Control flow discovery fix

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "simod"
-version = "5.1.5"
+version = "5.1.6"
 authors = [
     "Ihar Suvorau <ihar.suvorau@gmail.com>",
     "David Chapela <david.chapela@ut.ee>",

src/simod/control_flow/discovery.py

@@ -140,12 +140,11 @@ def add_bpmn_diagram_to_model(bpmn_model_path: Path):
     """
     global bpmn_layout_jar_path
 
-    args = [
-        "java",
-        "-jar",
-        '"' + str(bpmn_layout_jar_path) + '"',
-        '"' + str(bpmn_model_path) + '"'
-    ]
+    if is_windows():
+        args = ["java", "-jar", '"' + str(bpmn_layout_jar_path) + '"', '"' + str(bpmn_model_path) + '"']
+    else:
+        args = ["java", "-jar", str(bpmn_layout_jar_path), str(bpmn_model_path)]
+
     print_step(f"Adding BPMN diagram to the model: {args}")
     execute_external_command(args)
 

src/simod/simod.py

@@ -30,6 +30,7 @@
 from simod.resource_model.repair import repair_with_missing_activities
 from simod.resource_model.settings import HyperoptIterationParams as ResourceModelHyperoptIterationParams
 from simod.runtime_meter import RuntimeMeter
+from simod.settings.control_flow_settings import ProcessModelDiscoveryAlgorithm
 from simod.settings.simod_settings import SimodSettings
 from simod.simulation.parameters.BPS_model import BPSModel
 from simod.simulation.prosimos import simulate_and_evaluate
@@ -213,7 +214,10 @@ def run(self, runtimes: Optional[RuntimeMeter] = None):
             )
             # Instantiate event log to discover the process model with
             xes_log_path = self._best_result_dir / f"{self._event_log.process_name}_train_val.xes"
-            self._event_log.train_validation_to_xes(xes_log_path)
+            if best_control_flow_params.mining_algorithm is ProcessModelDiscoveryAlgorithm.SPLIT_MINER_V1:
+                self._event_log.train_validation_to_xes(xes_log_path, only_complete_events=True)
+            else:
+                self._event_log.train_validation_to_xes(xes_log_path)
             # Discover the process model
             discover_process_model(
                 log_path=xes_log_path,

