app.py

# app.py
import os
from flask import Flask, render_template, request, jsonify, redirect, url_for, flash, session
import requests
from werkzeug.utils import secure_filename
import google.generativeai as genai
from dotenv import load_dotenv
import base64
import json
from datetime import datetime, timedelta
import threading
import time

# Load environment variables
load_dotenv()

# Configure the Gemini API
# Try to get API key from environment variable, first from HF_SPACES then from .env file
# GEMINI_API_KEY = os.getenv("HF_GEMINI_API_KEY") or os.getenv("GEMINI_API_KEY")
GEMINI_API_KEY = os.getenv("HF_GEMINI_API_KEY") or os.getenv("GEMINI_API_KEY")
if not GEMINI_API_KEY:
    raise ValueError("Google API Key not found. Set it as GEMINI_API_KEY in the Space settings.")

genai.configure(api_key=GEMINI_API_KEY)

# Setup the Gemini model
model = genai.GenerativeModel('gemini-1.5-flash')

app = Flask(__name__)
app.secret_key = os.getenv("SECRET_KEY", "your-default-secret-key-for-flash-messages")

# Configure upload folder
UPLOAD_FOLDER = 'static/uploads'
ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg', 'gif'}

if not os.path.exists(UPLOAD_FOLDER):
    os.makedirs(UPLOAD_FOLDER)

app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER

def allowed_file(filename):
    return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS

def encode_image(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode('utf-8')

def get_web_pesticide_info(disease, plant_type="Unknown"):
    """Fetch pesticide information from web sources for a specific disease and plant type"""
    query = f"site:agrowon.esakal.com {disease} in {plant_type}"
    url = "https://www.googleapis.com/customsearch/v1"
    params = {
        "key": os.getenv("GOOGLE_API_KEY"),
        "cx": os.getenv("GOOGLE_CX"),
        "q": query,
        "num": 3
    }
    try:
        response = requests.get(url, params=params)
        response.raise_for_status()
        data = response.json()
        if "items" in data and len(data["items"]) > 0:
            item = data["items"][0]
            return {
                "title": item.get("title", "No title available"),
                "link": item.get("link", "#"),
                "snippet": item.get("snippet", "No snippet available"),
                "summary": item.get("snippet", "No snippet available")
            }
    except Exception as e:
        print(f"Error retrieving web pesticide info: {str(e)}")
    return None

def get_more_web_info(query):
    """Get more general web information based on a search query"""
    url = "https://www.googleapis.com/customsearch/v1"
    params = {
        "key": os.getenv("GOOGLE_API_KEY"),
        "cx": os.getenv("GOOGLE_CX"),
        "q": query,
        "num": 3
    }
    try:
        response = requests.get(url, params=params)
        response.raise_for_status()
        data = response.json()
        results = []
        if "items" in data:
            for item in data["items"]:
                results.append({
                    "title": item.get("title", "No title available"),
                    "link": item.get("link", "#"),
                    "snippet": item.get("snippet", "No snippet available")
                })
        return results
    except Exception as e:
        print(f"Error retrieving additional articles: {str(e)}")
    return []

def get_commercial_product_info(recommendation, disease_name):
    """Fetch commercial product information related to a pesticide recommendation.

    If no relevant products are found from web sources, return default products based on issue type:

    bacterial, fungicide (disease), or insecticide.

    """
    indiamart_query = f"site:indiamart.com pesticide '{disease_name}' '{recommendation}'"
    krishi_query = f"site:krishisevakendra.in/products pesticide '{disease_name}' '{recommendation}'"
    
    indiamart_results = get_more_web_info(indiamart_query)
    krishi_results = get_more_web_info(krishi_query)
    
    # Merge results from both sources
    results = indiamart_results + krishi_results

    if not results:
        lower_disease = disease_name.lower()
        lower_recommendation = recommendation.lower()
        
