Source-Web-23070-DataAnalysis/v2/scrapper/embeddings.js

import { readFileSync, writeFileSync, existsSync, mkdirSync } from 'fs';
import { v4 as uuidv4 } from 'uuid';

const COLLECTION_DATA = 'data-v2';


async function clearCollection(collection){
    // clear Qdrant collection
    const res_delete=await fetch(`http://localhost:6333/collections/${collection}`, {
        method: 'DELETE',
    });
    const data_delete = await res_delete.json();
    console.log(data_delete);

    const res=await fetch(`http://localhost:6333/collections/${collection}`, {
        method: 'PUT',
        body: JSON.stringify({
            vectors: {
                size: 1536,
                distance: "Cosine"
            },
            payload: {
                text: "string"
            }
        }),
    });
    const data = await res.json();
    console.log(data);

}

// function generateUUID() { // Public Domain/MIT
//     var d = new Date().getTime();//Timestamp
//     var d2 = (performance && performance.now && (performance.now()*1000)) || 0;
//     return 'xxxxxxxx-xxxx-4xxx-yxxx-xxxxxxxxxxxx'.replace(/[xy]/g, function(c) {
//         var r = Math.random() * 16;
//         return r | 0;
//     });
// }

export async function processData(filepath, clear=false){
    
    if(clear) await clearCollection(COLLECTION_DATA);

    console.log("Processing file for embeddings:", filepath);

    // fetch data from url
    const response = readFileSync(filepath);
    const json=await JSON.parse(response);

    json.output.forEach(async (item, index)=>{

        const text=jsonToText(item);
        // console.log(text);
        const embeddings=await textToEmbeddings(text);
        

        const uuid=uuidv4();
        
        await writeToQdrant(uuid, embeddings, {
            summry: item.summry,
            keywords: item.keywords,
            number: item.number,
            total: item.total,
            metadata: JSON.stringify(item.metadata),
        },COLLECTION_DATA);

    });
}


async function textToEmbeddings(text){
   // call embedding API
   const response=await fetch('https://api.openai.com/v1/embeddings', {
       method: 'POST',
       headers: {
           'Content-Type': 'application/json',
           'Authorization': `Bearer ${process.env.OPENAI_API_KEY}`
       },
       body: JSON.stringify({
           model: "text-embedding-3-small",
           input: text
       })
   });
   const result = await response.json();
//    console.log(result);

   return result.data[0]?.embedding;
}

function jsonToText(item){
   let text = "";
   
    text += `Summary: ${item.summry} `;
    text += `Keywords: ${item.keywords.join(", ")} `;
    text += `Order: ${item.number}/${item.total} `;
    text += `User: ${item.user} `;
    text += `Content: ${item.content.replace(/[\r\n]+/g, ' ')} `;
   
   return text;
}



async function writeToQdrant(id,embeddings, payload, collection){
    // write embeddings to Qdrant
    const res=await fetch(`http://localhost:6333/collections/${collection}/points?wait=true`, {
        method: 'PUT',
        headers: {
            'Content-Type': 'application/json',
        },
        body: JSON.stringify({
            points: [{
                id: id,
                vector: embeddings,
                payload: payload
            }]
        })
    });

    const data = await res.json();
    console.log(data);

    return data;
}

export async function searchByText(query){
    const queryEmbeddings=await textToEmbeddings(query);
    return await searchQdrant(queryEmbeddings);
}

export async function searchByTheme(themeId){
    // get theme embeddings from Qdrant
    const res=await fetch(`http://localhost:6333/collections/${COLLECTION_THEME}/points/${themeId}`, {
        method: 'GET',
        headers: {
            'Content-Type': 'application/json',
        }
    });

    if (!res.ok) {
        const errorData = await res.json();
        console.error("Search failed:", errorData);
        return null;
    }

    const data = await res.json();
    console.log(data);

    return await searchQdrant(data.result.vector);
}

async function searchQdrant(query_embeddings){
    // search Qdrant
    const res=await fetch(`http://localhost:6333/collections/${COLLECTION_DATA}/points/search`, {
        method: 'POST',
        headers: {
            'Content-Type': 'application/json',
        },
        body: JSON.stringify({
            vector: query_embeddings,
            limit: 50,
            with_payload: true,
            with_vector: true
        })
    });

    if (!res.ok) {
        const errorData = await res.json();
        console.error("Search failed:", errorData);
        return null;
    }

    const data = await res.json();
    console.log("Search Result:", data);

    return data.result;
}