(EAI-950) Segment - Send data feed to Segment

packages/chatbot-server-mongodb-public/.env.example

@@ -1,20 +1,21 @@
 PORT=3000
-OPENAI_ENDPOINT=https://<resource_name>.openai.azure.com/
-OPENAI_API_KEY=<api_key>
+OPENAI_ENDPOINT="https://<resource_name>.openai.azure.com/"
+OPENAI_API_KEY="<api_key>"
 OPENAI_VERIFIED_ANSWER_EMBEDDING_DEPLOYMENT="docs-chatbot-embedding-ada-002"
 OPENAI_RETRIEVAL_EMBEDDING_DEPLOYMENT="text-embedding-3-small"
-OPENAI_EMBEDDING_MODEL=text-embedding-ada-002
-OPENAI_EMBEDDING_MODEL_VERSION=2023-03-15-preview
-OPENAI_CHAT_COMPLETION_DEPLOYMENT=<deployment_name>
-OPENAI_CHAT_COMPLETION_MODEL_VERSION=2023-05-15
-MONGODB_CONNECTION_URI=<connection_uri>
-MONGODB_DATABASE_NAME=<docs-chatbot-ENV>
-VECTOR_SEARCH_INDEX_NAME=<index_name>
-NODE_ENV=development
-ALLOWED_ORIGINS=http://example.com,http://localhost:5173
-OPENAI_PREPROCESSOR_CHAT_COMPLETION_DEPLOYMENT=<deployment name>
+OPENAI_EMBEDDING_MODEL="text-embedding-ada-002"
+OPENAI_EMBEDDING_MODEL_VERSION="2023-03-15-preview"
+OPENAI_CHAT_COMPLETION_DEPLOYMENT="<deployment_name>"
+OPENAI_CHAT_COMPLETION_MODEL_VERSION="2023-05-15"
+MONGODB_CONNECTION_URI="<connection_uri>"
+MONGODB_DATABASE_NAME="<docs-chatbot-ENV>"
+VECTOR_SEARCH_INDEX_NAME="<index_name>"
+NODE_ENV="development"
+ALLOWED_ORIGINS="http://example.com,http://localhost:5173"
+OPENAI_PREPROCESSOR_CHAT_COMPLETION_DEPLOYMENT="<deployment name>"
 OPENAI_API_VERSION="2024-06-01"
 JUDGE_EMBEDDING_MODEL="text-embedding-3-small"
 JUDGE_LLM="gpt-4o-mini"
 BRAINTRUST_TRACING_API_KEY="<some api key>"
 BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME="chatbot-responses-dev"
+SEGMENT_WRITE_KEY="<your segment write key>"

packages/chatbot-server-mongodb-public/package.json

@@ -27,6 +27,7 @@
     "generate-eval-cases": "ts-node src/eval/bin/generateEvalCasesYamlFromCSV.ts"
   },
   "dependencies": {
+    "@segment/analytics-node": "^2.2.1",
     "@slack/web-api": "^7.8.0",
     "common-tags": "^1.8.2",
     "cookie-parser": "^1.4.6",
@@ -49,6 +50,7 @@
     "@types/common-tags": "^1.8.1",
     "@types/cookie-parser": "^1.4.6",
     "@types/express": "^4.17.21",
+    "@types/express-serve-static-core": "^5.0.6",
     "@types/jest": "^29.5.2",
     "@typescript-eslint/eslint-plugin": "^5.58.0",
     "@typescript-eslint/parser": "^5.58.0",
@@ -58,6 +60,7 @@
     "eslint-config-prettier": "^8.8.0",
     "eslint-plugin-jest": "^27.2.1",
     "jest": "^29.6.1",
+    "node-mocks-http": "^1.16.2",
     "nodemon": "^3.0.1",
     "prettier": "^2.8.7",
     "readline": "^1.3.0",

