app.py

# import os
# import subprocess
# import whisper
# import requests
# import tempfile
# import warnings
# import threading
# from flask import Flask, request, jsonify, send_file, render_template

# from dotenv import load_dotenv
# import requests




# warnings.filterwarnings("ignore", category=UserWarning, module="whisper")


# app = Flask(__name__)


# # Gemini API settings
# load_dotenv()
# API_KEY = os.getenv("FIRST_API_KEY")

# # Ensure the API key is loaded correctly
# if not API_KEY:
#     raise ValueError("API Key not found. Make sure it is set in the .env file.")

# GEMINI_API_ENDPOINT = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
# GEMINI_API_KEY = API_KEY

# # Load Whisper AI model at startup
# print("Loading Whisper AI model...")
# whisper_model = whisper.load_model("base")
# print("Whisper AI model loaded successfully.")

# # Define the "/" endpoint for health check
# @app.route("/", methods=["GET"])
# def health_check():
#     return jsonify({"status": "success", "message": "API is running successfully!"}), 200


# def process_video_in_background(video_file, temp_video_file_name):
#     """
#     This function is executed in a separate thread to handle the long-running
#     video processing tasks such as transcription and querying the Gemini API.
#     """
#     try:
#         transcription = transcribe_audio(temp_video_file_name)

#         if not transcription:
#             print("Audio transcription failed")
#             return

#         structured_data = query_gemini_api(transcription)

#         # Send structured data back or store it in a database, depending on your use case
#         print("Processing complete. Structured data:", structured_data)

#     except Exception as e:
#         print(f"Error processing video: {e}")

#     finally:
#         # Clean up temporary files
#         if os.path.exists(temp_video_file_name):
#             os.remove(temp_video_file_name)


# @app.route('/process-video', methods=['POST'])
# def process_video():
#     if 'video' not in request.files:
#         return jsonify({"error": "No video file provided"}), 400

#     video_file = request.files['video']

#     try:
#         # Save video to a temporary file
#         with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_video_file:
#             video_file.save(temp_video_file.name)
#             print(f"Video file saved: {temp_video_file.name}")

#             # Start the video processing in a background thread
#             threading.Thread(target=process_video_in_background, args=(video_file, temp_video_file.name)).start()

#             return jsonify({"message": "Video is being processed in the background."}), 202

#     except Exception as e:
#         return jsonify({"error": str(e)}), 500


# def transcribe_audio(video_path):
#     """
#     Transcribe audio directly from a video file using Whisper AI.
#     """
#     try:
#         print(f"Transcribing video: {video_path}")
#         result = whisper_model.transcribe(video_path)
#         return result['text']
#     except Exception as e:
#         print(f"Error in transcription: {e}")
#         return None


# def query_gemini_api(transcription):
#     """
#     Send transcription text to Gemini API and fetch structured recipe information.
#     """
#     try:
#         # Define the structured prompt
#         prompt = (
#             "Analyze the provided cooking video transcription and extract the following structured information:\n"
#             "1. Recipe Name: Identify the name of the dish being prepared.\n"
#             "2. Ingredients List: Extract a detailed list of ingredients with their respective quantities (if mentioned).\n"
#             "3. Steps for Preparation: Provide a step-by-step breakdown of the recipe's preparation process, organized and numbered sequentially.\n"
#             "4. Cooking Techniques Used: Highlight the cooking techniques demonstrated in the video, such as searing, blitzing, wrapping, etc.\n"
#             "5. Equipment Needed: List all tools, appliances, or utensils mentioned, e.g., blender, hot pan, cling film, etc.\n"
#             "6. Nutritional Information (if inferred): Provide an approximate calorie count or nutritional breakdown based on the ingredients used.\n"
#             "7. Serving size: In count of people or portion size.\n"
#             "8. Special Notes or Variations: Include any specific tips, variations, or alternatives mentioned.\n"
#             "9. Festive or Thematic Relevance: Note if the recipe has any special relevance to holidays, events, or seasons.\n"
#             f"Text: {transcription}\n"
#         )

#         payload = {
#             "contents": [
#                 {"parts": [{"text": prompt}]}
#             ]
#         }
#         headers = {"Content-Type": "application/json"}

#         # Send request to Gemini API
#         response = requests.post(
#             f"{GEMINI_API_ENDPOINT}?key={GEMINI_API_KEY}",
#             json=payload,
#             headers=headers
#         )
#         response.raise_for_status()

#         # Extract and return the structured data
#         data = response.json()
#         return data.get("candidates", [{}])[0].get("content", {}).get("parts", [{}])[0].get("text", "No result found")

#     except requests.exceptions.RequestException as e:
#         print(f"Error querying Gemini API: {e}")
#         return {"error": str(e)}


# if __name__ == '__main__':
#     app.run(debug=True)


# Above code is without polling and sleep

import os
import subprocess
import whisper
import requests
from flask import Flask, request, jsonify, render_template
import tempfile

app = Flask(__name__)
print("APP IS RUNNING, ANIKET")
# Gemini API settings
from dotenv import load_dotenv
# Load the .env file
load_dotenv()

print("ENV LOADED, ANIKET")

# Fetch the API key from the .env file
API_KEY = os.getenv("FIRST_API_KEY")

# Ensure the API key is loaded correctly
if not API_KEY:
    raise ValueError("API Key not found. Make sure it is set in the .env file.")