tests/assets/process_model_with_SplitMiner_self_loops.bpmn

@@ -0,0 +1,99 @@
+<?xml version="1.0" encoding="UTF-8"?><definitions xmlns="http://www.omg.org/spec/BPMN/20100524/MODEL" xmlns:dc="http://www.omg.org/spec/DD/20100524/DC" xmlns:bpmndi="http://www.omg.org/spec/BPMN/20100524/DI" xmlns:di="http://www.omg.org/spec/DD/20100524/DI" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" targetNamespace="http://www.omg.org/bpmn20" exporter="ProM. http://www.promtools.org/prom6" exporterVersion="6.3" xsi:schemaLocation="http://www.omg.org/spec/BPMN/20100524/MODEL BPMN20.xsd"><process id="proc_202125197">
+<startEvent id="node_bd0aff33-74bf-4708-b70c-537f73d32d2c" name=""/>
+<endEvent id="node_78c1c673-495d-4b8d-b05d-42638d5de46c" name=""/>
+<task id="node_a0945a12-7e36-4a15-92c6-4a4973eda9b9" name="W_Handle leads">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<task id="node_4f108168-54a6-456b-acf5-f98d4e902c68" name="W_Call after offers">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<task id="node_4dc21123-2cff-482e-b19d-7ef5d11bc12c" name="End"/>
+<task id="node_ffd70b5b-8db5-4b3f-ae73-3a5eede600a5" name="W_Call incomplete files">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<task id="node_db5c7824-e7d7-4008-8f23-dea554d6bc29" name="W_Validate application">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<task id="node_d2d9db90-141c-4dd9-a08d-c310e74bb1d4" name="W_Complete application">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<task id="node_7183cdb2-6d64-4af7-8110-cfbd65d39c70" name="Start"/>
+<task id="node_a6399e55-fa52-41cc-9bc5-d267db419a0c" name="W_Assess potential fraud">
+<standardLoopCharacteristics testBefore="false"/>
+</task>
+<exclusiveGateway id="node_9cb9e9ff-853b-4b1b-aa04-d463cb2e79e4" name="" gatewayDirection="Converging">
+<incoming>
+node_01d8c20e-52d6-4fbf-80d1-f7d458406138</incoming>
+<incoming>
+node_c450dccc-8229-4e76-b957-494c44fec17e</incoming>
+<outgoing>
+node_2b39336a-70f4-4dea-a952-b6f543777b5d</outgoing>
+</exclusiveGateway>
+<exclusiveGateway id="node_0fd985e7-d73d-4689-957a-515b0e17b2f0" name="" gatewayDirection="Converging">
+<incoming>
+node_4cf191ad-01b2-4038-8b26-18a164000b05</incoming>
+<incoming>
+node_24fc93fc-4b66-4f44-9e8e-a649a3fcdc12</incoming>
+<outgoing>
+node_135b1874-4857-4b7a-a81d-50beb311441c</outgoing>
+</exclusiveGateway>
+<exclusiveGateway id="node_2747f1fc-aa96-4232-96a1-45ccb73333b0" name="" gatewayDirection="Diverging">
+<incoming>
+node_cbce22a3-4f84-4bea-bc66-d6acc60d873a</incoming>
+<outgoing>
+node_4cf191ad-01b2-4038-8b26-18a164000b05</outgoing>
+<outgoing>
+node_6c2703c8-ee82-43e5-ba14-1721cf5138c9</outgoing>
+</exclusiveGateway>
+<exclusiveGateway id="node_477528f4-b023-46d5-9ba8-68ea7be789eb" name="" gatewayDirection="Diverging">
+<incoming>
+node_2ea0cee5-c9bf-4d90-bf38-82614ca84894</incoming>
+<outgoing>
+node_a1fdda3a-cf96-4980-a32c-cac63028ef50</outgoing>
+<outgoing>
+node_b96514e9-abb8-47da-9ce9-6ef16bd246db</outgoing>
+<outgoing>
+node_5679761c-c5d3-4e3a-8248-13fd7b48ab95</outgoing>
+</exclusiveGateway>
+<exclusiveGateway id="node_ca100f94-484b-4ec3-8f0c-3bcef06973c8" name="" gatewayDirection="Diverging">
+<incoming>
+node_6659ae33-5879-40ee-9b4c-12113190ead1</incoming>
+<outgoing>
+node_01d8c20e-52d6-4fbf-80d1-f7d458406138</outgoing>
+<outgoing>
+node_83305431-0a61-43bc-ac36-3acd90ad606c</outgoing>
+</exclusiveGateway>
+<exclusiveGateway id="node_7bae894c-932a-476f-b91e-911220376611" name="" gatewayDirection="Converging">
+<incoming>
+node_46301cea-0047-452b-a608-16a7e9f79cfe</incoming>
+<incoming>
+node_80986967-4875-41c2-828f-3e144d755b81</incoming>
+<incoming>
+node_5679761c-c5d3-4e3a-8248-13fd7b48ab95</incoming>