packages/chatbot-server-mongodb-public/src/config.ts

@@ -19,6 +19,7 @@ import {
   makeVerifiedAnswerGenerateUserPrompt,
   makeDefaultFindVerifiedAnswer,
   defaultCreateConversationCustomData,
+  defaultAddMessageToConversationCustomData,
 } from "mongodb-chatbot-server";
 import cookieParser from "cookie-parser";
 import { makeStepBackRagGenerateUserPrompt } from "./processors/makeStepBackRagGenerateUserPrompt";
@@ -37,6 +38,7 @@ import {
   makeCommentMessageUpdateTrace,
   makeRateMessageUpdateTrace,
 } from "./tracing/routesUpdateTraceHandlers";
+import { useSegmentIds } from "./middleware/useSegmentIds";
 export const {
   MONGODB_CONNECTION_URI,
   MONGODB_DATABASE_NAME,
@@ -62,6 +64,7 @@ const {
   BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME,
   SLACK_BOT_TOKEN,
   SLACK_COMMENT_CONVERSATION_ID,
+  SEGMENT_WRITE_KEY,
 } = process.env;
 
 const allowedOrigins = process.env.ALLOWED_ORIGINS?.split(",") || [];
@@ -225,42 +228,70 @@ const llmAsAJudgeConfig = {
   },
 };
 
+const segmentConfig = SEGMENT_WRITE_KEY
+  ? {
+      writeKey: SEGMENT_WRITE_KEY,
+    }
+  : undefined;
+
 export const config: AppConfig = {
   conversationsRouterConfig: {
     llm,
     middleware: [
       blockGetRequests,
       requireValidIpAddress(),
       requireRequestOrigin(),
+      useSegmentIds(),
       cookieParser(),
     ],
     createConversationCustomData: !isProduction
       ? createConversationCustomDataWithAuthUser
       : undefined,
+    addMessageToConversationCustomData: async (req, res) => {
+      const defaultCustomData = await defaultAddMessageToConversationCustomData(
+        req,
+        res
+      );
+      return {
+        ...defaultCustomData,
+        segmentUserId: res.locals.customData.segmentUserId ?? undefined,
+        segmentAnonymousId:
+          res.locals.customData.segmentAnonymousId ?? undefined,
+      };
+    },
     addMessageToConversationUpdateTrace:
-      makeAddMessageToConversationUpdateTrace(
-        retrievalConfig.findNearestNeighborsOptions.k,
-        { ...llmAsAJudgeConfig, percentToJudge: isProduction ? 0.1 : 1 }
-      ),
-    rateMessageUpdateTrace: makeRateMessageUpdateTrace(llmAsAJudgeConfig),
-    commentMessageUpdateTrace: makeCommentMessageUpdateTrace(
+      makeAddMessageToConversationUpdateTrace({
+        k: retrievalConfig.findNearestNeighborsOptions.k,
+        llmAsAJudge: {
+          ...llmAsAJudgeConfig,
+          percentToJudge: isProduction ? 0.1 : 1,
+        },
+        segment: segmentConfig,
+      }),
+    rateMessageUpdateTrace: makeRateMessageUpdateTrace({
+      llmAsAJudge: llmAsAJudgeConfig,
+      segment: segmentConfig,
+    }),
+    commentMessageUpdateTrace: makeCommentMessageUpdateTrace({
       openAiClient,
-      JUDGE_LLM,
-      SLACK_BOT_TOKEN !== undefined &&
+      judgeLlm: JUDGE_LLM,
+      slack:
+        SLACK_BOT_TOKEN !== undefined &&
         SLACK_COMMENT_CONVERSATION_ID !== undefined
-        ? {
-            token: SLACK_BOT_TOKEN,
-            conversationId: SLACK_COMMENT_CONVERSATION_ID,
-            llmAsAJudge: llmAsAJudgeConfig,
-            braintrust: BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME
-              ? {
-                  orgName: "mongodb-education-ai",
-                  projectName: BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME,
-                }
-              : undefined,
-          }
-        : undefined
-    ),
+          ? {
+              token: SLACK_BOT_TOKEN,
+              conversationId: SLACK_COMMENT_CONVERSATION_ID,
+              llmAsAJudge: llmAsAJudgeConfig,
+              braintrust: BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME
+                ? {
+                    orgName: "mongodb-education-ai",
+                    projectName: BRAINTRUST_CHATBOT_TRACING_PROJECT_NAME,
+                  }
+                : undefined,
+            }
+          : undefined,
+      segment: segmentConfig,
+    }),
     generateUserPrompt,
     systemPrompt,
     maxUserMessagesInConversation: 50,