How to Build a RAG-Powered (Retrieval-Augmented Generation) Android App with Gemini Embedding and Flash Models?

RAG-Powered Android App

Steps:
1 Prepare Document Data:
Extract text from a PDF using Python and PyPDF2, chunk it, generate embeddings with Gemini’s models/embedding-001, and export to rag_data.json.

2 Set Up Android Project:
Create an Android app with Gson and OkHttp dependencies, load rag_data.json from assets into chunks and embeddings, and add internet permission.

3 Implement RAG Logic:
Fetch query embeddings from embedding-001, compute cosine similarity to retrieve top-k chunks, and generate a response using gemini-1.5-flash.

4 Integrate and Test:
Run RAG in a background thread, display the response in a TextView, and test with logging and error handling.

Sample Prompts:

  1. What is the main topic of the document?
  2. What is History of AI?

I hope you like this video. For any questions, suggestions or appreciation please contact us at: https://programmerworld.co/contact/ or email at: programmerworld1990@gmail.com

Details:

Python code: Preparing document data

import google.generativeai as genai
import PyPDF2
import os
import numpy as np
from typing import List, Dict
import time
import json

# Configure your Google API key
API_KEY = "AIzaSyAGPOmHVpP4zLw-B9nNbvPYszkYOWb5uJ8"
genai.configure(api_key=API_KEY)

def extract_text_from_pdf(pdf_path: str) -> str:
    with open(pdf_path, 'rb') as file:
        reader = PyPDF2.PdfReader(file)
        text = ""
        for page in reader.pages:
            text += page.extract_text() + "\n"
    return text

def chunk_text(text: str, chunk_size: int = 500) -> List[str]:
    words = text.split()
    chunks = [' '.join(words[i:i + chunk_size]) for i in range(0, len(words), chunk_size)]
    return chunks

def generate_embeddings(chunks: List[str]) -> List[List[float]]:
    embeddings = []
    for chunk in chunks:
        result = genai.embed_content(
            model="models/embedding-001",
            content=chunk,
            task_type="retrieval_document"
        )
        embeddings.append(result['embedding'])
    return embeddings

class SimpleVectorStore:
    def __init__(self, chunks: List[str], embeddings: List[List[float]]):
        self.chunks = chunks
        self.embeddings = np.array(embeddings)

    def retrieve(self, query_embedding: List[float], top_k: int = 3) -> List[Dict[str, str]]:
        query_embedding = np.array(query_embedding)
        similarities = np.dot(self.embeddings, query_embedding) / (
            np.linalg.norm(self.embeddings, axis=1) * np.linalg.norm(query_embedding)
        )
        top_indices = similarities.argsort()[-top_k:][::-1]
        return [{"text": self.chunks[i], "similarity": similarities[i]} for i in top_indices]

def rag_query(query: str, vector_store: SimpleVectorStore) -> str:
    query_embedding = genai.embed_content(
        model="models/embedding-001",
        content=query,
        task_type="retrieval_query"
    )['embedding']

    retrieved_chunks = vector_store.retrieve(query_embedding)
    context = "\n\n".join([chunk["text"] for chunk in retrieved_chunks])

    prompt = f"""
    Context:
    {context}

    Question: {query}
    Answer based on the context provided above. If the answer isn't clear from the context, say so.
    """

    model = genai.GenerativeModel('gemini-1.5-flash')
    response = model.generate_content(prompt)
    return response.text

def main():
    pdf_path = "custom_document.pdf"
    print("Extracting text from PDF...")
    pdf_text = extract_text_from_pdf(pdf_path)
    print("Chunking text...")
    chunks = chunk_text(pdf_text)
    print("Generating embeddings...")
    embeddings = generate_embeddings(chunks)
    print("Creating vector store...")
    vector_store = SimpleVectorStore(chunks, embeddings)
    query = "What is the main topic of the document?"
    print(f"\nQuery: {query}")
    response = rag_query(query, vector_store)
    print(f"Response: {response}")