+<outgoing>
+node_c450dccc-8229-4e76-b957-494c44fec17e</outgoing>
+</exclusiveGateway>
+<sequenceFlow id="node_e3c70fd3-097e-49d2-8710-9cb4ec658576" name="" sourceRef="node_d2d9db90-141c-4dd9-a08d-c310e74bb1d4" targetRef="node_4f108168-54a6-456b-acf5-f98d4e902c68"/>
+<sequenceFlow id="node_e50aa326-4026-4f7f-8050-f2e5f4ef4286" name="" sourceRef="node_4dc21123-2cff-482e-b19d-7ef5d11bc12c" targetRef="node_78c1c673-495d-4b8d-b05d-42638d5de46c"/>
+<sequenceFlow id="node_08a48377-5512-4dfd-9e96-fde430046296" name="" sourceRef="node_bd0aff33-74bf-4708-b70c-537f73d32d2c" targetRef="node_7183cdb2-6d64-4af7-8110-cfbd65d39c70"/>
+<sequenceFlow id="node_cbce22a3-4f84-4bea-bc66-d6acc60d873a" name="" sourceRef="node_7183cdb2-6d64-4af7-8110-cfbd65d39c70" targetRef="node_2747f1fc-aa96-4232-96a1-45ccb73333b0"/>
+<sequenceFlow id="node_6c2703c8-ee82-43e5-ba14-1721cf5138c9" name="" sourceRef="node_2747f1fc-aa96-4232-96a1-45ccb73333b0" targetRef="node_a0945a12-7e36-4a15-92c6-4a4973eda9b9"/>
+<sequenceFlow id="node_6659ae33-5879-40ee-9b4c-12113190ead1" name="" sourceRef="node_4f108168-54a6-456b-acf5-f98d4e902c68" targetRef="node_ca100f94-484b-4ec3-8f0c-3bcef06973c8"/>
+<sequenceFlow id="node_83305431-0a61-43bc-ac36-3acd90ad606c" name="" sourceRef="node_ca100f94-484b-4ec3-8f0c-3bcef06973c8" targetRef="node_db5c7824-e7d7-4008-8f23-dea554d6bc29"/>
+<sequenceFlow id="node_2ea0cee5-c9bf-4d90-bf38-82614ca84894" name="" sourceRef="node_db5c7824-e7d7-4008-8f23-dea554d6bc29" targetRef="node_477528f4-b023-46d5-9ba8-68ea7be789eb"/>
+<sequenceFlow id="node_a1fdda3a-cf96-4980-a32c-cac63028ef50" name="" sourceRef="node_477528f4-b023-46d5-9ba8-68ea7be789eb" targetRef="node_a6399e55-fa52-41cc-9bc5-d267db419a0c"/>
+<sequenceFlow id="node_b96514e9-abb8-47da-9ce9-6ef16bd246db" name="" sourceRef="node_477528f4-b023-46d5-9ba8-68ea7be789eb" targetRef="node_ffd70b5b-8db5-4b3f-ae73-3a5eede600a5"/>
+<sequenceFlow id="node_80986967-4875-41c2-828f-3e144d755b81" name="" sourceRef="node_a6399e55-fa52-41cc-9bc5-d267db419a0c" targetRef="node_7bae894c-932a-476f-b91e-911220376611"/>
+<sequenceFlow id="node_46301cea-0047-452b-a608-16a7e9f79cfe" name="" sourceRef="node_ffd70b5b-8db5-4b3f-ae73-3a5eede600a5" targetRef="node_7bae894c-932a-476f-b91e-911220376611"/>
+<sequenceFlow id="node_5679761c-c5d3-4e3a-8248-13fd7b48ab95" name="" sourceRef="node_477528f4-b023-46d5-9ba8-68ea7be789eb" targetRef="node_7bae894c-932a-476f-b91e-911220376611"/>
+<sequenceFlow id="node_135b1874-4857-4b7a-a81d-50beb311441c" name="" sourceRef="node_0fd985e7-d73d-4689-957a-515b0e17b2f0" targetRef="node_d2d9db90-141c-4dd9-a08d-c310e74bb1d4"/>
+<sequenceFlow id="node_24fc93fc-4b66-4f44-9e8e-a649a3fcdc12" name="" sourceRef="node_a0945a12-7e36-4a15-92c6-4a4973eda9b9" targetRef="node_0fd985e7-d73d-4689-957a-515b0e17b2f0"/>
+<sequenceFlow id="node_4cf191ad-01b2-4038-8b26-18a164000b05" name="" sourceRef="node_2747f1fc-aa96-4232-96a1-45ccb73333b0" targetRef="node_0fd985e7-d73d-4689-957a-515b0e17b2f0"/>
+<sequenceFlow id="node_2b39336a-70f4-4dea-a952-b6f543777b5d" name="" sourceRef="node_9cb9e9ff-853b-4b1b-aa04-d463cb2e79e4" targetRef="node_4dc21123-2cff-482e-b19d-7ef5d11bc12c"/>
+<sequenceFlow id="node_c450dccc-8229-4e76-b957-494c44fec17e" name="" sourceRef="node_7bae894c-932a-476f-b91e-911220376611" targetRef="node_9cb9e9ff-853b-4b1b-aa04-d463cb2e79e4"/>
+<sequenceFlow id="node_01d8c20e-52d6-4fbf-80d1-f7d458406138" name="" sourceRef="node_ca100f94-484b-4ec3-8f0c-3bcef06973c8" targetRef="node_9cb9e9ff-853b-4b1b-aa04-d463cb2e79e4"/>
+</process>
+<bpmndi:BPMNDiagram id="id_161575504">
+<bpmndi:BPMNPlane bpmnElement="proc_202125197"/>
+</bpmndi:BPMNDiagram>
+</definitions>

