Improve Proactive code generation

Modules/IntentSpecification2WorkflowGenerator/api/api_main.py

@@ -134,6 +134,9 @@ def annotate_dataset_from_frontend():
 def download_proactive():
     graph = Graph().parse(data=request.json.get("graph", ""), format='turtle')
     ontology = Graph().parse(data=request.json.get('ontology', ''), format='turtle')
+    layout = request.json.get('layout', '')
+    label_column = request.json.get('label_column', '')
+    data_product_name = request.json.get('data_product_name', '')
 
     # Connect to Proactive
     gateway = proactive.ProActiveGateway(base_url="https://try.activeeon.com:8443", debug=False, javaopts=[], log4j_props_file=None,
@@ -185,21 +188,26 @@ def download_proactive():
         bucket = gateway.getBucket("ai-machine-learning")
 
         ################## Change file path (name) and label name (send both as parameters)
-        load_dataset_task = bucket.create_Import_Data_task(import_from="PA:USER_FILE", file_path="countries.csv", file_delimiter=",", label_column="IncomeGroup")
+        load_dataset_task = bucket.create_Import_Data_task(import_from="PA:USER_FILE", file_path=data_product_name + ".csv", file_delimiter=";", label_column=label_column)
         proactive_job.addTask(load_dataset_task)
 
         split_data_task = bucket.create_Split_Data_task()
         split_data_task.addDependency(load_dataset_task)
         proactive_job.addTask(split_data_task)
 
+        # Model depends on the layout, the rest is the same
+        scale_task = bucket.create_Scale_Data_task()
+        model_task = bucket.create_Support_Vector_Machines_task()
+        for key in layout:
+            if "decision_tree_predictor" in key:
+                model_task = bucket.create_Random_Forest_task()
+                break
         random_forest_task = bucket.create_Random_Forest_task()
         proactive_job.addTask(random_forest_task)
 
-        bucket.create_Support_Vector_Machines_task()
-
         train_model_task = bucket.create_Train_Model_task()
         train_model_task.addDependency(split_data_task)
-        train_model_task.addDependency(random_forest_task)
+        train_model_task.addDependency(model_task)
         proactive_job.addTask(train_model_task)
 
         download_model_task = bucket.create_Download_Model_task()

api/src/main/java/edu/upc/essi/dtim/odin/intents/IntentService.java

@@ -74,7 +74,20 @@ public String postIntent(String intentName, String problem, String projectID, St
         newIntent = saveIntent(newIntent);
 
         Project project = projectService.getProject(projectID);
-        project.addIntent(newIntent);
+        List<Intent> intentsOfProject = project.getIntents();
+
+        // Before adding the new intent, we remove all those that have no workflows. This is because we first create the
+        // intent and then the workflows, but an intent with no workflows (which can happen if user cancels the process)
+        // is of no use.
+        for (Intent intentInProject : project.getIntents()) {
+            if (intentInProject.getWorkflows().isEmpty()) {
+                intentsOfProject.remove(intentInProject);
+                break;
+            }
+        }
+
+        intentsOfProject.add(newIntent);
+        project.setIntents(intentsOfProject);
         projectService.saveProject(project);
 
         return newIntent.getIntentID();

frontend/src/components/intents/AbstractPlanner.vue

@@ -58,6 +58,7 @@ const handleSubmit = async() => {
   data.append("dataProductID", selectedDataProduct.id)
   
   await intentsStore.postIntent(projectID, data)
+  await intentsStore.getAllIntents(projectID, data) // Refresh the list of intents
 
   $q.loading.show({message: 'Materializing data product...'}) // Then, create the csv file from the dataProduct
   await dataProductsStore.materializeDataProduct(projectID, selectedDataProduct.id)

frontend/src/stores/intentsStore.js

@@ -17,6 +17,7 @@ export const useIntentsStore = defineStore('intents', {
     intent_graph: {}, // Graph definition of the current intent
     ontology: "", // Ontology of the system (graph)
     algorithmImplementations: [], // List of algorithms defined in the ontology
+    labelColumn: "", // Column over which a classification model will operate. ONLY FOR INTEGRATION WITH PROACTIVE. SHOULD BE REMOVED WHEN INTEGRATION DEPENDS ON THE GRAPH
 
     abstractPlans: [], // List of abstract plans (displayed in Logical Planner)
     logicalPlans: [], // List of logical plans (displayed in Workflow Planner)
@@ -82,6 +83,7 @@ export const useIntentsStore = defineStore('intents', {
     },
 
     async annotateDataset(data) {
+      this.labelColumn = data["label"] // SHOULD BE REMOVED EVENTUALLY
       try {
         const response = await intentsAPI.annotateDataset(data);
         notify.positive(`Dataset annotated`)
@@ -186,7 +188,10 @@ export const useIntentsStore = defineStore('intents', {
     },
 
     async downloadProactive(plan) {
-      const data = {"graph": plan.graph, "ontology": this.ontology, "layout": this.plan}
+      const currentIntent = this.intents.find(intent => intent.intentID === String(this.intentID)) // SHOULD BE REMOVED EVENTUALLY
+      const dataProductName = currentIntent.dataProduct.datasetName // SHOULD BE REMOVED EVENTUALLY
+      // At some point, the translation to the Proactive ontology should be done, and the API should only require the graph to make it
+      const data = {"graph": plan.graph, "ontology": this.ontology, "layout": plan.plan, "label_column": this.labelColumn, "data_product_name": dataProductName}
       try {
         const response = await intentsAPI.downloadProactive(data);
         FileSaver.saveAs(new Blob([response.data]), `${plan.id}.xml`);