    # Export chunks and embeddings to a JSON file
    data_to_export = {
        "chunks": chunks,
        "embeddings": embeddings
    }
    with open("rag_data.json", "w") as f:
        json.dump(data_to_export, f)
    print("Exported chunks and embeddings to rag_data.json")

    time.sleep(2)  # Keep this for gRPC cleanup

if __name__ == "__main__":
    main()

Input PDF file:

custom_documentDownload

Output embedding JSON file (rag_data.json):

{"chunks": ["Sample Document: Introduction to Artificial Intelligence Artificial Intelligence (AI) refers to the development of computer systems that can perform tasks that typically require human intelligence. These tasks include learning, problem -solving, decision -making, and perception. AI systems are powered by algorithm s and data, enabling them to mimic cognitive functions and improve over time through experience. History of AI The concept of AI dates back to the 1950s, with pioneers like Alan Turing proposing machines that could simulate human thought. Turing\u2019s famous \"Turing Test\" questioned whether a machine could exhibit intelligent behavior indistinguishable from a human. Th e term \"Artificial Intelligence\" was officially coined by John McCarthy in 1956 during the Dartmouth Conference, marking the birth of AI as a field of study. Early AI research focused on symbolic reasoning and rule-based systems, while modern AI leverages statistical methods and neural networks. Applications of AI AI is used in various fields such as healthcare, finance, and transportation. In healthcare, AI powers diagnostic tools that analyze medical images to detect diseases like cancer. In finance, it drives fraud detection systems by identifying unusual pattern s in transactions. In transportation, AI enables autonomous vehicles to navigate roads using sensors and real -time decision -making. Other applications include virtual assistants (e.g., Siri, Alexa), recommendation systems (e.g., Netflix, Spotify), and natu ral language processing tools. Future of AI The future of AI promises advancements in natural language processing, robotics, and ethical AI development. Researchers aim to create more sophisticated models that understand context and emotions better. Robotics powered by AI could revolutionize industr ies like manufacturing and elderly care. However, challenges remain, including ensuring fairness in AI decisions, maintaining transparency in how models work, and addressing job displacement caused by automation. The ethical implications of AI are a growing area of concern as its influence expands."], "embeddings": [[0.015937733, -0.049690623, -0.0204021, 0.0007430521, 0.05593546, 0.060835373, 0.034127835, 0.015798252, 0.014865388, 0.0647824, 0.004501347, 0.02073167, 0.041307747, 0.019409342, -0.02338441, -0.048979405, 0.00175639, 0.0494867, -0.05169524, -0.0029872488, 0.008235313, -0.04425001, 0.020743825, -0.021080898, 0.024118695, -0.026611507, -0.0024623415, -0.02872784, -0.0103659, 0.021297151, -0.035618663, 0.03648871, -0.029480629, 0.014664115, 0.0064022066, -0.040722784, 0.0067286775, 0.03197567, 0.022416105, 0.03604338, 0.008371223, -0.03324304, -0.022109516, 0.008300066, 0.009181495, -0.044486675, 