tests/test_control_flow/test_discovery.py

@@ -1,3 +1,4 @@
+import shutil
 import tempfile
 from pathlib import Path
 
@@ -6,7 +7,7 @@
 from pix_framework.discovery.gateway_probabilities import GatewayProbabilitiesDiscoveryMethod
 from pix_framework.io.bpmn import get_activities_names_from_bpmn
 
-from simod.control_flow.discovery import discover_process_model
+from simod.control_flow.discovery import discover_process_model, post_process_bpmn_self_loops
 from simod.control_flow.settings import HyperoptIterationParams
 from simod.settings.common_settings import Metric
 from simod.settings.control_flow_settings import ProcessModelDiscoveryAlgorithm
@@ -103,3 +104,48 @@ def test_discover_process_model_explicit_self_loops(entry_point, test_data):
         # Commented because SM2 doesn't sort the events, thus no parallelism
         # parallel_gateways = root.findall(".//bpmn:parallelGateway", namespaces=ns)
         # assert len(parallel_gateways) == 2, "There should only be two parallel gateways in this model"
+
+
+def test_transform_process_model_explicit_self_loops(entry_point):
+    with tempfile.TemporaryDirectory() as tmp_dir:
+        # Copy source model with self-loops
+        original_model_path = entry_point / "process_model_with_SplitMiner_self_loops.bpmn"
+        model_path = Path(tmp_dir) / "process_model_with_SplitMiner_self_loops.bpmn"
+        shutil.copy(original_model_path, model_path)
+        # Fix process model with self-loops in all activities except Start and End
+        post_process_bpmn_self_loops(model_path)
+        # Assert that no implicit self-loops are there
+        tree = etree.parse(model_path)
+        root = tree.getroot()
+        ns = {"bpmn": root.nsmap.get(None, "http://www.omg.org/spec/BPMN/20100524/MODEL")}
+        tasks = root.findall(".//bpmn:task", namespaces=ns)
+        for task in tasks:
+            assert task.find(
+                "bpmn:standardLoopCharacteristics",
+                namespaces=ns
+            ) is None, f"Task '{task.get('name')}' has an implicit self loop"
+            if task.get("name") == "Start":
+                # Find the incoming flow of the "Start" task
+                task_id = task.get("id")
+                sequence_flows = root.findall(".//bpmn:sequenceFlow", namespaces=ns)
+                incoming_flows = [flow for flow in sequence_flows if flow.get("targetRef") == task_id]
+                assert len(incoming_flows) == 1, f"Task 'Start' should have exactly one incoming flow"
+                # Assert that the source element of the incoming flow is the start event
+                incoming_flow_source = incoming_flows[0].get("sourceRef")
+                start_events = root.findall(".//bpmn:startEvent", namespaces=ns)
+                start_event_ids = {event.get("id") for event in start_events}
+                assert incoming_flow_source in start_event_ids, f"'Start' task was modified."
+            elif task.get("name") == "End":
+                # Find the outgoing flow of the "End" task
+                task_id = task.get("id")
+                sequence_flows = root.findall(".//bpmn:sequenceFlow", namespaces=ns)
+                outgoing_flows = [flow for flow in sequence_flows if flow.get("sourceRef") == task_id]
+                assert len(outgoing_flows) == 1, f"Task 'End' should have exactly one outgoing flow"
+                # Assert that the target element of the outgoing flow is the end event
+                outgoing_flow_target = outgoing_flows[0].get("targetRef")
+                end_events = root.findall(".//bpmn:endEvent", namespaces=ns)
+                end_event_ids = {event.get("id") for event in end_events}
+                assert outgoing_flow_target in end_event_ids, f"'End' task was modified."
+        # Verify number of gateways is original + 2 per self-loop activity
+        exclusive_gateways = root.findall(".//bpmn:exclusiveGateway", namespaces=ns)
+        assert len(exclusive_gateways) == 18, "There should only be 18 exclusive gateways in this model"