(EAI-1149, EAI-1259) Updates to user message front matter (#892)

* Normalize URL in user msg formatting

* Add labels to origin rules & use those for formatting user msg frontmatter

Author: hschawe

packages/chatbot-server-mongodb-public/src/processors/formatUserMessageForGeneration.test.ts

@@ -1,5 +1,6 @@
 import { formatUserMessageForGeneration } from "./formatUserMessageForGeneration";
 import { ConversationCustomData, logger } from "mongodb-rag-core";
+import { ORIGIN_RULES } from "mongodb-chatbot-server";
 
 beforeAll(() => {
   logger.error = jest.fn();
@@ -8,16 +9,19 @@ beforeAll(() => {
 describe("formatUserMessageForGeneration", () => {
   const userMessageText = "Hello, world!";
   const reqId = "test-request-id";
+  const testMongoDbPageUrl = "https://mongodb.com";
+  const resultMongoDbPageUrl = "mongodb.com";
 
   it("formats front matter correctly for mongodb.com origin", () => {
-    const origin = "https://mongodb.com";
     const result = formatUserMessageForGeneration({
       userMessageText,
       reqId,
-      customData: { origin } satisfies ConversationCustomData,
+      customData: {
+        origin: testMongoDbPageUrl,
+      } satisfies ConversationCustomData,
     });
     expect(result).toEqual(`---
-pageUrl: ${origin}
+pageUrl: ${resultMongoDbPageUrl}
 ---
 
 ${userMessageText}`);
@@ -31,7 +35,40 @@ ${userMessageText}`);
       customData: { origin } satisfies ConversationCustomData,
     });
     expect(result).toEqual(`---
-pageUrl: ${origin}
+pageUrl: learn.mongodb.com
+---
+
+${userMessageText}`);
+  });
+
+  it("normalizes a URL with trailing backslash", () => {
+    const origin = testMongoDbPageUrl + "/docs/pageName/";
+    const result = formatUserMessageForGeneration({
+      userMessageText,
+      reqId,
+      customData: {
+        origin,
+      } satisfies ConversationCustomData,
+    });
+    expect(result).toEqual(`---
+pageUrl: ${resultMongoDbPageUrl + "/docs/pageName"}
+---
+
+${userMessageText}`);
+  });
+
+  it("normalizes a URL with query", () => {
+    const origin =
+      "https://learn.mongodb.com/courses/mongodb-for-sql-experts?param1=value1&param2=value2";
+    const result = formatUserMessageForGeneration({
+      userMessageText,
+      reqId,
+      customData: {
+        origin,
+      } satisfies ConversationCustomData,
+    });
+    expect(result).toEqual(`---
+pageUrl: learn.mongodb.com/courses/mongodb-for-sql-experts
 ---
 
 ${userMessageText}`);
@@ -67,7 +104,7 @@ ${userMessageText}`);
       } satisfies ConversationCustomData,
     });
     expect(result).toEqual(`---
-client: MongoDB VS Code plugin
+client: MongoDB VS Code extension
 ---
 
 ${userMessageText}`);
@@ -92,6 +129,25 @@ client: ${expectedClientLabel}
 ${userMessageText}`);
   });
 
+  it("does not add client front matter for unlabelled mongodb originCodes", () => {
+    const unlabelledOriginCodes: string[] = [];
+    ORIGIN_RULES.reduce((acc, rule) => {
+      if (!rule.label) unlabelledOriginCodes.push(rule.code);
+      return acc;
+    }, unlabelledOriginCodes);
+
+    unlabelledOriginCodes.forEach((originCode) => {
+      const result = formatUserMessageForGeneration({
+        userMessageText,
+        reqId,
+        customData: {
+          originCode,
+        } satisfies ConversationCustomData,
+      });
+      expect(result).toEqual(userMessageText);
+    });
+  });
+
   it("logs a warning and does not add pageUrl if origin is malformed", () => {
     const malformedOrigin = "http://not a url";
     const result = formatUserMessageForGeneration({
@@ -119,19 +175,18 @@ ${userMessageText}`);
   });
 