0.00013025786, 0.036705945, 0.043699875, -0.061026238, 0.030195666, 0.024267554, 0.08894829, 0.0028238026, 0.0010572604, -0.04599659, 0.025037382, 0.015956644, -0.012850152, 0.010750755, -0.0092878835, 0.013064177, -0.012717866, 0.0054410263, 0.012139852, -0.04487596, -0.056222692, 0.018665195, 0.036022894, 0.0007064748, 0.06049654, 0.0054333922, 0.08654995, -0.012002155, -0.0398773, -0.07271985, -0.03050464, 0.026083704, 0.03753472, -0.029052755, -0.013759981, -0.026409278, -0.04711357, -0.044707566, -0.05994972, 0.019444847, 6.912404e-05, -0.030967508, -0.03660031, 0.01612776, -0.035901815, 0.040098023, 0.043284886, -0.032243546, -0.0075980285, 0.061743252, -0.017150588, -0.043187395, 0.0052927965, -0.041648347, -0.0003984908, -0.020512585, -0.06376851, 0.0369314, 0.0387055, 0.020558184, -0.012336664, 0.07740225, 0.009710032, 0.059394155, -0.043346696, -0.011663661, -0.013236541, -0.02946507, 0.05073721, -0.015902925, 0.003983061, -0.0012480431, 0.033538993, 0.01357407, 0.0030091542, -5.738158e-05, 0.06705514, -0.014455231, 0.004190682, 0.06077048, 0.0072714896, 0.014704358, 0.030258518, -0.01325368, 0.02309376, -0.025445024, -0.014987877, 0.0057038465, 0.034884263, 0.09502264, 0.04071962, 0.00695927, 0.035410013, 0.020899177, 0.009136633, -0.010116534, -0.007016623, 0.02180859, -0.019940283, 0.022875443, -0.06398348, -0.0014229885, 0.04992978, -0.06373337, 0.0015378344, 0.006679453, -0.05333586, -0.021301636, 0.056033663, 0.02073949, -0.009454011, 0.07411683, 0.021824801, 0.03633786, 0.009664076, 0.010216537, 0.016735462, 0.06768287, -0.04581567, -0.047274772, 0.0016150668, -0.026179891, 0.014153067, 0.0172116, -0.013210634, 0.044761594, -0.0013930007, -0.062453393, -0.02727489, -0.061817367, -0.0018438917, 0.017063046, -0.02036668, -0.006428096, -0.03683801, -0.022479918, 0.03659679, 0.019906605, 0.028995853, 0.0003945944, 0.053864058, -0.034257345, -0.06734754, 0.008313122, 0.0017552855, -0.014964498, -0.03535838, 1.1424553e-05, 0.025452448, 0.07250688, 0.05883478, 0.03916208, 0.0061630104, -0.017803451, 0.021428367, 0.06500971, -0.019534871, -0.060047086, -0.005530852, 0.0033090585, 0.0660379, -0.0077277333, -0.070355386, 0.049699314, -0.05304278, 0.024587676, 0.0034271968, 0.01679895, 0.0071134917, -0.0020015922, 0.023699218, -0.0076949187, 0.019677768, -0.025051823, -0.008719723, 0.0061841, -0.064047776, 0.046122093, -0.008883343, 0.052943565, -0.023913383, 0.024415245, 0.02907757, -0.04285641, -0.016512824, 0.073953815, 0.011839396, -0.015737407, 0.064493105, 0.0071743196, 0.009867819, 0.037423916, 0.04324142, 0.054122593, -0.03992351, 0.018336473, 0.0496274, 0.019233147, -0.053372253, -0.024768358, -0.058523644, 0.022124853, 0.024964062, 0.036711246, -0.027265575, -0.01956039, -0.015557396, 0.0047413725, -0.04075764, 0.0386848, -0.05072684, 0.0116405925, -0.012607731, -0.0006128997, 0.023374066, 0.014387301, 0.01986946, 0.010927249, -0.008451515, -0.03137258, -0.04829457, -0.07822872, 0.017459549, 0.05125215, -0.03377331, -0.08442833, 0.05544913, 0.0058780718, 0.04611517, 0.011229092, 0.0017561985, 0.03190164, 0.015662117, -0.024430783, 0.0005791091, -0.012369232, 0.036429815, -0.011309613, -0.016431445, 0.023900557, -0.053013287, -0.029711781, 