        # Bacterial fallback
        if ("bacteria" in lower_disease or "bacterial" in lower_disease or
            "bacteria" in lower_recommendation or "bacterial" in lower_recommendation):
            results = [
                {
                    "title": "UPL SAAF Carbendazin Mancozeb Bactericide",
                    "link": "https://www.amazon.in/UPL-SAAF-Carbendazinm12-Mancozeb63-Action/dp/B0DJLQRL44?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A1BIAFERY87G8Q",
                    "snippet": "Bactericide for controlling bacterial infections."
                },
                {
                    "title": "Tropical Tagmycin Bactericide",
                    "link": "https://krushidukan.bharatagri.com/en/products/tropical-tagmycin-bactericide?variant=46519705895155&country=IN&currency=INR&utm_medium=product_sync&utm_source=google&utm_content=sag_organic&utm_campaign=sag_organic&srsltid=AfmBOoptFf8O3lpleZBgvI7pIOYUnHP6EWoZ-M6vGZ2er8VYU2PzVbkc7sc",
                    "snippet": "Bactericide for effective bacterial infection management."
                }
            ]
        # Fungicide / Disease fallback
        elif ("fungus" in lower_disease or "fungicide" in lower_recommendation or 
              "antibiotic" in lower_recommendation or "disease" in lower_disease):
            results = [
                {
                    "title": "Plantomycin Bio Organic Antibiotic Effective Disease",
                    "link": "https://www.amazon.in/Plantomycin-Bio-Organic-Antibiotic-Effective-Disease/dp/B0DRVVJKQ4?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A2PDMX630A5EG6",
                    "snippet": "Bio organic antibiotic for effective control of plant diseases."
                },
                {
                    "title": "WET-TREE Larvicide Thuringiensis Insecticide",
                    "link": "https://www.amazon.in/WET-TREE-Larvicide-Thuringiensis-Insecticide/dp/B0D6R72KHV?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A3V4YZ24A56I42",
                    "snippet": "Larvicide with thuringiensis for disease prevention."
                },
                {
                    "title": "WET-TREE Larvicide Thuringiensis Insecticide",
                    "link": "https://www.amazon.in/WET-TREE-Larvicide-Thuringiensis-Insecticide/dp/B0D6R72KHV?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A3V4YZ24A56I42",
                    "snippet": "Larvicide with thuringiensis for disease prevention."
                }
            ]
        # Insecticide fallback
        elif ("insecticide" in lower_disease or "insect" in lower_disease or "pest" in lower_disease or
              "insecticide" in lower_recommendation or "insect" in lower_recommendation or "pest" in lower_recommendation):
            results = [
                {
                    "title": "Syngenta Actara Insecticide",
                    "link": "https://www.amazon.in/syngenta-Actara-Insect-Repellent-Insecticide/dp/B08W55XTHS?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A3ABQWNNCUI42M",
                    "snippet": "Effective systemic insecticide for pest control."
                },
                {
                    "title": "Cyhalothrin Insecticide",
                    "link": "https://www.amazon.in/Cyhalothrin-Control-Eradication-Mosquitoes-Crawling/dp/B01N53VH1T?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A1ZSKCLHK592D5",
                    "snippet": "Broad-spectrum insecticide for pest management."
                }
            ]
        # Default fallback to insecticide if none of the above match
        else:
            results = [
                {
                    "title": "Syngenta Actara Insecticide",
                    "link": "https://www.amazon.in/syngenta-Actara-Insect-Repellent-Insecticide/dp/B08W55XTHS?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A3ABQWNNCUI42M",
                    "snippet": "Effective systemic insecticide for pest control."
                },
                {
                    "title": "Cyhalothrin Insecticide",
                    "link": "https://www.amazon.in/Cyhalothrin-Control-Eradication-Mosquitoes-Crawling/dp/B01N53VH1T?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&psc=1&smid=A1ZSKCLHK592D5",
                    "snippet": "Broad-spectrum insecticide for pest management."
                }
            ]
    