   it("adds both pageUrl and client front matter if both are present", () => {
-    const origin = "https://mongodb.com";
     const originCode = "VSCODE";
-    const expectedClientLabel = "MongoDB VS Code plugin";
+    const expectedClientLabel = "MongoDB VS Code extension";
     const result = formatUserMessageForGeneration({
       userMessageText,
       reqId,
       customData: {
-        origin,
+        origin: testMongoDbPageUrl,
         originCode,
       } satisfies ConversationCustomData,
     });
     expect(result).toEqual(`---
-pageUrl: ${origin}
+pageUrl: ${resultMongoDbPageUrl}
 client: ${expectedClientLabel}
 ---
 

packages/chatbot-server-mongodb-public/src/processors/formatUserMessageForGeneration.ts

@@ -1,5 +1,6 @@
 import { updateFrontMatter, ConversationCustomData } from "mongodb-rag-core";
-import { originCodes } from "mongodb-chatbot-server";
+import { originCodes, ORIGIN_RULES } from "mongodb-chatbot-server";
+import { normalizeUrl } from "mongodb-rag-core/dataSources";
 import { z } from "zod";
 import { logRequest } from "../utils";
 
@@ -19,6 +20,19 @@ type FormatUserMessageForGenerationParams = {
   customData: ConversationCustomData;
 };
 
+interface OriginCodeLabels {
+  /** Code to label mapping. */
+  [code: string]: string;
+}
+
+// Some origin codes have a label to add to the front matter
+const originCodeLabels = ORIGIN_RULES.reduce((acc, rule) => {
+  if (rule.label !== undefined) {
+    acc[rule.code] = rule.label;
+  }
+  return acc;
+}, {} as OriginCodeLabels);
+
 export function formatUserMessageForGeneration({
   userMessageText,
   reqId,
@@ -52,7 +66,7 @@ export function formatUserMessageForGeneration({
         url.hostname === "mongodb.com" ||
         url.hostname.endsWith(".mongodb.com")
       ) {
-        frontMatter.pageUrl = parsedCustomData.origin;
+        frontMatter.pageUrl = normalizeUrl({ url: parsedCustomData.origin });
       }
     } catch (e) {
       logRequest({
@@ -62,10 +76,10 @@ export function formatUserMessageForGeneration({
       });
     }
   }
-  if (parsedCustomData.originCode === "VSCODE") {
-    frontMatter.client = "MongoDB VS Code plugin";
-  } else if (parsedCustomData.originCode === "GEMINI_CODE_ASSIST") {
-    frontMatter.client = "Gemini Code Assist";
+
+  const originLabel = originCodeLabels[parsedCustomData.originCode ?? ""];
+  if (originLabel) {
+    frontMatter.client = originLabel;
   }
 
   if (Object.keys(frontMatter).length === 0) {

packages/mongodb-chatbot-server/src/processors/addCustomData.ts

@@ -41,15 +41,28 @@ export type OriginCode = (typeof originCodes)[number];
 interface OriginRule {
   regex: RegExp;
   code: OriginCode;
+  /**
+    Name used to label the origin when formatting the user message front matter. 
+    Leave empty if no label should be used.
+    */
+  label?: string;
 }
 
-const ORIGIN_RULES: OriginRule[] = [
+export const ORIGIN_RULES: OriginRule[] = [
   { regex: /learn\.mongodb\.com/, code: "LEARN" },
   { regex: /mongodb\.com\/developer/, code: "DEVELOPER" },
   { regex: /mongodb\.com\/docs/, code: "DOCS" },
   { regex: /mongodb\.com\//, code: "DOTCOM" },
-  { regex: /google-gemini-code-assist/, code: "GEMINI_CODE_ASSIST" },
-  { regex: /vscode-mongodb-copilot/, code: "VSCODE" },
+  {
+    regex: /google-gemini-code-assist/,
+    code: "GEMINI_CODE_ASSIST",
+    label: "Gemini Code Assist",
+  },
+  {
+    regex: /vscode-mongodb-copilot/,
+    code: "VSCODE",
+    label: "MongoDB VS Code extension",
+  },
 ];
 
 function getOriginCode(origin: string): OriginCode {