            addrtype = AddrType.IPv4
+        else:
+            addrtype = AddrType.NAME
+        eidx += 1
+
+    try:
+        port = int(url.path[eidx:])
+    except ValueError:
+        raise ValueError('Invalid port number')
+
+    # parse query parameters
+    # reject unknown parameters (currently all of them)
+    qs = parse_qs(url.query)
+    if len(qs):
+        raise ValueError('Invalid query string')
+
+    return SocketURLInfo(host=host, port=port, addrtype=addrtype)
+
 class SocketBackend(Backend):
     def __init__(self, destination: str, create: bool):
         self.version = None
         self.prev_version = None
         self.destination = destination
-        self.url = urlparse(self.destination)
-        self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        (host, port) = self.url.path.split(':')
-        logging.info('Initialized socket backend')
-        self.sock.connect((host, int(port)))
+        self.url = parse_socket_url(destination)
+        if self.url.addrtype == AddrType.IPv6:
+            self.sock = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
+        else: # TODO NAME is assumed to be IPv4 for now
+            self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        logging.info('Initialized socket backend, connecting to {}:{} (addrtype {})...'.format(
+                self.url.host, self.url.port, self.url.addrtype))
+        self.sock.connect((self.url.host, self.url.port))
         logging.info('Connected to {}'.format(destination))
 
     def _send_packet(self, typ: int, payload: bytes) -> None:

backup/test_backup.py

@@ -1,5 +1,6 @@
 from backend import Backend
 from filebackend import FileBackend
+import socketbackend
 from flaky import flaky
 from pyln.testing.fixtures import *  # noqa: F401,F403
 from pyln.testing.utils import sync_blockheight
@@ -279,3 +280,38 @@ def test_compact(bitcoind, directory, node_factory):
     l1.rpc.backup_compact()
     tmp = tempfile.TemporaryDirectory()
     subprocess.check_call([cli_path, "restore", bdest, tmp.name])
+
+def test_parse_socket_url():
+    with pytest.raises(ValueError):
+        # fail: invalid url scheme
+        socketbackend.parse_socket_url('none')
+        # fail: no port number
+        socketbackend.parse_socket_url('socket:127.0.0.1')
+        socketbackend.parse_socket_url('socket:127.0.0.1:')
+        # fail: unbracketed IPv6
+        socketbackend.parse_socket_url('socket:::1:1234')
+        # fail: no port number IPv6
+        socketbackend.parse_socket_url('socket:[::1]')
+        socketbackend.parse_socket_url('socket:[::1]:')
+        # fail: invalid port number
+        socketbackend.parse_socket_url('socket:127.0.0.1:12bla')
+        # fail: unrecognized query string key
+        socketbackend.parse_socket_url('socket:127.0.0.1:1234?dummy=value')
+
+    # IPv4
+    s = socketbackend.parse_socket_url('socket:127.0.0.1:1234')
+    assert(s.host == '127.0.0.1')
+    assert(s.port == 1234)
+    assert(s.addrtype == socketbackend.AddrType.IPv4)
+
+    # IPv6
+    s = socketbackend.parse_socket_url('socket:[::1]:1235')
+    assert(s.host == '::1')
+    assert(s.port == 1235)
+    assert(s.addrtype == socketbackend.AddrType.IPv6)
+
+    # Hostname
+    s = socketbackend.parse_socket_url('socket:backup.local:1236')
+    assert(s.host == 'backup.local')
+    assert(s.port == 1236)
+    assert(s.addrtype == socketbackend.AddrType.NAME)