0.013283226, -0.06636632, -0.028312866, -0.0044030487, -0.002637525, -0.08722542, -0.077643335, 0.020380896, -0.0115109775, 0.044436984, 0.040254474, -0.041377578, -0.016876306, -0.06668526, 0.00517847, -0.06631592, -0.0246784, -0.008839854, 0.02837776, -0.03193792, 0.024396565, 0.05067571, 0.04628903, -0.0589773, -0.039378867, -0.012344068, 0.07259791, 0.07994252, 0.021630451, -0.001081549, -0.00552696, 0.021120338, 0.030826252, 0.051611647, 0.012984188, -0.017994417, -0.04069849, 0.043664422, -0.03439143, 0.013621143, -0.0043348926, 0.012852082, -0.029259088, -0.0028064342, -0.047893044, 0.018586196, -0.01844385, 0.013625844, -0.11570167, 0.03394006, 0.01750292, -0.008058954, 0.011012508, 0.021228703, -0.017727114, -0.013536644, 0.026722452, 0.018608596, -0.08612556, -0.001815377, 0.086088754, -0.0054167397, 0.030782707, 0.051644508, -0.0493394, -0.010569267, -0.00025885765, -0.004875795, 0.044858832, -0.009041357, 0.09971623, -0.020451976, -0.027086664, 0.039829213, -0.023165608, -0.017565817, -0.024190255, -0.012250289, -0.0006819277, 0.047523163, -0.0067309006, 0.07020531, 0.025200514, -0.02797345, -0.0036610565, -0.037956573, 0.02860782, -0.022278793, -0.024428299, -0.05687103, 0.018811999, 0.004646171, -0.04951052, -0.014922714, 0.07843073, 0.04389206, -0.00089073874, 0.0054843314, -0.0051236134, 0.05634164, -0.027470073, 0.049337145, -0.014935135, 0.01125057, 0.100979894, -0.00036375507, 0.0034577502, -0.03856103, -0.03823353, -0.04507229, 0.0016518702, 0.048400387, -0.03177638, -0.019780912, -0.006582053, -0.0141180465, -0.013500944, -0.03010358, 0.044978272, -0.01982992, -0.013967415, 0.04356579, 0.009341139, -0.0092959115, 0.008178898, -0.0475293, -0.046589285, -0.011702044, 0.011692903, -0.0019787399, -0.0023206756, 0.025687642, -0.029119054, -0.032207325, 0.012919956, -0.0076698554, -0.07871898, -0.018896496, 0.014371966, 0.0102103, 0.028057354, 0.018049635, 0.027571365, -0.041244313, 0.0341185, -0.0132847885, 0.009983149, -0.030771801, -0.00802973, 0.025323084, -0.049184617, -0.008115789, 0.01897837, -0.0009168715, 0.0081048235, -0.044833012, -0.0627945, -0.036278456, -0.01569583, -0.053250417, 0.020824011, -0.1010228, 0.014328087, -0.03278986, -0.04869567, -0.024637906, -0.027096026, -0.0320477, 0.009735668, 0.030937275, -0.020631185, -0.003415371, 0.045160368, -0.07016449, 0.0060208514, -0.07534053, 0.025457505, -0.057314463, 0.010813801, -0.027304871, -0.039581034, 0.044135723, 0.018805286, -0.052193917, 0.057560883, 0.0026461673, -0.039432835, -0.027079953, -0.046797697, 0.054896712, -0.097453065, 0.009493537, 0.018215142, 0.03559076, 0.041505467, 0.009291867, -0.037611373, -0.008220406, 0.05739598, 0.018572247, -0.0071714544, 0.027492383, -0.0007672517, 0.029340424, -0.01908262, -0.050116904, -0.034887534, 0.0015773056, -0.03752532, 0.083446674, 0.044683754, 0.043232914, -0.004991584, 0.009663692, -0.010543812, 0.0098278485, 0.091826245, -0.044635363, 0.04331855, -0.006996937, -0.03972313, -0.012431925, 0.02074892, 0.031899042, -0.019880988, 0.0077145384, 0.06840948, -0.036762588, 0.018286794, -0.0060506593, 0.028786978, 0.0020494899, 0.010927087, -0.017445471, -0.117860846, -0.010804658, 0.04260833, -0.04460682, 0.004313246, 0.029645534, -0.06371792, -0.025484318, -0.034365837, 