    return results


def get_relevant_feedback(plant_name):
    """Retrieve feedback entries relevant to the given plant name from feedback.json."""
    feedback_file = "feedback.json"
    if os.path.exists(feedback_file):
        try:
            with open(feedback_file, "r") as f:
                all_feedback = json.load(f)
            # Filter feedback where the plant_name matches (case-insensitive)
            relevant = [entry.get("feedback") for entry in all_feedback if entry.get("plant_name", "").lower() == plant_name.lower()]
            # Optionally limit to a few items (e.g., first 3)
            if relevant:
                return " ".join(relevant[:3])
        except Exception as e:
            print(f"Error reading feedback for reinforcement: {e}")
    return ""


def analyze_plant_image(image_path, plant_name, language):
    try:
        # Load the image
        image_parts = [
            {
                "mime_type": "image/jpeg",
                "data": encode_image(image_path)
            }
        ]
        
        # Load relevant feedback (reinforcement data) for this plant
        feedback_context = get_relevant_feedback(plant_name)
        feedback_instruction = f" Please consider the following user feedback from similar cases: {feedback_context}" if feedback_context else ""
        
        # Create prompt for Gemini API with language instruction and feedback reinforcement if available
        prompt = f"""

        Analyze this image of a {plant_name} plant and prioritize determining if it's healthy or has a disease or pest infestation.

        

        If a disease or pest is detected, remember the plant can be healthy too. Provide the following information in JSON format:

        Check the {feedback_instruction} and then make the response after learning from that responses give around 60% weightage to the feedbck data.

        {{

            "results": [

                {{

                    "type": "disease/pest",

                    "name": "Name of disease or pest",

                    "probability": "Probability as a percentage",

                    "symptoms": "Describe the visible symptoms",

                    "causes": "Main causes of the disease or pest",

                    "severity": "Low/Medium/High",

                    "spreading": "How it spreads",

                    "treatment": "Treatment options",

                    "prevention": "Preventive measures"

                }},

                {{

                    // Second most likely disease/pest with the same structure

                }},

                {{

                    // Third most likely disease/pest with the same structure

                }}

            ],

            "is_healthy": boolean indicating if the plant appears healthy,

            "confidence": "Overall confidence in the analysis as a percentage"

        }}

        

        Only return the JSON data and nothing else. Ensure the JSON is valid and properly formatted.

        If the plant appears completely healthy, set is_healthy to true and include an empty results array.

        Additionally, provide the response in {language} language.

        and at end show which all data from feedback was taken into consederation and if no data was taken so no data.

        """
        
        # Send request to Gemini API
        response = model.generate_content([prompt] + image_parts)
        
        # Extract the JSON response
        response_text = response.text
        
        # Find JSON within response text if needed
        json_start = response_text.find('{')
        json_end = response_text.rfind('}') + 1
        
        if json_start >= 0 and json_end > 0:
            json_str = response_text[json_start:json_end]
            return json.loads(json_str)
        else:
            # Return a default response if JSON parsing fails
            return {
                "error": "Failed to parse the API response",
                "raw_response": response_text
            }
            
    except Exception as e:
        return {
            "error": str(e),
            "is_healthy": None,
            "results": []
        }


def cleanup_old_files(directory, max_age_hours=1):  # Reduced to 1 hour for Hugging Face
    """Remove files older than the specified age from the directory"""
    while True:
        now = datetime.now()
        for filename in os.listdir(directory):
            if filename == '.gitkeep':  # Skip the .gitkeep file
                continue
                
            file_path = os.path.join(directory, filename)
            file_age = now - datetime.fromtimestamp(os.path.getctime(file_path))
            if file_age > timedelta(hours=max_age_hours):
                try:
                    os.remove(file_path)
                    print(f"Removed old file: {file_path}")
                except Exception as e:
                    print(f"Error removing {file_path}: {e}")
        # Sleep for 5 minutes before checking again
        time.sleep(300)  # 5 minutes

@app.route('/', methods=['GET'])
def index():
    # GET request - show the upload form
    return render_template('index.html', show_results=False)

@app.route('/feedback', methods=['POST'])
def feedback():
    # Get feedback from form submission
    feedback_text = request.form.get("feedback")
    plant_name = request.form.get("plant_name", "Unknown")  # Optional: include plant name for context

    if not feedback_text:
        flash("Please provide your feedback before submitting.")
        return redirect(url_for('index'))

