Load MANUFDB lazily

scapy/all.py

@@ -11,6 +11,8 @@
 from scapy.config import *
 from scapy.dadict import *
 from scapy.data import *
+# MANUFDB is lazily loaded, so it isn't picked up by the * import above
+from scapy.data import MANUFDB
 from scapy.error import *
 from scapy.themes import *
 from scapy.arch import *

scapy/data.py

@@ -627,18 +627,33 @@ def select_path(directories, filename):
         "services",
     ))
     ETHER_TYPES = load_ethertypes()
-    MANUFDB = load_manuf()
 else:
     IP_PROTOS = load_protocols("/etc/protocols")
     TCP_SERVICES, UDP_SERVICES, SCTP_SERVICES = load_services("/etc/services")
     ETHER_TYPES = load_ethertypes("/etc/ethertypes")
-    MANUFDB = load_manuf(
-        select_path(
-            ['/usr', '/usr/local', '/opt', '/opt/wireshark',
-             '/Applications/Wireshark.app/Contents/Resources'],
-            "share/wireshark/manuf"
+
+
+def __getattr__(attr):
+    # type: (str) -> Any
+    """
+    Lazily load MANUFDB on first access, as it is large and only used to
+    resolve MAC addresses.
+    """
+    if attr != "MANUFDB":
+        raise AttributeError("module %r has no attribute %r" % (__name__, attr))
+    if WINDOWS:
+        manufdb = load_manuf()
+    else:
+        manufdb = load_manuf(
+            select_path(
+                ['/usr', '/usr/local', '/opt', '/opt/wireshark',
+                 '/Applications/Wireshark.app/Contents/Resources'],
+                "share/wireshark/manuf"
+            )
         )
-    )
+    globals()["MANUFDB"] = manufdb
+    return manufdb
+
 
 BLUETOOTH_CORE_COMPANY_IDENTIFIERS = load_bluetoothids()
 

test/regression.uts

@@ -1065,6 +1065,23 @@ assert in6_addrtovendor("fe80::0200:0fff:fe01:0203").lower().startswith("next")
 assert conf.manufdb.lookup("00:00:0F:01:02:03") == ('Next', 'Next, Inc.')
 assert "00:00:0F" in conf.manufdb.reverse_lookup("Next")
 
+= Test manuf DB is lazily loaded
+~ manufdb
+import subprocess
+out = subprocess.check_output([
+    sys.executable, "-c",
+    "import scapy.data; print('MANUFDB' in vars(scapy.data))"
+])
+assert out.strip() == b"False"
+import scapy.data
+assert scapy.data.MANUFDB is conf.manufdb
+assert "MANUFDB" in vars(scapy.data)
+try:
+    scapy.data.DOES_NOT_EXIST
+    assert False
+except AttributeError:
+    pass
+
 = Test multiple wireshark's manuf formats
 ~ manufdb