0.039286356, -0.048986245, -0.0022667223, -0.00073199923, -0.016565606, -0.010210421, -0.0010090302, 0.018692797, -0.02195638, 0.029425928, 0.0028283575, 0.04839934, 0.039316483, -0.031129254, 0.008511486, 0.021532277, -0.10140012, 0.017440228, -0.045239262, -0.014356461, -0.0006704565, -0.003675826, -0.047367416, 0.032411683, -0.012287991, -0.0078509655, -0.020727087, 0.0016333379, -0.020598425, -0.020938404, 0.027403027, 0.033502463, 0.029717032, 0.059931103, 0.001264345, -0.033374254, -0.020484531, 0.04942524, -0.042713333, -0.0114802625, -0.044165462, 0.020019066, -0.019690126, 0.03163057, -0.023107199, -0.021289922, 0.02055969, -0.039849162, -0.02369738, 0.072991244, 0.017919794, -0.011995686, 0.017090427, 0.010134111, -0.016831394, 0.031115562, 0.036736, 0.012217628, 0.0016613844, -0.06303714, 0.046404876, -0.030550165, 0.004221511, 0.0015106994, 0.035962638, 0.019923486, -0.036546458, -0.01274042, 0.023699276, -0.039390557, -0.056512266, 0.005445429, -0.0023738767, 0.032482892, 0.032642107, -9.127115e-06, 0.016064115, 0.00048732743, 0.019533297, 0.002003949, -0.055820957, -0.014017039, -0.0013552338, -0.028416937, -0.057297293, 0.07549086, 0.03002154, -0.0344023, -0.049031887, 0.012205708, 0.019521939, 0.038936846, 0.025922872, 0.023197515, 0.0027053559, -0.041828368, -0.011932597, 0.09492208, 0.07031678, 0.05003787, 0.06771662, -0.013675302, -0.02507077, -0.047122367, -0.032464422, -0.018087894, 0.024937652, -0.00084780063, 0.016263701, -0.06959095, 0.0064508244, 0.002332378, -0.011606934, 0.03548693, 0.039297428, -0.023541367, -0.08397377, -0.047677506, 0.0035153893, -0.031850506, 0.014992461, 0.04920879, -0.026219416, 0.0149139995, -0.003956472, -0.0077915904, -0.023765916, 0.05076665, 0.00068453775, -0.013344515, 0.021755088, 0.014721706, 0.009876042, -0.008728505, -0.029906679, -0.017609712, -0.06329055, -0.009052265, 0.033339575, -0.07146805, 0.055614427, 0.042565517, -0.027625734, 0.047788728, 0.0073321434, -0.0018334655, 0.058487877, 0.023336174, 0.0010983209, 0.0060013416, 0.017578518, -0.021726642, 0.016368028, -0.017886557, 0.0072122314, 0.0072680414, -0.014004148, -0.04558729, -0.024214637, 0.020501517, -0.02785427, 0.02573613, 0.043737296, 0.033565212, -0.006454003, -0.019322738, -0.03945689, 0.006336293, 0.051637273, 0.008084477, -0.062842764, 0.003810282, 0.041968573, 0.012026803, -0.014858716, -0.00031981067, 0.05912095, -0.0026807087, 0.04228106, 0.045894567, -0.04709101, 0.00897077, 0.009761777, -0.014156779, -0.06513087, 0.0134507725, 0.008747292, -0.01693257, 0.03207501, 0.046277817, 0.02978224, 0.034059193, -0.026005294, -0.070972174, 0.07830663, -0.055013753, -0.005208802, -0.0054895882, 0.016601022, 0.032797925, -0.039050337, -0.015307272, 0.059401408, -0.034186713, 0.036989942, 0.01034918, 0.06314615, -0.021075713, -0.095132895, 0.0067423834, -0.018803598, -0.012408568, 0.076515876, 0.056670092, -0.056374375, -0.01929818, 0.0014459653, 0.04872634, -0.03402996, -0.04095938, -0.05226812, -0.010170962, 0.045151826, 0.063585676, -0.031659152, -8.947953e-05, -0.015878012, -0.009755888, -0.0077300514, -0.039799515, -0.011702895, 0.0026955942, -0.017514782, 0.041243166, 0.02918135, -0.026672442, 0.023746604]]}

