support fields with "AllowSubtypes"

Allow fields in structs that are subtype of a class.

asyncua/ua/ua_binary.py

@@ -13,7 +13,7 @@ from dataclasses import is_dataclass, fields
 from asyncua import ua
 from .uaerrors import UaError
 from ..common.utils import Buffer
-from .uatypes import type_is_list, type_is_union, type_from_list, types_from_union
+from .uatypes import ExtensionObject, type_is_list, type_is_union, type_from_list, types_from_union, type_allow_subclass
 
 logger = logging.getLogger('__name__')
 
@@ -326,6 +326,8 @@ def struct_to_binary(obj):
 @functools.lru_cache(maxsize=None)
 def create_type_serializer(uatype):
     """Create a binary serialization function for the given UA type"""
+    if type_allow_subclass(uatype):
+        return extensionobject_to_binary
     if type_is_list(uatype):
         return create_list_serializer(type_from_list(uatype))
     if hasattr(Primitives, uatype.__name__):
@@ -613,6 +615,7 @@ def _create_dataclass_deserializer(objtype):
     for field in fields(objtype):
         optional_enc_bit = 0
         field_type = resolved_fieldtypes[field.name]
+        subtypes = type_allow_subclass(field.type)
         # if our member has a switch and it is not set we will need to skip it
         if type_is_union(field_type):
             optional_enc_bit = 1 << enc_count
@@ -623,7 +626,7 @@ def _create_dataclass_deserializer(objtype):
     def decode(data):
         kwargs = {}
         enc = 0
-        for field, optional_enc_bit, deserialize_field in field_deserializers:
+        for field, optional_enc_bit, deserialize_field, subtypes in field_deserializers:
             if field.name == "Encoding":
                 enc = deserialize_field(data)
             elif optional_enc_bit == 0 or enc & optional_enc_bit:

asyncua/ua/uatypes.py

@@ -52,6 +52,8 @@ def type_is_union(uatype):
 def type_is_list(uatype):
     return get_origin(uatype) == list
 
+def type_allow_subclass(uatype):
+    return get_origin(uatype) not in [Union, list, None]
 
 def types_from_union(uatype, origin=None):
     if origin is None:
@@ -68,12 +70,16 @@ def types_from_union(uatype, origin=None):
 def type_from_list(uatype):
     return get_args(uatype)[0]
 
+def type_from_allow_subtype(uatype):
+    return get_args(uatype)[0]
 
 def type_string_from_type(uatype):
     if type_is_union(uatype):
         uatype = types_from_union(uatype)[0]
     elif type_is_list(uatype):
         uatype = type_from_list(uatype)
+    elif type_allow_subclass(uatype):
+        uatype = type_from_allow_subtype(uatype)
     return uatype.__name__
 
 

schemas/generate_model_from_nodeset.py

@@ -4,7 +4,7 @@ Generate address space code from xml file specification
 from xml.etree import ElementTree
 from logging import getLogger
 from dataclasses import dataclass, field
-from typing import Any, List
+from typing import Any, List, Optional
 import re
 from pathlib import Path
 
@@ -96,7 +96,7 @@ class Field:
 @dataclass
 class Struct:
     name: str = None
-    basetype: str = None
+    basetype: Optional[str] = None
     node_id: str = None
     doc: str = ""
     fields: List[Field] = field(default_factory=list)
@@ -123,7 +123,7 @@ class Enum:
     data_type: str = None
     fields: List[Field] = field(default_factory=list)
     doc: str = ""
-    is_option_set: bool = False 
+    is_option_set: bool = False
 
 
 
@@ -210,6 +210,8 @@ def reorder_structs(model):
                 ok = False
         if ok:
             _add_struct(s, newstructs, waiting_structs, types)
+
+
     if len(model.structs) != len(newstructs):
         _logger.warning('Error while reordering structs, some structs could not be reinserted: had %s structs, we now have %s structs', len(model.structs), len(newstructs))
         s1 = set(model.structs)
@@ -230,6 +232,9 @@ def nodeid_to_names(model):
     ids["22"] = "ExtensionObject"
 
     for struct in model.structs:
+        if struct.basetype is not None:
+            if struct.basetype.startswith("i="):
+                struct.basetype = ids[struct.basetype[2:]]
         for sfield in struct.fields:
             if sfield.data_type.startswith("i="):
                 sfield.data_type = ids[sfield.data_type[2:]]
@@ -270,6 +275,16 @@ def split_requests(model):
     model.structs = structs
 
 
+def get_basetypes(el) -> List[str]:
+    # return all basetypes
+    basetypes = []
+    for ref in el.findall("./{*}References/{*}Reference"):
+        if ref.get("ReferenceType") == "HasSubtype" and \
+           ref.get("IsForward", "true") == "false" and \
+           ref.text != "i=22":
+            basetypes.append(ref.text)
+    return basetypes
+
 class Parser:
     def __init__(self, path):
         self.path = path
@@ -341,8 +356,14 @@ class Parser:
             doc=doc,
             node_id=el.get("NodeId"),
         )
+        basetypes = get_basetypes(el)
+        if basetypes:
+            struct.basetype = basetypes[0]
+            if len(basetypes) > 1:
+                print(f'Error found mutliple basetypes for {struct} {basetypes}')
         for sfield in el.findall("./{*}Definition/{*}Field"):
             opt = sfield.get("IsOptional", "false")
+            allow_subtypes = True if sfield.get("AllowSubTypes", "false") == 'true' else False
             is_optional = True if opt == "true" else False
             f = Field(
                 name=sfield.get("Name"),
@@ -351,6 +372,7 @@ class Parser:
                 array_dimensions=sfield.get("ArayDimensions"),
                 value=sfield.get("Value"),
                 is_optional=is_optional,
+                allow_subtypes=allow_subtypes
             )
             if is_optional:
                 struct.has_optional = True

schemas/generate_protocol_python.py

@@ -4,7 +4,7 @@ import datetime
 BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 
 IgnoredEnums = ["NodeIdType"]
-IgnoredStructs = ["QualifiedName", "NodeId", "ExpandedNodeId", "FilterOperand", "Variant", "DataValue",
+IgnoredStructs = ["QualifiedName", "NodeId", "ExpandedNodeId", "Variant", "DataValue",
                   "ExtensionObject", "XmlElement", "LocalizedText"]
 
 
@@ -61,7 +61,7 @@ class CodeGenerator:
         self.write('')
         self.write('from datetime import datetime')
         self.write('from enum import IntEnum, IntFlag')
-        self.write('from typing import Union, List, Optional')
+        self.write('from typing import Union, List, Optional, Type')
         self.write('from dataclasses import dataclass, field')
         self.write('')
         self.write('from asyncua.ua.uatypes import FROZEN')
@@ -110,7 +110,10 @@ class CodeGenerator:
         self.write('')
         self.iidx = 0
         self.write('@dataclass(frozen=FROZEN)')
-        self.write(f'class {obj.name}:')
+        if obj.basetype:
+            self.write(f'class {obj.name}({obj.basetype}):')
+        else:
+            self.write(f'class {obj.name}:')
         self.iidx += 1
         self.write('"""')
         if obj.doc:
@@ -141,7 +144,8 @@ class CodeGenerator:
                 typestring = f"List[{typestring}]"
             if field.is_optional:
                 typestring = f"Optional[{typestring}]"
-
+            if field.allow_subtypes:
+                typestring = f"Type[{typestring}]"
             if field.name == field.data_type:
                 # variable name and type name are the same. Dataclass do not like it
                 hack_names.append(field.name)