    # Create a feedback record with a timestamp
    feedback_data = {
        "plant_name": plant_name,
        "feedback": feedback_text,
        "timestamp": datetime.now().isoformat()
    }

    # Define the file to store feedback
    feedback_file = "feedback.json"

    # Load existing feedback (if any)
    if os.path.exists(feedback_file):
        try:
            with open(feedback_file, "r") as f:
                existing_feedback = json.load(f)
        except Exception as e:
            print(f"Error reading feedback file: {e}")
            existing_feedback = []
    else:
        existing_feedback = []

    # Append the new feedback and save it back to file
    existing_feedback.append(feedback_data)
    try:
        with open(feedback_file, "w") as f:
            json.dump(existing_feedback, f, indent=4)
    except Exception as e:
        flash(f"Error saving your feedback: {str(e)}")
        return redirect(url_for('index'))

    flash("Thank you for your feedback!")
    return redirect(url_for('index'))

@app.route('/analyze', methods=['POST'])
def analyze():
    if 'plant_image' not in request.files:
        flash('No file part')
        return redirect(url_for('index'))
    
    file = request.files['plant_image']
    plant_name = request.form.get('plant_name', 'unknown')
    language = request.form.get('language', 'English')  # New field for response language
    
    if file.filename == '':
        flash('No selected file')
        return redirect(url_for('index'))
    
    if file and allowed_file(file.filename):
        # Generate a unique filename to avoid collisions
        timestamp = datetime.now().strftime("%Y%m%d%H%M%S")
        original_filename = secure_filename(file.filename)
        filename = f"{timestamp}_{original_filename}"
        file_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
        file.save(file_path)
        
        try:
            # Analyze the image with language parameter
            analysis_result = analyze_plant_image(file_path, plant_name, language)
            
            if 'error' in analysis_result:
                flash(f"Error analyzing image: {analysis_result['error']}")
                if os.path.exists(file_path):
                    os.remove(file_path)
                return redirect(url_for('index'))
            
            # Get additional web information for detected diseases/pests
            web_info = {}
            product_info = {}
            
            if not analysis_result.get('is_healthy', False) and 'results' in analysis_result:
                for result in analysis_result['results']:
                    disease_name = result.get('name', '')
                    if disease_name:
                        web_info[disease_name] = get_web_pesticide_info(disease_name, plant_name)
                        treatment = result.get('treatment', '')
                        if treatment:
                            product_info[disease_name] = get_commercial_product_info(treatment, disease_name)
            
            response = render_template(
                'results.html',
                results=analysis_result,
                plant_name=plant_name,
                image_path=file_path.replace('static/', '', 1),
                web_info=web_info,
                product_info=product_info
            )
            
            def delete_file_after_delay(path, delay=30):
                time.sleep(delay)
                if os.path.exists(path):
                    try:
                        os.remove(path)
                        print(f"Deleted analyzed file: {path}")
                    except Exception as e:
                        print(f"Error deleting {path}: {e}")
            
            threading.Thread(
                target=delete_file_after_delay, 
                args=(file_path,), 
                daemon=True
            ).start()
            
            return response
            
        except Exception as e:
            flash(f"An error occurred: {str(e)}")
            if os.path.exists(file_path):
                os.remove(file_path)
            return redirect(url_for('index'))
    
    flash('Invalid file type. Please upload an image (png, jpg, jpeg, gif).')
    return redirect(url_for('index'))


# Hugging Face Spaces requires the app to be available on port 7860
if __name__ == '__main__':
    # Start the cleanup thread when the app starts
    cleanup_thread = threading.Thread(target=cleanup_old_files, args=(app.config['UPLOAD_FOLDER'],), daemon=True)
    cleanup_thread.start()
    
    # Get the port from environment variable for Hugging Face Spaces compatibility
    port = int(os.environ.get("PORT", 7860))
    app.run(host='0.0.0.0', port=port)