Android Project Files:

package com.programmerworld.rag_poweredandroidapp;

import android.os.Bundle;
import android.view.View;
import android.widget.Button;
import android.widget.EditText;
import android.widget.TextView;
import androidx.appcompat.app.AppCompatActivity;

public class MainActivity extends AppCompatActivity {

    private EditText queryInput;
    private TextView resultText;
    private Button submitButton;


    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);

        queryInput = findViewById(R.id.query_input);
        resultText = findViewById(R.id.result_text);
        submitButton = findViewById(R.id.submit_button);

        submitButton.setOnClickListener(new View.OnClickListener() {
            @Override
            public void onClick(View v) {
                String query = queryInput.getText().toString();
                if (!query.isEmpty()) {
                    generateWithRAG(query);
                }
            }
        });
    }

    // Generate response with RAG
    private void generateWithRAG(String query) {

        RagSystem rag = new RagSystem(this);
        new Thread(() -> {
            String response = rag.query(query);
            runOnUiThread(() -> {
                resultText.setText(response);
            });
        }).start();

    }
}

package com.programmerworld.rag_poweredandroidapp;


import android.content.Context;
import android.util.Log;
import com.google.gson.Gson;
import com.google.gson.reflect.TypeToken;
import java.io.InputStreamReader;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import okhttp3.MediaType;
import okhttp3.OkHttpClient;
import okhttp3.Request;
import okhttp3.RequestBody;
import okhttp3.Response;

public class RagSystem {
    private static final String TAG = "RagSystem";
    private static final String API_KEY = "AIzaSyAGPOmHVpP4zLw-B9nNbvPYszkYOWb5uJ8"; // Replace with your Google API key
    private static final MediaType JSON = MediaType.parse("application/json; charset=utf-8");
    private List<String> chunks;
    private List<double[]> embeddings;
    private OkHttpClient client;

    public RagSystem(Context context) {
        client = new OkHttpClient();
        loadData(context);
    }

    private void loadData(Context context) {
        try {
            InputStreamReader reader = new InputStreamReader(context.getAssets().open("rag_data.json"));
            Gson gson = new Gson();
            Map<String, Object> data = gson.fromJson(reader, new TypeToken<Map<String, Object>>(){}.getType());
            chunks = (List<String>) data.get("chunks");
            List<List<Object>> rawEmbeddings = (List<List<Object>>) data.get("embeddings");
            embeddings = new ArrayList<>();
            for (List<Object> embedding : rawEmbeddings) {
                double[] arr = new double[embedding.size()];
                for (int i = 0; i < embedding.size(); i++) {
                    arr[i] = ((Number) embedding.get(i)).doubleValue();
                }
                embeddings.add(arr);
            }
            Log.i(TAG, "Loaded " + chunks.size() + " chunks and embeddings.");
        } catch (Exception e) {
            Log.e(TAG, "Error loading data: " + e.getMessage());
        }
    }

    private double cosineSimilarity(double[] vec1, double[] vec2) {
        double dotProduct = 0.0, normA = 0.0, normB = 0.0;
        for (int i = 0; i < vec1.length; i++) {
            dotProduct += vec1[i] * vec2[i];
            normA += vec1[i] * vec1[i];
            normB += vec2[i] * vec2[i];
        }
        return dotProduct / (Math.sqrt(normA) * Math.sqrt(normB));
    }

