(DOCSP-33419) [ANALYTICS] Operational Views (#192)

scripts/package.json

@@ -7,6 +7,8 @@
     "createQualityTestsYaml": "npm run build && node ./build/createQualityTestsYaml.js",
     "scrubMessages": "npm run build && node ./build/scrubMessages.js",
     "analyzeMessages": "npm run build && node ./build/analyzeMessages.js",
+    "createView/messages_by_rating": "npm run build && node ./build/createView/messages_by_rating.js",
+    "createView/top_250_references": "npm run build && node ./build/createView/top_250_references.js",
     "build": "npm run clean && tsc",
     "postbuild": "cp ./src/MessageAnalysis.d.ts ./build/",
     "clean": "rm -rf ./build",

scripts/src/createView/messages_by_rating.ts

@@ -0,0 +1,90 @@
+import { MongoClient, ObjectId } from "mongodb";
+import { assertEnvVars } from "chat-core";
+import { Message } from "chat-server";
+
+import "dotenv/config";
+
+const { MONGODB_DATABASE_NAME, MONGODB_CONNECTION_URI } = assertEnvVars({
+  MONGODB_DATABASE_NAME: "",
+  MONGODB_CONNECTION_URI: "",
+});
+
+type Rating = true | false | null;
+
+export type MessageByRating = {
+  _id: Rating;
+  rating: Rating;
+  count: number;
+  uniqueIps: string[];
+  conversationIds: ObjectId[];
+  messages: (Message & {
+    conversationId: ObjectId;
+    messageId: ObjectId;
+    index: number;
+  })[];
+};
+
+(async function main() {
+  const client = await MongoClient.connect(MONGODB_CONNECTION_URI);
+  try {
+    const db = client.db(MONGODB_DATABASE_NAME);
+    await db.createCollection<MessageByRating>("messages_by_rating", {
+      viewOn: "conversations",
+      pipeline: [
+        {
+          $match: {
+            messages: {
+              $elemMatch: {
+                role: "user",
+              },
+            },
+          },
+        },
+        {
+          $unwind: {
+            path: "$messages",
+            includeArrayIndex: "messageIndex",
+          },
+        },
+        {
+          $match: {
+            "messages.role": "assistant",
+          },
+        },
+        {
+          $group: {
+            _id: "$messages.rating",
+            count: {
+              $sum: 1,
+            },
+            uniqueIps: {
+              $addToSet: "$ipAddress",
+            },
+            conversationIds: {
+              $addToSet: "$_id",
+            },
+            messages: {
+              $push: {
+                $mergeObjects: [
+                  "$messages",
+                  {
+                    conversationId: "$_id",
+                    messageId: "$messages.id",
+                    index: "$messageIndex",
+                  },
+                ],
+              },
+            },
+          },
+        },
+        {
+          $addFields: {
+            rating: "$_id",
+          },
+        },
+      ],
+    });
+  } finally {
+    await client.close();
+  }
+})();

scripts/src/createView/top_250_references.ts

@@ -0,0 +1,101 @@
+import { MongoClient, ObjectId } from "mongodb";
+import { assertEnvVars } from "chat-core";
+import { Message } from "chat-server";
+
+import "dotenv/config";
+
+const { MONGODB_DATABASE_NAME, MONGODB_CONNECTION_URI } = assertEnvVars({
+  MONGODB_DATABASE_NAME: "",
+  MONGODB_CONNECTION_URI: "",
+});
+
+const NUM_REFERENCES = 250;
+
+export type TopReference = {
+  _id: string;
+  count: number;
+  conversationIds: ObjectId[];
+  messages: (Message & {
+    conversationId: ObjectId;
+    messageId: ObjectId;
+    index: number;
+  })[];
+};
+
+(async function main() {
+  const client = await MongoClient.connect(MONGODB_CONNECTION_URI);
+  try {
+    const db = client.db(MONGODB_DATABASE_NAME);
+    await db.createCollection<TopReference>(
+      `top_${NUM_REFERENCES}_references`,
+      {
+        viewOn: "conversations",
+        pipeline: [
+          {
+            $match: {
+              "messages.role": "user",
+            },
+          },
+          {
+            $unwind: {
+              path: "$messages",
+              includeArrayIndex: "messageIndex",
+            },
+          },
+          {
+            $match: {
+              "messages.role": "assistant",
+              "messages.references.0": {
+                $exists: true,
+              },
+            },
+          },
+          {
+            $addFields: {
+              references: "$messages.references",
+            },
+          },
+          {
+            $unwind: {
+              path: "$references",
+              includeArrayIndex: "referenceIndex",
+            },
+          },
+          {
+            $group: {
+              _id: "$references.title",
+              count: {
+                $sum: 1,
+              },
+              conversationIds: {
+                $addToSet: "$_id",
+              },
+              messages: {
+                $push: {
+                  $mergeObjects: [
+                    "$messages",
+                    {
+                      conversationId: "$_id",
+                      messageId: "$messages.id",
+                      index: "$messageIndex",
+                    },
+                  ],
+                },
+              },
+            },
+          },
+          {
+            $sort: {
+              count: -1,
+            },
+          },
+          {
+            $limit: NUM_REFERENCES,
+          },
+        ],
+      }
+    );
+  } finally {
+    await client.close();
+  }
+})();