GEMINI_API_ENDPOINT = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
GEMINI_API_KEY = API_KEY


# Load Whisper AI model at startup
print("Loading Whisper AI model..., ANIKET")
whisper_model = whisper.load_model("base")  # Choose model size: tiny, base, small, medium, large
print("Whisper AI model loaded successfully, ANIKET")


# Define the "/" endpoint for health check
@app.route("/", methods=["GET"])
def health_check():
    return jsonify({"status": "success", "message": "API is running successfully!"}), 200

@app.route("/mbsa")
def mbsa():
    return render_template("mbsa.html")

@app.route('/process-video', methods=['POST'])
def process_video():
    print("GOT THE PROCESS VIDEO REQUEST, ANIKET")
    """
    Flask endpoint to process video:
    1. Extract audio and transcribe using Whisper AI.
    2. Send transcription to Gemini API for recipe information extraction.
    3. Return structured data in the response.
    """
    
    if 'video' not in request.files:
        return jsonify({"error": "No video file provided"}), 400

    video_file = request.files['video']
    print("VIDEO FILE NAME: ", video_file)
    try:
        print("SAVING THE FILE TEMPO, ANIKET")
        # Step 1: Save video to a temporary file
        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_video_file:
            video_file.save(temp_video_file.name)
            print(f"Video file saved: {temp_video_file.name}")

            
            # Step 2: Extract audio from video using ffmpeg (waiting for completion)
            audio_path = extract_audio(temp_video_file.name)
            print("AUDIO PATH FROM LINE 221, ANIKET", audio_path)
            
            if not audio_path:
                return jsonify({"error": "Audio extraction failed"}), 500

            print("STARTING TRANSCRIPTION, GOT THE .WAV AUDIO PATH THAT WAS STORED TEMPO, ANIKET")
            # Step 3: Transcribe the audio using Whisper AI (waiting for completion)
            transcription = transcribe_audio(audio_path)
)
            print("BEFORE THE transcription FAILED ERROR, CHECKING IF I GOT THE TRANSCRIPTION", transcription)

            if not transcription:
                return jsonify({"error": "Audio transcription failed"}), 500
            
            print("GOT THE transcription")

            print("Starting the GEMINI REQUEST TO STRUCTURE IT")
            # Step 4: Generate structured recipe information using Gemini API (waiting for completion)
            structured_data = query_gemini_api(transcription)
            
            print("GOT THE STRUCTURED DATA", structured_data)
            # Step 5: Return the structured data
            return jsonify(structured_data)

    except Exception as e:
        return jsonify({"error": str(e)}), 500

    finally:
        # Clean up temporary files
        if os.path.exists(temp_video_file.name):
            os.remove(temp_video_file.name)


def extract_audio(video_path):
    """
    Extract audio from video using ffmpeg and save as WAV file.
    """
    try:
        # Define the audio output path
        audio_path = video_path.replace(".mp4", ".wav")
        command = [
            "ffmpeg",
            "-i", video_path,
            "-q:a", "0",
            "-map", "a",
            audio_path
        ]

        # Run the command and wait for it to finish (synchronous)
        subprocess.run(command, check=True)
        print(f"Audio extracted to: {audio_path}")
        return audio_path

    except Exception as e:
        print(f"Error extracting audio: {e}")
        return None


def transcribe_audio(audio_path):
    """
    Transcribe audio using Whisper AI.
    """
    print("CAME IN THE transcribe audio folder")
    try:
         
        # Transcribe audio using Whisper AI
        print("Transcribing audio...")
        result = whisper_model.transcribe(audio_path)
        print("THE RESULTS ARE", result)
        
        return result.get("text", "").strip()

    except Exception as e:
        print(f"Error in transcription: {e}")
        return None


def query_gemini_api(transcription):
    """
    Send transcription text to Gemini API and fetch structured recipe information.
    """
    try:
        # Define the structured prompt
        prompt = (
            "Analyze the provided cooking video transcription and extract the following structured information:\n"
            "1. Recipe Name: Identify the name of the dish being prepared.\n"
            "2. Ingredients List: Extract a detailed list of ingredients with their respective quantities (if mentioned).\n"
            "3. Steps for Preparation: Provide a step-by-step breakdown of the recipe's preparation process, organized and numbered sequentially.\n"
            "4. Cooking Techniques Used: Highlight the cooking techniques demonstrated in the video, such as searing, blitzing, wrapping, etc.\n"
            "5. Equipment Needed: List all tools, appliances, or utensils mentioned, e.g., blender, hot pan, cling film, etc.\n"
            "6. Nutritional Information (if inferred): Provide an approximate calorie count or nutritional breakdown based on the ingredients used.\n"
            "7. Serving size: In count of people or portion size.\n"
            "8. Special Notes or Variations: Include any specific tips, variations, or alternatives mentioned.\n"
            "9. Festive or Thematic Relevance: Note if the recipe has any special relevance to holidays, events, or seasons.\n"
            f"Text: {transcription}\n"
        )

        # Prepare the payload and headers
        payload = {
            "contents": [
                {
                    "parts": [
                        {"text": prompt}
                    ]
                }
            ]
        }
        headers = {"Content-Type": "application/json"}

        # Send request to Gemini API and wait for the response
        print("Querying Gemini API...")
        response = requests.post(
            f"{GEMINI_API_ENDPOINT}?key={GEMINI_API_KEY}",
            json=payload,
            headers=headers,
            timeout=60  # 60 seconds timeout for the request
        )
        response.raise_for_status()

        # Extract and return the structured data
        data = response.json()
        return data.get("candidates", [{}])[0].get("content", {}).get("parts", [{}])[0].get("text", "No result found")

    except requests.exceptions.RequestException as e:
        print(f"Error querying Gemini API: {e}")
        return {"error": str(e)}