    private List<String> retrieveChunks(double[] queryEmbedding, int topK) {
        List<Map<String, Object>> similarities = new ArrayList<>();
        for (int i = 0; i < embeddings.size(); i++) {
            double similarity = cosineSimilarity(queryEmbedding, embeddings.get(i));
            Map<String, Object> entry = new HashMap<>();
            entry.put("index", (double) i);
            entry.put("similarity", similarity);
            similarities.add(entry);
        }
        similarities.sort((a, b) -> Double.compare((double) b.get("similarity"), (double) a.get("similarity")));
        List<String> retrieved = new ArrayList<>();
        for (int i = 0; i < Math.min(topK, similarities.size()); i++) {
            int index = ((Double) similarities.get(i).get("index")).intValue();
            retrieved.add(chunks.get(index));
        }
        return retrieved;
    }

    private double[] getQueryEmbedding(String query) throws Exception {
        String url = "https://generativelanguage.googleapis.com/v1beta/models/embedding-001:embedContent?key=" + API_KEY;
        String json = "{\"content\": {\"parts\": [{\"text\": \"" + query.replace("\"", "\\\"") + "\"}]}, \"task_type\": \"retrieval_query\"}";
        RequestBody body = RequestBody.create(json, JSON);
        Request request = new Request.Builder().url(url).post(body).build();

        Log.d(TAG, "Attempting to send request to: " + url);
        Log.d(TAG, "Request JSON: " + json);

        try (Response response = client.newCall(request).execute()) {
            String responseBody = response.body().string();
            Log.d(TAG, "Response Code: " + response.code());
            Log.d(TAG, "Embedding Response Body: " + responseBody);

            if (!response.isSuccessful()) {
                throw new Exception("API request failed with code " + response.code() + ": " + responseBody);
            }

            Gson gson = new Gson();
            EmbeddingResponse embeddingResponse = gson.fromJson(responseBody, EmbeddingResponse.class);

            if (embeddingResponse == null || embeddingResponse.embedding == null || embeddingResponse.embedding.values == null) {
                throw new Exception("No embedding values found in response");
            }

            return embeddingResponse.embedding.values;
        } catch (Exception e) {
            Log.e(TAG, "Network error: " + e.getMessage(), e);
            throw e;
        }
    }

    private String generateResponse(String query, String context) throws Exception {
        String url = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent?key=" + API_KEY;
        String prompt = "Context:\n" + context + "\n\nQuestion: " + query + "\nAnswer based on the context.";
        String json = "{\"contents\": [{\"parts\": [{\"text\": \"" + prompt.replace("\"", "\\\"") + "\"}]}]}";
        RequestBody body = RequestBody.create(json, JSON);
        Request request = new Request.Builder().url(url).post(body).build();

        try (Response response = client.newCall(request).execute()) {
            String responseBody = response.body().string();
            Log.d(TAG, "Generate Request JSON: " + json);
            Log.d(TAG, "Generate Response Body: " + responseBody);
            Gson gson = new Gson();
            Map<String, Object> result = gson.fromJson(responseBody, new TypeToken<Map<String, Object>>(){}.getType());
            List<Map<String, Object>> candidates = (List<Map<String, Object>>) result.get("candidates");
            if (candidates == null || candidates.isEmpty()) {
                throw new Exception("No candidates found in response");
            }
            Map<String, Object> candidate = candidates.get(0);
            Map<String, Object> content = (Map<String, Object>) candidate.get("content");
            if (content == null) {
                throw new Exception("No content found in candidate");
            }
            List<Map<String, String>> parts = (List<Map<String, String>>) content.get("parts");
            if (parts == null || parts.isEmpty()) {
                throw new Exception("No parts found in content");
            }
            return parts.get(0).get("text");
        }
    }

    public String query(String userQuery) {
        try {
            double[] queryEmbedding = getQueryEmbedding(userQuery);
            List<String> relevantChunks = retrieveChunks(queryEmbedding, 3);
            String context = String.join("\n\n", relevantChunks);
            return generateResponse(userQuery, context);
        } catch (Exception e) {
            Log.e(TAG, "Error processing query: " + e.getMessage());
            return "Error: " + e.getMessage();
        }
    }

    private static class EmbeddingResponse {
        public EmbeddingData embedding;

