Add relation model. Add calculate score initial version

agent/nodes/produceRanking.ts

@@ -1,9 +1,50 @@
 import { GraphNode } from "@langchain/langgraph";
 import { MessagesState } from "../state";
-import { AIMessage, HumanMessage } from "@langchain/core/messages";
+import { BaseMessage } from "@langchain/core/messages";
+
+type Priority = keyof typeof mapping;
+
+const mapping = {
+  VERYHIGH: 1.0,
+  HIGH: 0.75,
+  MEDIUM: 0.5,
+  LOW: 0.25,
+  VERYLOW: 0.0
+} as const;
+
+function mapResponse(value: string): number {
+  const upper = value.toUpperCase() as Priority;
+
+  if (upper in mapping) {
+    return mapping[upper];
+  }
+
+  return 0;
+}
+
+function getLastMessageContaining(
+  messages: BaseMessage[],
+  searchString: string
+): string {
+  for (let i = messages.length - 1; i >= 0; i--) {
+    const content = messages[i].content;
+    if (typeof content === "string" && content.includes(searchString)) {
+      return content;
+    }
+  }
+  return "";
+}
 
 export const produceRanking: GraphNode<typeof MessagesState> = async (state) => {
-  //TODO: produce ranking here
+  //TODO: what should these weights be
+  let conf = getLastMessageContaining(state.messages, "CONFIDENCE")?.split(":")[1] //TODO: we can better error handle here
+  let ragas = getLastMessageContaining(state.messages, "RAGAS")?.split(":")[1] //TODO: we can genericify this too surely
+  let rel = getLastMessageContaining(state.messages, "RELATION")?.split(":")[1]
+
+  let result = mapResponse(conf) * Number.parseFloat(ragas) * mapResponse(rel)
   
-  return { messages: [ new AIMessage(state.messages?.length.toString() ?? "0")] };
+  let current = state.proposedTriggerEvent;
+  current[state.proposedTriggerEventIndex].score = result;
+
+  return  { proposedTriggerEvent: current };
 };