modulerization

dist/index.js

@@ -3,21 +3,19 @@ dotenv.config();
 import express from "express";
 import multer from "multer";
 import cors from "cors";
-import OpenAI from "openai";
 import { v4 as uuidv4 } from "uuid";
 import { uploadToS3 } from "./uploads3.js";
 import { PORT, PROMPT } from "./constant.js";
 import { client } from "./MongoDbsetup.js";
+import { openAiVisionApi } from "./openAIVisionAPi.js";
+import { openAiEmbeddings } from "./openAiEmbeddingsApi.js";
+import { vectorQuerySchema } from "./vectorQuerySchema.js";
 const corsOptions = {
  origin: "http://localhost:3000",
 };
 const app = express();
 const upload = multer(); // Using multer's default memory storage
 app.use(cors(corsOptions)); // This will enable all CORS requests. For production, configure this properly.
-const openai = new OpenAI({
- organization: process.env.OPENAI_ORGANISATION,
- apiKey: process.env.OPENAI_API_KEY,
-});
 // Create a MongoClient with a MongoClientOptions object to set the Stable API version
 app.post("/upload", upload.array("images"), async (req, res) => {
  if (!req.files || (Array.isArray(req.files) && req.files.length === 0)) {
@@ -28,77 +26,36 @@ app.post("/upload", upload.array("images"), async (req, res) => {
  try {
  const results = await Promise.all(uploadPromises);
  const allPromise = results.map(async (url) => {
- const response = await openai.chat.completions.create({
- model: "gpt-4-vision-preview",
- temperature: 0,
- messages: [
- {
- role: "system",
- content: "the system only speaks in JSON. Do not generate output that isn’t in properly formatted JSON.",
- },
- {
- role: "user",
- content: [
- {
- type: "text",
- text: PROMPT,
- },
- {
- type: "image_url",
- image_url: {
- url: url,
- },
- },
- ],
- },
- ],
- });
+ const response = await openAiVisionApi(url, PROMPT);
  const stringArray = response.choices[0].message.content.split("json");
  const jsonString = stringArray.join("");
  const trimmedString = jsonString.replace(/`|\n/g, "");
  const jsonData = JSON.parse(trimmedString);
  return { jsonData, url, id: uuidv4() };
  });
- const data = await Promise.all(allPromise)
- .then((results) => {
- // Process API call results
- return results;
- })
- .catch((error) => {
- console.error("Error:", error);
- });
+ const data = await Promise.all(allPromise);
  if (data) {
- try {
- await client.connect();
- console.log("Connected successfully to MongoDB");
- const database = client.db("Cluster0"); // Replace with your database name
- const collection = database.collection("image_metadata");
- const allPromises = data.map(async (item) => {
- const data = await openai.embeddings.create({
- model: "text-embedding-3-small",
- input: item.jsonData.description,
- encoding_format: "float",
- });
- const document = {
- subjects: item.jsonData.subjects,
- attributes: item.jsonData.attributes,
- themes: item.jsonData.themes,
- contexts: item.jsonData.contexts,
- description: item.jsonData.description,
- image_url: item.url,
- id: item.id,
- embeddings: data.data[0].embedding,
- };
- console.log(document.description);
- return collection.insertOne(document);
- });
- const insertedRecords = await Promise.all(allPromises);
- console.log("records inserted");
- res.status(200).send("data inserted");
- }
- catch (error) {
- console.log("Error", error);
- }
+ await client.connect();
+ console.log("Connected successfully to MongoDB");
+ const database = client.db("Cluster0"); // Replace with your database name
+ const collection = database.collection("image_metadata");
+ const allPromises = data.map(async (item) => {
+ const data = await openAiEmbeddings(item.jsonData.description);
+ const document = {
+ subjects: item.jsonData.subjects,
+ attributes: item.jsonData.attributes,
+ themes: item.jsonData.themes,
+ contexts: item.jsonData.contexts,
+ description: item.jsonData.description,
+ image_url: item.url,
+ id: item.id,
+ embeddings: data.data[0].embedding,
+ };
+ return collection.insertOne(document);
+ });
+ const insertedRecords = await Promise.all(allPromises);
+ console.log("records inserted");
+ res.status(200).send("data inserted");
  }
  }
  catch (error) {
@@ -113,34 +70,11 @@ app.get("/search", async (req, res) => {
  const database = client.db("Cluster0"); // Replace with your database name
  const collection = database.collection("image_metadata");
  const queryString = req.query.query;
- const queryVector = await openai.embeddings.create({
- model: "text-embedding-3-small",
- input: queryString,
- encoding_format: "float",
- });
+ const queryVector = await openAiEmbeddings(queryString);
  const query = queryVector.data[0].embedding;
- const agg = [
- {
- $vectorSearch: {
- index: "PlotSemanticSearch",
- path: "embeddings",
- queryVector: query,
- numCandidates: 100,
- limit: 100,
- },
- },
- {
- $project: {
- _id: 0,
- image_url: 1,
- description: 1,
- score: {
- $meta: "vectorSearchScore",
- },
- },
- },
- ];
- const result = await collection.aggregate(agg).toArray();
+ const result = await collection
+ .aggregate(vectorQuerySchema(query))
+ .toArray();
  const scoreThreshold = 65;
  res.status(200).send({
  data: result