        static class EmbeddingData {
            public double[] values;
        }
    }
}
plugins {
    alias(libs.plugins.android.application)
}

android {
    namespace = "com.programmerworld.rag_poweredandroidapp"
    compileSdk = 35

    defaultConfig {
        applicationId = "com.programmerworld.rag_poweredandroidapp"
        minSdk = 33
        targetSdk = 35
        versionCode = 1
        versionName = "1.0"

        testInstrumentationRunner = "androidx.test.runner.AndroidJUnitRunner"
    }

    buildTypes {
        release {
            isMinifyEnabled = false
            proguardFiles(
                getDefaultProguardFile("proguard-android-optimize.txt"),
                "proguard-rules.pro"
            )
        }
    }
    compileOptions {
        sourceCompatibility = JavaVersion.VERSION_11
        targetCompatibility = JavaVersion.VERSION_11
    }
}

dependencies {

    implementation(libs.appcompat)
    implementation(libs.material)
    implementation(libs.activity)
    implementation(libs.constraintlayout)
    testImplementation(libs.junit)
    androidTestImplementation(libs.ext.junit)
    androidTestImplementation(libs.espresso.core)

    implementation("com.google.ai.client.generativeai:generativeai:0.9.0")
    implementation("com.google.guava:guava:33.3.1-android")

    implementation("com.google.code.gson:gson:2.10.1")
    implementation("com.squareup.okhttp3:okhttp:4.10.0")

}
<?xml version="1.0" encoding="utf-8"?>
<manifest xmlns:android="http://schemas.android.com/apk/res/android"
    xmlns:tools="http://schemas.android.com/tools">

    <uses-permission android:name="android.permission.INTERNET"/>

    <application
        android:allowBackup="true"
        android:dataExtractionRules="@xml/data_extraction_rules"
        android:fullBackupContent="@xml/backup_rules"
        android:icon="@mipmap/ic_launcher"
        android:label="@string/app_name"
        android:roundIcon="@mipmap/ic_launcher_round"
        android:supportsRtl="true"
        android:theme="@style/Theme.RAGpoweredAndroidApp"
        tools:targetApi="31">
        <activity
            android:name=".MainActivity"
            android:exported="true">
            <intent-filter>
                <action android:name="android.intent.action.MAIN" />

                <category android:name="android.intent.category.LAUNCHER" />
            </intent-filter>
        </activity>
    </application>

</manifest>
<?xml version="1.0" encoding="utf-8"?>
<LinearLayout xmlns:android="http://schemas.android.com/apk/res/android"
    android:layout_width="match_parent"
    android:layout_height="match_parent"
    android:orientation="vertical"
    android:padding="16dp">

    <EditText
        android:id="@+id/query_input"
        android:layout_width="match_parent"
        android:layout_height="69dp"
        android:hint="Enter your query"
        android:inputType="text" />

    <Button
        android:id="@+id/submit_button"
        android:layout_width="match_parent"
        android:layout_height="wrap_content"
        android:layout_gravity="center"
        android:layout_marginTop="16dp"
        android:text="Submit" />

    <TextView
        android:id="@+id/result_text"
        android:layout_width="match_parent"
        android:layout_height="wrap_content"
        android:text="Response will appear here"
        android:textSize="16sp"
        android:layout_marginTop="16dp" />

</LinearLayout>

Screenshots:

Python code in VSCode:

(virtualpython) C:\Tools\Python\RAG_Gemini>python script.py
Extracting text from PDF...
Chunking text...
Generating embeddings...
Creating vector store...

Query: What is the main topic of the document?
Response: The main topic of the document is Artificial Intelligence (AI).  The document provides an introduction to AI, its history, applications, and future prospects, including ethical considerations.

Exported chunks and embeddings to rag_data.json
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1740584815.369497   12828 init.cc:232] grpc_wait_for_shutdown_with_timeout() timed out.

Leave a Reply

Your email address will not be published. Required fields are marked *

wpChatIcon
    wpChatIcon