Implement native BPMN workflows and guided modes

src/govoplan_workflow/backend/bpmn.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 from collections import Counter
 from dataclasses import dataclass
 from typing import Literal
-from xml.etree.ElementTree import ParseError
+from xml.etree.ElementTree import Element, ParseError
 
 from defusedxml.ElementTree import fromstring
 from defusedxml.common import DefusedXmlException
@@ -27,33 +27,45 @@ NATIVE_EXECUTION_ELEMENTS = frozenset(
     {
         "startEvent",
         "endEvent",
+        "task",
         "manualTask",
         "userTask",
         "serviceTask",
-        "businessRuleTask",
-        "receiveTask",
         "sendTask",
-        "exclusiveGateway",
-        "parallelGateway",
-        "sequenceFlow",
+        "receiveTask",
         "intermediateCatchEvent",
-        "intermediateThrowEvent",
+        "sequenceFlow",
     }
 )
 
 NATIVE_MAPPING_ELEMENTS = NATIVE_EXECUTION_ELEMENTS | frozenset(
     {
-        "task",
+        "definitions",
+        "process",
+        "collaboration",
+        "documentation",
+        "extensionElements",
+        "incoming",
+        "outgoing",
+        "conditionExpression",
         "scriptTask",
+        "serviceTask",
+        "businessRuleTask",
+        "receiveTask",
+        "sendTask",
         "callActivity",
         "subProcess",
         "transaction",
         "adHocSubProcess",
+        "exclusiveGateway",
+        "parallelGateway",
         "inclusiveGateway",
         "eventBasedGateway",
         "complexGateway",
         "boundaryEvent",
         "eventSubProcess",
+        "intermediateCatchEvent",
+        "intermediateThrowEvent",
         "dataObject",
         "dataObjectReference",
         "dataStoreReference",
@@ -64,6 +76,16 @@ NATIVE_MAPPING_ELEMENTS = NATIVE_EXECUTION_ELEMENTS | frozenset(
         "textAnnotation",
         "association",
         "group",
+        "dataInputAssociation",
+        "dataOutputAssociation",
+        "choreography",
+        "choreographyTask",
+        "callChoreography",
+        "subChoreography",
+        "conversation",
+        "callConversation",
+        "subConversation",
+        "conversationLink",
     }
 )
 
@@ -116,7 +138,7 @@ def bpmn_support_level(element_type: str) -> SupportLevel:
     return "interchange_only"
 
 
-def inspect_bpmn_xml(xml: str) -> BpmnInspection:
+def parse_bpmn_xml(xml: str) -> Element:
     encoded = xml.encode("utf-8")
     if not encoded:
         raise BpmnInspectionError("BPMN XML is empty")
@@ -134,6 +156,15 @@ def inspect_bpmn_xml(xml: str) -> BpmnInspection:
         raise BpmnInspectionError(
             "BPMN document root must be bpmn:definitions in the BPMN 2.0 model namespace"
         )
+    if sum(1 for _item in root.iter()) > MAX_BPMN_ELEMENTS:
+        raise BpmnInspectionError(
+            f"BPMN document exceeds the {MAX_BPMN_ELEMENTS}-element inspection limit"
+        )
+    return root
+
+
+def inspect_bpmn_xml(xml: str) -> BpmnInspection:
+    root = parse_bpmn_xml(xml)
 
     diagnostics: list[BpmnDiagnostic] = []
     elements: list[BpmnElementInventoryItem] = []
@@ -149,10 +180,6 @@ def inspect_bpmn_xml(xml: str) -> BpmnInspection:
     while stack:
         element, parent_type, parent_id = stack.pop()
         visited += 1
-        if visited > MAX_BPMN_ELEMENTS:
-            raise BpmnInspectionError(
-                f"BPMN document exceeds the {MAX_BPMN_ELEMENTS}-element inspection limit"
-            )
         element_namespace, element_type = _qualified_name(element.tag)
         element_id = _bounded_attribute(element.attrib.get("id"), 255)
         if element_namespace == BPMN_MODEL_NAMESPACE:
@@ -268,17 +295,24 @@ def _collect_references(
     attributes: dict[str, str],
     references: list[tuple[str, str | None, str]],
 ) -> None:
-    fields: tuple[str, ...]
-    if element_type == "sequenceFlow":
-        fields = ("sourceRef", "targetRef")
-    elif element_type == "messageFlow":
-        fields = ("sourceRef", "targetRef", "messageRef")
-    elif element_type == "participant":
-        fields = ("processRef",)
-    elif element_type == "lane":
-        fields = ("partitionElementRef",)
-    else:
-        fields = ()
+    fields_by_element = {
+        "sequenceFlow": ("sourceRef", "targetRef"),
+        "messageFlow": ("sourceRef", "targetRef", "messageRef"),
+        "association": ("sourceRef", "targetRef"),
+        "participant": ("processRef",),
+        "lane": ("partitionElementRef",),
+        "boundaryEvent": ("attachedToRef",),
+        "dataInputAssociation": ("sourceRef", "targetRef"),
+        "dataOutputAssociation": ("sourceRef", "targetRef"),
+        "dataObjectReference": ("dataObjectRef",),
+        "dataStoreReference": ("dataStoreRef",),
+        "messageEventDefinition": ("messageRef", "operationRef"),
+        "signalEventDefinition": ("signalRef",),
+        "errorEventDefinition": ("errorRef",),
+        "escalationEventDefinition": ("escalationRef",),
+        "compensateEventDefinition": ("activityRef",),
+    }
+    fields = fields_by_element.get(element_type, ())
     for field in fields:
         value = attributes.get(field)
         if value:
@@ -306,4 +340,5 @@ __all__ = [
     "NATIVE_MAPPING_ELEMENTS",
     "bpmn_support_level",
     "inspect_bpmn_xml",
+    "parse_bpmn_xml",
 ]