dist/openAIVisionAPi.d.ts

@@ -0,0 +1 @@
+export declare const openAiVisionApi: (url: string, prompt: string) => Promise<import("openai/resources").ChatCompletion>;

dist/openAIVisionAPi.js

@@ -0,0 +1,28 @@
+import openai from "./openAiSetup";
+export const openAiVisionApi = async (url, prompt) => {
+ return await openai.chat.completions.create({
+ model: "gpt-4-vision-preview",
+ temperature: 0,
+ messages: [
+ {
+ role: "system",
+ content: "the system only speaks in JSON. Do not generate output that isn’t in properly formatted JSON.",
+ },
+ {
+ role: "user",
+ content: [
+ {
+ type: "text",
+ text: prompt,
+ },
+ {
+ type: "image_url",
+ image_url: {
+ url: url,
+ },
+ },
+ ],
+ },
+ ],
+ });
+};

dist/openAiEmbeddingsApi.d.ts

@@ -0,0 +1 @@
+export declare const openAiEmbeddings: (text: string) => Promise<import("openai/resources").CreateEmbeddingResponse>;

dist/openAiEmbeddingsApi.js

@@ -0,0 +1,8 @@
+import openai from "./openAiSetup";
+export const openAiEmbeddings = async (text) => {
+ return await openai.embeddings.create({
+ model: "text-embedding-3-small",
+ input: text,
+ encoding_format: "float",
+ });
+};

dist/openAiSetup.d.ts

@@ -0,0 +1,3 @@
+import OpenAI from "openai";
+declare const openai: OpenAI;
+export default openai;

dist/openAiSetup.js

@@ -0,0 +1,6 @@
+import OpenAI from "openai";
+const openai = new OpenAI({
+ organization: process.env.OPENAI_ORGANISATION,
+ apiKey: process.env.OPENAI_API_KEY,
+});
+export default openai;

dist/vectorQuerySchema.d.ts

@@ -0,0 +1,20 @@
+export declare const vectorQuerySchema: (query: number[]) => ({
+ $vectorSearch: {
+ index: string;
+ path: string;
+ queryVector: number[];
+ numCandidates: number;
+ limit: number;
+ };
+ $project?: undefined;
+} | {
+ $project: {
+ _id: number;
+ image_url: number;
+ description: number;
+ score: {
+ $meta: string;
+ };
+ };
+ $vectorSearch?: undefined;
+})[];

dist/vectorQuerySchema.js

@@ -0,0 +1,24 @@
+export const vectorQuerySchema = (query) => {
+ const agg = [
+ {
+ $vectorSearch: {
+ index: "PlotSemanticSearch",
+ path: "embeddings",
+ queryVector: query,
+ numCandidates: 100,
+ limit: 100,
+ },
+ },
+ {
+ $project: {
+ _id: 0,
+ image_url: 1,
+ description: 1,
+ score: {
+ $meta: "vectorSearchScore",
+ },
+ },
+ },
+ ];
+ return agg;
+};